predicting resin pockets and blemishes in radiata pine lumber from log properties trevor g. jones1*, jun li yang2, don l. mcconchie3 and geoffrey m. downes4 1 the new zealand institute for plant & food research ltd, private bag 11600, palmerston north 4442, new zealand. 2 melbourne university, po box 4399, victoria 3052, australia. 3 wood quality focus, po box 176, lake tekapo 7945, new zealand 4 forest quality, po box 293 huonville, tasmania 7109, australia *corresponding author: trevor.jones@plantandfood.co.nz (received for publication 26 october 2021; accepted in revised form 11 april 2022) abstract background: resin pockets and blemishes in pruned logs of radiata pine (pinus radiata d.don) can reduce the value of clear and moulding grades of lumber. external resin features (erf) on the bark of the logs have proved an effective method of predicting the incidence of resin pockets in the lumber. resin canals have been associated with resin blemishes in radiata pine, and could prove useful in improving the prediction of the grade recovery of lumber. methods: pruned butt logs of radiata pine trees from two forests in the north island, new zealand, were selected for low, moderate, and severe levels of external resin features (erf) on the bark, and for low, average, and high resin canal diameter, frequency and brightness from breast height increment cores. the relationships were evaluated between these properties, and the lumber resin features and grade recovery of the logs. results: the number of resin pockets, the blemish rating, and the percentage of boards with resin streaks and resinous heartwood increased, and the recovery of clears and moulding grade boards and the lumber value declined, with the severity of the erf class of the logs. multiple regression models gave good predictions of the grade recovery and loss of lumber value, using the log erf class, volume, heartwood content, and number of type 1 resin pockets on the ends of the logs, as independent variables. the resin canal properties did not improve the regression models. resin blemishes were associated with type 2 resin pockets, and were more frequent in the forest where false growth rings were present. this suggests the constitutive resin flow from resin canals, rather than the resin canal size and frequency, is more important in determining the incidence of resin blemishes. conclusions: the prediction of the grade recovery of lumber using the erfs of radiata pine logs, was supplemented by the log volume, heartwood content and number of type 1 resin pockets on the ends of the logs. the environmental factors that drive the constitutive resin enrichment of resin canals, such as drought conditions that give rise to false growth rings, could be useful in improving the prediction of grade recovery for forest stands. new zealand journal of forestry science jones et al. new zealand journal of forestry science (2022) 52:14 https://doi.org/10.33494/nzjfs522022x196x e-issn: 1179-5395 published on-line: 3/05/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access the forests in these regions can vary from 0 to 3 resin pockets per square metre (m2) of sawn surface area. the recovery of clears and moulding grades of lumber is reduced as the incidence of resin pockets increases in the pruned logs, with a loss in value as the boards are downgraded to shop and commons grades of lumber. the lumber from pruned logs of radiata pine with low to severe levels of resin pockets can be reduced in introduction resin pockets are found in radiata pine (pinus radiata d.don) plantations in all regions of new zealand, with a higher prevalence in northland, hawkes bay, nelson, marlborough, and canterbury (clifton 1969; cown 1973; park 2004, woollons et al. 2008). the incidence of resin pockets in lumber produced from pruned logs from keywords: resin pockets, blemishes, radiata pine, resin canals, lumber mailto:trevor.jones@plantandfood.co.nz http://creativecommons.org/licenses/by/4.0/), jones et al. new zealand journal of forestry science (2022) 52:14 page 2 value by 20 to 68%, due to the presence of resin pockets and other resinous defects on the surface of the boards (hughes 2007). the presence in lumber of 0.7 to 2.0 resin pockets/m2 of sawn surface area was found to reduce the value of the lumber from pruned logs of radiata pine by 7 to 34% (mcconchie 2003). the loss in lumber value has led to the development of methods for the visual grading of pruned logs for resin pockets and blemishes. the resin pockets and blemishes that occur on the ends of the logs have been evaluated as a method for predicting the incidence of resin pockets in the lumber (ridoutt et al. 1999). the predictive models that were developed for radiata pine logs were limited by the low incidence of resin pockets on the ends of the logs. this prevented the establishment of consistent relationships between the log end assessments and the grade recovery of lumber. the external resin features (erf) on the bark of the logs have been a more effective method of predicting the incidence of resin pockets in the lumber (mcconchie, 2003). the presence of external resin bleeding, lesions, and galls on the surface of the logs, can be used to segregate the radiata pine logs for the grade recovery of lumber. the volume of mouldings and better grades of lumber declined by 12, 26, and 55% for pruned logs of radiata pine with low, moderate, and severe classifications of erf on the bark of the logs (mcconchie, 2003). resin blemishes have been associated with the frequency of resin canals in the logs of radiata pine, with a higher incidence of resin blemishes on the ends of logs, among forest sites with higher resin canal frequency (yang et al. 2007). resin canal frequency appears to be weakly linked to the external resin features (erf) on the bark of radiata pine trees (yang et al. 2007; ananais et al. 2010), and the resin pockets and resinous patches on the ends of radiata pine logs (yang et al. 2007), which suggests it may be useful in predicting the incidence of blemishes in the lumber. in this study, the use of external resin features (erf) on the bark of the logs was evaluated in combination with the resin canals from increment cores, and log end properties, to determine if these measurements could improve the prediction of the grade recovery of lumber. the pruned butt logs of radiata pine trees selected for low, moderate, and severe erf classes, and a range of resin canal properties, were sawn into randomwidth boards and graded for appearance lumber. the relationships were evaluated between the log erf classes, the resin canal, and log end properties, and the lumber resin features and grade recovery. methods external resin features trees of 26-year-old radiata pine trees were selected for low, moderate, and severe levels of external resin features (erf) on the bark of pruned butt logs, from plantation stands in northland and coastal hawke’s bay (figure 1, tables 1 and 2). ninety trees in parengarenga forest, northland, and eighty-seven trees in tangoio forest, coastal hawke’s bay, were selected based on the presence of external resin bleeding, lesions, and galls on the bark of the pruned butt logs, using the erf classification in the wqi field guide (mcconchie, 2003). resin canals the vertical resin canal properties of all the selected trees were measured using breast height (1.4 m) increment cores, taken from the outer ten growth rings from the bark. one 12 mm diameter increment core was sampled from each tree, and air-dried at room temperature. a silviscan strip (evans, 1994), was cut from each core, and one transverse surface was polished with 240, 320 and 1200 grit abrasive paper, and scanned at 600 dpi using a flat-bed scanner. the scanning parameters were kept the same for all the strips (image brightness = 140, image contrast = 170) to ensure the resin canal brightness could be compared between the strips. figure 1: radiata pine trees of: (a) low; (b) moderate; and (c) severe external resin feature (erf) classes on the bark of the pruned butt logs, at parengarenga forest in northland. (a) (b) (c) table 1: description of the parengarenga and tangoio forest sites descriptor parengarenga forest tangoio forest latitude 34° 36’ 29’’ s 39° 20’ 31’’ s longitude 172° 54’ 05’’ e 176° 53’ 45’’ e elevation, m 60 80 slope, ° 3 23 average annual temperature, °c 15.6 13.8 average annual rainfall, mm 1296 1284 soil series and type pinaki sand tokeawa silt loam the diameter, frequency, and brightness of the resin canals were measured using the idl imaging routine developed at csiro (chen et al. 2008, figure 2). resin canal size was estimated by drawing a circle over each canal and calculating the circle area using the software. resin canal diameter was derived from circle area. the average greyscale intensity of the pixels within the boundary of each resin canal was used to represent the brightness of the resin canal. log selection the butt logs of thirty trees were selected for sawmilling from each of parengarenga and tangoio forest, based on the erf classes of the logs and the resin canal properties of the breast height increment cores. ten trees were selected from each of the low, moderate and severe erf classes, and for each erf class the trees were selected for combinations of low, average, and high resin canal diameter, frequency and brightness. log measurements the trees were felled in april 2009 and the pruned butt logs for sawmilling were assessed for small and large end diameter, length, acoustic velocity, heartwood content, resin pockets, and blemishes. log volume, excluding bark, was calculated using smalian’s formula (ellis, 1995). acoustic velocity was measured using a director hm200 (fibre-gen, new zealand) connected to a laptop computer and operated in supervisor mode, to ensure the correct resonance peak was used in calculating the log acoustic velocity. heartwood content, resin pockets, and blemishes were measured using 50 mm thick discs, cut from both ends of the logs. one transverse surface of the discs was cut smooth using a disc-surfacing saw, then wetted with water, mounted in a camera booth on a back board with calibration pins, and photographed using a digital camera (lee and brownlie, 2009). resin pockets and blemishes in the disc photos were classified by type, and the dimensions, location, and ring position were measured using an idl imaging routine developed at csiro (ottenschlaeger et al. 2012). disc photos were corrected to a constant scale, using the back board calibration pins, and the resin pockets classified as type 1 or 2 (figure 3). xy positions of the resin pockets were recorded, the annual rings identified, and the within-ring position of the resin pockets estimated as a percentage of the ring width. sawing and grading of boards the pruned butt logs were sawn into 40 mm thick random-width boards, using a sawing-for-grade cutting pattern (figure 4). boards were visually graded in the green rough-sawn condition, using the wwpa factory grading rules (wwpa 1991), and assigned to the grades: clears, mouldings, no. 1 shop, no. 2 shop, no. 3 shop, and commons. boards were assessed for: width, length, actual grade (including resin features), adjusted grade (excluding resin features), number of type 1 and type 2 resin pockets, blemish rating (0 – 3), presence of resin streaks, resinous latewood, and galls on the worst longgrain face and edge combined. jones et al. new zealand journal of forestry science (2022) 52:14 page 3 table 1: description of the study sites descriptor parengarenga forest tangoio forest compartment/ stand 259/1, 261/1 280/1, 281/1 planted 1982 1982 planted stocking, s/ha 1794 1st prune 1988 to 2.2 m 1988 to 2.2 m 2nd prune 1988 to 4.2 m 1990 to 4.2 m 3rd prune* 1991, 1990 to 6.3 m 1993 to 6.7 m waste thinning 1988 to 538 s/ha production thinning* 1995 to 229 s/ha, 1994 to 293 s/ha 1997 to 296, 256 s/ha table 2: stand history of the parengarenga and tangoio forest stands figure 2: a view of the idl imaging routine used to obtain resin canal information on the 2 mm wide polished transverse surface of a strip from tangoio forest. annual growth rings are marked on the top strip, and show the frequent occurrence of false growth rings at tangoio forest. resin canals are identified and their area and brightness measured. * information given in consecutive order for compartment/stand. the frequencies of type 1 and 2 resin pockets in the boards were calculated per square metre (m²) of surface area, for the boards in each log, forest and erf class. volume-weighted average blemish rating of the boards was calculated for each log, forest and erf class. the number of boards with blemish ratings (0 – 3), resin streaks, resinous latewood, and galls were calculated as a percentage of the total volume of boards in each log, forest and erf class. jones et al. new zealand journal of forestry science (2022) 52:14 page 4 grade recovery actual and adjusted grade recoveries of clears, mouldings, no. 1 shop, no. 2 shop, no. 3 shop, and commons, were calculated as a percentage of the total volume of boards in each log, forest and erf class. lumber value lumber value of the actual grade and adjusted grade was calculated for each board, based on its volume and grade using the prices (nz$): clears and moulding $780/m3, no. 1 shop $490/m3, no. 2 shop $380/m3, no. 3 shop $290/m3, commons $155/m3. lumber value was summed for all the boards in each log, forest and erf class. value per cubic metre ($/m³) was calculated by dividing the total value of the lumber by the volume. value loss of the resin pockets and blemishes was the difference between the value of actual and adjusted grades, expressed as a percentage of adjusted grade. average 2001 prices for 5/4 random width lumber at the mill door in new zealand were used to provide a comparison with mcconchie and turner (2002). statistical analyses log variables of parengarenga and tangoio forest were compared using one-way analysis of variance (anova) with tukey’s multiple comparison test, using genstat statistical software (vsn international ltd, hemel hempstead, uk) and the model: yij = μ + fi + tj(i) + eij where: yij denotes the log variable measured on the butt log of tree j in forest i; µ is the overall population mean; fi represents the effect of forest (fixed), tj(i) the effect of trees (random), and eij represents the error term for the log measurements. figure 3: the presence of type 1 and 2 resin pockets in radiata pine logs. left: type 1 resin pocket, a narrow lens shaped cavity filled with resin and callus, occurring within a single growth ring. right: type 2 resin pocket, showing cambial damage and an occlusion scar, and centripetal resin bleeding towards the pith. figure 4: the sawing-for-grade cutting pattern used to saw the pruned butt logs into 40-mm thick random-width boards. type 1 resin pocket type 2 resin pocket correlations between log and board variables were calculated using sas statistical software (sas institute, inc., cary, nc, usa) and the procedure proc corr. stepwise regression equations for the actual grade recovery of clears + mouldings, actual grade value, and lumber value loss, were fitted using log variables as potential independent variables. sas procedure proc reg was used for this analysis. a p-level of 0.05 was used for selecting variables to enter the regression, or for removing variables. results resin canal properties the size and frequency of the resin canals in the outer ten growth-rings at breast height (1.4 m) differed for parengarenga forest, northland, and tangoio forest, hawke’s bay (table 3). resin canals of trees in parengarenga forest were on average larger in diameter, of higher frequency per unit area, and of slightly lower brightness, compared with trees in tangoio forest. log properties volume of the logs, and numbers of type 1 and 2 resin pockets on ends of the logs were similar at parengarenga and tangoio forest. acoustic velocity of the logs was higher at parengarenga forest, and number of resin blemishes on ends of the logs was higher at tangoio forest (table 4). acoustic velocity was higher on average for low erf class logs, compared with moderate and severe erf class logs (table 4). the trend was more evident for the logs at parengarenga forest, but there was a weak but significant correlation between acoustic velocity and erf class for logs from both forests (table 5). number of type 2 resin pockets and blemishes on the ends of the logs increased with the severity of the erf class of the logs (table 4). there were moderate correlations between the number of type 2 resin pockets and blemishes on ends of the logs, and erf class of the logs (table 5). board properties presence of resin features on the sawn surface of the random-width boards was strongly influenced by the erf class of the logs (table 6). numbers of type 1 and 2 resin pockets, blemish rating, and percentage of boards with resin streaks, and galls, increased with severity of the erf class of logs. there were weak to moderate correlations between these resin features on the surfaces of boards, and erf classes of the logs (table 7). the number of type 2 resin pockets and blemishes on the ends of the logs showed weaker correlations with resin features on the surface of the boards, compared with erf class of the logs (table 7). the number of type 1 resin pockets on the ends of the logs, showed a stronger correlation with number of type 1 resin pockets on the surfaces of boards, compared with erf class of the logs. resin canal properties of breast height increment cores was only very weakly correlated with resin features on surfaces of the boards (table 7). grade recovery actual grade recoveries of random-width boards showed a substantial reduction in the volume of clears and moulding grades, and a corresponding increase in jones et al. new zealand journal of forestry science (2022) 52:14 page 5 erf class diameter (mm) frequency (canals/cm²) brightness (greyscale, 0 – 255) parengarenga forest low 0.22 b 27 ab 204 bc moderate 0.24 a 31 a 205 abc severe 0.22 b 31 a 202 c tangoio forest low 0.20 c 17 c 208 abc moderate 0.20 c 17 c 208 ab severe 0.20 c 22 bc 211 a table 3: resin canal properties for the log erf classes of parengarenga and tangoio forest. average breast height increment core values followed by the same letter do not differ significantly (p > 0.05). table 4: log properties for the log erf classes of parengarenga and tangoio forest. average log values followed by the same letter do not differ significantly (p > 0.05). erf class volume (m³) acoustic velocity (km/s) heartwood (%) parengarenga forest low 1.0a 3.5a 34a moderate 1.1a 3.2 bc 33a severe 1.2a 3.3 b 30a tangoio forest low 0.9a 3.2 bcd 21a moderate 0.9a 3.0 d 22a severe 0.9a 3.1 cd 22a type 1 resin pockets (number/ log*) type 2 resin pockets (number/ log*) resin blemishes (number/ log*) parengarenga forest low 0 b 1 c 5 c moderate 2ab 2 bc 5 c severe 2ab 5ab 11 bc tangoio forest low 1ab 1 bc 3 c moderate 2ab 5abc 15ab severe 4a 8a 20a * average number of resin pockets and blemishes on the large and small ends of the logs combined. no.3 shop and commons grades, with an increase in severity of the erf class of the logs (table 8). recoveries of clears + moulding grades for the low, moderate, and severe erf class logs were: 55, 19, and 7 percent of the total board volume at parengarenga forest, and 38, 19, and 5 percent at tangoio forest. moderate erf class logs produced very few boards of clears grade, and there were no boards of clears grade from severe erf class logs. adjusted grade recovery of the random-width boards showed that if resin features on the surfaces of boards were excluded, the low, moderate, and severe erf class logs had similar board grade recoveries (table 8). recovery of clears grade boards was however, much higher for logs from parengarenga forest, compared with those from tangoio forest. this was explained by larger defect cores and pruned branches in logs from tangoio forest, which had the effect of downgrading many boards from tangoio forest to no. 3 shop grade. jones et al. new zealand journal of forestry science (2022) 52:14 page 6 log properties volume acoustic velocity heartwood erf class type 1 resin pockets type 2 resin pockets resin blemish resin canal diameter resin canal freq. acoustic velocity -0.05 heartwood 0.16 0.13 erf class 0.12 -0.30* -0.05 type 1 resin pockets 0.07 -0.25 -0.20 0.42** type 2 resin pockets 0.12 -0.28* -0.21 0.54** 0.50** resin blemishes -0.02 -0.36** -0.24 0.54** 0.40** 0.87** resin canal diameter 0.48** 0.25 0.11 0.10 0.02 -0.13 -0.22 resin canal frequency 0.31* 0.24 0.07 0.18 0.04 -0.10 -0.15 0.39** resin canal brightness 0.20 -0.26* -0.15 0.05 0.18 0.17 0.14 -0.05 -0.02 table 5: correlation coefficients (r) of the log properties, for the combined parengarenga and tangoio forest logs (n = 60). * p < 0.05, ** p < 0.01 erf class type 1 resin pockets (number/m²) type 2 resin pockets (number/m²) blemish rating (grade, 0 – 3) parengarenga forest low 0.1 b 0.1 c 0.8 cd moderate 0.4 b 0.3 c 1.5 a b severe 0.6 a b 1.0 ab 2.0 a tangoio forest low 0.4 b 0.1 c 0.6 d moderate 0.8 a b 0.6 b c 1.2 b c severe 1.4 a 1.2 a 2.0 a resin streaks (% of boards) resinous latewood (% of boards) galls ( % of boards) parengarenga forest low 53 b 6 a 0 b moderate 79 a 23 a 5 a b severe 92 a 34 a 7 a b tangoio forest low 52 b 7 a 1 b moderate 94 a 23 a 3 b severe 98 a 13 a 13 a table 6: board resin pockets and blemishes for the log erf classes of parengarenga and tangoio forest. average board values followed by the same letter do not differ significantly (p > 0.05). comparison of actual and adjusted grade recoveries for the low erf class logs, showed that when resin features were excluded the recovery of clears grade boards doubled, recovery of mouldings grade boards decreased, and overall, recovery of clears + mouldings grade boards increased by 10 and 6 percent of total board volume for parengarenga and tangoio forest, respectively. for moderate erf class logs recoveries of clears + mouldings grade boards increased by 44 and 21 percent of the total board volume for parengarenga and tangoio forest, with the increase occurring in clears grade boards. for severe erf class logs the recoveries of clears + mouldings grade boards increased by 66 and 56 percent of the total board volume for parengarenga and tangoio forest, with increases in both the clears and moulding grade boards. lumber value resin features on the surface of the random-width boards reduced total lumber values by 38 percent for parengarenga forest logs, and 31 percent for tangoio forest logs. actual grade values of the lumber were similar for parengarenga and tangoio forest, but adjusted grade values (resin features excluded) were lower for tangoio forest, due to the larger defect cores and pruned branches in the tangoio forest logs (table 9). moderate and severe erf class logs were effective in segregating for lumber value (table 9). there was greater loss of lumber value with increased severity of the log erf class. losses of lumber value for the low, moderate, and severe erf class logs were 9, 42, and 57 percent at parengarenga forest, and 6, 28 and 57 percent at tangoio forest. jones et al. new zealand journal of forestry science (2022) 52:14 page 7 log properties type 1 resin pockets2 type 2 resin pockets2 resin blemish rating resin streaks resinous latewood galls volume -0.17 -0.13 0.15 0.10 0.02 0.19 acoustic velocity -0.23 -0.20 -0.17 -0.34** -0.04 -0.13 heartwood -0.16 -0.14 0.25 0.16 0.02 -0.11 erf class 0.38** 0.68** 0.80** 0.65** 0.28* 0.47** type 1 resin pockets1 0.59** 0.50** 0.20 0.12 -0.13 0.26* type 2 resin pockets1 0.34* 0.64** 0.50** 0.43** 0.05 0.34* resin blemishes 0.27* 0.59** 0.50** 0.43** 0.10 0.33* resin canal diameter -0.17 -0.09 0.08 -0.05 0.07 0.10 resin canal frequency -0.20 0.04 0.09 -0.05 0.23 0.27* resin canal brightness 0.17 0.23 -0.02 0.02 -0.15 0.03 table 7: correlation coefficients (r) of the board resin features and log properties, for the combined parengarenga and tangoio forest logs (n = 60). * p < 0.05, ** p < 0.01 1 number of resin pockets on the large and small ends of the logs combined. 2 number of resin pockets per m² of board surface, worst face and edge. erf class clears moulding no.1 shop no.2 shop no.3 shop commons actual grade recovery, % parengarenga forest low 22 33 1 0 30 14 moderate 3 16 0 0 50 31 severe 0 7 0 2 54 37 tangoio forest low 14 24 9 2 43 8 moderate 1 18 3 2 52 24 severe 0 5 0 1 42 52 adjusted grade recovery, % parengarenga forest low 47 18 3 0 22 10 moderate 46 17 3 0 25 9 severe 54 19 1 0 19 7 tangoio forest low 28 16 8 1 40 7 moderate 25 15 9 2 41 8 severe 32 29 2 0 27 10 table 8: actual and adjusted board grade recoveries for the log erf classes of parengarenga and tangoio forest. the average board grade values are a percentage of the total board volume. prediction of grade recovery and lumber value linear regression models gave good predictions of actual grade recoveries of clears + moulding grade boards, actual grade value, and loss of lumber value (table 10). predictions of actual grade recovery, and loss of lumber value, were weaker using numbers of resin pockets and blemishes on the ends of the logs, compared with erf classes of logs. resin canal properties of the logs showed no significant correlations with actual grade recovery, or loss of lumber value. multiple regression models gave improved grade recovery and value predictions using log erf class, volume, heartwood content, and number of type 1 resin pockets on ends of logs as independent variables (table 11). erf class of log was the main contributor to the models, with the other log variables providing only small reductions in residual sums of squares. the exclusion of erf class of log from the models resulted in poor predictions, with the number of type 1 and 2 resin pockets and blemishes on the ends of the logs proving to be relatively ineffective predictive variables in the absence of erf class. discussion weaker relationships between resin features on the surface of the boards and those on ends of the logs could be attributed to the small number of the resin pockets and blemishes that were sampled on the log ends. type 1 resin pockets rarely extend more than 50100 mm longitudinally, and while type 2 resin pockets can extend for more than 200-250 mm (ottenschlaeger et al. 2012), the distances are short compared with 5 m lengths of pruned butt logs. resin bleeding and occlusion scars of type 2 resin pockets, and lesions and galls that accumulate on the surface of the bark during the life of a tree, provide good indications of the severity of the internal resin defects that are present in the wood (mcconchie 2003; mcconchie et al. 2007; ottenschlaeger et al. 2012). losses of lumber value for the moderate and severe erf class logs from parengarenga and tangoio forest (table 9) were greater than the 14 and 34 percent losses for the moderate and severe erf class logs from tikitere forest in the bay of plenty (mcconchie 2003). logs from tikitere forest had a lower number of resin pockets on log ends, compared with logs from parengarenga and tangoio forests. hughes (2007) found an increase in loss of lumber value with increasing numbers of resin pockets on log ends, with logs from nelson forests having similar value losses to those of logs from parengarenga and tangoio forests, when compared with the same numbers of resin pockets on the log ends. this suggests that resin pockets on the log ends could be useful indicators of the relative loss of lumber value among radiata pine plantation forests and stands. jones et al. new zealand journal of forestry science (2022) 52:14 page 8 erf class actual grade value ($/m³) adjusted grade value ($/m³) lumber value loss (%) parengarenga forest low 545 601 9 moderate 341 593 42 severe 275 642 57 tangoio forest low 482 511 6 moderate 358 498 28 severe 246 579 57 table 9: actual and adjusted board grade values for the log erf classes of parengarenga and tangoio forest. log properties actual grade recovery of clears + mouldings actual grade value adjusted grade value lumber value loss volume 0.00 0.00 0.39** 0.00 acoustic velocity 0.10** 0.08* 0.04 0.03 heartwood 0.02 0.03 0.00 0.04 erf class 0.44** 0.54** 0.02 0.71** type 1 resin pockets1 0.14** 0.14** 0.00 0.18** type 2 resin pockets1 0.14** 0.17** 0.02 0.24** resin blemishes1 0.13** 0.18** 0.00 0.21** resin canal diameter 0.01 0.01 0.28** 0.00 resin canal frequency 0.00 0.00 0.21** 0.00 resin canal brightness 0.00 0.00 0.00 0.00 table 10: coefficients of determination (r2) for prediction of the board grade recovery from the log properties, using the combined parengarenga and tangoio forest logs (n = 60). * p < 0.05, ** p < 0.01 1 number of resin pockets and blemishes on the large and small ends of the logs combined. higher incidences of type 1 resin pockets on surfaces of boards of the moderate and severe erf class logs from tangoio forest (table 6), could be attributed to frequent occurrences of false growth rings at this site (figure 1). false growth rings caused by drought conditions have been associated with presence of type 1 resin pockets in canterbury (cown, 1973) and are considered zones of weakness that split tangentially along the false growth ring latewood/earlywood boundary. they are caused by axial compressive stresses (bariska & kučera, 1985), that occur with bending of tree stems, and this has been shown to increase the incidence of type 1 resin pockets (watt et al. 2009; jones et al. 2013). there was a relationship between type 2 resin pockets on the log ends and blemish rating and resin streaks on the surface of the boards (table 7). ottenschlaeger et al. (2012) observed that the association between type 2 resin pockets and blemishes, and resin streaks, occurs as a result of the centripetal transport of resin from the type 2 resin pockets towards the pith, and is a wound response to cambial damage at the time of the resin pocket formation. the blemish rating and percentage of boards with resin streaks increased for the moderate and severe erf logs from parengarenga and tangoio forest, which suggests that the severity of the cambial damage by the type 2 resin pockets increased in these logs. incidences of resin blemishes on the ends of the logs were strongly associated with type 2 resin pockets (table 4), with both occurring in larger numbers on the ends of the logs from tangioio forest (table 5). this suggests there was greater resin flow in response to cambial damage at this site. resin flow can increase with vertical resin canals of larger diameter and higher frequency (blanche et al. 1992; hood & sala, 2015), and in response to drought conditions (blanche et al. 1992; lombardero et al. 2000; zas et al. 2020; rissanen et al. 2021). constitutive resin enrichment of the resin canals and resin flow are greater when drought conditions limit growth (lombardero et al. 2000; rissanen et al. 2021), as occurred with the formation of false growth rings at tangoio forest. resin enrichment of the resin canals appeared to be a stronger driver of resin blemish formation than the larger diameter and frequency of the resin canals at parengarenga forest. this suggests that the environmental and genetic factors which increase the size and frequency of the resin canals (reid & watson 1965; rigling et al. 2003; govina et al. 2021; rissanen et al. 2021), may not increase the constitutive resin flow, and could explain the lack of relationships between the resin canal properties and resin pockets and blemishes. conclusions external resin feature (erf) class of the pruned butt logs provided an effective method of segregating logs for differences in the number of type 1 and 2 resin pockets and blemishes on the surface of the boards, the actual grade recovery, and the loss of lumber value associated with the resin features. resin pockets and blemishes on the ends of the logs were less effective at predicting the actual grade recovery and lumber value, due to the small number of the resin features on the log ends, compared with the resin bleeding, lesions and galls assessed by erf classes on the surface of the bark. resin canal properties of logs did not improve predictions of actual grade recoveries and the losses of lumber value associated with resin features, but inclusion of the log heartwood content, the number of type 1 resin pockets on log ends, and log volume, improved predictions to a small extent when combined with the erf classes of the logs. jones et al. new zealand journal of forestry science (2022) 52:14 page 9 log properties actual grade recovery of clears + mouldings actual grade value adjusted grade value lumber value loss erf class included model 0.49 0.71 0.53 0.82 acoustic velocity 0.07 volume 0.07 0.41 heartwood 0.04 0.06 0.08 erf class 0.45 0.55 0.05 0.72 type 1 resin pockets1 0.03 0.02 erf class excluded model 0.29 0.44 0.53 0.47 acoustic velocity 0.07 0.07 volume 0.05 0.41 heartwood 0.07 0.11 0.13 type 1 resin pockets1 0.07 0.05 0.06 type 2 resin pockets1 0.15 0.28 resin blemishes1 0.21 1 number of resin pockets and blemishes on the large and small ends of the logs combined. table 11: coefficients of determination (r²) for the regression models, using log properties for the prediction of log grade recovery. partial r2 values given for the log variables. jones et al. new zealand journal of forestry science (2022) 52:14 page 10 competing interests the authors declare that they have no competing interests. acknowledgements we thank bob shirley (pf olsen ltd) and brian garnett (pan pac forest products ltd) for the selection of the radiata pine plantation stands and the logging at parengarenga forest, northland, and tangoio forest, hawke’s bay. winston liew and maria ottenschlaeger (csiro) for the resin canal measurements of the breast height increment cores. john lee and rod brownlie (scion) for the log end disc surfacing and photography, and maria ottenschlaeger (csiro) for the image analysis of the disc photos. the titc sawmill at waiariki polytechnic for sawing the logs to random-width boards, and the assistance of teresa mcconchie, walt povey and pat hodgkiss in the grading of the boards. graeme young (tenon ltd) provided support and advice in the design of the study. the research was funded by the wood quality initiative (wqi) ltd. authors' contributions gd and jy initiated the research and designed the study. dm organised the experimental work of tree selection, logging, sawmilling and lumber grading. tj managed the study, assisted with the experimental work, carried out the data analysis, and wrote the manuscript. all authors read and approved the final manuscript. references ananías, r.a., lastra, j., salvo, l., contreras, h., barría, c., & peredo, m. (2010). preliminary study of the resin canals in radiata pine. maderas. ciencia y tecnología 12(2), 135-142. https://doi.org/10.4067/s0718221x2010000200008 bariska, m., & kučera, l.j. (1985). on the fracture morphology in wood. part 2: macroscopical deformations upon ultimate axial compression in wood. wood science and technology, 19(1), 19-34. https://doi.org/10.1007/bf00354750 blanche, c.a., lorio, p.l., sommers, r.a., hodges, j.d., & nebeker, t.e. (1992). seasonal cambial growth and development of loblolly pine: xylem formation, inner bark chemistry, resin ducts, and resin flow. forest ecology and management 49, 151-165. https://doi.org/10.1016/0378-1127(92)90167-8 chen, f., yang, j.l., & downes, g. (2008). a visual information assessment tool for resin canal identification and property measurement. iawa journal 29(4), 397-408. https://doi. org/10.1163/22941932-90000194 clifton, n.c. (1969). resin pockets in canterbury radiata pine. new zealand journal of forestry 14(1), 38-49. cown, d.j. (1973). resin pockets: their occurrence and formation in new zealand forests. new zealand journal of forestry 18(2), 233-251. ellis, j.c. (1995). measuring logs. in d. hammond (ed.), forestry handbook, 3rd ed (pp. 107-108). wellington: new zealand institute of forestry (inc.). evans r. (1994). rapid measurement of the transverse dimensions of tracheids in radial wood sections from pinus radiata. holzforschung 48(2), 168-172. https://doi.org/10.1515/hfsg.1994.48.2.168 govina, j.k., apiolaza, l.a., & altaner, c.m. (2021). variation and genetic parameters of axial resin canal features in clones and families of pinus radiata. new forests 52, 167-176. https://doi. org/10.1007/s11056-020-09780-8 hood, s., & sala, a. (2015). ponderosa pine resin defences and growth: metrics matter. tree physiology 35, 1223-1235. https://doi.org/10.1093/treephys/ tpv098 hughes, m. (2007). resin assessment a valuable tool. new zealand tree grower, august, 25-27. jones, t.g., downes, g.m., watt, m.s., kimberley, m.o., culvenor, d.s., ottenschlaeger, m., estcourt, g., & xue, j. (2013). effect of stem bending and soil moisture on the incidence of resin pockets in radiata pine. new zealand journal of forestry science, 43: 10. https://www.scionresearch. com/__data/assets/pdf_file/0003/65127/11795395-43-10_jones.pdf lee, j., & brownlie, r. (2009). assessment of the end grain of log ends and log discs. wood processing newsletter no. 43. new zealand forest research institute, rotorua. lombardero, m.j., ayres, m.p., lorio, p.l., & ruel, j.j. (2000). environmental effects on constitutive and inducible resin defences of pinus taeda. ecology letters 3, 329-339. https://doi.org/10.1046/ j.1461-0248.2000.00163.x mcconchie, d.l. (2003). field guide to assist recognition and classification of resinous defects on the bark of radiata pine. wqi report no. app 12. rotorua, new zealand: wood quality initiative. mcconchie, d.l., cown, d., & donaldson, l. (2007). resin and other features in logs and lumber. field guide. rotorua, new zealand: wood quality initiative. mcconchie, d.l, & turner, j. (2002). how external resin on radiata pine logs relates to clearwood quality and value. wood processing newsletter no. 32. rotorua: new zealand forest research institute. ottenschlaeger, m., downes, g.m., bruce, j., & jones, t.g. (2012). type 1 and 2 resin pockets in new zealand radiata pine: how do they differ? new zealand journal of forestry science 42, 39-46. https://doi.org/10.4067/s0718-221x2010000200008 https://doi.org/10.4067/s0718-221x2010000200008 https://doi.org/10.1007/bf00354750 https://doi.org/10.1016/0378-1127(92)90167-8 https://doi.org/10.1163/22941932-90000194 https://doi.org/10.1163/22941932-90000194 https://doi.org/10.1515/hfsg.1994.48.2.168 https://doi.org/10.1007/s11056-020-09780-8 https://doi.org/10.1007/s11056-020-09780-8 https://doi.org/10.1093/treephys/tpv098 https://doi.org/10.1093/treephys/tpv098 https://www.scionresearch.com/__data/assets/pdf_file/0003/65127/1179-5395-43-10_jones.pdf https://www.scionresearch.com/__data/assets/pdf_file/0003/65127/1179-5395-43-10_jones.pdf https://www.scionresearch.com/__data/assets/pdf_file/0003/65127/1179-5395-43-10_jones.pdf https://doi.org/10.1046/j.1461-0248.2000.00163.x https://doi.org/10.1046/j.1461-0248.2000.00163.x jones et al. new zealand journal of forestry science (2022) 52:14 page 11 park, j.c. (2004). the incidence of resin pockets. new zealand journal of forestry 49, 32. reid, r.w., & watson, j.a. (1965). sizes, distributions, and numbers of vertical resin ducts in lodgepole pine. canadian journal of botany 44, 519-525. https:// doi.org/10.1139/b66-062 ridoutt, b.g., mcconchie, d.l., & ball, r.d. (1999). predicting resin pockets in radiata pine logs from blemishes on log ends. wood and fiber science 31(4), 434-440. rigling, a., brühlhart, h., bräker, o.u., forster, t., & schweingruber, f.h. (2003). effects of irrigation on diameter growth and vertical resin duct production in pinus sylvestris l. on dry sites in the central alps, switzerland. forest ecology and management 175, 285-296. https://doi.org/10.1016/s03781127(02)00136-6 rissanen, k., hölttä, t., bäck, j., rigling, a., wermelinger, b., & gessler, a. (2021). drought effects on carbon allocation to resin defences and on resin dynamics in old-grown scots pine. environmental and experimental botany 185. https://doi. org/10.1016/j.envexpbot.2021.104410 watt, m.s., downes, g., jones, t., ottenschlaeger, m., leckie, a.c., smaill, s.j., kimberley, m.o., & brownlie, r. (2009). effect of stem guying on the incidence of resin pockets. forest ecology and management, 258(9), 1913-1917. https://doi. org/10.1016/j.foreco.2009.07.032 woollons, r., manley, b., park, j. (2008). factors influencing the formation of resin pockets in pruned radiata pine butt logs from new zealand. new zealand journal of forestry science, 38(2/3), 323-334. wwpa (1991). western lumber grading rules, portland, or, usa: western wood products association. yang, j.l., downes, g.m., chen, f., & cown, d.j. (2007). investigation of within-tree and between-site resin canal variation in radiata pine. in: proceedings of iufro all-division 5 conference, forest products and environment: a productive symbiosis, taipei, taiwan, october 2007, p. 179. zas, r., touza, r., sampedro, l., lario, f.j., bustingorri, g., & lema, m. (2020). variation in resin flow among maritime pine populations: relationship with growth potential and climatic responses. forest ecology and management, 474: 118351. https:// doi.org/10.1016/j.foreco.2020.118351 https://doi.org/10.1139/b66-062 https://doi.org/10.1139/b66-062 https://doi.org/10.1016/s0378-1127(02)00136-6 https://doi.org/10.1016/s0378-1127(02)00136-6 https://doi.org/10.1016/j.envexpbot.2021.104410 https://doi.org/10.1016/j.envexpbot.2021.104410 https://doi.org/10.1016/j.foreco.2009.07.032 https://doi.org/10.1016/j.foreco.2009.07.032 https://doi.org/10.1016/j.foreco.2020.118351 https://doi.org/10.1016/j.foreco.2020.118351 measuring harvest residue accumulations at new zealand’s steepland log-making sites campbell harvey school of forestry, university of canterbury, new zealand *corresponding author: campbell.harvey@pg.canterbury.ac.nz (received for publication 5 september 2021; accepted in revised form 26 march 2022) abstract background: when harvesting plantation forests of pinus radiata (d.don) in new zealand, large residue piles commonly accumulate on or adjacent to processing sites. while the merchantable volume that is transported to market is carefully measured, little is known of the quantity of the piled, residual material. a working knowledge of residues is becoming more important as it is not only a potentially merchantable product for the bioenergy market, but when stored in perpetuity it can present a risk of self-ignition, and specifically on steep slopes, it presents a mobilisation risk if not stored correctly. methods: the area, bulk volume and depth of residue piles at 16 recently harvested steepland sites were measured from a wide geographic spread across new zealand. unmanned aerial vehicle imagery was used to build georeferenced photogrammetric models of residue piles (94 per cent of the studied volume). pile area was determined from interpreting boundaries from orthophotos and volumes determined by interpolating the obscured terrain surfaces on duplicate photogrammetric models. the remaining 6 per cent of pile volume was measured with handheld gps tracking of the perimeters and on-site estimation of average pile depth. results: for a mean harvest area of 18.9 ha, there was a mean of 2.4 piles per harvest site, 2600 m3 bulk volume and 2900 m2 of area covered. for every hectare harvested, a bulk volume of 170 m3 is piled at the landing, or alternatively, 0.23 m3 of bulk pile volume per tonne harvested. the manual terrain interpolation methodology was tested against collecting georeferenced pre-harvest terrain surfaces, yielding an average difference of 19% across two sites and six residue piles. conclusions: this research demonstrates the ability to investigate the bulk volume and site coverage of landing residue piles with equipment and software tools available to today’s forester. mean values for pile area and volume are presented to reflect the current state of knowledge and can be a reference point for future initiatives. new zealand journal of forestry science harvey new zealand journal of forestry science (2022) 52:12 https://doi.org/10.33494/nzjfs522022x186x e-issn: 1179-5395 published on-line: 13/04/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access to market, is accurately measured and can be reconciled against the inventory data that is typically available to forestry companies (gordon 2005). a recent study detailed the residues left in the cutover, which showed the median volume of course woody debris remaining was 88 m3/ha (harvey & visser 2022). however, little is known about the residues that are left behind at the landings (processing areas). the majority of plantations currently being harvested have been tended under a clearwood regime; a result of markets and common practice in the 1990s (maclaren & knowles 2005). in recent times there has been an increasing proportion of stands transitioned into introduction most plantation-grown radiata pine (pinus radiata d.don) is typically harvested between the ages of 25 and 35 years in new zealand, depending on a range of factors including market conditions, site and stand management (maclaren 1993). currently, the average felling age is 29.5 years (mpi 2020) with an estimated average total recoverable volume (trv) of 585 m3/ha (clearwood regime) or 593 m3/ha (framing regime) across new zealand (mpi 2015), with comparatively lower stocking levels and larger tree sizes in clearwood regimes (maclaren 1993). the recovered volume, taken as logs keywords: plantation forestry, slash, harvesting operations, biomass. http://creativecommons.org/licenses/by/4.0/), harvey new zealand journal of forestry science (2022) 52:12 page 2 framing regimes across new zealand (mpi 2020). the regime change is expected to increase final crop stocking levels and decrease average piece sizes at harvest with time. goulding (2005) estimated that in an average stand, 85 per cent of the total standing volume will be merchantable, leaving residues that will range from about 10 per cent in good condition, well-tended stands, to over 20 per cent for untended stands on moderately steep terrain. murphy (1982) showed that value loss at harvest due to stem breakage increases with increasing diameter at breast height (dbh); one consideration when assessing alternative regimes. breakage at any point along the stem results in a change in the objective function for maximising the stem’s value; establishing a new, lower optimum for the remaining value. stem breakage means a logging contractor needs to manage more stem ‘pieces’ and inevitably less log volume will leave the site; however, depending on where the breakpoint is along the stem, the value loss may only be trivial (murphy 1982). several studies have shown that most harvested trees are broken during felling in new zealand and that the typical break height (for the first break) is around two-thirds of the height of the stem (fraser et al. 1997; lambert 1996; twaddle 1987). breakage typically results in the generation of un-merchantable woody debris, where the final section of stem before (and after) the break cannot conform to any available log grade specifications. managing this additional debris provides only indirect benefits to logging contractors (e.g. tidy work area, reduced hazards) and can result in considerably reduced production. harvesting systems can be categorised as either cut-to-length (ctl) or whole tree harvesting (wth). in ctl systems the trees are processed in the cutover where the residues are left and only the logs extracted to roadside or a landing for subsequent transportation. in wth the stems are extracted to a landing where they are processed into logs. wth remains the preferred harvest system on steep slopes as it offers greater productivity and value recovery by enabling a larger number of log sorts that meet both domestic and international demand. even with the widespread adoption of mechanisation on steep terrain and expanding slope limits for ctl, wth still remains the preferred extraction option for steep slopes (berkett 2012; raymond 2018; visser 2018). where markets for the piled material either do not exist or extraction is deemed uneconomic, or no alternative management is applied (e.g. incineration), harvesting residues resulting from wth accumulate at landings (figure 1). residues are not only the branches and tops, but also stem offcuts from felling breakage and trimming (hall 1994). the poor form of some radiata pine crops means larger diameter segments will also accumulate at the landing. however, few market opportunities have developed to make use of the convenient accumulation of residues at steepland landings (visser et al. 2019). large residue piles remaining after harvest not only exclude land from re-establishment, but they can also present ongoing management problems. piles generate heat internally by decomposition and those where the rate of heat generation is greater than heat shedding (typically deep piles) are prone to self-combustion (buggeln & rynk 2002). piles located on unstable, steep figure 1: an example of a residue pile at an operational landing in a radiata pine plantation. terrain can mobilise and cause significant impacts on the natural and built environment (phillips et al. 2012). while there are risks posed by storing large residue piles on steep terrain over long periods of time, the resulting accumulation at landings from wth systems can benefit biomass extraction programmes due to easy access to the material. over the years, ‘fit-for-purpose’ management approaches have developed to ensure that permanent residue piles pose acceptable risk while they decompose in situ (hall 1998; visser et al. 2018). the best management practices (bmps) applicable at a particular site principally depend on the stability of the underlying soils and values at risk (nzfoa 2020). bmps range from piling on a natural terrain bench adjacent the landing, to end-hauling loads to a nearby, unused landing with several other management options in between. research was conducted in the 1990s on various aspects of new zealand’s landing residue piles, including work studies on the management of them (hall 1993a, 1993b, 1994, 1998, 1999; hall & mcmahon 1997). various methods for bulk volume measurement have been used previously, including measuring dimensions of individual woody residues as-cut (hall 1994), volumes of piles using broad approximations of geometry (hardy 1996) or measuring cross-sections of piles as they are deconstructed with heavy machinery (p. hall, personal communication, 14 april 2021). hall (1993a) established that the typical solid volume of residues discarded into the piles was approximately 4 per cent of total extracted volume (tev) for hauler operations. additionally, the relative proportion of the various components of the tree hat made up the pile mass were reported through a detailed study of log-making residues, as the material was produced. this study showed that 66 per cent of the pile volumes were made up of woody stem sections, with the remainder being branch material (hall 1994). this study was a snapshot of harvesting residue production at the time and a benchmark for future change. there is a gap in recent literature and operational knowledge around the physical characteristics of landing piles. with a national increase in harvest mechanisation and greater incentive for industrial process heat users to transition to renewable energy sources (climate change commission 2021), an upto-date knowledge of the resource is necessary for the forest owners and managers aiming to make material available for the developing bioenergy market. new tools are now available for measuring piles with complex shapes. structure from motion (sfm) photogrammetry has become an increasingly useful tool for detailed measurement and terrain modelling, with assessments of piles (of any material) an established research and commercial application of the technology (tucci et al. 2019). model construction pipelines use the known camera dimensions, identifying tie points on overlapping digital images to precisely define the camera pose in each image. known camera locations and common points on overlapping images contribute to calculations of geometry by the principle of motion parallax and a ‘cloud’ of points (point cloud) of the scene is constructed, with each point assigned a location in three dimensions. with georeferencing, point clouds can additionally be given geographic or projected coordinates for use in mapping. davis (2015) investigated the use of photogrammetry with imagery captured from an unmanned aerial vehicle (uav) to assess the volume of small (<10 m3 each) residue piles on near-flat cutovers. measurement of small accumulations of woody debris in natural and modelled fluvial systems has also been completed, with a key focus on sfm workflows (spreitzer et al. 2019, 2020). a common limitation is the estimation of surfaces occluded by piled material. this is typically handled by automated interpolation of datum surfaces for small piles or simple (i.e. flat) ground (ajayi & ajulo 2021; davis 2015), and manual inference/interpolation for more complex datum surfaces (spreitzer et al. 2019). both result in model error. where resources permit, calculating the temporal change of georeferenced surface models can reduce or eliminate surface estimation, ensuring highly accurate models of all relevant surfaces, as demonstrated by baldi et al. (2007). this research aims to provide the latest benchmark for the bulk volume of harvest residues accumulating at new zealand’s steepland landings by using modern and accessible measurement methods, demonstrating and discussing modern procedures that a forest owner/ manager may use to gain a better understanding of their own resource. improved understanding of landing residue volumes promises to assist marketing the material, and/or decisions concerning containment where residues are to remain on site in perpetuity. methods sixteen recently harvested steepland sites were made available by participating forest companies for this research. all sites were radiata pine plantations, managed under typical silvicultural regimes and covered a wide geographic spread across new zealand (table 1). forest managers provided data on regime, pre-harvest inventory, volume of each log grade sold, and harvesting method where available. except for stands mh and gn that were grown under framing regimes, all other sites were clearwood regimes. the area and bulk volume were measured for all landing residue piles associated with a harvest area at each of the study sites. three techniques were used for measuring the landing residue piles. the first technique (applied to the majority of piles) made use of photogrammetric models derived from uav photography (example see figure 2). two consumer-grade uavs were used for this study. the specifications for each model are detailed in table 2. the second technique made use of portable global positioning system (gps) tracking of pile perimeters (±6m accuracy) coupled with estimation of average pile depth. the second methodology was used on five small piles only – accounting for approximately 6 per cent of the total volume surveyed. the third was used on two sites only as validation for the first method. this involved collecting georeferenced preand postharvest photogrammetric models of the sites. harvey new zealand journal of forestry science (2022) 52:12 page 3 table 1: description of the study sites the image capture methodology adapted as fieldwork progressed. for the initial three sites (gt, mh & tp) images of residue piles were captured by manually controlling the position of the uav camera, firstly capturing images in a wide arc around the residue pile(s) then directly overhead of the pile(s), ensuring significant overlap between images. the process is described in riedinger & harvey (2021). image capture was refined for the remaining sites by using the pix4dcapture flight control application (pix4d s.a., prilly, switzerland). preprogrammed flights standardised image capture (see figure 2), ensuring image overlap exceeded 60 per cent at take-off elevation. overlap, flight extents and uav height were set to provide coverage beyond each pile’s extent and a ground sample distance lower than 3 cm/ pixel at the take-off elevation. georeferencing was used to ensure accurate dimensioning of all except two models (sites gt and mh). four ground control points (gcps) were arranged around residue piles, in locations visible to the uav camera sensor. gps coordinates of each gcp were averaged over 60 seconds using a trimble zephyr 3 rover receiver (trimble, sunnyvale, ca, u.s.a.) and subsequently post-processed to 5-15 cm accuracy using local base-station datasets. photogrammetric models were constructed using agisoft metashape (agisoft llc, st. petersburg, russia) harvey new zealand journal of forestry science (2022) 52:12 page 4 according to the flowchart in figure 3. input data for each model were the aerial photos (including geotag information) and post-processed gcp waypoints. gcp centres were located on aerial imagery prior to assignment of post-processed waypoints. ‘medium’ site code region extraction system harvest area (ha) stand age (yrs) extracted volume per hectare (m3/ha) gj canterbury waitaha ground-based 8.7 29 472 gt cable 31.0 30 546 mh ground-based 12.6 no data no data gn tasman te tai-o-aorere cable 9.5 29 611 mg ground-based 36.8 25 392 ht gisborne te tai rāwhiti cable 25.3 27 553 pk cable 23.0 25 507 ma cable 16.7 28 594 mc cable 8.3 27 866 pe cable 6.9 26 507 hf cable 13.9 27 553 mo marlborough te tauihu-o-te-waka ground-based 41.1 no data no data tp ground-based 21.2 27 407 pc wellington te whanga-nui-a-tara mixed 8.2 28 746 rk cable 6.1 26 795 tk otagoōtākou cable 33.5 33 841 table 1: harvesting site details. figure 2: example agisoft metashape point cloud of a residue pile and surrounding terrain. blue squares show the location and orientation of the camera on a grid above the pile. resolution point clouds (standard agisoft metashape setting) were constructed in agisoft metashape, which were then downsampled to output 0.1 m resolution digital elevation models (dems), along with orthophotos of varying resolutions; dependant on the limitations imposed on the pre-programmed flightpath. the dem for each pile was imported into roadeng9 terrain (softree, vancouver, b.c., canada) and a triangular irregular network (tin) model generated. a duplicate model of each residue pile was created and terrain obscured by the pile was estimated by manual interpolation due to the unique geometry of most sites. manual interpolation required features such as the fill batter top edge or fill batter bottom edge to be manually projected underneath the pile, forming an estimated terrain surface. the resulting difference between the unaltered, original tin (with pile surface) and the duplicate tin (with the interpolated terrain surface) yielded the bulk volume measure of each pile. maximum pile depths were additionally calculated by the difference in elevation between the interpolated surfaces and the unaltered, original surfaces using cloudcompare software (2.5d volume function, www. cloudcompare.org). histograms of pile depth on a 0.1 m raster grid were filtered for depths >0.1 m to eliminate noise on the pile boundaries. maximum depth was established at the 90 per cent threshold to also eliminate noise at the upper threshold (random woody residues harvey new zealand journal of forestry science (2022) 52:12 page 5 uav model camera sensor positioning system rated max. flight time per charge dji mavic pro 1/2.3” cmos effective pixels: 12.35 million gps/glonass 27 min dji mavic 2 pro 1” cmos effective pixels: 20 million gps+glonass 31 min table 2: specifications of uavs used to capture images of landing residue piles. source: dji user manuals figure 3: surface and volume calculation pipeline. poking up out of the pile). average pile depth was calculated for each site by dividing total bulk volume by the total pile plan area. to establish a measure of the manual interpolation method’s accuracy, two sites were scanned after construction (but prior to harvest), then again postharvest; these two sites contained six residue piles. the manual terrain interpolation method was completed ‘blind’ (prior to constructing the pre-harvest terrain model) and for each residue pile, two volume measures were calculated: one by computing the difference between the interpolated terrain surface and the postharvest (pile) surface, and the other between the georeferenced pre-harvest datum (terrain) surface and the post-harvest (pile) surface. results the study sites represent a typical range of harvesting systems employed to clearfell steepland forests in new zealand; from tracked ground-based to cable hauler. pile areas and volumes differ significantly between sites (table 3). to compare in more equal terms, pile areas are also expressed in pile area per hectare harvested, as the harvest areas range in size from 6.1 to 41.1 ha. of note is that several sites had residue piles that (when combined) covered approximately 0.5 ha each. not all of the pile area is lost planting area however as landing surfaces are seldom replanted in new zealand operations due to the need for soil rehabilitation and ongoing nutrient management (hall 2000). model accuracy was estimated using the agisoft postprocessing report feature and the user specified gcp locations. except at three sites, the rmse in the x-direction ranged from 1.4-4.9 cm, in the y-direction from 0.3-6.2 cm and in the z-direction from 0.2-5.9 cm for all sites. site tp used gcps but no automated flight control, gaining an x/y/z rmse of 4.5/8.3/3.2 cm. sites gt and mh neither used automated flight control, nor gcps, therefore gained x/y/z rmses for estimated camera locations of 85/59/68 cm and 78/72/40 cm, respectively. similar to pile area, bulk pile volumes varied significantly among study sites. table 4 details the volume measured, and the directly comparable metrics of bulk volume per hectare harvested and bulk volume per tonne harvested. the measures of average pile depth and maximum pile depth (see table 3) indicate that for most harvests, there is little difficulty in achieving a pile height less than 3 m to align with current industry guidelines. table 5 provides the summary statistics from table 3 and table 4. each site has been considered a data point in generating the mean values. a reduced dataset size is indicated where data could not be provided by the hosting forest manager. harvey new zealand journal of forestry science (2022) 52:12 page 6 site code total # of piles combined pile area (m2) pile area per hectare harvested (m2/ha) mean pile depth (m) max pile depth at the 90% threshold (m) gj 1 1360 160 1.6 4.3 gt 2 4730 150 0.7 2.2 gn 3 1500 160 0.7 1.7 ht 3 3680 150 0.8 1.7 hf 4 3450 250 0.7 2.1 mh 2 3730 300 0.5 2.8 mo 1 2630 60 2.8 5.5 tp 2 1040 50 0.9 2.1 pk 1 4910 210 0.9 2.3 pe 2 1340 190 0.9 2.2 ma 3 3750 220 0.5 1.6 mc 2 2680 320 0.7 1.9 mg 1 1480 40 1.4 3.7 pc 2 1920 230 1.2 3.1 rk 1 2280 370 0.9 2.3 tk 7 5560 170 0.8 table 3: ground covered by residue piles and depth statistics. the assessment of accuracy for the six piles over two sites (sites pe and hf) found that the mean difference between the volumes yielded by the manual interpolation method and the georeferenced pre-harvest datum surface method was 19%, with a range of 49%. five of the six pile volumes were underestimated by the manual interpolation method – where the assumption is made that the georeferenced datum surface method is correct and the datum surface (landing shape/height etc.) remains constant between pre-and-post-harvest data collection visits. discussion one of the goals of this study was to provide an upto-date benchmark of residue pile volumes. previous benchmarks were provided by earlier studies. for example, a 1993 study investigated the retrieval of residue piles at four hauler landings, which involved surveying the bulk volume of material moved and that beyond the machine’s reach (hall 1993a). the mean bulk volume and mean tev were 1400 m3 and 6694 m3, respectively. by assuming the density of freshly harvested radiata pine is 1 t/m3, the estimated bulk volume of residue harvey new zealand journal of forestry science (2022) 52:12 page 7 table 4: volumes of residue piles. site code combined bulk volume (m3) bulk vol. per hectare harvested (m3/ha) bulk vol. per tonne harvested (m3/t) gj 2190 250 0.53 gt 3520 110 0.21 gn 1100 120 0.19 ht 2770 110 0.20 hf 2250 160 0.29 mh 1880 150 no data mo 7235 180 no data tp 940 40 0.11 pk 4474 190 0.38 pe 1104 160 0.32 ma 1831 110 0.18 mc 1958 240 0.27 mg 2096 60 0.15 pc 2317 280 0.38 rk 2160 350 0.45 tk 4396 130 0.16 attribute value harvest area (ha) 18.9 number of piles per site 2.4 bulk pile volume per site (m3) 2600 pile area per site (m2) 2900 pile area per hectare harvested (m2/ha) 190 bulk pile volume per hectare harvested (m3/ha) 170 pile depth (m) 0.92 max. pile depth (at the 90% threshold) (m) *15 sites 2.6 bulk vol. / tonne harvested (m3/t) *14 sites 0.23 table 5: mean values for all parameters across all sites (n=16). piles in 1993 was approximately 0.21 m3/t harvested. a subsequent study in 1994 measured the solid volume of branch and stem material discarded from six log making operations on radiata pine cable hauler sites and three ground-based harvesting sites (hall 1994). for the hauler sites (assuming that they were steepland harvests and therefore comparable) mean measured solid log making residue volumes were 13.8 per cent of tev, excluding the douglas-fir (pseudotsuga menziesii) stand datum. making the same assumption on radiata pine density and assuming a bulk density of 0.25 t/ m3 for woody residues (visser et al. 2010), the comparative bulk density figure for residues in the 1994 study is 0.55 m3/t. that places the 1994 result at the upper end of the range measured in this study. recognising that a series of assumptions underlie these comparisons, it is significant that the bulk volume per tonne harvested the 1994 result is more than double the result from one year earlier. this study corroborates the results in the study by hall (1993), although market conditions, harvesting machinery and harvest practice have changed markedly during the intervening years. managing residue pile area requires balancing of competing interests. soil area covered by a residue pile(s) can represent an opportunity cost by reducing land area available for establishing the succeeding crop. however, concentrating residues by piling high carries an increased risk of self-ignition. self-ignition thresholds in radiata pine residue piles requires further research to expand the working knowledge, however anecdotal evidence has formed the basis for the current bmp for residue pile height (3 m) (nzfoa 2020). for the piles measured in this study, the average maximum depth is below the current target bmp for height. increasing the average pile height may allow more land area to be replanted, however high stacking can only be done on stable ground (nzfoa 2020), with the understanding that increasing pile height on a slope decreases its stability and increases risk of mass movement. the results show that on these study sites, maintaining low overall pile depth takes priority over maximising replanted area. further insight into the results of individual sites should consider operational factors. additional reasons for a high or low pile area per hectare may be explainable by the log market conditions, landing layout, machinery used, management instructions or terrain form. such finer details are beyond the scope of this study and may require much larger datasets to establish meaningful conclusions. new zealand plantations are now trending more towards framing regimes. the majority of sites measured in this study were clearwood regimes, as is typical of most 1990s crops. framing regimes when compared to clearwood (typically) have higher trv, higher stocking and smaller piece size. reduced piece size has been shown to lower the breakage rates (murphy 1982), however with more stems per hectare the effect may be negated. whether the physical differences between regimes results in a measurable difference in landing residue volumes (all other factors controlled) is yet to be established. this research was completed soon after a number of major mass mobilisation events (cave et al. 2017). as such the measurements made for this study may already reflect changes to practices for both creating and storing harvest residue piles on or near landings in steep terrain. it is recognised that the current strong emphasis on both minimising environmental risk as well as creating biomass market opportunities for renewable energies may have already resulted in changed residue management practices (dale 2019; visser et al. 2018). while this study cannot predict those changes, it can serve as a benchmark and reference point to measure future developments against. this research demonstrates the ability to investigate the bulk volume of residue piles with the modern equipment and software tools available (or cloudcomputing substitutes) to today’s forest manager. previous methods required measurement in association with heavy machinery or estimation of both the terrain below and the surface of the pile (hardy 1996), making use of the tools available at the time. the method employed in this study improves on previous methods by better modelling the pile surface. one way for improving terrain estimation would be to site residue piles on as flat ground as possible. whilst clearly advantageous for measurement accuracy, several factors precluded the viability of the idea for this study. with additional time, more accurate results would be obtained by establishing a georeferenced datum surface of the completed landing formation, prior to piling with harvest residues – as used for the validation of the method employed in this study. where serious consideration is given to removal of the product, accuracy may become increasingly important, justifying modelling landings (and surrounds) preharvest. results from this project can assist forest managers to predict the bulk volume of residues that may accumulate at a steepland landing during a wth operation. estimation of volumes is recommended by guidance documents for current legislation (mpi 2018) to ensure storage capacity is adequate – or alternatives are planned for. these results are advantageous for feasibility studies on a forest’s ability to supply a biomass market with landing residues. finally, this study sets the latest benchmark for residue volumes as harvesting machinery, methodology and markets develop over time. there is little, recently published information on the volumes of harvest residues discarded at new zealand’s steepland landings. this study addresses the question, but much more can be done at a finer scale with the resources and data available to commercial operators. it is intended that this study provides accessible ideas and tools to foresters who are looking to supply (but not currently supplying) a biomass market. it is important that the industry collects information on the material as security of supply is critical to business cases for heat users considering conversion from fossil fuels to residual biomass. while international log markets continue to demand small-diameter or industrial logs, residual biomass will play a vital role in meeting bioenergy demand locally. the procedures discussed in this paper harvey new zealand journal of forestry science (2022) 52:12 page 8 require limited training and many can be completed with cloud-computing services, reducing computing capacity issues. conclusions this is a renewed look at residual biomass accumulations at landings and demonstrates how an investigation could be conducted in a forest or forest estate with tools readily available to today’s forester. it sets the latest benchmark for landing residue pile volumes in new zealand’s steepland plantations. markets for harvest residues are developing, regime change is occurring, and innovations to harvest systems are promising to reduce the production of broken/low quality material. the information provided on current steepland pile volumes offers new zealand forestry companies, forest owners and the biomass market, a better understanding about the current availability of the material in steepland plantations and therefore potential for increased utilisation. competing interests the author declares that they have no competing interests. acknowledgements the author would like to acknowledge the support by the forest companies and managers who were part of this study and hopes these methods and results may assist developments in the market for steepland landing residues. this project is part of a phd research programme, and the author acknowledges the guidance by professor rien visser of the school of forestry, university of canterbury and peter hall of scion. funding this research project is supported by forest growers research ltd. and the nzif foundation. the author gratefully acknowledges the support received from both organisations. references ajayi, o.g., & ajulo, j. (2021). investigating the applicability of unmanned aerial vehicles (uav) photogrammetry for the estimation of the volume of stockpiles. quaestiones geographicae, 40(1), 2538. https://doi.org/10.2478/quageo-2021-0002 baldi, p., coltelli, m., fabris, m., marsella, m., & tommasi, p. (2007). high precision photogrammetry for monitoring the evolution of the nw flank of stromboli volcano during and after the 2002–2003 eruption. bulletin of volcanology, 70(6), 703-715. https://doi.org/10.1007/s00445-007-0162-1 berkett, h. (2012). an examination of the current slope gradients being experienced by ground based forest machines in new zealand plantation forests. masters in forestry science thesis, university of canterbury. retrieved from https://go.exlibris. link/rw94d4by buggeln, r., & rynk, r. (2002). self-heating in yard trimmings: conditions leading to spontaneous combustion. compost science & utilization, 10(2), 162-182. https://doi.org/10.1080/106565 7x.2002.10702076 cave, m., davies, n., & langford, j. (2017). gdc report. cyclone cook slash investigation. 121 p. gisborne, nz: gisborne district council. retrieved from h t t p s : / / w w w. g d c . g o v t . n z / _ _ d a t a / a s s e t s / p d f _ f i l e / 0 0 1 3 / 1 0 4 0 8 / c y c l o n e c o o ks l a s h investigation-2017-report.pdf climate change commission. (2021). ināia tonu nei: a low emissions future for aotearoa. ccc report. retrieved from https://ccc-production-media. s3.ap-southeast-2.amazonaws.com/public/inaiatonu-nei-a-low-emissions-future-for-aotearoa/ i n a i a t o n u n e i a l o we m i s s i o n s f u t u re f o r aotearoa.pdf dale, r. (2019). workshop report: harvest residue mangement on erosion prone land. fgr workshop report. 18 p. rotorua, nz: forest growers research (fgr). retrieved from https://fgr.nz/documents/ download/7601?612598755 davis, b. (2015). refinement of a drone-based method for estimating coarse woody debris and biomass residue following forest harvest. (msc.), university of georgia, georgia, usa. retrieved from https:// getd.libs.uga.edu/pdfs/davis_brian_j_201708_ ms.pdf fraser, d., palmer, d., mcconchie, m., & evanson, t. (1997). breakage in manually-felled clearfell radiata pine. liro project report 63, 42 p. rotorua, nz: logging industry research organisation new zealand (liro). retrieved from https://fgr.nz/documents/ download/4122?1322711812 gordon, a.d. (2005). forest sampling and inventory. in m. colley (ed.), forestry handbook (4 ed., pp. 132135). tauranga, new zealand: the new zealand institute of forestry (inc.). goulding, c.j. (2005). measurement of trees. in m. colley (ed.), forestry handbook (4 ed., pp. 145-148). tauranga, new zealand: the new zealand institute of forestry (inc.). hall, p. (1993a). dismantling of accumulations of logging residue around hauler landings. liro report 18(6), 6 p. rotorua, nz: logging industry research organisation new zealand (liro). retrieved from https://fgr.nz/documents/ download/4578?2046499472 hall, p. (1993b). skid site rehabilitation. liro report 18(8), 6 p. rotorua, nz: logging industry research organisation new zealand (liro). harvey new zealand journal of forestry science (2022) 52:12 page 9 https://doi.org/10.2478/quageo-2021-0002 https://doi.org/10.1007/s00445-007-0162-1 https://go.exlibris.link/rw94d4by https://go.exlibris.link/rw94d4by https://doi.org/10.1080/1065657x.2002.10702076 https://doi.org/10.1080/1065657x.2002.10702076 https://www.gdc.govt.nz/__data/assets/pdf_file/0013/10408/cyclone-cook-slash-investigation-2017-report.pdf https://www.gdc.govt.nz/__data/assets/pdf_file/0013/10408/cyclone-cook-slash-investigation-2017-report.pdf https://www.gdc.govt.nz/__data/assets/pdf_file/0013/10408/cyclone-cook-slash-investigation-2017-report.pdf https://ccc-production-media.s3.ap-southeast-2.amazonaws.com/public/inaia-tonu-nei-a-low-emissions-future-for-aotearoa/inaia-tonu-nei-a-low-emissions-future-for-aotearoa.pdf https://ccc-production-media.s3.ap-southeast-2.amazonaws.com/public/inaia-tonu-nei-a-low-emissions-future-for-aotearoa/inaia-tonu-nei-a-low-emissions-future-for-aotearoa.pdf https://ccc-production-media.s3.ap-southeast-2.amazonaws.com/public/inaia-tonu-nei-a-low-emissions-future-for-aotearoa/inaia-tonu-nei-a-low-emissions-future-for-aotearoa.pdf https://ccc-production-media.s3.ap-southeast-2.amazonaws.com/public/inaia-tonu-nei-a-low-emissions-future-for-aotearoa/inaia-tonu-nei-a-low-emissions-future-for-aotearoa.pdf https://ccc-production-media.s3.ap-southeast-2.amazonaws.com/public/inaia-tonu-nei-a-low-emissions-future-for-aotearoa/inaia-tonu-nei-a-low-emissions-future-for-aotearoa.pdf https://fgr.nz/documents/download/7601?612598755 https://fgr.nz/documents/download/7601?612598755 https://getd.libs.uga.edu/pdfs/davis_brian_j_201708_ms.pdf https://getd.libs.uga.edu/pdfs/davis_brian_j_201708_ms.pdf https://getd.libs.uga.edu/pdfs/davis_brian_j_201708_ms.pdf https://fgr.nz/documents/download/4122?1322711812 https://fgr.nz/documents/download/4122?1322711812 https://fgr.nz/documents/download/4578?2046499472 https://fgr.nz/documents/download/4578?2046499472 harvey new zealand journal of forestry science (2022) 52:12 page 10 retrieved from https://fgr.nz/documents/ download/4574?1197501549 hall, p. (1994). waste wood at logging landings. liro report 19(15), 14 p. rotorua, nz: logging industry research organisation new zealand (liro). retrieved from https://fgr.nz/documents/ download/4514?1060028902 hall, p. (1998). logging residue at landings. new zealand forestry, 43(1), 30-32. hall, p. (1999). effect of logmaking on production of stem wood residue at landings. liro report 24 (4), 4 p. rotorua, nz: liro forestry solutions. retrieved from https://fgr.nz/documents/ download/4310?879758742 hall, p. (2000). skid site rehabilitation tree growth trials, results to age 7. liro report 25(1), 10 p. rotorua, nz: liro forestry solutions. retrieved from https:// fgr.nz/documents/download/4284?696343852 hall, p., & mcmahon, s. (1997). logging residue at hauler landings results from an industry survey. liro report 22(2), 8 p. rotorua, nz: liro forestry solutions. retrieved from https://fgr.nz/ documents/download/4394?565924820 hardy, c.c. (1996). guidelines for estimating volume, biomass, and smoke production for piled slash. general technical report pnw-gtr-364. 28 p. portland, or, usa: u.s. department of agriculture. retrieved from https://www.fs.fed.us/pnw/pubs/ pnw_gtr364.pdf harvey, c., & visser, r. (2022). characterisation of harvest residues on new zealand’s steepland plantation cutovers. new zealand journal of forestry science, 52:7, 12 p. https://doi.org/10.33494/ nzjfs522022x174x lambert, m.j. (1996). stem breakage of pinus radiata during mechanical felling in kinleith forest, central north island, new zealand. master of forestry science thesis, university of canterbury. retrieved from https://go.exlibris.link/9dgdhqry maclaren, j.p. (1993). radiata pine growers’ manual fri bulletin no. 184 (pp. 144). rotorua, new zealand: forest research institute. maclaren, j.p., & knowles, r.l. (2005). silviculture of radiata pine. in m. colley (ed.), forestry handbook (4 ed., pp. 119-121). tauranga, new zealand: the new zealand institute of forestry (inc.). mpi, ministry for primary industries. (2015). 2015 yield tables. retrieved from: https://www.mpi.govt.nz/ dmsdocument/30954-2015-yield-tables mpi, ministry for primary industries. (2018). harvesting (regulations 62-71). wellington, nz: ministry for primary industries (mpi). retrieved from https:// www.mpi.govt.nz/dmsdocument/27948-nes-pf5.6-harvesting-regulations-62-71 mpi, ministry for primary industries. (2020). national exotic forest description as at 1 april 2020. wellington, nz: ministry for primary industries (mpi). retrieved from https://www.mpi.govt.nz/ dmsdocument/43540-2020-nefd-report murphy, g. (1982). value savings from alternative felling patterns on steep country. lira report 7(8), 4 p. rotorua, nz: logging industry research association inc. retrieved from https://fgr.nz/ documents/download/5053?973453426 nzfoa, new zealand forest owners association. (2020). forest practice guides managing processing slash on landings. new zealand forest owners association. retrieved from https://docs.nzfoa.org.nz/site/ assets/files/1509/6-1_harvest-slash_managingprocessing-slash-on-landings-2-0.pdf phillips, c., marden, m., & basher, l. (2012). plantation forest harvesting and landscape response what we know and what we need to know. new zealand journal of forestry, 56(4), 4-12. raymond, k. (2018). pgp steepland harvesting – a collaborative research and development programme. new zealand journal of forestry, 63(3), 4 p. retrieved from http://www.nzjf.org. nz/free_issues/nzjf63_3_2018/166de66f-78e042a2-8fa1-5453573fdc3d.pdf riedinger, l., & harvey, c. (2021). structure-from-motion photogrammetry as a tool for harvest residue pile measurement. new zealand journal of forestry, 65(4), 6-11. spreitzer, g., tunnicliffe, j., & friedrich, h. (2019). using structure from motion photogrammetry to assess large wood (lw) accumulations in the field. geomorphology, 346. https://doi.org/10.1016/j. geomorph.2019.106851 spreitzer, g., tunnicliffe, j., & friedrich, h. (2020). porosity and volume assessments of large wood (lw) accumulations. geomorphology, 358. https:// doi.org/10.1016/j.geomorph.2020.107122 tucci, g., gebbia, a., conti, a., fiorini, l., & lubello, c. (2019). monitoring and computation of the volumes of stockpiles of bulk material by means of uav photogrammetric surveying. remote sensing, 11(12). https://doi.org/10.3390/rs11121471 twaddle, a. (1987). piece size characteristics of managed stands after felling. lira report 9(1), 4 p. rotorua, nz: logging industry research association inc. (lira). retrieved from https://fgr.nz/documents/ download/5210?2137134046 visser, r. (2018). benchmarking harvesting cost and productivity – 2017 update. fgr report htn1002, 4 p. rotorua, nz: forest growers research (fgr). retrieved from https://fgr.nz/documents/ download/7091?936289211 https://fgr.nz/documents/download/4514?1060028902 https://fgr.nz/documents/download/4514?1060028902 https://fgr.nz/documents/download/4310?879758742 https://fgr.nz/documents/download/4310?879758742 https://fgr.nz/documents/download/4284?696343852 https://fgr.nz/documents/download/4284?696343852 https://fgr.nz/documents/download/4394?565924820 https://fgr.nz/documents/download/4394?565924820 https://www.fs.fed.us/pnw/pubs/pnw_gtr364.pdf https://www.fs.fed.us/pnw/pubs/pnw_gtr364.pdf https://doi.org/10.33494/nzjfs522022x174x https://doi.org/10.33494/nzjfs522022x174x https://go.exlibris.link/9dgdhqry https://www.mpi.govt.nz/dmsdocument/30954-2015-yield-tables https://www.mpi.govt.nz/dmsdocument/30954-2015-yield-tables https://www.mpi.govt.nz/dmsdocument/27948-nes-pf-5.6-harvesting-regulations-62-71 https://www.mpi.govt.nz/dmsdocument/27948-nes-pf-5.6-harvesting-regulations-62-71 https://www.mpi.govt.nz/dmsdocument/27948-nes-pf-5.6-harvesting-regulations-62-71 https://www.mpi.govt.nz/dmsdocument/43540-2020-nefd-report https://www.mpi.govt.nz/dmsdocument/43540-2020-nefd-report https://fgr.nz/documents/download/5053?973453426 https://fgr.nz/documents/download/5053?973453426 https://docs.nzfoa.org.nz/site/assets/files/1509/6-1_harvest-slash_managing-processing-slash-on-landings-2-0.pdf https://docs.nzfoa.org.nz/site/assets/files/1509/6-1_harvest-slash_managing-processing-slash-on-landings-2-0.pdf https://docs.nzfoa.org.nz/site/assets/files/1509/6-1_harvest-slash_managing-processing-slash-on-landings-2-0.pdf http://www.nzjf.org.nz/free_issues/nzjf63_3_2018/166de66f-78e0-42a2-8fa1-5453573fdc3d.pdf http://www.nzjf.org.nz/free_issues/nzjf63_3_2018/166de66f-78e0-42a2-8fa1-5453573fdc3d.pdf http://www.nzjf.org.nz/free_issues/nzjf63_3_2018/166de66f-78e0-42a2-8fa1-5453573fdc3d.pdf https://doi.org/10.1016/j.geomorph.2019.106851 https://doi.org/10.1016/j.geomorph.2019.106851 https://doi.org/10.1016/j.geomorph.2020.107122 https://doi.org/10.1016/j.geomorph.2020.107122 https://doi.org/10.3390/rs11121471 https://fgr.nz/documents/download/5210?2137134046 https://fgr.nz/documents/download/5210?2137134046 https://fgr.nz/documents/download/7091?936289211 https://fgr.nz/documents/download/7091?936289211 harvey new zealand journal of forestry science (2022) 52:12 page 11 visser, r., hall, p., & raymond, k. (2010). good practice guide: production of wood fuel from forest landings. (technical guide 9.0), 44 p. wellington, nz: the energy efficiency & conservation authority (eeca). retrieved from https://ir.canterbury. ac.nz/bitstream/handle/10092/5545/12626362_ g o o d % 2 0 p r a c t i c e % 2 0 g u i d e % 2 0 w o o d % 2 0 fuel%20final.pdf ?sequence=1&isallowed=y visser, r., harrill, h., & baek, k. (2019). biomass recovery operations in new zealand: a review of the literature. fgr report h041, 32 p. rotorua, nz: forest growers research (fgr). retrieved from https://fgr.nz/documents/download/8199 visser, r., spinelli, r., & brown, k. (2018). best practices for reducing harvest residues and mitigating mobilisation of harvest residues in steepland plantation forests., 53 p. gisborne, nz: gisborne district council (gdc). retrieved from https://www.gdc.govt .nz/__data/assets/pdf_ file/0012/13710/best-practices-for-reducingharvest-residues-and-mitigating-mobilisation-ofharvest-residues-in-steepland-plantation-forests. pdf https://ir.canterbury.ac.nz/bitstream/handle/10092/5545/12626362_good%20practice%20guide%20wood%20fuel%20final.pdf?sequence=1&isallowed=y https://ir.canterbury.ac.nz/bitstream/handle/10092/5545/12626362_good%20practice%20guide%20wood%20fuel%20final.pdf?sequence=1&isallowed=y https://ir.canterbury.ac.nz/bitstream/handle/10092/5545/12626362_good%20practice%20guide%20wood%20fuel%20final.pdf?sequence=1&isallowed=y https://ir.canterbury.ac.nz/bitstream/handle/10092/5545/12626362_good%20practice%20guide%20wood%20fuel%20final.pdf?sequence=1&isallowed=y https://fgr.nz/documents/download/8199 https://www.gdc.govt.nz/__data/assets/pdf_file/0012/13710/best-practices-for-reducing-harvest-residues-and-mitigating-mobilisation-of-harvest-residues-in-steepland-plantation-forests.pdf https://www.gdc.govt.nz/__data/assets/pdf_file/0012/13710/best-practices-for-reducing-harvest-residues-and-mitigating-mobilisation-of-harvest-residues-in-steepland-plantation-forests.pdf https://www.gdc.govt.nz/__data/assets/pdf_file/0012/13710/best-practices-for-reducing-harvest-residues-and-mitigating-mobilisation-of-harvest-residues-in-steepland-plantation-forests.pdf https://www.gdc.govt.nz/__data/assets/pdf_file/0012/13710/best-practices-for-reducing-harvest-residues-and-mitigating-mobilisation-of-harvest-residues-in-steepland-plantation-forests.pdf https://www.gdc.govt.nz/__data/assets/pdf_file/0012/13710/best-practices-for-reducing-harvest-residues-and-mitigating-mobilisation-of-harvest-residues-in-steepland-plantation-forests.pdf effect of chemical modification and heat treatment on biological durability and dimensional stability of pinus roxburghii sarg. ajmal samani, sauradipta ganguly* and sanjeet kumar hom wood preservation discipline, forest products division, forest research institute, dehradun 248006, india *corresponding author: sauradipta.ganguly@fridu.edu.in (received for publication 11 november 2020; accepted in revised form 23 november 2021) abstract background: depleting supplies of wood species with inherent natural durability has resulted in the focus being shifted to non-durable plantation grown and imported timber. despite its abundant availability and better treatability, the use of pinus roxburghii is limited to packing cases, crates, shutters, door and window frame, carpentry and joinery items due to its nondurable nature. hence, to promote use of such timber for applications such as decking, cladding and facade elements chemical modification with a combination of citric acid and sodium hypophosphite, and heat treatment were explored to improve its service life. methods: chemical modification was performed using a water solution of citric acid (6.9%) and sodium hypophosphite (6.5%) followed by curing at 140°c for 8 hrs. dimensional stability was determined by estimating the volumetric swelling coefficient and anti-swelling efficiency (ase) of treated and control samples. durability against fungus and termites was evaluated using a soil block bioassay and termite mound test as per standard methods. results: both chemical modification and heat treatment of p. roxburghii resulted in enhanced dimensional stability and biological durability compared to the untreated controls. chemical modification and heat treatment resulted in 23.05% and 18.37% volumetric ase, respectively. results showed that a highly perishable species became significantly more durable after chemical modification, exhibiting 5–6 times less mass loss by termites in comparison to the controls. wood samples modified with citric acid showed excellent resistance to both white and brown rot fungi and exhibited 14-15 times less reduction in mass compared with untreated samples. conclusions: citric acid chemical modification is an environment friendly process that improved the dimensional stability as well as resistance against biodegradation. these studies may provide valuable inputs to establish this mode of chemical modification as a cost-effective alternative to other chemicals for wood preservation. the concentrations of the chemicals and temperature for fixation may be varied to establish an optimum combination for best output. new zealand journal of forestry science samani et al. new zealand journal of forestry science (2021) 51:15 https://doi.org/10.33494/nzjfs512021x143x e-issn: 1179-5395 published on-line: 21/12/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access applications, wood is either used outdoors or indoors. when exposed to outdoor conditions and not in direct contact with the ground, in either open or under shade conditions, wood needs subsequent protection for a better service life (militz 2015; humar et al. 2017). wood and wood-based products used in buildings and introduction wood is an excellent construction material and is gaining attention worldwide for use in a range of different applications due to its positive environmental credentials, good workability, ease of availability and sustainable nature. in construction and other common keywords: biological durability, chemical modification, citric acid, dimensional stability, oligoporus placenta, sodium hypophosphite, trametes versicolor mailto:sauradipta.ganguly@fridu.edu.in http://creativecommons.org/licenses/by/4.0/), samani et al. new zealand journal of forestry science (2021) 51:15 page 2 other constructions outdoors are subjected to a wide range of environmental conditions in service (kržišnik et al. 2018) which may inflict different levels of moisture stress and biodeterioration (isaksson et al. 2011). present day research has shed light on the fact that the water excluding efficiency (wee) of wood in service is of paramount importance to predict its performance in service in addition to its inherent durability which may be attributed to the presence of biocides or the biologically active extractives in wood (brischke et al. 2006; meyer-veltrup et al. 2017). depleting supplies of wood species with inherent natural durability has resulted in the focus being shifted to non-durable plantation grown and imported timber amongst indian researchers (ganguly& tripathi 2018; samani et al. 2019; ganguly et al. 2020; hom et al. 2020a). the non-durable nature of such species calls for particular attention before putting them to use. despite its abundant availability and better treatability, the use of pinus roxburghii is limited to packing cases, crates, shutters, door and window frame, carpentry and joinery items. for local building purposes, p. roxburghii is the least preferred species due to its poor inherent durability and resulting extremely high susceptibility to decaying agents, when compared with other conifers. hence, to promote use of such non-durable timber for applications such as decking, cladding and facade elements, proper modification or treatment techniques must be adopted to ensure its continuous service in use. wood modification alters the physical and chemical properties of wood to enhance its performance and durability in service (sandberg et al. 2017; samani et al. 2019; samani et al. 2020). the present day focus on wood modification and protection is on environmentally benign agents (ganguly 2018; samani et al. 2020; treu et al. 2020; ganguly et al. 2021) keeping in mind their impact on nature and the environment. through a better understanding of the mode of action, cost-effective treatment methods can be explored (singh and singh, 2012). chemical modification (cm) was reported to stabilise wood, improve its moisture performance, photodegradation and biodeterioration and make its performance more desirable in service (rowell 1983; donath et al. 2004; hansmann et al. 2005; hom et al. 2020b). cm implies etherification, esterification or acetylation between some chemical and hydroxyl (-oh) group of wood. important parameters for successful modification are temperature, processing time, type of catalyst and wood species (despot et. al. 2008). the citric acid (ca) modification is a two-step esterification and takes place in the presence of heat and is often considered an eco-friendly and cost-effective alternative to impart biological durability (bd) (hasan et al. 2006 & 2007; despot et al. 2008; treu et al. 2020) and dimensional stability (ds) (samani et al. 2020). in the first step of ca modification, a cyclic anhydride is formed between two adjacent carboxylic acid groups of ca molecules, due to the loss of one molecule of water. in the second step, this cyclic anhydride reacts with the wood hydroxyl group to form an ester linkage between the two substrates. if the reaction temperature is sufficiently high and/or the reaction time is long, a second carboxylic group on the citric acid can form another bond with the wood resulting in the formation of a crosslink (mcsweeny et al. 2006). sodium hypophosphite (shp) arguably is the best catalyst used for cm using ca as in its presence anhydride intermediates are formed faster and degradation of cellulose is minimal at the time of curing, although the mode of action and results are not fully understood (feng et al. 2014). sefc et al. (2009) reported fir (abies alba mill.) and beech (fagus sylvatica l.) wood modified by ca using shp and monosodium phosphate as catalyst returned excellent anti swelling efficiency (ase) with both the catalysts. the average ase was about 54% in fir wood and about 40% in beech wood which can be termed satisfactory as far as a successful modification is concerned. heat treatment (ht) is also reported for its positive contribution to wood research by improving ds and bd (esteves and pereira 2009) which however may vary depending on treatment types, temperatures, species (militz 2002) and even the orientation of wood. reduction in the highly reactive -oh groups (weiland & guyonnet 2003) is the primary reason for the enhanced ds of ht wood. significantly higher bd of ht wood was reported previously by dirol and guyonnet (1993) at higher temperatures where modified specimens of spruce (picea abies (l.) h. karst.), fir (abies spp.) and poplar (populus spp.) showed an ml of less than 1% at temperatures between 200°c-260°c whereas 40% ml of the untreated set was recorded against several rotting fungi. higher temperatures and longer duration of treatment were previously reported to impart better bd by kim et al. (1998). based on the above cited literature, wood samples of pinus roxburghii were modified using cm and ht to assess the performance of modified wood in terms of ds and bd which may enhance its performance in service. materials and methods seasoned planks of p. roxburghii were converted into samples of dimension 50 × 50 × 50 mm3 for the dimensional stability (ds) test, 19 × 19 × 19 mm3 for the soil block assay (sba) and 100 × 25 × 6 mm3 for the termite mound test (tmt). samples were free of any visual defect or deformation and to ascertain homogeneity in data, all samples were prepared from the same part of the plank. a total of three sets of samples were prepared for ds, sba and tmt and each set had six replicates (table 1). the samples treated with ca (6.9%) and shp (6.5%) (t1) by applying an initial vacuum of 30 mm hg for 10 min and then the vessel was filled with the treating solution. samples were maintained under the same vacuum for 3 h. the vacuum was released after 3 h and samples were allowed to remain soaked in the solution for 18 h under atmospheric pressure (despot et al. 2008). after completion of the impregnation cycle, samples were removed from the treatment solution, air dried for 48 h at room condition and kept in the laboratory oven at 140°c for 8 h for thermo condensation (hasan et al. 2012). the second set (t2) was kept in the oven at140°c for 8 h to see the effect of heat. the final set (t3) consisted of untreated control samples. https://www.sciencedirect.com/science/article/pii/s0926669018302255?via%3dihub&amp;bib0165 weight percent gain (wpg) the weight percent gain (wpg) of the specimens after cm and ht (loss) was calculated using the following equation: wpg= (wt – wo)/wo*100 (1) where wo and wt are oven dried (od) weights of unmodified and cm/ht samples, respectively. dimensional stability (ds) ds was determined by comparing the total volumetric swelling (vs) of treated and control samples. the samples of all sets were oven dried (od) at 103±2°c, and both od weight and initial dimensions were recorded. swelling was determined by a water immersion test. all samples were submerged in distilled water and were kept under vacuum for 30 minutes and allowed to soak for 24 h (sefc et al. 2009). weight and dimensions of saturated samples were recorded. the swelling coefficient was determined as: s (%) = (vs-vo)/vo*100 (2) where vs is the volume of soaked sample and vo is the volume of oven dried sample. the anti-swelling efficiency (ase) and water excluding efficiency (wee%) were calculated as follows: ase (%) = (su –sm)/su*100 (3) where su and sm are swelling coefficients of control and cm/ht wood samples respectively. wee(%) = (wc-wt)/wc*100 (4) where wc= mass of water absorbed by untreated controls (g) and wt=water absorbed by cm/ht samples (g). soil block bioassay(sba) the sba was carried out as per is 4873 (anon. 2008). one brown rot and one white rot fungi were chosen for the sba study namely oligoporous placenta murr. (op) (fri culture no. 180) and trametes versicolor linn. (tv) (fri culture no. 651). fresh cultures of the fungi were obtained from the collection of the forest pathology group at the forest research institute, india. the test blocks of size 19×19×19 mm3and the feeder blocks of size 4 × 19 × 35 mm3 were prepared along the length of grain. bombax ceiba sapwood was used for preparation of the feeder blocks. the test blocks were oven dried at 103±2°c and then conditioned at 75% relative humidity (rh) until constant weights(w1) were achieved. w1 was considered as the weight before incubation. sieved, air-dried garden soil (125g) having ph 5.07.0 was placed (compacted by tapping) in screw capped bottles. 130% of water holding capacity of the soil in these test bottles was maintained by adding distilled water (44 ml). two feeder blocks were placed directly on the surface of the soil. the prepared bottles were closed loosely and sterilised in an autoclave at a pressure of 1 kg/cm2 for 30 min. sterilised culture bottles were cooled and the fungal inoculum from freshly grown culture, approximately 8–10 mm in diameter, was placed on the edge of the feeder blocks in culture bottles. post inoculation, bottles were incubated in bod (biochemical oxygen demand) with slightly loosened lids at 25±2oc and 70±4% rh for approximately three weeks, until the feeder blocks were completely covered by the test fungi. two test blocks were placed on feeder blocks in contact with mycelium along the cross-sectional face in each culture bottle. the bottles containing the test blocks were incubated for a period of 12 weeks in the incubator maintained at 25±2°c and an rh of about 70±4%. at the end of the incubation period the blocks were removed from the culture bottles, cleaned off from the adhering mycelium with a soft tissue and dried at room temperature for 3-4 days. subsequently the blocks were again dried in a hot air oven and then conditioned at 75% rh until the constant weights (w2) were obtained. calculation of mass loss (ml) ml (%) was calculated from the conditioned weight of the blocks before and after sba. ml(%) = (w1-w2)/w1*100 (5) samani et al. new zealand journal of forestry science (2021) 51:15 page 3 table 1: description of the study sites set treatment name treatment conditions number of samples assessed for each property* total ds sba tmt rs ts vs t1 cm: ca+shp citric acid (6.9%) and sodium hypophosphite (6.5%) 6 6 6 6 6 6 36 t2 ht: heat heat treatment 140oc for 8 hrs 6 6 6 6 6 6 36 t3 c: control control 6 6 6 6 6 6 36 total 18 18 18 18 18 18 108 table 1: treatments and sample distribution * r = radial, t = tangential, v = volumetric, s = swelling coefficient; tmt = termite mound test; sba = soil block bioassay; ds = dimensional stability where w1= conditioned weight of the blocks before test and w2= conditioned weight of the blocks after the test. termite mound test (tmt) the tmt was conducted as per is 4833 (anon. 1993). cm, ht and control blocks were buried at different places inside a termite mound of odontotermes obesus (rambur 1842) at the start of may. blocks were removed from the mound in november when activity of termites almost ceased due to fall in temperature. the blocks were examined for termite attack and reinstalled in the following may to have exposure to termites for two successive termite seasons. after removal from the termite mound, blocks were cleaned of mud and debris and evaluated visually to ascertain and quantify damage by termites. efficacy of treatments in terms of protection against termites was evaluated by ml % (wood consumed) as reported by previous researchers (shukla 1977; kumar & dev 1993) (table 2). statistical analysis data generated were analysed statistically using ibm spss software version 25. results and discussion the weight gains of 9.25% by p. roxburghii after cm can be attributed to crosslinking of ca in the cell wall by substitution of oh groups (table 3). the finding was similar or better than that of sefc et al. (2009), for beech wood using 6.9% ca + 6.5% shp for different curing temperatures and times. the weight loss observed due to ht was caused by several factors such as species, rate samani et al. new zealand journal of forestry science (2021) 51:15 page 4 of dehydration reactions, degradation of amorphous polysaccharides, wood density, treatment condition etc. (maclean 1951; 1953; stamm 1956; hillis 1975; esteves et al. 2007; candelier et al. 2016). in softwoods the flow of the liquid between adjacent tracheids is through the pits between the cell walls in a transverse direction. during the curing in case of cm followed by thermo-condensation or during ht, the pits become aspirated, reducing the permeability. pit aspiration can be attributed to formation of hydrogen bonds between adjacent cellulose chains and presence of resinous compounds. the cross linking of ca with wood polymers and subsequent thermo condensation might have increased the aspiration resulting in lower wpg values. the molecular volume of the substituted group along with a degree of substitution of the hydroxyl groups influences mee of modified wood thus reducing de-aspiration by cm as compared to ht (chang & chang 2002). this hinders water to create a hydrogen bond with oh groups of wood hence increasing wee as observed (hom et al. 2020b; lehringer et al. 2009). ml after ht depends on wood species, temperature of modification, treatment time and heating medium (esteves & pereira 2009). esteves et al. (2007) had reported higher ml values of hardwood than softwood after identical treatment. the mass loss values obtained for p. roxburghii in the present study were in conformity with previous works by these authors. lower ml values of softwoods after ht can be due to the presence of lignin, which is more abundant than in hardwoods, and is less susceptible to degradation at lower temperatures. it can be observed from table 3 and figure 1 that the impact of cm and ht on p. roxburghii in terms of mass loss (%) termite resistance class 0-6 very resistant (class i) 7-16 resistant (class ii) 17-30 moderately resistant (class iii) 31-50 51 and above poorly resistant (class iv+) perishable (class iv-) table 2: classification of wood into various resistance classes after tmt treatment weight gain radial swelling tangential swelling volumetric swelling wee (%) t1 (cm: ca+shp) 9.25 (+ 0.58) 3.75 (+0.08)a 5.44 (+0.15)m 9.76 (+0.45)x 21.13 (+ 1.02) t2 (ht: heat) -0.42 (+ 0.22) 3.68 (+0.17)a 5.97 (+0.02)n 10.32 (+0.54)y 13.23 (+ 0.76) t3 (c: control) 0.00 4.67 (+0.10)b 6.98 (+0.07)o 12.94 (+0.22)z table 3: mean weight percentage gain/loss due to treatment and swelling in water immersion test of pinus roxburghii (standard errors are given in parenthesis) (different letters denote different homogeneous groups as per duncan analysis. duncan analyses were performed separately for all above mentioned parameters and reported accordingly. a,b denote different groups for radial swelling; m,n,o denote different groups for tangential swelling and x,y,z denote different groups for volumetric swelling). swelling was non-significant but both cm and ht have significantly reduced swelling as compared to nonmodified controls. it was observed that the swelling coefficient was lowest in the radial direction for ht wood, but for tangential and volumetric samples, cm modified wood resulted in lowest swelling. the swelling for non-modified samples ranged between 4.18-13.36 while the swelling observed for t3 was between 4.67 12.93, which was significantly higher than t1 and t2 with swelling values ranging between 3.75-9.76 and 3.68-10.32. the swelling obtained due to cm was similar to values reported for radiata pine (pinus radiata d.don) modified using acetic anhydride (hom et al. 2020b). the result obtained by ht was lower than the values reported by tripathi and bhoru (2014) for p. roxburghii subjected to varying temperatures for thermal modification. it is observed that the impact of ht and cm were similar on ase in the radial direction (table 4).in the tangential direction, cm samples performed better than ht samples and had significantly higher ase. the ase observed in cm samples was higher in the tangential direction than in the radial section, although the values were nearly comparable. while all the reasons for such behaviour could not be elucidated, it is likely attributed to the anatomical structures of different sets where presence of resin and earlywood and latewood percentage could have affected the mode of treatments and returned such unusual patterns for t2. the volumetric ase observed for cm pinus roxburghii was significantly higher than for ht samples. the ase observed for t1 ranged between 20-23% which was similar to the ase for scots pine (pinus sylvestris l.) treated with a combination of 18% activated glucose and 8.2% ca (guo et al. 2019). for ht samples, the ase observed in this study was significantly better than previous results obtained by tripathi and bhoru (2014) for pinus roxburghii. soil block assay enhanced biological durability after cm at higher weight percentage gain was reported by despot et al. (2008) and essoua essoua et al. (2016). similarly, at a wpg of 9.25, chemically modified p. roxburghii samples (ml% 3.47) exhibited 14 times better performance (figure 2) than the untreated samples when exposed to op (ml% 51.33). the cm set showed 15 times less mass loss against tv (ml% 3.15) in comparison to the untreated controls (ml% 48.35). the primary reason why the use of p. roxburghii is limited is its highly perishable nature. enhanced biological durability against decay fungi using chemical modification may render it fit for several application such as cladding, decking and facade elements as the use of p. roxburghii can improve the aesthetics. the mass loss samani et al. new zealand journal of forestry science (2021) 51:15 page 5 figure 1: swelling co-efficient (s) of pinus roxburghii samples in different directions after different treatments (see table 1 for treatment details; se values are mentioned after +). treatment ase radial ase tangential ase volumetric t1 (cm: ca+shp) 19.77 (+ 1.12) 22.06 (+ 0.81) 23.05 (+ 0.67) t2 (ht: heat) 21.19 (+ 0.93) 14.54 (+ 1.06) 18.37 (+ 1.4) table 4: anti-swelling efficiency of modified pinus roxburghii wood (standard errors of the means are given in parenthesis) results from all three treatments showed that oligoporus placenta was marginally more severe on p. roxburghii than trametes versicolor although apart from t2, the mass loss values were highly comparable between the two fungi. the reduction in ml% exhibited in the present study is in accordance with previous findings by larnøy et al. (2018) with ca and sorbitol modified pine wood. beck (2020) reported leaching of some substrate from ca modified wood with sorbitol which might have resulted in some mass loss as well as reducing the virulence of test fungi within the culture jars. additionally, enhanced durability after cm with ca can be attributed to the cross linking of ca on -oh groups of cell wall components of wood which made it unrecognisable as a food source to the decaying fungi. previous studies have highlighted the importance of reduction in the moisture content in wood cell wall to be the most effective and critical factor in fungal decay resistance mechanism (thybring 2013) as it reduces the diffusion of oxidative fungal metabolites (ringman et al. 2019) which might also be the case here as the wee of cm set was remarkably higher than the ht and control sets. another theory highlighted by zelinka et al. (2016) was that after modification of wood, micro pores in the wood cell wall may not remain accessible as they often become closed, preventing or reducing enzymes entering the cell wall. heat treatment results in degradation of wood polymers like hemicellulose. subsequent release of acetic acid may inflict some chemical changes in wood. the primary reason for increased bd against decay fungi was reported to be the formation of several molecules after ht, such as furfural which reticulates with lignin in wood and becomes unrecognisable by the decaying fungi (weiland & guyonnet 2003). lesser accessibility of free -oh groups after ht and reduced equilibrium moisture content (emc) also make it perform better against bio-deterioration. in the present study, ht reduced biodeterioration by fungi but not to a greater extent. similar findings with ht wood were reported by (kamdem et al. 2002) where heat treated samples of maritime pine (pinus pinaster aiton), poplar (populus spp.) and norway spruce (picea abies) showed improved biological durability but not up to the expected level for use with ground contact. an 8% reduction in mass loss from exposure to the white rot fungi trametes versicolor was observed in the ht set and the difference observed was statistically significant in comparison to the controls. for the brown rot fungi oligoporus placenta, the ht set again performed significantly better with a mass loss of 33.85%, which was approximately 18% less than the value observed in the control set (51.33%). kim et al. (1998) reported the positive impact of treatment time and temperature increment on bd of treated wood although kamdem et al. (2002) concluded that mechanical strength reduction at higher temperatures was more. therefore, for a highly perishable species like p. roxburghii, heat treatment at such low temperature levels might not be the best without additional treatments for wood protection but can be ideal for low durability species where direct exposure to high humidity can be avoided. the utility of chemical modification by subsequent curing with heat treatment was reported by salman et al. (2017) where impregnation of aqueous solutions of vinylic monomers before thermal modification improved the termite durability of heat treated wood. in the same study, impregnation of a 10% aqueous solution of maleic anhydride/polyglycerol adduct or polyglycerol methacrylate followed by thermal modification at 220°c led to significantly improved durability against termites, where ht controls at 220°c were strongly attacked. in the present study chemical modification performed better and exhibited about six times less mass loss than the other two treatments (figure 3). after the termite mound test, the durability class of the chemically modified set was elevated to class ii whereas the other two sets remained at class iv. similar findings for ca treated wood have been previously reported (lee et al. 2020; treu et al. 2020). treu et al. (2020) attributed potential leaching of the chemicals (ca+sorbitol) and subsequent changes in wood-water relations after modification as samani et al. new zealand journal of forestry science (2021) 51:15 page 6 figure 2: mass loss of pinus roxburghii samples subjected to different treatments (see table 1 for treatment details) following a soil block assay where they were exposed to oligoporus placenta and trametes versicolor the primary reason for such results. another possible explanation of such enhanced biological durability against termites is the decreased level of protozoa in certain termite species after being exposed to acetylated wood which leads to starvation after one or two weeks as highlighted by duarte et al. (2017) and may be the reason justifying the higher mortality of termites observed by treu et al. (2020). salman et al. (2016, 2017) predicted the utility of chemical modification followed by heat curing for its remarkable properties against biodeterioration by termites and the present findings substantiate their claim. they attributed the radical formations after thermal degradations as the primary cause of improved biological durability after chemical modification, whereas the modification of wood cell wall polymers with citric acid could make it inadequate as a nutrition source for insects. however, the authors also highlighted that the synergistic effect of cm and ht in combination are better at higher temperatures. this must explain the poor biological durability shown by the heat treated set of p. roxburghii where surprisingly ml% observed (67.06) was more than in the untreated control sets (63.13). not all reasons for this result can be elucidated but it is reasonable to conclude that heat treatment at low temperatures for a short duration fails to improve biological durability of p. roxburghii in ground contact. nunes et al. (2004) also reported insignificant improvement in biological durability of heat treated wood against termites. citric acid chemical modification being an environment friendly process improved the dimensional stability as well as resistance against biodegradation of pinus roxburghii. heat treatment was explored in a laboratory oven as a cost-effective alternative and also showed improvement in reducing the swelling and resistance against fungi but failed to impart any resistance against termites. while the results for both treatments were promising, further investigations with varying concentrations of chemicals used and thermocondensation temperature and medium are required. research for eco-friendly chemical modification processes holds promise for subsequent adaptation at industrial scale. however, further trails in varying climates and other decaying agents must be conducted to elucidate on the exact mechanism and to predict exact efficacy in service. conclusion citric acid chemical modification being an environment friendly process improved the dimensional stability as well as resistance against biodegradation of pinus roxburghii. heat treatment was explored in a laboratory oven as a cost-effective alternative and also showed improvement in reducing the swelling and resistance against fungi but failed to impart any resistance against termites. while the results for both treatments were promising, further investigations with varying concentrations of chemicals used and thermocondensation temperature and medium are required. research for eco-friendly chemical modification processes holds promise for subsequent adaptation at industrial scale. however, further trails in varying climates and other decaying agents must be conducted to elucidate on the exact mechanism and to predict exact efficacy in service. list of abbreviations ca citric acid shp sodium hypophosphite cm chemical modification /chemically modified ht heat treatment /heat treated ml mass loss ase anti swelling efficiency bd biological durability ds dimensional stability sba soil block bioassay op oligoporus placenta tv trametes versicolor nd non-durable tmt termite mound test od oven dried wpg weight percent gain wee water excluding efficiency emc equilibrium moisture content samani et al. new zealand journal of forestry science (2021) 51:15 page 7 figure 3: percent mass loss observed after the termite mound test of pinus roxburghii subjected to different treatments (see table 1 for treatment details) acknowledgements the authors are thankful to the director, fri, india for providing necessary support and icfre, dehradun, india for proving funding to conduct the research. the authors also duly acknowledge the contribution and help received from the late subhrajit ghosh in the preparation of this manuscript. conflicts of interest the authors hereby declare that there is no conflict of interest associated with this manuscript to the best of their knowledge. authors' contributions as: funding, conceptualisation and laboratory resources; sg: draft preparation, data analysis, satistical analysis, correspondence, editing; skh: conceptualisation, execution, draft preparation, editing. references anon. (2008). methods of field testing of preservatives in wood. is 4833. new delhi, india: bureau of indian standards. anon. (1993). methods of laboratory testing of wood preservatives against fungi and borers (powder post beetles) part 1 determination of threshold values of wood preservatives against fungi. is 4873 (part 1). new delhi, india: bureau of indian standards. beck, g. (2020). leachability and decay resistance of wood polyesterified with sorbitol and citric acid. forests, 11: 650. https://doi.org/10.3390/ f11060650 brischke, c., bayerbach, r., & rapp, a.o. (2006). decayinfluencing factors: a basis for service life prediction of wood and wood-based products, wood material science & engineering, 1(3-4), 91-107. https://doi. org/10.1080/17480270601019658 candelier, k., thevenon, m., petrissans, a., dumarcay, s., gerardin, p., & petrissans, m. (2016). control of wood thermal treatment and its effects on decay resistance: a review. annals of forest science, 73, 571-583. https://doi.org/10.1007/s13595-0160541-x chang, h.t.,& chang, s.t. (2002). moisture excluding efficiency and dimensional stability of wood improved by acylation. bioresource technology, 85(2), 201-204. https://doi.org/10.1016/s09608524(02)00085-8 despot, r., hasan, m., jug, m., & sefc, b. (2008). biological durability of wood modified by citric acid. drvna industrija, 59(2), 55-59. dirol, d., & guyonnet, r. (1993). the improvement of wood durability by rectification process. international research group on wood preservation [irg/wp 93-40015]. stockholm, sweden: irg secretariat. donath s., militz h., & mai c. (2004). wood modification with alcoxysilanes. wood science and technology, 38(7), 555-566. https://doi.org/10.1007/s00226004-0257-1 duarte, s., duarte, m., borges, p.a.v., & nunes, l. (2016). dietary-driven variation effects on the symbiotic flagellate protist communities of the subterranean termite reticulitermes grassei clément. journal of applied entomology, 141, 300-307. https://doi. org/10.1111/jen.12331 essouaessoua, g.g., blanchet, p., landry, v., & beauregard, r. (2016). pine wood treated with a citric acid and glycerol mixture: biomaterial performance improved by a bio-byproduct. bioresources, 11(2), 3049-3072. https://doi.org/10.15376/ biores.11.2.3049-3072 esteves, b.m., domingos, i.j., & pereira, h.m. (2007). improvement of technological quality of eucalyptus wood by heat treatment in air at 170°c-200°c. forest products journal, 57(1/2), 47-52. esteves, b.m., & pereira, h.m. (2009). wood modification by heat treatment: a review. bioresources, 4(1), 370-404. https://doi.org/10.15376/ biores.4.1.esteves feng, x., xiao, z., sui, s., wang, q., & xie, y. (2014). esterification of wood with citric acid: the catalytic effects of sodium hypophosphite (shp). holzforschung, 68(4), 427-433. https://doi. org/10.1515/hf-2013-0122 ganguly, s. (2018). conserving wood biodiversity with the help of wood science and technology. international research journal of environmental sciences, 7(11), 42-44. ganguly, s. & tripathi, s. (2018). study on effect of microwave treatment on wood permeability and preservative retention in imported timber. journal of agroecology and natural resource management, 5(1), 34-40. ganguly, s., tripathi, s., tiwari, p., sumi, a., & kanyal, r. (2020). screening of azadirachta indica seed oil against sap-stain and mould fungi in imported tectona grandis and southern yellow pine wood through fumigation. journal of tropical forest science, 32(2), 114-124. https://doi. org/10.26525/jtfs32.2.114 ganguly, s., hom, s.k., tripathi, s., ghosh, s., kanyal, r., & samani, a. (2021). quantitative evaluation of microwave irradiation on short-rotation plantation wood species. maderas ciencia y tecnología. 23(25), 1-14. https://doi.org/10.4067/s0718221x2021000100425 guo, w., xiao, z., wentzel, m., emmerich, l., xie, y., & militz, h. (2019). modification of scots pine with activated glucose and citric acid: physical and samani et al. new zealand journal of forestry science (2021) 51:15 page 8 https://doi.org/10.3390/f11060650 https://doi.org/10.3390/f11060650 https://doi.org/10.1080/17480270601019658 https://doi.org/10.1080/17480270601019658 https://doi.org/10.1007/s13595-016-0541-x https://doi.org/10.1007/s13595-016-0541-x https://doi.org/10.1016/s0960-8524(02)00085-8 https://doi.org/10.1016/s0960-8524(02)00085-8 https://doi.org/10.1007/s00226-004-0257-1 https://doi.org/10.1007/s00226-004-0257-1 https://doi.org/10.1111/jen.12331 https://doi.org/10.1111/jen.12331 https://doi.org/10.15376/biores.11.2.3049-3072 https://doi.org/10.15376/biores.11.2.3049-3072 https://doi.org/10.15376/biores.4.1.esteves https://doi.org/10.15376/biores.4.1.esteves https://doi.org/10.1515/hf-2013-0122 https://doi.org/10.1515/hf-2013-0122 https://doi.org/10.26525/jtfs32.2.114 https://doi.org/10.26525/jtfs32.2.114 https://doi.org/10.4067/s0718-221x2021000100425 https://doi.org/10.4067/s0718-221x2021000100425 mechanical properties, bioresources, 14(2), 34453458. hansmann c., weichslberger g., & gindl w. (2005). a two-step modification treatment of solid wood by bulk modification and surface treatment. wood science and technology, 39(6), 502-511. https:// doi.org/10.1007/s00226-005-0002-4 hasan, m., despot, r., katovic, d. bischof vukusic, s., bonger, a. & jambrekovic, v. (2007). citric acidpromising agent for increased biological durability of wood. 3rd european conference on wood modification, cardiff uk, october 15-16, pp. 275278. hasan, m., despot, r., trajkovic, j. & sefe, b. (2006). role of modification processes at increasing biological durability of wood. international conference european union-challenges and perspectives for the wood-processing industry, zagreb, october 13. proceedings, pp. 153-158. hasan, m., despot, r., šefc, b. &ištok, i., sedlar, t. & lacic, r. (2012). optimisation of modification of beech wood by citric acid. in: d. jones, h. militz, m. petrič, f. pohleven, m. humar and m. pavlič (eds.), sixth european conference on wood modification, ljubljana, slovenia. ecwms6. hillis, w.e. (1975). the role of wood characteristics in high temperature drying. journal of the institute of wood science, 7(2), 60-67. hom s.k., ganguly s., bhoru y.u., & samani a. (2020b). effect of chemical modification on dimensional stability of pinus radiata d. don using acetic anhydride. journal of forest science, 66, 208-217. https://doi.org/10.17221/13/2020-jfs hom, s.k., ganguly, s., samani, a. & tripathi, s. (2020a). improvement in fire retardancy with doublestep chemical modification on pinus radiata d. don using dimethyl methylphosphonate with propylene oxide and maleic anhydride, international wood products journal, 11(3), 138-145. https://doi.org/ 10.1080/20426445.2020.1765624 humar, m., kržišnik, d., lesar, b., thaler, n., ugovšek, a., zupančič, k., & žlahtič, m. (2017). thermal modification of wax-impregnated wood to enhance its physical, mechanical, and biological properties. holzforschung, 71(1), 57-64. https:// doi.org/10.1515/hf-2016-0063 isaksson, t., thelandersson, s., brischke, c., & jermer, j. (2011). service life of wood in outdoor above ground applications-engineering design guideline [report tvbk-3060]: lund, sweden: division of structural engineering, lund university. accessed 20 august 2020 from: http://www.kstr.lth.se/ fileadmin/kstr/pdf_files/guideline/tvbk3067. pdf kamdem, d.p., pizzi, a., & jermannaud, a. (2002). durability of heat-treated wood. holz als rohund werkstoff, 60(1), 1-6. https://doi.org/10.1007/ s00107-001-0261-1 kim, g., yun, k., & kim, j. (1998). effect of heat treatment on the decay resistance and the bending properties of radiata pine sapwood, material und organismen, 32(2), 101-108. kržišnik, d., lesar, b., thaler, n., & humar, m. (2018). influence of natural and artificial weathering on the colour change of different wood and woodbased materials. forests, 9(8), 488. https://doi. org/10.3390/f9080488 kumar, s. & dev, i. (1993). wood preservation in india. [icfre publication-28]. dehradun, india: indian council of forestry research and education. larnøy, e., karaca, a., gobakken., l.r., & hill, c.a.s. (2018). polyesterification of wood using sorbitol and citric acid under aqueous conditions, international wood products journal, 9(2), 66-73. https://doi.org/10.1 080/20426445.2018.1475918 lee, s.h., md tahir, p., lum, w.c., tan, l. p., bawon, p., park, b. d., osman al edrus, s.s., & abdullah, u.h. (2020). a review on citric acid as green modifying agent and binder for wood. polymers, 12(8): 1692. https://doi.org/10.3390/polym12081692 lehringer c., schwarze f.w.m.r., & militz h. (2009): a review on promising approaches for liquid permeability improvement in softwoods. wood and fiber science, 4(4), 373-385. maclean, j.d. (1951). rate of disintegration of wood under different heating conditions. american wood preservers’ association proceedings, 47, 155-168. maclean, j.d. (1953). effect of steaming on the strength of wood. american wood-preservers’ association proceedings, 49, 88-11. mcsweeny, j.d., rowell, r.m. & min, s.h. (2006). effect of citric acid modification of aspen wood on sorption of copper ion. journal of natural fibers, 3(1), 4358. https://doi.org/10.1300/j395v03n01_05 meyer-veltrup, l., brischke, c., alfredsen, g. humar, m., flæte, p. o., isaksson, t.,brelid, p. l., westin, m., & jermer, j.(2017). the combined effect of wetting ability and durability on outdoor performance of wood: development and verification of a new prediction approach. wood science and technology, 51(3), 615-637. https://doi.org/10.1007/s00226017-0893-x militz, h. (2002). heat treatment of wood: european processes and their background. [irg/wp 0240241]. cardiff, wales: international research group on wood preservation. militz, h. (2015). wood modification in europe in the year 2015: a success story? eighth european conference on wood modification, helsinki, finland samani et al. new zealand journal of forestry science (2021) 51:15 page 9 https://doi.org/10.1007/s00226-005-0002-4 https://doi.org/10.1007/s00226-005-0002-4 https://doi.org/10.17221/13/2020-jfs https://doi.org/10.1080/20426445.2020.1765624 https://doi.org/10.1080/20426445.2020.1765624 https://doi.org/10.1515/hf-2016-0063 https://doi.org/10.1515/hf-2016-0063 http://www.kstr.lth.se/fileadmin/kstr/pdf_files/guideline/tvbk3067.pdf http://www.kstr.lth.se/fileadmin/kstr/pdf_files/guideline/tvbk3067.pdf http://www.kstr.lth.se/fileadmin/kstr/pdf_files/guideline/tvbk3067.pdf https://doi.org/10.1007/s00107-001-0261-1 https://doi.org/10.1007/s00107-001-0261-1 https://doi.org/10.1080/20426445.2018.1475918 https://doi.org/10.1080/20426445.2018.1475918 https://doi.org/10.3390/polym12081692 https://doi.org/10.1300/j395v03n01_05 https://doi.org/10.1007/s00226-017-0893-x https://doi.org/10.1007/s00226-017-0893-x samani et al. new zealand journal of forestry science (2021) 51:15 page 10 nunes, l., nobre, t., & rapp, a. (2004). thermally modified wood in choice tests with subterranean termites. reinbeck, ia, usa: cost e37. ringman, r., beck, g., & pilgård, a. (2019). the importance of moisture for brown rot degradation of modified wood: a critical discussion. forests, 10(6): 522. https://doi.org/10.3390/f10060522 rowell r.m. (1983). chemical modification of wood. forest product abstracts, 6, 363-382. salman, s., thévenon, m. f., pétrissans, a., dumarçay, s., candelier, k., & gérardin, p. (2017). improvement of the durability of heat-treated wood against termites. maderas. ciencia y tecnología, 19(3), 317-328. https://doi.org/10.4067/s0718221x2017005000027 salman, s., petrissans, a., thevenon, m.f., dumarçay, s., & gerardin, p. (2016). decay and termite resistance of pine blocks impregnated with different additives and subjected to heat treatment. european journal of wood and wood products,74(1), 37-42. https:// doi.org/10.1007/s00107-015-0972-3 samani a., ganguly s., kanyal r., & tripathi s. (2019). effect of microwave pre-treatment on preservative retention and treatability of melia composita wood. journal of forest science, 65(10), 391-396. https:// doi.org/10.17221/39/2019-jfs samani, a., hom, s.k., bhoru, y.u. & ganguly, s. (2020). dimensional stability of wood modified by citric acid. indian forester, 146(5), 455-458. sandberg, d., kutnar, a., & mantanis, g. (2017). wood modification technologies a review. iforest, 10(6), 895-908. https://doi.org/10.3832/ifor2380-010 sefc, b., trajkovi, j., hasan, m., katovi, d., vuku, s.b. & martina, f. (2009). dimensional stability of wood modified by citric acid using different catalysts. drvna industrija, 60(1), 23-26. shukla, k.s. (1977). preliminary investigations on the termite resistance of wood wool boards. journal of the timber development association (india), 23(1), 21-23. singh, t., & singh, a.p. (2012). a review on natural products as wood protectant. wood science and technology, 46, 851-870. https://doi.org/10.1007/ s00226-011-0448-5 stamm, a.j. (1956). thermal degradation of wood and cellulose. industrial and engineering chemistry, 48(3), 413-417. https://doi.org/10.1021/ ie51398a022 thybring, e.e. (2013). the decay resistance of modified wood influenced by moisture exclusion and swelling reduction. international biodeterioration & biodegradation, 82, 87-95. https://doi. org/10.1016/j.ibiod.2013.02.004 treu, a., nunes, l., & larnøy, e. (2020). macrobiological degradation of esterified wood with sorbitol and citric acid. forests, 11(7), 776-786. https://doi. org/10.3390/f11070776 tripathi, s. & bhoru, y.u. (2014). changes in dimensional stability of pinus roxburghii sarg. and mangifera indica l. woods after retification. international wood products journal, 5(4), 179-185. https://doi. org/10.1179/2042645314y.0000000062 weiland, j., & guyonnet, r. (2003). study of chemical modifications and fungi degradation of thermally modified wood using drift spectroscopy. holz als rohund werkstoff, 61, 216-220. https://doi. org/10.1007/s00107-003-0364-y zelinka, s.l., ringman, r., pilgård, a., thybring, e.e., jakes, j.e., & richter, k. (2016). the role of chemical transport in the brown-rot decay resistance of modified wood. international wood products journal, 7(2), 66-70. https://doi.org/10.1080/204 26445.2016.1161867 https://doi.org/10.3390/f10060522 https://doi.org/10.4067/s0718-221x2017005000027 https://doi.org/10.4067/s0718-221x2017005000027 https://doi.org/10.17221/39/2019-jfs https://doi.org/10.17221/39/2019-jfs https://doi.org/10.3832/ifor2380-010 https://doi.org/10.1007/s00226-011-0448-5 https://doi.org/10.1007/s00226-011-0448-5 https://doi.org/10.1021/ie51398a022 https://doi.org/10.1021/ie51398a022 https://doi.org/10.1016/j.ibiod.2013.02.004 https://doi.org/10.1016/j.ibiod.2013.02.004 https://doi.org/10.3390/f11070776 https://doi.org/10.3390/f11070776 https://doi.org/10.1179/2042645314y.0000000062 https://doi.org/10.1179/2042645314y.0000000062 https://doi.org/10.1007/s00107-003-0364-y https://doi.org/10.1007/s00107-003-0364-y https://doi.org/10.1080/20426445.2016.1161867 https://doi.org/10.1080/20426445.2016.1161867 flight activity of woodand bark-boring insects at new zealand ports stephen m. pawson1, 2*, jessica. l. kerr1, chanatda somchit3, carl w. wardhaugh3 1 scion (new zealand forest research institute), 10 kyle street, riccarton, christchurch, new zealand 2. current address: school of forestry, university of canterbury, christchurch, new zealand 3. scion (new zealand forest research institute), 49 sala street, rotorua, new zealand *corresponding author: steve.pawson@canterbury.ac.nz (received for publication 2 october 2020; accepted in revised form 15 december 2020) abstract background: barkand wood-boring forest insects spread via international trade. surveys frequently target new arrivals to mitigate establishment. alternatively, monitoring pest activity in exporting countries can inform arrival and establishment risk. methods: we report >3 years data from daily sampling of barkand wood-boring insects that are associated with recently felled pinus radiata d.don at five new zealand ports. results: average catch differed between ports and months with arhopalus ferus (mulsant), hylurgus ligniperda f., and hylastes ater (paykull) comprising 99.6% of the total catch. arhopalus ferus was absent during winter with hylastes ater and hylurgus ligniperda activity between june and august representing 3.5 and 3.7% of total catch, respectively. maximum temperature and wind speed influenced flight activity of all three species but not universally across all ports. flight activity transitioned to a nonlinear pattern above 20°c. arhopalus ferus has a unimodal flight risk period between late-september and late-april. hylastes ater was also unimodal except in dunedin where it was bimodal like hylurgus ligniperda was in all regions with spring and midto late-summer activity periods. although hylastes ater was observed during winter, the probability of a flight event during winter was between 0 and 0.02 per week. hylurgus ligniperda flight probability was zero in dunedin and low at all other ports from may to august. conclusions: modelling seasonal changes in flight probability can inform risk-based phytosanitary measures. we demonstrate the utility of maximum temperature and seasonality as a predictor of wood commodity infestation risk. such predictors allow national plant protection organisations to develop standards that protect the post-treatment phytosanitary security of individual consignments. new zealand journal of forestry science pawson et al. new zealand journal of forestry science (2020) 50:14 https://doi.org/10.33494/nzjfs502020x132x e-issn: 1179-5395 published on-line: 23/12/2020 © the author(s). 2020 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access wood packaging material (haack et al. 2014). the cryptic nature of barkand wood-boring pests make them difficult to detect cost-effectively at the border. surveillance for the purposes of detecting recent incursions in the importing country can reduce the number of successful establishments because it can facilitate eradication, i.e., early detection increases the probability of a successful eradication programme (tobin et al. 2014). there are a number of tools and strategies used in surveillance programmes; however, trapping programmes that use introduction barkand wood-boring insects have spread widely as a result of international trade (brockerhoff et al. 2006; haack 2006). protected within solid wood products, or within live plants (liebhold et al. 2012), they move with commodities and are likely to continue to colonise new areas as trade volumes increase (liebhold et al. 2017). these risks have been reduced via the successful implementation of ispm 15 that has resulted in a reduction in the movement of wood boring insects in keywords: biosecurity, ecological risks, export logs, forest entomology, phytosanitary, quarantine, regulated pests, trade http://creativecommons.org/licenses/by/4.0/), pawson et al. new zealand journal of forestry science (2020) 50:14 page 2 a synthetic chemical lure (or combination of lures) are commonly used to survey for barkand wood-borer activity at ports and their immediate surrounds due to their high efficacy and relatively low cost (allison et al. 2018; chase et al. 2018; flaherty et al. 2018; rassati et al. 2014). such surveys characterise the phenology of the native fauna and existing exotic species (brockerhoff et al. 2006; wylie et al. 2008) and has reported range expansion of species both within and outside their native biogeographic range (rassati et al. 2018). such surveys have also documented new establishments of exotic species (rassati, faccoli, petrucco toffolo et al. 2015). import trade volumes were positively correlated with the number of alien wood-boring beetle species detected during a surveillance survey at italian ports (rassati, faccoli, petrucco toffolo et al. 2015), and the volume (or value) of imported goods was determined to be a strong predictor of barkand wood-boring beetle interceptions (haack 2001; huang et al. 2012). these findings suggest that import data can be used to direct limited surveillance resources to commodities and sites that represent the greatest likelihood of interceptions and, hence, the greatest risk of a new alien species incursion. wood commodity trade pathways, e.g. logs, are complex systems that can be widely variable over time (piel et al. 2008). although logs are known to present specific phytosanitary risks (usda 1992), little is known regarding changes in the phytosanitary risk of consignments from different regions and at different times of the year. similarly, factors within the exporting country that can influence the phytosanitary risk profile of individual consignments can be temporally and spatially variable and are poorly understood. surveillance for new exotic organisms that arrive at ports (or transitional facilities) around the world is undertaken in some countries to inform management actions that are intended to prevent the establishment of new species. however, surveillance to monitor populations of resident pest species of wood commodities during the period between harvest and export is less common; however, postharvest phytosanitary risks are a significant component of any pest risk assessment conducted by countries importing wood commodities. by comparison, there is extensive monitoring of pest populations of horticultural crops in areas of production that is a critical component of risk profiling to inform the implementation of riskbased, or systems, approaches to pest management and phytosanitary measures (moore et al. 2016; walker et al. 2017). in a forestry context, pre-export surveys have the potential to identify geographic and temporal differences in the potential phytosanitary risk profile of specific export wood consignments. pre-export surveys are particularly important when considering systems-based approaches to the management of phytosanitary risk. systems approaches use multiple measures applied at different points (and times) within the supply chain that in combination may reduce or eliminate the need for single end-point phytosanitary treatments, such as fumigation, when the phytosanitary risk is shown to be negligible (allen et al. 2017). a systems approach may incorporate risk-based approaches to phytosanitary measures, e.g., a pest free place of production ((pfpp), ippc 1999) or an area of low pest prevalence ((alpp), ippc 2005), that may apply to one or more particular geographic regions or at certain times of the year, such as winter. however, phytosanitary measures, including systems approaches, must have appropriate procedures to maintain the phytosanitary security of a consignment and, hence, the validity of the phytosanitary certification issued by the national plant protection organisation (ippc 2016). supporting evidence to underpin the implementation of an alpp as part of a systems approach to phytosanitary measures requires an understanding of the changing infestation risk profile that may occur along the supply chain for individual wood consignments. this is important because different mitigation measures are required at different points along the supply chain. for instance, measures that minimise the infestation of freshly felled trees in a plantation (debarking, rapid extraction, etc.) are very different to those employed on a port (application of a phytosanitary treatment, appropriate storage, etc.). understanding the population dynamics of forest insects at major wood exporting ports provides information to support alternative riskbased phytosanitary treatments as it is one place where individual export consignments can have long residence times as they await shipment. in 2017, new zealand exported 19.2 million m3 of roundwood logs, primarily pinus radiata d.don, from plantation forests (mpi 2019), which is equivalent to 14.4% of the 2016 world trade in industrial roundwood (fao 2016). two thirds of new zealand’s log exports passed through the five ports we surveyed: whangarei, tauranga, napier, nelson and dunedin, of which tauranga and whangarei are the two largest log-exporting ports (mpi & statistics nz, unpublished data). we report here the results from three years of continuous trapping to assess the flight activity of seven barkor wood-boring species (arhopalus ferus (mulsant), hylastes ater (paykull), hylurgus ligniperda f., mitrastethus baridioides redtenbacher, pachycotes peregrinus (chapuis), prionoplus reticularis white, and sirex noctilio (f.)) at the five major logand timberexporting ports in new zealand. these species are the most common barkand wood-boring species that are associated with recently felled pinus radiata in new zealand. actual abundance in managed plantation forest stands ranges from very abundant (hylurgus ligniperda), to comparatively rare (s. noctilio) (pawson, unpublished data). these data can be used to identify potential areas and/or temporal periods when the activity of these key pest species is low or non-existent (i.e. no phytosanitary risk) within the port environment. methods three flight intercept traps were deployed at each of five ports in new zealand (additional file: fig. a1). ports were sampled at the following cities, from north to south: whangarei, tauranga, napier, nelson, and dunedin. traps were placed in operational log storage yards and separated by at least 50 m. because of safety requirements, trap placement was dictated by the availability of zones within the log yards that were not subject to traffic and log yard operations (additional file: fig. a2). traps at each port were established at different times ranging from 25 july to the 21 november 2013 and the traps were removed from all sites on 28 september 2016. although 13 of the 15 traps were operated continuously at the same location for the duration of the study, two of the traps at port nelson had to be moved in january and may 2016, respectively, because the log yard operations were expanded (additional file: table a1). insects were sampled using black cross-vane flight intercept traps (kerr et al. 2017). traps were made from 600 × 210 mm mulflutetm polypropylene sheets (mulford international, christchurch, new zealand), topped with a mulflutetm (210 × 210 mm) rain cover, and a black funnel (216 mm diameter) that directed catch into our collection system (figure 1). traps were baited with separate 150 ml dispensers (450 × 50 mm, 150 μm polyethylene tubing (accord plastics, masterton, nz) with felt strips) of ethanol (nuplex specialties nz, mt. wellington, new zealand) and pinechem 500 (lawter (nz), mt. maunganui, new zealand. (note: this is a discontinued product that can be made on special request to the manufacturer)) a mixture of alphaand beta-pinene that was shown by our mass spectrometry analysis to consist of an average concentration of 71% alpha-pinene and 18% beta-pinene and other minor monoterpenes. the release rates of ethanol were ~0.02 g/day and pinechem 500 ~0.76 g/day, respectively. release rates were calculated on the basis of ambient temperature conditions with 36 daily measurements between january and february 2013. these semio-chemicals were known to be the best available attractants at the time of the trapping programme for the species of forest insect targeted by our trapping programme (kerr et al. 2017). flight intercept traps were connected to a cylindrical aluminium housing with an internal motor that rotated a circular carousel of plastic containers that would contain the insects caught by the trap (6286ptcl sq pet jar 58mm 233ml, stowers, new zealand). the motor automatically activated at 24-h intervals and moved the carousel to the next (unused) plastic container thereby separating the trap catches every 24 hours (figure 1). cross-vane traps were hung from a metal y-post at a top height of ~1.5 m with a piece of 50-mm-diameter pvc pipe that connected the trap funnel to the body of the cylindrical aluminium housing. the flight intercept trap and plastic containers were coated with alpha-cypermethrin (ripcord plus, basf new zealand limited, auckland, new zealand) every 4 weeks to kill insects and minimise the potential for insects to move between plastic containers within the carousel. traps were visited for maintenance and to collect samples at fortnightly intervals between 1 april to 30 september and monthly from 1 october to 31 march. trap contents were removed, and the numbers of: a. ferus; hylurgus ligniperda; hylastes ater; prionoplus reticularis; s. noctilio; pachycotes peregrinus; and m. baridioides were recorded, if present. in cases where traps malfunctioned between maintenance periods, the total trap catch within the carousel was pooled to note the total number of insects collected rather than daily catches. in total this affected 1,925 catch days of a total sampling period of 16,974 days. pawson et al. new zealand journal of forestry science (2020) 50:14 page 3 figure 1: flight intercept trap attached to a bespoke separator system for monitoring daily insect activity (top image). internal carousel (bottom left image) and sample containers (bottom right image). raw meteorological data (temperature, relative humidity, wind speed) were used to analyse the effect of local weather on insect flight activity. meteorological stations were present on four of the five ports with data for dunedin (port otago) sourced from musselburgh (-45.90129, 170.5147), the closest meteorological station situated 12 km from the port. the raw meteorological data were summarised with the extensible time series package (r-xts) (ryan & ulrich 2018) using r version 3.4.1 (r development core team 2017) to provide daily (12:00 am to 23:59 pm) and evening (20:00 pm to 23:59 pm) summaries for each variable. monthly export log volumes were recorded for the individual ports. the monthly export log volumes were provided by the new zealand ministry for primary industries (ministry for primary industries and statistics new zealand, unpublished data). the relationship between monthly log volume and trap catch was assessed as the quantity of logs stored is expected to influence the strength of the semio-chemical odour (pinenes and ethanol) plume emanating from the port. hence, it may influence the relative attraction of the site to dispersing beetles. landscape composition, particularly the forest cover surrounding a port, has been shown to influence the abundance of barkand wood-boring beetles in traps at ports (rassati et al. 2018). to assess this we calculated the percent cover of “exotic forest’ and “forest – harvested” classes of the new zealand landcover database version 4 that documents landcover in 2012 (landcare research 2015). cover was calculated in a 5 km radius of each trap and averaged for each of the three traps at a given port. no trend was observed in a graphical analysis of flight activity (trap catch) as a function of landscape forest context surround the ports (additional file: fig. b12) hence forest cover was not included in further analyses. analysis/modelling details the phenological data we collected were analysed with respect to climatic variables (temperature, wind speed, and humidity) and export log volumes to determine key drivers that promote or suppress flight activity. we used these data to make predictions of the probability of flight activity for each species at each port. pawson et al. new zealand journal of forestry science (2020) 50:14 page 4 effect of seasonality, meteorology, and export volume on flight activity the effects of seasonality and meteorology on speciesspecific flight activity (i.e. trap count) were analysed using generalized additive models (gams). total trap catch varied between species (table 1) and flight activity models were only generated for the three most abundant species, a. ferus, hylastes ater, and hylurgus ligniperda. the total trap catch data required to model the flight activity of all other species was insufficient because of the low numbers trapped of the duration of the study. daily trap catch was transformed into catch per 100 trap day and summed across each port at weekly intervals between 10 july 2013 and 28 september 2016. transformation of daily catch data into units of catch per 100 trap days permits communication of low catch rates that would otherwise be expressed as small fractions of an individual during the sampling period. as an example of this transformation, if 100 traps were established and then these traps were checked on a daily basis, then the total observed catch on any given day amongst those traps would reflect the catch per 100 trap days. to analyse species-specific seasonal changes in flight activity, gams included a port effect and a season (weeks of the year; week) effect. the port effect allows for variation in flight activity between ports, whereas weeks of the year represents the seasonal trend. an interaction term for ‘port’ and ‘week’ was used to account for differences in the way that trap counts varied over time at different ports. because export volumes were collinear with other variables, they were eliminated from the model. see additional file for specific details of seasonal gam models and figs. b1-b3 showing a scatter plot of daily catch versus monthly export volume. to analyse the effect of meteorology on flight activity, individual gam models were constructed separately for each species by meteorological variable. the meteorological variables were averaged within the weekly trapping period as follows: average daily maximum temperature (°c); average evening maximum temperature (°c); average daily instantaneous wind speed (m-1s-1); average evening instantaneous wind speed (m-1s-1); average daily humidity (%) and average evening humidity (%). daily averages were used to model port species total catch total trap days arhopalus ferus hylastes ater hylurgus ligniperda mitrastethus baridioides pachycotes peregrinus prionoplus reticularis sirex noctilio whangarei 111 20 632 0 0 1 2 766 2,806 tauranga 166 33 721 0 0 0 0 920 3,258 napier 758 118 1,104 1 0 0 2 1,983 2,833 nelson 628 50 657 0 0 0 3 1,338 3,160 dunedin 111 429 50 0 0 14 0 604 2,991 total catch 1,774 650 3,164 1 0 15 7 5,611 15,048 table 1: trap catch of forest insects captured at five ports in the north and south islands of new zealand between the 10 july 2013 and 28 september 2016. flight activity of hylastes ater and hylurgus ligniperda because both species are predominantly diurnal with evening averages used to model the nocturnal a. ferus (pawson et al. 2017). see additional file for specific details of individual meteorological gam models. daily probability of flight at ports the probability of flight as a function of season was analysed separately for a. ferus, hylastes ater and hylurgus ligniperda at each port. we first transformed the daily capture data into binary presence/absence data by re-coding all trap catches greater than zero, then aggregated this daily binary data on a weekly temporal resolution to increase the number of trapdays on which the probability of flight was estimated. this was performed by summing the daily indicators of beetles in each trap (i.e. either 0 or 1) across weekly (7-day) intervals, and across all traps at each port (i.e. 3 traps), providing a daily probability of flight for each week expressed as a binomial proportion (out of a maximum of 21 trap-days). the minimum possible value of 0 is indicative of no single capture during that week at that port, while the maximum possible value of 21 is indicative of positive trap capture in every trap on every day of that week at a port. across all ports the average week comprised 19.4 trap-days due to the malfunctioning of some traps. our approach reduces environmental stochasticity (e.g. rainy days) or biased data (e.g. carousel malfunctions) compared with the alternative of modelling a maximum of 3 trap-days on a daily resolution. gams with a binomial error and logit link including a first-autoregressive covariance structure were used to analyse the impact of season on the probability of flight of a. ferus, hylastes ater, and hylurgus ligniperda. species-specific models included port and week (weeks of the year), and their interaction. the model parameters were estimated using penalised quasi-likelihood. since the response variable was the presence of beetles in traps out of a maximum of 21 trap-days, a gam with a binomial distribution was used to analyse these data. in a binomial gam with ni > 1, over-dispersion can occur. in the analysis of the binomial gams of daily probability of flight, a binomial response showed evidence of overdispersion. the penalised quasi-likelihood was therefore applied, as per recommendations by wood (2006). in all cases, cyclic cubic regression splines were used. penalties were based on the second-order derivatives and the automatic smoothing parameter selection was obtained through minimisation of a generalised cross validation (gcv) and the unbiased risk estimator (ubre) (wood 2006). graphical tools such as pearson residual plots were used to test for model validation. auto-correlation plots were used to assess temporal autocorrelation. for hylurgus ligniperda, over-dispersion was detected, and the standard errors were corrected using a quasibinomial model. results over the three years of trapping on five ports around new zealand, 5,611 individuals of the seven target forest insects were captured (table 1). over half of these (56.4%) were hylurgus ligniperda, while another 31.6% were a. ferus and 11.6% were hylastes ater. very few sirex, prionoplus and mitrastethus, and no pachycotes were collected. as a result, species-level analyses were restricted to the three most abundant species, a. ferus, hylastes ater, and hylurgus ligniperda that differed in average catch both between ports and months (additional file: fig. a3). arhopalus ferus was captured in higher numbers in napier and nelson, while hylastes ater was most abundant in dunedin. hylurgus ligniperda showed the opposite pattern to hylastes ater and was the most commonly caught species at every port except dunedin, where only 50 individuals were collected over three years of trapping. impact of season on flight activity the flight activity of all three focal species (a. ferus, hylastes ater, and hylurgus ligniperda) varied significantly with season across all of the ports sampled (p < 0.001; additional file: table b1). the monthly export volume was strongly collinear with the ‘port’ variable and was removed from the model (additional file). the estimated flight activity of a. ferus was highest in february across all ports, but very low from early april until early november (figure 2; additional file: fig. b1). no a. ferus were observed in traps between 26 april and 8 october and only 30 (1.7% of total catch) were caught between 1 april and 1 november across all years. the two bark beetle species had low estimated activity during the southern hemisphere winter period (1 june to 30 august). hylastes ater exhibited two significant peaks in estimated flight activity in dunedin where it was most abundant; the first in april and the second in october (figure 2; additional file: fig. b2). hylastes ater had a single spring peak in napier but activity was generally low at other ports with minor autumn activity (figure 2). total actual trap count of hylastes ater across all ports between june and august was only 23 (3.5% of total catch) across all years. hylurgus ligniperda also displayed a bimodal pattern in estimated activity, with the first peak in late summer (februarymarch) and another in late spring (october-november) across most ports, although this second peak was most pronounced in napier (figure 2; additional file: fig. b3). the flight activity of hylurgus ligniperda in dunedin was low throughout the year. the total actual trap count for hylurgus ligniperda across all ports between june and august was only 118 (3.7% of total catch) across all years. impact of weather on flight activity maximum temperature affected the flight activity of a. ferus, hylastes ater and hylurgus ligniperda over the entire trapping period at most ports (additional file: table b2). only a. ferus at tauranga and hylastes ater at whangarei and napier showed no effect of maximum temperature on flight activity (additional file: table b2). the flight activity of all three species showed similar trends across all of the ports, with activity increasing in a nonlinear pattern once maximum temperatures reached about 20°c (fig. 3, additional file: figs. b4-b6). however, instances where maximum temperature was not a significant factor, or where the relationship was pawson et al. new zealand journal of forestry science (2020) 50:14 page 5 linear, occurred when a particular species was captured in relatively low numbers at a particular port. average wind speed affected the activity of a. ferus (f=32.09, p<0.001) and hylastes ater (f=23.72, p<0.001) (additional file: table b3), but did not affect the activity of hylurgus ligniperda over the entire trapping period (f=0.56, p>0.05; additional file: table b3). there was no interaction between the effect of average wind speed and port. the fitted function indicated that the flight activity of a. ferus and hylastes ater peaks during calm conditions (0 m-1s-1) and declines rapidly with increasing wind speed (additional file: fig. b7). for a. ferus, this decline continues to the highest recorded wind speeds, but the activity of hylastes ater begins to increase again above a wind speed of approximately 7 m-1s-1 (additional file: fig. b7). humidity did not have an effect on the seasonal flight activity of any of our focal forest insect species (all p>0.05, additional file: table b4). probability of flight a gam with a binomial distribution best described the relationship between the probability of flight and weeks of the year for a. ferus and hylastes ater, whereas the quasi-binomial gam best fitted the probability of hylurgus ligniperda flight. for a. ferus and hylurgus ligniperda, weeks of the year affected the probability of flight over the entire trapping period for all ports (p<0.05, additional file: b, table b5). for hylastes ater, weeks of the year affected the probability of flight over the entire trapping period for all ports, except in whangarei (p < 0.001; additional file: table b5). the probability trends for a. ferus at each port were very similar in shape, but flight activity in napier and nelson was almost twice that of whangarei at the same time of year (figure 4). the plot shows that the probability of flight for the five ports increased steeply from late september (excluding dunedin, which increased gradually from early november) and reached a maximum level in early february before decreasing sharply. on the basis of three years sampling, very little or no a. ferus flight activity is predicted from late-april until early-october (additional file: table b6). the probability model showed that hylastes ater displays a bimodal seasonality pattern. the probability of flight in dunedin, where abundances were highest, peaked in mid-april and late october (figure 4). the probability of flight for hylastes ater was low in the summer months (december and january), and close to zero through pawson et al. new zealand journal of forestry science (2020) 50:14 page 6 figure 2: estimated flight activity per 100 trap days with 95% confidence intervals of arhopalus ferus, hylastes ater and hylurgus ligniperda by port throughout the year using the quasi-poisson gam approach. month the winter (june to august) (additional file: table b6). hylurgus ligniperda displayed a similar bimodal seasonal pattern in the probability of activity, with peaks in early february and late september (figure 4). the probability of flight was low from may to august at all ports for hylurgus ligniperda, but only reached zero in dunedin (additional file: table b6). although low, hylurgus ligniperda was the most likely species to be present during the winter period. discussion major sea ports and airports are a focal point of trade between countries. consequently, these facilities are also the main entry and exit points for insects associated with trade. although most port surveillance programmes that target forest insects are designed to detect and report new organisms as they enter a country (brockerhoff et al. 2006; rassati, faccoli, marini et al. 2015; rassati, faccoli, petrucco toffolo et al. 2015), detection is unlikely to be perfect (skarpaas & økland 2009). knowledge of which pest species are present on a port, and when they are active, can also be used to identify when and where export phytosanitary risks are greatest. this permits pawson et al. new zealand journal of forestry science (2020) 50:14 page 7 the application of appropriate treatments that are commensurate to the identified phytosanitary risk and ensures appropriate precautions can be made to prevent re-infestation after treatment. conversely, phenological data of forest pests at ports could be used to define periods when export phytosanitary risk is minimal. if the phytosanitary risks can be proven that they do not exceed the maximum pest limit of the importing country then phytosanitary treatments may potentially be avoided. the flight activity of new zealand’s most significant forest insect pests of bulk wood exports varied substantially between species and ports. maximum temperature was a significant predictor of flight activity, with little activity occurring when temperatures were below 15°c. consequently, little or no flight activity occurred during the colder winter months (june to august), particularly at the more southerly ports and for a. ferus and hylastes ater, which displayed short, predictable peaks in activity at certain times of the year. our collection data and predictive models show that a. ferus is not active at any of the five ports we sampled from may to september, while hylastes ater is mostly inactive across the country from june to august. the figure 3: estimated flight activity at ports per 100 trap days with 95% confidence intervals by maximum evening temperature for arhopalus ferus and maximum daily temperature for hylastes ater and hylurgus ligniperda using additive modelling. table 2: confusion matrix pawson et al. new zealand journal of forestry science (2020) 50:14 page 8 figure 4: predicted probability of flight and 95% confidence intervals of arhopalus ferus, hylastes ater and hylurgus ligniperda by port throughout the year using a gam approach. month seasonality of hylurgus ligniperda varied slightly with latitude. individuals were collected during every month of the year at the two most northerly ports (whangarei and tauranga), but were inactive from may to august at the most southerly port of dunedin. these results suggest that the seasonality of hylurgus ligniperda is affected in part by temperature, with some adult beetles emerging whenever the climate is suitable for flight activity. arhopalus ferus and hylastes ater by contrast, displayed the same strict seasonality patterns across the country, suggesting that warmer temperatures during winter are not sufficient to trigger emergence or induce flight activity of adult individuals. using our detailed phenological data on the most significant forest insect pests of concern in p. radiata forests in new zealand, we are able to identify both ports and times of the year when the probability of infestation of logs in a port environment is low. this could allow for the potential implementation of an alpp. we found that the probability of flight activity was close to zero during june and july for all of our target species at dunedin, nelson, and napier, and also at the northernmost ports for a. ferus and hylastes ater. the probability of flight activity for hylurgus ligniperda at whangarei and tauranga was also low, but above zero. these data indicate an alpp exists at export ports for the lower two-thirds of new zealand during at least the middle of winter (june, july). however, the implementation of an alpp is reliant on the delivery of logs from the forest to the port that have an infestation level below that of a maximum pest limit that is required by trading partners. this may be achievable during winter if ‘just in time’ practices ensure that logs are harvested and removed from the forest when meteorological conditions determine that flight activity, and hence post-harvest infestation, is unlikely (meurisse et al. unpublished data). haack (2001), huang et al. (2012), and; rassati et al. (2015b) showed that there was a relationship between the volume (and value) of imported goods and the number of barkand wood-boring beetle interceptions. from an export perspective, however, we found no relationship between the volume of wood exported from a port (on a monthly basis) and the numbers of forest insects we caught in our traps (additional file: fig. b11). we had expected to see an increase in trap catch when log volumes increased because greater volumes of stored wood will produce a larger odour plume that should be more attractive to dispersing insects. there are alternative explanations that we have not tested, including site-specific effects, type of trap (see below) and competition between odour plumes (i.e., logs piles versus traps) that could explain this result. a potential concern for trading partners may be that we utilised only one type of trap because taxonomic bias between different trap designs is a known confounding factor when monitoring flying insects (hanula et al. 2011; kerr et al. 2017). however, we are confident that our trap type, its colour and the lures that we used were the best available to capture our target pest species (brockerhoff et al. 2006; kerr et al. 2017). the semiochemicals we used are proven to be highly attractive to these wood-boring species (brockerhoff et al. 2006) because they mimic the volatiles emitted by dead and dying trees that these insects (all deadwood feeders) use to locate new hosts. however, the relative attraction of different species to these lures has yet to be formally quantified. the flight intercept trap (fit) design is also particularly good at capturing beetles because the target pest species in our study react to barriers by dropping to the ground (or in the case of a fit, into a collection container). abiotic factors, e.g. aspect, slope, proximity to deadwood sources, and or physical structures are also known to influence trap catch (brockerhoff et al. 2006, 2017). however, an intensive study that defines factors that would maximise the capture rates of forest insects within ports has not been done. our traps were positioned where they would not interfere with normal port operations, but these trap placements may not have been optimal for the detection of insects. alternative trapping locations, such as adjacent land, among nearby trees and in wood waste sites may yield slightly different results (rassati, faccoli, marini et al. 2015; rassati, faccoli, petrucco toffolo et al. 2015). such trapping could potentially assist in defining phytosanitary risk at a port by estimating risks posed by source populations within the broader landscape that are within the dispersal capabilities of target species (meurisse & pawson 2017). additional file appendix a: site information and collection of meteorological variables. appendix b: generalized additive models (gams) of the effects of season, weather, and volume on flight activity of forest insects. competing interests the authors declare that they have no competing interests. authors’ contributions sp conceived the study and prepared the manuscript with cw. trap network was managed by jk. cs completed the analysis. all authors contributed to, reviewed, and approved final manuscript. acknowledgements this work was funded by the ministry for business, innovation, and employment (cx04x1204) and stakeholders in methyl bromide reduction. the authors thank the participating port companies, david finchett (northport) and brett wilton (bimaris) for meteorological data tauranga, port of napier, and port nelson. the trap network was maintained by brooke o’connor, liam wright, tia uaea, and mark west. andre van haandel, orla harris, caitlan penny, caroline gous, and lydia hale are thanked for contributions to sample sorting. david palmer is thanked for production of landcover estimates around each port. pawson et al. new zealand journal of forestry science (2020) 50:14 page 9 references allen, e., noseworthy, m., & ormsby, m. (2017). phytosanitary measures to reduce the movement of forest pests with the international trade of wood products. biological invasions, 19(11), 3365-3376. https://doi.org/10.1007/s10530-017-1515-0 allison, j., strom, b., sweeney, j., & mayo, p. (2018). trap deployment along linear transects perpendicular to forest edges: impact on capture of longhorned beetles (coleoptera: cerambycidae). journal of pest science, 92, 299-308. https://doi.org/10.1007/ s10340-018-1008-7 brockerhoff, e.g., bain, j., kimberley, m., & knížek, m. (2006). interception frequency of exotic bark and ambrosia beetles (coleoptera: scolytinae) and relationship with establishment in new zealand and worldwide. canadian journal of forest research, 36(2), 289-298. https://doi.org/10.1139/x05-250 brockerhoff, e.g., chinellato, f., faccoli, m., kimberley, m., & pawson, s.m. (2017). effects of elevation and aspect on the flight activity of two alien pine bark beetles (coleoptera: curculionidae, scolytinae) in recently-harvested pine forests. forest ecology and management, 384, 132-136. https://doi. org/10.1016/j.foreco.2016.10.046 brockerhoff, e.g., jones, d., kimberley, m.o., suckling, d., m., & donaldson, t. (2006). nation-wide survey for invasive wood-boring and bark beetles (coleoptera) using traps baited with pheromones and kairomones. forest ecology and management, 228, 234-240. https://doi.org/10.1016/j. foreco.2006.02.046 chase, k.d., stringer, l.d., butler, r.c., liebhold, a.m., miller, d.r., shearer, p.w., & brockerhoff, e.g. (2018). multiple-lure surveillance trapping for ips bark beetles, monochamus longhorn beetles, and halyomorpha halys (hemiptera: pentatomidae). journal of economic entomology, 111(5), 22552263. https://doi.org/10.1093/jee/toy190 fao. (2016). global forest products facts and figures (2016) (pp. 1-16). rome: food and agricultural organisation. flaherty, l., gutowski, j.m.g., hughes, c., mayo, p., mokrzycki, t., pohl, g., silk, p., van rooyen, k., & sweeney, j. (2018). pheromone-enhanced lure blends and multiple trap heights improve detection of bark and wood-boring beetles potentially moved in solid wood packaging. journal of pest science, 92, 309-325. https://doi.org/10.1007/s10340-0181019-4 haack, r.a. (2001). intercepted scolytidae (coleoptera) at u.s. ports of entry: 1985-2000. integrated pest management reviews, 6(3-4), 253-282. https:// doi.org/10.1023/a:1025715200538 haack, r.a. (2006). exotic barkand wood-boring coleoptera in the united states: recent establishments and interceptions. canadian journal of forest research, 36(2), 269-288. https:// doi.org/10.1139/x05-249 haack, r.a., britton, k.o., brockerhoff, e.g., cavey, j.f., garrett, l.j., kimberley, m., lowenstein, f., nuding, a., olson, l.j., turner, j., & vasilaky, k.n. (2014). effectiveness of the international phytosanitary standard ispm no. 15 on reducing wood borer infestation rates in wood packaging material entering the united states. plos one, 9(5), e96611. https://doi.org/10.1371/journal.pone.0096611 hanula, j.l., ulyshen, m.d., & horn, s. (2011). effect of trap type, trap position, time of year, and beetle density on captures of the redbay ambrosia beetle (coleoptera: curculionidae: scolytinae). journal of economic entomology, 104(2), 501-508. https:// doi.org/10.1603/ec10263 huang, d., zhang, r., kim, k.c., & suarez, a.v. (2012). spatial pattern and determinants of the first detection locations of invasive alien species in mainland china. plos one, 7(2), e31734. https:// doi.org/10.1371/journal.pone.0031734 ippc. (1999). ispm10: requirements for the establishment of pest free places of production and pest free production sites. cpm-2 adopted, revised 2015. rome, ippc, fao. ippc. (2005). ispm22: requirements for the establishment of areas of low pest prevalence. cpm-7 adopted, revised 2015. rome: ippc, fao. ippc. (2016). ispm 7: phytosanitary certification system. 2011-04 cpm-6 adopted revised 2016. rome, ippc, fao. kerr, j.l., kelly, d., bader, m.k.f., & brockerhoff, e.g. (2017). olfactory cues, visual cues, and semiochemical diversity interact during host location by invasive forest beetles. journal of chemical ecology, 43(1), 17-25. https://doi.org/10.1007/s10886-0160792-x landcare research (2015). lcdb v4.1 land cover database version 4.1, mainland new zealand. retrieved from https://lris.scinfo.org.nz/ l ayer/48423-l cdb-v41-l and-cover-databaseversion-41-mainland-new-zealand/ liebhold, a.m., brockerhoff, e.g., garrett, l.j., parke, j.l., & britton, k.o. (2012). live plant imports: the major pathway for forest insect and pathogen invasions of the us. frontiers in ecology and the environment, 10(3), 135-143. https://doi.org/10.1890/110198 liebhold, a.m., brockerhoff, e.g., & kimberley, m. (2017). depletion of heterogeneous source species pools predicts future invasion rates. journal of applied ecology, 54(6), 1968-1977. https://doi. org/10.1111/1365-2664.12895 meurisse, n., & pawson, s. (2017). quantifying dispersal of a non-aggressive saprophytic bark beetle. plos pawson et al. new zealand journal of forestry science (2020) 50:14 page 10 https://doi.org/10.1007/s10530-017-1515-0 https://doi.org/10.1007/s10340-018-1008-7 https://doi.org/10.1007/s10340-018-1008-7 https://doi.org/10.1139/x05-250 https://doi.org/10.1016/j.foreco.2016.10.046 https://doi.org/10.1016/j.foreco.2016.10.046 https://doi.org/10.1016/j.foreco.2006.02.046 https://doi.org/10.1016/j.foreco.2006.02.046 https://doi.org/10.1093/jee/toy190 https://doi.org/10.1007/s10340-018-1019-4 https://doi.org/10.1007/s10340-018-1019-4 https://doi.org/10.1023/a:1025715200538 https://doi.org/10.1023/a:1025715200538 https://doi.org/10.1139/x05-249 https://doi.org/10.1139/x05-249 https://doi.org/10.1371/journal.pone.0096611 https://doi.org/10.1603/ec10263 https://doi.org/10.1603/ec10263 https://doi.org/10.1371/journal.pone.0031734 https://doi.org/10.1371/journal.pone.0031734 https://doi.org/10.1007/s10886-016-0792-x https://doi.org/10.1007/s10886-016-0792-x https://lris.scinfo.org.nz/layer/48423‐lcdb‐v41‐land‐cover‐database‐version‐41‐mainland‐new‐zealand/ https://lris.scinfo.org.nz/layer/48423‐lcdb‐v41‐land‐cover‐database‐version‐41‐mainland‐new‐zealand/ https://lris.scinfo.org.nz/layer/48423‐lcdb‐v41‐land‐cover‐database‐version‐41‐mainland‐new‐zealand/ https://doi.org/10.1890/110198 https://doi.org/10.1111/1365-2664.12895 https://doi.org/10.1111/1365-2664.12895 one, 12(4), e0174111. https://doi.org/10.1371/ journal.pone.0174111 moore, s.d., kirkman, w., & hattingh, v. (2016). verification of inspection standards and efficacy of a systems approach for thaumatotibia leucotreta (lepidoptera: tortricidae) for export citrus from south africa. journal of economic entomology, 109(4), 1564-1570. https://doi.org/10.1093/jee/ tow139 ministry for primary industries (2019). exports by forestry product years ended 31 december 1980 to 2017. ministry for primary industries, wellington. retrieved 16 october 2019 from https://www. teururakau.govt.nz/news-and-resources/opendata-and-forecasting/forestry/wood-productmarkets/ pawson, s.m., marcot, b.g., & woodberry, o.g. (2017). predicting forest insect flight activity: a bayesian network approach. plos one, 12(9), e0183464. https://doi.org/10.1371/journal.pone.0183464 piel, f., gilbert, m., de cannière, c., & grégoire, j.c. (2008). coniferous round wood imports from russia and baltic countries to belgium. a pathway analysis for assessing risks of exotic pest insect introductions. diversity and distributions, 14(2), 318-328. https:// doi.org/10.1111/j.1472-4642.2007.00390.x r development core team. (2017). r: a language and environment for statistical computing. version 3.4.1. r foundation for statistical computing, vienna, austria: http://www.r-project.org/ rassati, d., faccoli, m., marini, l., haack, r.a., battisti, a., & petrucco toffolo, w. (2015). exploring the role of wood waste landfills in early detection of nonnative wood-boring beetles. journal of pest science, 88(3), 563-572. https://doi.org/10.1007/s10340014-0639-6 rassati, d., faccoli, m., petrucco toffolo, e., battisti, a., & marini, l. (2015). improving the early detection of alien wood-boring beetles in ports and surrounding forests. journal of applied ecology, 52, 50-58. https://doi.org/10.1111/1365-2664.12347 rassati, d., haack, r.a., knížek, m., & faccoli, m. (2018). national trade can drive range expansion of bark and wood-boring beetles. journal of economic entomology, 111(1), 260-268. https://doi. org/10.1093/jee/tox308 rassati, d., petrucco toffolo, e., roques, a., battisti, a., & faccoli, m. (2014). trapping wood boring beetles in italian ports: a pilot study. journal of pest science, 87(1), 61-69. https://doi.org/10.1007/s10340013-0499-5 ryan, j.a., & ulrich, j.m. (2018). xts: extensible time series. version 0.10-2. https://cran.r-project.org/ package=xts skarpaas, o., & økland, b. (2009). timber import and the risk of forest pest introductions. journal of applied ecology, 46(1), 55-63. https://doi.org/10.1111/ j.1365-2664.2008.01561.x tobin, p.c., kean, j.m., suckling, d.m., mccullough, d.g., herms, d.a., & stringer, l.d. (2014). determinants of successful arthropod eradication programs. biological invasions, 16(2), 401-414. https://doi. org/10.1007/s10530-013-0529-5 usda. (1992). pest risk assessment of the importation of pinus radiata and douglas-fir logs from new zealand. [miscellaneous publication no 1508]. united states department of agriculture forest service. p. 63. walker, j.t.s., suckling, d.m., & wearing, c.h. (2017). past, present, and future of integrated control of apple pests: the new zealand experience. annual review of entomology, 62(1), 231-248. https://doi. org/10.1146/annurev-ento-031616-035626 wood, s.n. (2006). generalized additive models: an introduction with r. boca raton, fl, usa: chapman & hall/crc. wylie, f.r., griffiths, m., & king, j. (2008). development of hazard site surveillance programs for forest invasive species: a case study from brisbane, australia. australian forestry, 71(3), 229-235. https://doi.org/10.1080/00049158.2008.106750 40 pawson et al. new zealand journal of forestry science (2020) 50:14 page 11 https://doi.org/10.1371/journal.pone.0174111 https://doi.org/10.1371/journal.pone.0174111 https://doi.org/10.1093/jee/tow139 https://doi.org/10.1093/jee/tow139 https://www.teururakau.govt.nz/news-and-resources/open-data-and-forecasting/forestry/wood-product-markets/ https://www.teururakau.govt.nz/news-and-resources/open-data-and-forecasting/forestry/wood-product-markets/ https://www.teururakau.govt.nz/news-and-resources/open-data-and-forecasting/forestry/wood-product-markets/ https://www.teururakau.govt.nz/news-and-resources/open-data-and-forecasting/forestry/wood-product-markets/ https://doi.org/10.1371/journal.pone.0183464 https://doi.org/10.1111/j.1472-4642.2007.00390.x https://doi.org/10.1111/j.1472-4642.2007.00390.x http://www.r-project.org/ https://doi.org/10.1007/s10340-014-0639-6 https://doi.org/10.1007/s10340-014-0639-6 https://doi.org/10.1111/1365-2664.12347 https://doi.org/10.1093/jee/tox308 https://doi.org/10.1093/jee/tox308 https://doi.org/10.1007/s10340-013-0499-5 https://doi.org/10.1007/s10340-013-0499-5 https://cran.r-project.org/package=xts https://cran.r-project.org/package=xts https://doi.org/10.1111/j.1365-2664.2008.01561.x https://doi.org/10.1111/j.1365-2664.2008.01561.x https://doi.org/10.1007/s10530-013-0529-5 https://doi.org/10.1007/s10530-013-0529-5 https://doi.org/10.1146/annurev-ento-031616-035626 https://doi.org/10.1146/annurev-ento-031616-035626 https://doi.org/10.1080/00049158.2008.10675040 https://doi.org/10.1080/00049158.2008.10675040 logging damage to residual trees during sustainable harvesting of uneven-age stands in the hyrcanian forests of iran afraz iranparast bodaghi1, mehrdad nikooy2, ramin naghdi2, farzam tavankar3,* 1 department of forestry, university campus 2, university of guilan, rasht, iran 2 department of forestry, university of guilan, somehsara, iran 3department of forestry, khalkhal branch, islamic azad university, khalkhal, iran *corresponding author: tavankar@aukh.ac.ir (received for publication 13 september 2018; accepted in revised form 7 april 2020) abstract background: the frequency of wounded trees and intensity of wounds during logging operations can have serious impacts on stand growth and forest sustainability. the aims of the study were to evaluate and compare stand (tree and regeneration) damage level, wound characteristics, and damage types occurring when using a cable skidder in salvage logging and selection cutting. methods: this study was conducted on four sites of mixed uneven-aged hardwood stands in the hyrcanian forests of iran. at two sites, salvage logging was applied (sl1 and sl2), and low-intensity selection cutting (sc1 and sc2) was applied at the other two. a systematic plot sampling design was used on the study area for damage assessment. results: the percentage of regenerating trees damaged was 2.8, 2.1, 4.3, 1.4 %, in sl1, sl2, sc1, and sc2, respectively. the corresponding percentage of damaged trees was 4.3, 3.7, 4.9, 1.7 %. most of the damage (48–79%) to the stand occurred during the winching stage at all the sites. most of the wounds were located on the bole (51–78%). the average wound height and wound size at selection cutting sites were significantly higher than at the salvage logging sites. the incidence of high-intensity wounds at the salvage logging sites (55% at sl1 and 57% at sl2) was higher than at the selection cutting sites (24% in sc1 and 30% in sc2). regenerating beech (fagus orienalis lipsky) and alder (alnus subcordata c.a.mey) had the highest incidence of damage. the number of damaged trees increased with increasing winching distance. conclusion: damage levels in stands during salvage logging and low-intensity selection cutting are lower (about a quarter) than the damage level to residual trees (12–23%) and regeneration (5–11%) from conventional selection cutting in unevenaged mixed hardwood stands in the hyrcanian forests. because of the ecological and conservation value of deadwood, if the incidence of wind-fallen trees is low, the wood should be left in forest stands due to the high cost of salvage logging and the damage caused to residual and regenerating trees. new zealand journal of forestry science bodaghi et al. new zealand journal of forestry science (2020) 50:1 https://doi.org/10.33494/nzjfs502020x15x e-issn: 1179-5395 published on-line: 25/04/2020 © the author(s). 2020 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access (whitman et al. 1997; sist et al. 2003; nikooy et al. 2010; behjou 2014). the damage caused to residual trees could endanger the goal of single-tree selection cutting because of a decrease in growth, increased incidence of decaying wood, and higher probability of mortality for severely damaged trees (tavankar et al. 2017a). it should be noted that damage does not always lead to loss introduction forest managers are concerned about the potential damage to residual trees from cyclic harvest re-entries into the same stand in forests managed under a selection cutting system (picchio et al. 2012; tavankar et al. 2015a). one of the most important challenges of single-tree selection cutting is damage to residual trees keywords: cable skidder, logging damage, salvage logging, selection cutting, uneven-aged forest bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 2 of quality or disease entry. quality loss and fungi attack risk depend on wound intensity, tree size and species (tavankar et al. 2015b, 2017c). after a tree is wounded, decay fungi may be confined to compartments within the tree through a process called compartmentalisation, but the ability to compartmentalise varies by tree species (guyon et al. 2017). for example, fresh wounds on douglas-fir (pseudotsuga menziesii [mirbel] franco) or ponderosa pine (pinus ponderosa lawson & c. lawson) are often covered with resin. however, wounds on nonresinous tree species, such as true firs (abies spp.), generally result in more decay than do wounds on resinous species, such as pines (pinus spp.). globally, salvage logging is becoming more prominent for active forest restoration purposes (mercurio et al. 2010; frattaroli et al. 2017) and to recover economic value from timber in disturbed forests (kärhä et al. 2018). logging damage to residual trees during selection cutting and salvage logging, can cause substantial loss of potential revenue at the final harvest due to reduced timber quality (kiser 2011), reduced recovery of logs from damaged trees, and reduced tree growth (vasiliauskas 2001). in some cases, logging damage to residual trees can result in their death (tavankar et al. 2017b). the frequency of wounded trees and severity of wounds during logging operations may have detrimental impacts on stand growth and forest sustainability. logging wounds on residual trees often become an entry points for fungal invasion (vasiliauskas 2001), particularly in wounds that are close to ground level or in tree roots (bettinger & kellogg 1993; camp 2002). wounding can result in stem distortion and substantial losses of the volume and value of the final product (meadows 1993; lo monaco et al. 2015). the most common (>90%) type of tree wounding is logging damage (marchi et al. 2014; tavankar et al. 2017a). wound characteristics such as size, location, and severity are the main factors determining wound-healing ability and the impact on diameter growth of trees (tavankar et al. 2015b). bole wounds on residual trees are among the most common types of cable-skidder logging damage (nikooy et al. 2010; jourgholami et al. 2012; tavankar et al. 2015a). the amount of diameter growth reduction in decayed wounds in beech (fagus orientalis lipsky) trees was reported to be 15.3% (tavankar et al. 2015b). in an 85-year-old stand of spruce (picea abies (l.) karst.), in belarus, which 6% of trees had 5-year-old logging injuries, saw-log yield decreased by 11 m3 ha-1 (kovbasa 1996). financial revenues at final harvesting in spruce (picea abies (l.) karst.) stands decreased by 7–20% due to wound decay in austria (steyrer 1992). the extent and intensity of residual tree damage during logging operation depend on the harvest intensity (fjeld & granhus 1998; gullison & hardner 1993; han et al. 2000; sist et al. 2003; behjo 2014; tavankar et al. 2015a), site characteristics and topography (picchio et al. 2012; tavankar et al. 2015a), type and sizes of harvesting equipment (bragg et al. 1994; spinelli et al. 2010; marchi et al. 2014), pattern of roads and skid trails (gullison & hardner 1993; nikooy & ershadifar 2012; danilović et al. 2015), and experience and education level of logging workers (nikooy et al. 2010). hyrcanian forests in iran cover 1.8 million ha of the northern slopes of alborz mountain from coastal of caspian sea to an elevation of 2,800 m. high biodiversity (growing about 148 woody species) and extensive wildlife habitat highlight the ecological importance of these forests. the structure of these forests is mixed broadleaves and they are mainly managed as unevenaged high forests. the hyrcanian forests have been managed by different silvicultural treatments: shelter wood-cutting from 1970 to 2000, selection cutting from 2000 to 2014, and restricted cutting to damaged and fallen trees from 2016 to the present. the main natural disturbance in the hyrcanian mountain forests in the northern iran is wind damage and some trees in these forests are blown over every year. in order to minimise the impacts of biological agents (fungi and insects) and to make active forest restoration, forest managers apply salvage logging in damaged stands. another purpose of salvage logging is to earn income from the forest; however, it is not planned. in fact, the extensive removal of dead-wood resources by post-disturbance salvage logging may decrease biodiversity (thorn et al. 2017). salvage logging following windstorms removes the storm-created pit and mount system, leading to homogenised structures, altered microsite diversity, and altered assemblages of vascular plants (waldron et al. 2014). damage level to residual trees during conventional selection cutting by chainsaw and cable skidder logging operation in hyrcanian forests were previously reported (behjou 2014; nikooy et al. 2010; tavankar et al. 2017a). however, there is little information about stand damage during salvage logging and low-intensity selection cutting in the hyrcanian forests. the aims of the research were to evaluate and compare stand (tree and regeneration) damage level, wound characteristics, damaged types by salvage logging and selection cutting. methods the study was conducted in district no. 3 of the nav watershed forest area between 37° 38’ 34” to 37° 42’ 21” n and 48° 48’ 44” to 48° 52’ 30” e, in guilan province. average rainfall is 1,050 mm yr-1, with the heaviest rainfall in the summer and autumn. the soils are alfisols, well-drained, and the texture varies between sandy clay loam to clay loam. the average daily temperature ranges from a few degrees below 0°c in december, january, and february, and up to +26°c during the summer. dendrometric stand characteristics before logging operations commenced are shown in table 1. logging damage to residual trees were evaluated in four parcels in this district. in two parcels (no. 319 and 331) salvage logging was applied (sl1 and sl2), and in two other parcels (no. 318 and 321) low intensity selection cutting (sc1 and sc2) was applied. dendrometric stand characteristics before logging, skid trail and corridors features, and dendrometric characteristics of extracted wood are shown in tables 1, 2, and 3, respectively. logging operations were carried out in summer 2016. bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 3 logging site item sl1 sl2 sc1 sc2 elevation (a.s.l. m) 1,300–1,450 1,250–1,380 1,300–1,450 1,270–1,390 area (ha) 33 29 32 27 tree density (stems ha-1) 279 277 170 281 mean dbh (cm) 20.8 21.1 21.3 19.6 mean basal area (m2 ha-1) 23.7 24.6 22.3 24.8 mean tree height (m) 19.7 18.6 20.2 19.3 standing volume (m3 ha-1) 299 250 196 205 mean ground slope (%) 25.2 31.0 23.0 24.8 tree species composition (%) beech (57), hornbeam (19), maple (9), alder (5), other sp. (10) beech (65), maple (20), alder (7), hornbeam (5), other sp. (3) beech (60), alder (20), maple (8), hornbeam (8), other sp. (4) beech (69), hornbeam (19), maple (6), alder (3), other sp. (3) table 1: dendrometric stand characteristics before logging in four logging sites (sl: salvage logging, sc: selection cutting) logging site item sl1 sl2 sc1 sc2 skid trail length density (m ha-1) 22.7 19.6 20.1 22.5 mean skid trail slope (%) 8.3 10.2 12.3 10.9 mean winching strip length (m) 46.4 41.7 27.5 23.3 mean winching strip slope (%) 39.9 41.0 31.5 38.6 mean number of logs per turn (n) 1.9 1.8 1.8 1.9 mean log volume per turn (m3) 8.6 8.8 8.2 8.1 logging site extracted wood sl1 sl2 sc1 sc2 parameters of extracted trees total number (stem) 46 38 40 30 number density (stem ha-1) 1.4 1.3 1.3 1.1 total volume (m3) 308 240 348 120 volume density (m3 ha-1) 6.7 5.8 8.7 4.0 mean dbh (cm) 87.6 83.1 53.6 51.4 mean bole length (m) 17.2 17.6 16.5 15.8 mean volume (m3) 6.4 6.1 4.3 4.2 parameter of extracted logs total number 112 90 91 68 mean length (m) 6.8 7.1 6.6 6.5 mean small diameter (cm) 34.5 35.3 41.2 39.1 mean large diameter (cm) 65.4 69.4 69.3 68.4 mean volume (m3) 4.5 4.9 4.7 4.2 table 2: skid trail and corridors features in four logging sites (sl: salvage logging, sc: selection cutting) table 3: dendrometric characteristics of extracted wood in four logging sites (sl: salvage logging, sc: selection cutting) harvesting intensity in low selection cutting sites was light (4.4% and 1.9% of standing volume) (tables 1 and 3) in comparison with conventional selection cutting in these forests. trees were selected on the basis of closeto-nature silvicultural aims, i.e. creation of canopy gaps, especially in dense areas, regulation of light intensity on the forest floor to facilitate natural regeneration of beech trees, regulation of tree species composition, improve tree quality, and provide economic returns. ground-based timber extraction methods were applied in all stands. skid trail and corridor features are shown in table 2. felling was done by chainsaw and timber extraction was done with a timberjack 450c wheeled skidder (mass 9.8 t, width 3.8 m, length 6.4 m). the selected trees, which were scattered throughout the study area, were felled and cut into merchantable lengths (mostly 7.8 m, but occasionally 5.2 m) or to a 20 cm diameter under bark. logs were then winched from the felling site (downhill) toward the skidder and skidded to roadside landings. the skidding operation was limited to the constructed skid trails. before felling, the skid trails were planed and constructed. the skidding team consisted of a driver, a chaser, and a feller. the winching operation was controlled manually. data collection the study used a systematic sampling design to evaluate stand damage (meadows 1993; lotfalian et al. 2008; nikooy et al. 2010). a grid with 100-m x 100-m spacing was laid out across the stand and a circular 0.1 ha plots placed at each grid intersection. the species, diameter at breast height (dbh) and condition (healthy or damaged) of all trees (dbh ≥7.5 cm) and natural regeneration (dbh <7.5 cm) were measured and recorded in each plot. the following parameters were recorded for each damaged tree: cause of damage (i.e. from felling, winching and skidding); damage location (i.e. crown, bole or root); and depth of damage. because bark thickness differs among species found in the forest, the depth of damage was based on the nature of damaged tissue, i.e. i) low, bark removed, ii) medium, phloem damaged, and iii) high, wood damaged. the height of the wound centre from ground level was measured and the wound size was determined by measuring the maximum length, width and the ellipsoid surface area (picchio et al. 2011). the shortest distance between the stumps of the cut trees and the skid trail was measured and denoted as the winching distance (wd). a count of damaged trees was undertaken at up to a distance of 5 m from each side of the winching route center. data analysis the proportion of residual trees damaged in each sample plot (pi) was calculated as follows: pi = nd/nt (1) where, nd is number of damaged trees, and nt is total number of trees in the sample plot. to account for the non-normality of these data, they were transformed (pt) prior to analysis as follows: pt = arcsin √pi (2) bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 4 following checking for normality (kolmogorov– smirnov test) and homogeneity of variance (levene test), anova and duncan tests were applied to analyse the effect of different factors on the percentage of damaged trees and regeneration, wound size, and wound height. regression analysis and pearson correlation test were applied to test the relationship between number of damaged trees and winching distance at the logging sites. a non-parametric test was used to analyze the effect of logging methods on the wound severity, wound location, damage percentage at logging stages. all analyses were performed using spss 20 (ibm, ny, usa). results anova results showed that there were significant differences in the percentage of regeneration and trees damaged across the different logging sites. the percentage of damaged regeneration in sl1, sl2, sc1, and sc2 was 2.8, 2.1, 4.3, and 1.4 %, respectively (fig. 1a) whereas the percentage of damaged trees in sl1, sl2, sc1, and sc2 were 4.3, 3.7, 4.9, and 1.7 %, respectively (fig. 1b). duncan multiple tests showed that the percentage of damaged regeneration and trees at sc1 were significantly higher than at the other sites (fig. 1a, and fig. 1b). anova results also showed that there were significant difference in wound height and size among sites (table 4). average wound height and size in the selection cutting sites (62.8 cm and 330 cm2 in sc1 and 62.3 cm and 311.1 cm2 in sc2) were significantly higher than at the salvage logging sites (38.7 cm and 208.3 cm2 in sl1 versus 40.4 cm and 206.1 cm2, respectively). most damage (48–79%) to stands (both trees and regeneration) across all the all logging sites occurred during the winching operation (fig. 2a). the extent of winching damage at the salvage logging sites (79% in sl1 and 73% in sl2) was considerably higher than at the selection cutting sites (50% in sc1 and 48% in sc2). deep wounds were most abundant at the salvage logging sites and shallow wounds were most common at the selection felling sites. the number of serious wounds in the salvage logging sites (55% in sl1 and 57% in sl2) was higher than at the selection cutting sites (24% in sc1 and 30% in sc2). significant differences in the percentage of damage at different logging stages and wound severity were found among logging sites (table 5). the most frequent damage location across all logging sites was the bole (fig. 3a). the extent of bole damage at the salvage logging sites (76% in sl1 and 78% in sl2) was higher than selection cutting sites (55% in sc1 and 51% in sc2). crown damage was observed only in the selection cutting sites. root damage was more common at the salvage logging sites (23.6% in sl1 and 22% in sl2) compared with the selection cutting sites (14.2% in sc1 and 11.9% in sc2). most damage occurred to residual trees in the 20–40 cm diameter class across all the logging sites (fig. 3-b). more damage to larger trees (dbh >60 cm) was recorded at the selection cutting sites than at the salvage logging sites. the chi-square tests showed that logging methods had a significant effect on damage location and the size of damaged trees (table 5). bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 5 figure 1: damaged regeneration (a) and trees (b), height (c) and size (d) of wounds in in four logging sites (sl: salvage logging, sc: selection cutting) variable ss df ms f p value regeneration 0.131 3 0.044 21.99 0.000 tree 0.151 3 0.050 80.14 0.000 wound height 47,705 3 15,901 18.9 0.000 wound size 1,165,499 3 388,499 102.6 0.000 table 4: anova results for analysis of logging methods on damage percentage to regeneration and trees, and height and size of bole wounds figure 2: stand damage during logging stages (a), and intensity of bole wounds (b) in in four logging sites (sl: salvage logging, sc: selection cutting) beech (fagus orienalis) was the most common regenerating species that was damaged across all the logging sites (fig. 4a). damaged beech regeneration at the selection cutting sites (4.3% in sc1 and 4.5% in sc2) was more common than at the salvage logging sites (3.3% in sl1 and 3.0% in sl2). the most commonly damaged tree species across all logging sites was alder (alnus subcordata c.a.mey) (fig. 4b). in this case the percentage of damaged alder trees at the salvage logging sites (8.2% in sl1 and 7.6% in sl2) was higher than at the selection cutting sites (5.6% in sc1 and 4.2% in sc2). results of the regression analysis and pearson correlation tests showed that there was a significant positive correlation between winching distance and number of damaged trees (table 6). not unexpectedly, the number of damaged trees increased when winching distance increased in the all logging sites (fig. 5). discussion results of this study showed that less than 5% of regeneration and residual trees (i.e. 2.1 to 2.8% of regeneration and 3.7 to 4.3% of residual trees in salvage logging sites, and 1.4 to 4.3% of regeneration and 1.7 to 4.9% of residual trees in the selection cutting sites) were damaged using the harvesting methods employed. this was considerably lower than damage levels found in conventional selection cutting operations in the hyrcanian forests. for example, tavankar et al. (2015a) found damage levels of 12 to 23 percent to residual trees and 5 to 11 percent to regeneration. other studies in caspian forests also found higher levels of damage to residual trees following selection cutting, e.g. 15.5% by lotfalian et al. (2008), 19% by naghdi et al. (2008), 19.7% by nikooy et al. (2010) and 16.9% by tavankar et al. (2015b). even higher levels of damage (16–32%) to residual trees were found in a cut-to-length thinning operation in a young redwood (sequoia sempervirens (d. don) endle.) forest in northern california (hwang et al. 2018). silva et al. (2018) evaluate the bole and crown damage and the incidence of leaning trees in the residual stand following timber harvesting in a secondary atlantic rainforest in southern brazil. they found that on average, 26% of the residual trees suffered some kind of damage, with 12.1% suffering moderate or severe damage. yilmaz and akay (2008) in turkish forests found that 14% of residual trees were damaged during felling and skidding operation. the study carried out by ficklin et al. (1997) showed that about 22% of the residual trees was damaged by skidding operation using wheeled skidder. in a study conducted by hartsoug (2003) in north-eastern californian forests, it was shown that 23% of residual trees were damaged during ground-based logging. hosseini et al. (2000) compared damage to natural regeneration caused by two logging systems (skidder and cable system) in the caspian forests of iran. they showed that damage to regeneration during log skidding were significantly more severe than in log yarding (11% vs. 5%). whitman et al. (1997) showed that damage level to seedlings in a single tree selection cutting operation in northern belize was 15%. the amount of damage to regeneration and trees obtained from our study was lower than previous studies in hardwood selection cuttings. the low intensity of logging operations in our bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 6 variable df chi-square p value logging stage 9 58.8 0.000 wound intensity 6 40.7 0.000 damage location 6 83.8 0.000 damaged dbh 9 26.5 0.002 table 5: results of chi-square tests to analyze the effect of logging methods on the wound severity, wound location, damage percentages at logging stages figure 3: damage location (a) and dbh (b) of damaged trees in in four logging sites (sl: salvage logging, sc: selection cutting) bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 7 figure 4: damage to regeneration (a) and tree (b) species in in four logging sites (sl: salvage logging, sc: selection cutting) logging site n equation se r2 adjusted f value p value sl1 78 y = 1.1117x0.615 0.211 0.586 109.8 0.000 sl2 66 y = 0.6672x0.7197 0.228 0.709 159.3 0.000 sc1 78 y = 0.6277x0.7961 0.216 0.761 245.7 0.000 sc2 72 y = 0.1802x1.1828 0.258 0.731 194.2 0.000 table 6: results of regression analysis for relationship between number of damaged trees and winching distance in the logging sites study is likely to be the main reason of low levels of damage observed in the residual stands. intensity, size and location of bole wounds are important factors in bole decay (tavankar & bonyad 2014; câmpu & borz, 2017). results from our study showed that the average wound height and size at the selection cutting sites were significantly greater than at the salvage logging sites. this is most likely due to absence of the felling stage in salvage logging sites. previous studies have shown that the size and height of felling wounds were higher than winching and skidding wounds (tavankar et al. 2017b). one of the most important factors determining the healing rate of injured trees, is wound height (vasiliauskas 2001; tavankar et al. 2017b). the wounds near the ground are more likely to be exposed to wood rot-fungi than other wounds (camp 2002; metslaid et al. 2018). the average wound sizes in selection cutting sites (sc1 and sc2) and salvage logging sites (sl1 and sl2) were 330 cm2, 311 cm2, 208 cm2, and 206 cm2, respectively. in the years after the wound is created, diameter growth (tavankar et al. 2015b; hecht 2015) and the quality of the wood (bonyad & tavankar 2016) are more affected by the size of the wound. wounds larger than 1000 cm2 are less likely to heal, ultimately leading to wood decay (tavankar & bonyad 2014). the size of the bole wounds created in during the winching and skidding stages has been reported as being smaller compared with those created during the felling stage (tavankar et al. 2015a; behjou 2014). trees with injuries in the lower parts of stem and the root collar have the highest risk of decay. at all the logging sites, most of tree and regeneration damage (48–79%) occurred during the winching stage. winching is the major cause of damage to regeneration during cable skidder logging (nikooy et al. 2010; picchio et al. 2011; badraghi el al. 2015). the share of winching damage at the salvage logging sites (79% in sl1 and 73% in sl2) was more than at the selection cutting sites (50% in sc1 and 48% in sc2). picchio et al. (2012) and table 2: confusion matrix marchi et al. (2014) note that snatch blocks can help to preserve regeneration and considerably reduce overall impact. most bole wounds were deep (wood damaged) at the salvage logging sites, while at the selection cutting sites they were mostly light (bark damaged). light wounds rarely lead to decay in the hyrcanian forests. previous studies have shown that tree species, age, wound size and height play the fundamental rule in the decay of the bole severe wounds (tavankar & bonyad 2014; hecht 2015; smith et al. 1994; vasiliauskas 2001; kartoolinejad et al. 2017). for trees with deep wounds in hyrcanian forests, tavankar and bonyad (2014) found that 28% developed subsequent decay. the extent of high intensity wounds at the salvage logging sites (55 and 57%) was higher than at the selection cutting sites (24% in sc1 and 30% in sc2). the bole was the most frequent location for damage at all the logging sites. the frequency of bole damage at the salvage logging sites (76% in sl1 and 78% in sl2) was higher than at the selection cutting sites (55% in sc1 and 51% in sc2). furthermore, the extent of root damage at the salvage logging sites (23.6% in sl1 and 22% in sl2) was higher than at the selection cutting sites (14.2% in sc1 and 11.9% in sc2). all of crown damage and bole wounds higher than 2 m from the ground occurred in the tree felling stage. trees with severely damaged crowns are more likely to die in the future. most of the damage occurred to beech (fagus orienalis) regeneration at all the logging sites. pure and mixed beech communities, are the most dominant forest communities in the natural caspian forests (marvie-mohadjer 2006). these communities constitute approximately 17.6% of the area, 30% of standing volume and 23.6% of stem numbers in the iranian caspian forests (amiri et al. 2013). at all sites, the most commonly-damaged tree species was alder tree (alnus subcordata). previous research by tavankar et al. (2017b) found that the wound healing rate of this species ranged from 0 to 17.3 cm2 yr-1 with a mean of 4.95 cm2 yr-1, and the diameter growth of bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 8 figure 5: relationship between number of damaged trees and winching distance in in four logging sites (sl: salvage logging, sc: selection cutting) wounded trees was 13.3% lower compared with sound trees. the significant positive correlation between winching distance and number of damaged trees was consistent with results of other studies in a range of forest types (e.g. bettinger & kellogg 1993; lotfalian et al. 2008; naghdi et al. 2008; nikooy et al. 2010; jourgholami et al. 2012; marchi et al. 2014). overall, one of the key requirements for a logging method in a sustainable forest management context is that it should result in minimal damage to the residual stand and soil (naghdi et al. 2016). the study reported here provides important information on the nature and extent of damage in hyrcanian forests following selection and salvage logging. coupled with previous information on tree growth and wound decay, our results provide important information on the likely future impacts for treated stands. conclusions results from our study showed that damage levels to residual trees during salvage logging and low intensity selection cutting were lower (about a quarter) than levels previously observed during conventional selection cutting operations. our study showed that the severity of wounds on residual trees from salvage logging was higher than from selection cutting. most of tree wounds occurred during the log winching stage. the present study confirms that with careful planning and execution, it is possible to decrease mechanised harvesting damage. careful planning of skid trails and landings, using directional felling to inflict the smallest impact on the surrounding forest and skilled harvester labour are key to decreasing logging damage. worker training and supervision are necessary to reduce logging damage. pre-harvest planning to identify the winching corridors prior to log extraction can decrease stand and soil damage. residual stand damage is a natural consequence of selective cutting and has been considered to be a problem in the management of uneven-aged forest stands. however, it is an avoidable risk and through minimising the level of damage subsequent stand growth and product quality can be improved. list of abbreviations anova: analysis of variance; a.s.l.: above sea level; dbh: diameter at breast height; sc: selection cutting; se: standard error; sl: salvage logging; spss: statistical package for social sciences; ss: sum of squares; df: degree of freedom; ms: mean of squares. competing interests the authors declare that they have no competing interests. authors’ contributions aib was the primary author. aib, mn, and ft undertook the fieldwork. mn and rn provided critical revisions of the manuscript. ft compiled the data into spreadsheets and completed the statistical analyses. all authors read and approved the final manuscript. acknowledgements we are grateful to university of guilan for providing equipment for data collection. we also acknowledge the shafarood forest company for providing required data and information of the study area. references amiri, m., rahmani, r., sagheb-talebi, kh., & habashi, h. (2013). dynamic and structural characteristics of a natural unlogged oriental beech (fagus orientalis lipsky) stand during a 5-year period in shast kalate forest, northern iran. the international journal of environmental resources research, 1(2), 107–129. badraghi, n., erler, j., & hosseini, s.a. (2015). residual damage in different ground logging methods alongside skid trails and winching strips. journal of forest science, 61(12), 526–534. behjou, f.k. (2014). effects of wheeled cable skidding on residual trees in selective logging in caspian forests. small-scale forestry, 13(3), 367– 376. bettinger, p., & kellogg, l. (1993). residual stand damage from cut-to-length thinning of second growth timber in the cascade range of western oregon. forest product journal, 43(11–12), 59–64. bonyad, a., & tavnkar, f. (2016). investigation on resistance of different tree species to logging wounds (case study: district 1 of asalem-nav forest). journal of forest and wood products, 68(4), 741–756. (in persian). bragg, w., ostrofsky, w., & hoffman, b. (1994). residual tree damage estimates from partial cutting simulation. forest products journal, 44(7–8), 19– 22. camp, a. (2002). damage to residual trees by four mechanized harvest systems operating in small diameter, mixed conifer forests and steep slopes in northeastern washington: a case study. western journal of applied forestry, 17(1), 14–22. câmpu, v.r., & borz, s.a. (2017). amount and structure of tree damage when using cut-to-length system. environmental engineering and management journal, 16(9), 2053–2061. danilović, m., kosovski, m., gačić, d., stojnić, d., & antonić, s. (2015). damage to residual trees and regeneration during felling and timber extraction in mixed and pure beech stands. šumarski list, 5, 253–262. ficklin, r.l., dwyer, j.p., cutter, b.e., & draper, t. (1997). residual tree damage during selection cuts using two skidding systems in the missouri ozarks. proceedings of 11th central hardwood forest conference; columbia, missouri, 1997 mar 23–26; st paul, mn. columbia, mo: north central forest experiment station. pp. 36–46. bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 9 fjeld, d., & granhus, a. (1998). injuries after selection harvesting in multi-storied spruce stands – the influence of operating systems and harvest intensity. international journal of forest engineering, 9(2), 33–40. doi:10.1080/08435243. 1998.10702716 frattaroli, a.r., pirone, g., di cecco, v., console, c., contu, f., & mercurio, r. (2017). beech-wood restoration in the gran sasso and monti della laga national park (central apennines, italy). plant sociology, 54(1), 11–18. gullison, r., & hardner, j. (1993). the effects of road design and harvest intensity on forest damage caused by selective logging: empirical results and a simulation model from the bosque chimanes, bolivia. forest ecology and management, 59(1–2), 1–14. doi.org/10.1016/0378-1127(93)90067-w guyon, j., cleaver, c., jackson, m., saavedra, a., zambino, p. 2017. a guide to identifying, assessing, and managing hazard trees in developed recreational sites of the northern rocky mountains and the intermountain west. usda forest service, report number r1-17-31, 74 p. han, h.s., kellogg, l.d., fillip, g.m., & brown, t.d. (2000). scar closure and future timber value losses from thinning damage in western oregon. forest products journal, 50, 36–42. hartsough, b. (2003). economics of harvesting to maintain high structural diversity and resulting damage to residual trees. western journal of applied forestry, 18, 133–142. hecht, u., kohnle, u., nill, m., grüner, j., & metzler, b. (2015). bark wounds caused by felling are more susceptible to discoloration and decay than wounds caused by extraction in european beech. annals of forest science, 72(6), 731–740. doi: 10.1007/s13595-014-0432-y hosseini, s.m., majnounian, b., & nieuwenhuis, m. (2000). damage to natural regeneration in the hyrcanian forests of iran: a comparison of two typical timber extraction operations. journal of forest engineering, 11, 69–73. hwang, k., han, h.s., marshall, s., & page-dumroese, d. (2018). amount and location of damage to residual trees from cut-to-length thinning operations in a young redwood forest in northern california. forests, 9(6), 1–12. https://doi. org/10.3390/f9060352 jourgholami, m., rizvandi, v., & majnounian, b. (2012). evaluating the extent, patterns, size and distribution of tree scars following skidding operation (case study: kheyrud forest). iranian journal of forest, 4(3), 187–196. (in persian). kärhä, k., anttonen, t., poikela, a., palander, t., laurén, a., peltola, h., & nuutinen, y. (2018). evaluation of salvage logging productivity and costs in windthrown norway spruce-dominated forests. forest, 9(5), 1–22. https://doi.org/10.3390/ f9050280 kartoolinejad, d., najafi, a. & kazemi-najafi, s. (2017). long-term impacts of ground skidding on standing trees: assessment of decay using stress waves. environmental engineering & management journal, 16(10), 2283–2291. kiser, j. (2011). histochemical and geometric alterations of sapwood in costal douglas-fir following mechanical damage during commercial thinning. silva fennica, 45(4), 729–741. doi.org/10.14214/ sf.447 kovbasa, n.p. (1996). distribution and spreading of wound rot in belarus spruce stands and measures to limit the losses. ph.d. dissertation belarusian plant protection research institute, priluki, minsk. lo monaco, a., calienno, l., pelosi, c., balletti, f., agresti, g., & picchio, r. (2015). technical properties of beech wood from aged coppices in central italy. iforest, 8(1), 82–88. doi:10.3832/ifor1136-007 lotfalian, m., parsakho, a., & majnounian, b. (2008). a method for economic evaluation of forest logging damages on regeneration and stand (case study: alandan and waston serries). journal of environmental science and technology, 10(2), 51– 62. (in persian) marchi, e., picchio, r., spinelli, r., verani, s., venanzi, r., & certini, g. (2014). environmental impact assessment of different logging methods in pine forests thinning. ecological engineering, 70, 429– 436. marvie-mohadjer, m. (2006). silviculture. tehran: tehran university publishing. (in persian). meadows, j.s., (1993). logging damage to residual trees following partial cutting in a green ash-sugarberry stand in the mississippi delta. proceedings of the 9th central hardwood forest conference; 1993 mar 8–10. indiana, usda forest service. pp. 248–60. mercurio, r., contu, f., & scarfò, f. (2010). new approaches concerning forest restoration in a protected area of central italy: an introduction. scandinavian journal of forest research, 25(8), 115–120. metslaid, m., granhus, a., scholten, j., fjeld, d., & solheim, h. (2018). long-term effects of single-tree selection on the frequency and population structure of root and butt rot in uneven-sized norway spruce stands. forest ecology and management, 409, 509–517. https://doi.org/10.1016/j.foreco.2017.11.050 naghdi, r., nikooy, m., ghajar, e., & ershadifar, m. (2016). a practical linear model for estimation of tree falling direction error in mountainous forests of northern iran. ecopersia, 4(3), 1505–1516. bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 10 naghdi, r., rafatnia, n., bagheri, i., & hemati, v. (2008). evaluation of residual damage in felling gaps and extraction routes in single selection method (siyahkal forest). iranian journal of forest and poplar research, 16(1), 87–98. (in persian). nikooy, m., & ershadifar, m. (2012). effects of skid trail planning, landing construction and directional felling on normal selective logging in caspian forest. dubrovnik, croatia, 8–12 oct. 2012. nikooy, m., rashidi, r., & kocheki, g. (2010). residual trees injury assessment after selective cutting in broadleaf forest in shafaroud. caspian journal of environmental sciences, 8(2), 173–179. picchio, r., neri, f., petrini, e., verani, s., marchi, e., & certini, g. (2012). machinery-induced soil compaction in thinning two pine stands in central italy. forest ecology and management, 285, 38–43. doi.org/10.1016/j.foreco.2012.08.008 picchio, r., neri, f., maesano, m., savelli, s., sirna, a., blasi, s., baldini, s., & marchi, e. (2011). growth effects of thinning damage in a corsican pine (pinus laricio poiret) stand in central italy. forest ecology and management, 262, 237–243. doi.org/10.1016/j. foreco.2011.03.028 sist, p., sheil, d., kartawinata, k., & priyadi, h. (2003). reduced impact logging in indonesian borneo: some results confirming the need for new silvicultural prescriptions. forest ecology and management, 179(1), 415–427. doi.org/10.1016/ s0378-1127(02)00533-9 smith, h.c., miller, g.w., & schuler, t.m. (1994). closure of logging wounds after 10 years. usda forest service, research paper ne-692. spinelli, r., magagnotti, n., & nati, c. (2010). benchmarking the impact of traditional small-scale logging systems used in mediterranean forestry. forest ecology and management, 260, 1997–2001. doi.org/10.1016/j.foreco.2010.08.048 steyrer, g. (1992). extent and economical significance of stem decays in forestry enterprise. australian journal of forest science, 109, 221–249. silva, d.a., piazza, g., fantini, a.c., & vibrans, a.c. (2018). forest management in a secondary atlantic rainforest: assessing the harvest damage. advances in forestry science, 4(4), 187–193. tavankar, f., bonyad, a. e., nikooy, m., picchio, r., venanzi, r., & calienno, l. (2017a). damages to soil and tree species by cable-skidding in caspian forests of iran. forest systems, 26(1), 1–9. doi:10.5424/ fs/2017261-09100 tavankar, f., nikooy, m., picchio, r., bonyad, a., & venanzi, r. (2017b). effects of logging wounds on caucasian alder trees (alnus subcordata c.a. mey.) in iranian caspian forests. croatian journal of forest engineering, 38(1), 73–82. tavankar, f., bonyad, a., & majnounian, b. (2015a). affective factors on residual tree damage during selection cutting and cable-skidder logging in the caspian forests, northern iran. ecological engineering, 83(8), 505–512. doi:10.1016/j. ecoleng.2015.07.018 tavankar, t., bonyad, a., marchi, e., venanzi, r., & picchio, r. (2015b). effect of logging wounds on diameter growth of beech (fagus orientalis lipsky) trees following selection cutting in caspian forest of iran. new zealand journal of forestry science, 45:19, 1–7. doi: 10.1186/s40490-015-0052-9 tavankar, f., picchio, r., nikooy, m., lo monaco, a., venanzi, r., & iranparast bodaghi, a., (2017c): healing rate of logging wounds on broadleaf trees in hyrcanian forest with some technological implications. drewno, 60, 65–80. doi: 10.12841/ wood.1644-3985.200.05 tavankar, f., & bonyad, a. (2014). long-term effects of logging damages on quality of residual trees in the asalem nav forest. journal of environmental studies, 40(1), 39–50. (in persian) thorn, s., claus bässler, c., svoboda, m., & müller, j. (2017). effects of natural disturbances and salvage logging on biodiversity – lessons from the bohemian forest. forest ecology and management, 388, 113–119. doi: 10.1016/j.foreco.2016.06.006 vasiliauskas, r. (2001). damage to trees due to forestry operations and its pathological significance in temperate forest: a literature review. forestry, 74, 319–336. waldron, k., ruel, j.c., gauthier, s., de grandpré, l., & peterson, c.j. (2014). effects of post-windthrow salvage logging on microsites, plant composition and regeneration. applied vegetation sciences, 17, 323–337. whitman, a., brokaw, n., & hagan, h. (1997). forest damage caused by selection logging of mahogany in northern belize. forest ecology and management, 92(1–3), 87–96. doi.org/10.1016/s03781127(96)03941-2 yilmaz, m., & akay, a. (2008). stand damage of a selection cutting system in an uneven aged mixed forest of cimendagi in kahramanmarz turkey. international journal of natural engineering science, 2, 77–82. bodaghi et al. new zealand journal of forestry science (2020) 50:1 page 11 taper and individual tree volume equations of eucalyptus varieties under contrasting irrigation regimes juan carlos valverde1*, rafael rubilar1,2, alex medina3, oscar mardones4, verónica emhart3, daniel bozo1, yosselin espinoza1 and otavio campoe5 1 cooperativa de productividad forestal, departamento de silvicultura, fac. ciencias forestales, universidad de concepción, concepción, chile 2 centro nacional de excelencia para la industria de la madera (cenamad), pontificia universidad católica de chile, santiago, chile 3 bioforest s.a., km 15 s/n camino a coronel, coronel, concepción, chile 4 forestal mininco s.a., avenida alemania 751, los ángeles, chile 5 departamento de cîencias florestais, universidade federal de lavras, lavras, mg 3037, brazil *corresponding author: juvalverdeo@udec.cl (received for publication 27 august 2021; accepted in revised form 9 may 2022) abstract background: compatible taper and volume equations are key for traditional growth and yield and current process-based or hybrid models. however, most equations do not consider variables such as genotype, water regime and their interaction, limiting the development of general equations for species or regions. our research investigated taper and individual tree volume equations for eight eucalyptus genotypes (e. nitens, e. badjensis, e. smithii, e. camaldulensis x globulus and two varieties of low and high productivity of e. globulus and e. nitens x globulus), all materials are growing under summer irrigated vs. no irrigated conditions. methods: a 7-year old eucalyptus plantation experiment was sampled considering four representative trees per genotype x water regime combination treatment. four non-linear taper equations were evaluated: kozak (2004), kozak et al. (1969), ormerod (1973) and max and burkhart (1976). in addition, total and merchantable volume was evaluated with the schumacher and hall (1933) equation. the effect of genotype, irrigation regime and interaction were evaluated for each equation. then, the best taper equation was selected from adjusted coefficient of determination, mean square error, and aic and bic parameters. finally, the validation of evaluations was carried out with the leave-one-out jackknife method. results: genotype, irrigation regime, or the interaction were not statistically significant for all evaluated taper volume equations and a generalised model equation was obtained. the best taper equation was kozak (2004) which showed the best fit and adaptation to irregular boles. regarding volume equations, all showed a trend to underestimate volume (total and merchantable) in trees with a volume greater than 0.22 m3. validation of the equations showed reduced bias suggesting that the equations can be used to predict taper and volume regardless of eucalyptus genotype x irrigation regimen combinations. conclusions: our results suggest a negligible or minor effect of irrigation (water resource availability) and genotype (for tested taxas and genotypes) on taper and individual tree volume equations. a generalised taper and volume equation (total and merchantable) may be used for all tested genotypes, regardless of water regime (site water availability). this generalised model would simplify eucalyptus estimates required for stand management and projection. new zealand journal of forestry science valverde et al. new zealand journal of forestry science (2022) 52:15 https://doi.org/10.33494/nzjfs522022x181x e-issn: 1179-5395 published on-line: 25/05/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access 2010; arias-aguilar et al. 2020). taper equations have been developed for estimating individual tree diameter at different heights and have been based on simple independent variables such as total height, heights of interest, and diameter at breast height (dbh, at introduction the development of equations that describe the shape of a tree bole is essential for estimating wood volume, carbon sequestration and genetic selection of varieties with the best form for industry needs (vallejos et al. keywords: non-linear equations, model, water availability, allometrics, tree improvement. http://creativecommons.org/licenses/by/4.0/), valverde et al. new zealand journal of forestry science (2022) 52:15 page 2 1.3 m), allowing us to mathematically represent the shape of a tree bole (husch et al. 1993; nogueira et al. 2008). furthermore, taper equations may to comprise linear and non-linear models (goodwin 2009). linear models are characterised by their simplicity in application but lack precision (garber & maguire 2003). on the other hand, non-linear models have been widely implemented because they can be adapted to species with irregular bole shapes, and their mathematical relationships improve model precision with unbiased parameters that increase accuracy (mctague & weiskittel 2021). the development of taper and volume equations is highly relevant for eucalyptus given their large scale of forest plantations and industrial use worldwide (muhairwe 1999; son et al. 2009). the species present wide climatic adaptability (booth 2013) and intensive cultivation of eucalyptus in plantation forestry allows high timber and biomass yields for commercial purposes (lizarralde et al. 2008). these characteristics of the species have led to the development of intensive tree improvement programs that seek continuous increases in productivity of volume or biomass (hall et al. 2020), wood properties (hung et al. 2015), and resistance to pests (brennan et al. 2001). genetic improvement programs consider that a cylindrical shape is a valuable individual trait for the robustness of volume estimates (miguel et al. 2011) and sawtimber yield production. shiver and brister (1992) recommend for eucalyptus saligna the use of the equations of ormerod (1973) and max and burkhart (1976) for accuracy and ease of use. instead, osler et al. (1996) developed several studies with eucalyptus regnans showing that it is possible to use non-linear equations for taper analysis, and the kozak et al. (1969) non-linear equation successfully presented the best adjustment for juvenile trees. son et al. (2009) found that, for eucalyptus pellita, the kozak (2004) equation produced the best taper model fits (r2>0.90), generating individual tree volume estimates with no significant differences from destructive analyses. studies developed by souza et al. (2018), with three 10-yearold eucalyptus clonal varieties, showed that non-linear equations had the best fit for the bole taper profile (r2>0.88), with the kozak (2004) equation had the best performance for all varieties. interestingly, taper and volume models evaluated for genotypes exposed to different soil water availability regimes in the same site are scarce, and existing studies have mainly focused on coniferous species of the genus pinus (li & weiskittel 2010; lu et al. 2018) or effects of water availability usually have been investigated more commonly across sites for productivity purposes but not for investigating taper. souza et al. (2018) report that taper and volume equations must be genotype-specific since they vary with genotype. in contrast, scolforo et al. (2019) determined that it is possible to fit generalised equations for eucalyptus regardless of clone. therefore, the objective of our study was to evaluate the effect of genotype, irrigation regime and genotype x irrigation regime interaction on taper and volume equations (total and merchantable) for highly genetically improved eucalyptus genotypes, including e. globulus and e. nitens x globulus hybrids of high and low productivity and one of each e. nitens, e. badjensis, e. smithii genotypes and one e. camaldulensis x globulus hybrid. methods study site, genotypes and irrigation treatments the study was developed at a nursery facility located in the bio-bio region of chile close to yumbel town (37˚8´0.01´´ s, 72˚ 27´34.70´´ w) (figure 1). the site figure 1: location of the study site. has an average annual temperature of 13.8˚c, with a yearly rainfall of 1252 mm. the topography is flat, soils are classified as dystric xeropsamments (ciren 1999), and the previous land use was a pinus radiata d.don nursery hedge area. the site was planted from july to august 2013 after subsoiling to 80 cm deep. the site was established with a factorial design with three replicates; the first factor was the water regime (high irrigation vs. a low irrigation treatment). the second factor consisted of the genotypes (30 top-ranking selected from cmpc and arauco genetic improvement programs). finally, the combination of factors within each replica was randomly distributed according to rubilar et al. (2020) (30 genotypes x 2 irrigation treatments x 3 replicates). trees were planted at a 3 x 2 m (1666 trees ha-1) spacing, and genotype plots consisted of 5 x 5 trees, with an internal measurement plot of 3 x 3 trees. a summary of annual rainfall from a weather station located at the site and annual additions from each irrigation treatment before first harvesting sampling at the site are presented in table 1. a complete description of the site and silvicultural treatments are described in rubilar et al. (2020). to fulfil the purpose of our study, considering budget and operational limitations, only a subset of eight genotypes in both irrigation treatments from the 30 available genotypes initially established in the experiment were sampled and considered in our study. the final selected genotypes included two eucalyptus globulus (high yield-egh vs low yield egl), two e. nitens x globulus hybrids (high yield engh vs low yield engl), and one of each e. nitens (en), e. camedulensis x globulus (ecg), e. badjensis (eb) and e. smithii (es) genotypes. cumulative stand growth at age 7 (march 2020) for each selected genotype is presented in figure 2. genotypes selection was based on their operational use and high level of productivity, as detailed in rubilar et al. (2020). individual tree sampling individual tree sampling was carried out in january 2020 with three trees per genotype and irrigation treatment selected and 2021 when one additional tree per genotype and irrigation treatment was selected. individual trees were selected to represent the diameter distribution of each selected genotype under each irrigation treatment valverde et al. new zealand journal of forestry science (2022) 52:15 page 3 table 1: description of the study sites year rainfall (mm year-1) low irrigation (mm year-1) high irrigation (mm year-1) 2014 1302 18 55 2015 1102 83 384 2016 782 195 552 2017 972 50 837 2018 1162 68 295 2019 833 97 163 table 1: annual rainfall and irrigation regime treatments water additions in mm year -1 at the experimental site from 2014 to 2019. figure 2: cumulative mean stand volume of selected eight genotypes for each irrigation treatment before first sampling. (4 trees x 8 genotypes x 2 irrigation treatments = 64 trees). individual tree diameter was measured at 1.3 m above ground level (dbh) for each sampled tree before harvesting. subsequently, each tree was cut as close as possible to the ground level, and all its branches were removed, and diameter was measured from the base of the tree and every two metres increments along the bole up to its maximum height until a minimum diameter of 5 cm. analysis of cumulative dbh, height and volume for selected genotypes and irrigation treatments an initial analysis of variance (anova) was carried out to determine if there were significant differences among genotypes and irrigation regimes for dbh, total height and total and merchantable volume (estimated with the smalian formula, see detail in the next section). in addition, anova analyses were carried out using a tukey test to test for differences among treatments for each variable. the analyses were conducted in r software version 4.1.1 (r core team 2021). a bole profile analysis was conducted for each genotype x irrigation regime interaction, for which the relative diameter (d/d, is the ratio between any particular diameter at a specific height and dbh) and relative height (h/h, is the ratio between any specific height and total height). analysis of the d/d and h/h relationship were made according to what has been proposed by li and weiskittel (2010) using origin pro 2020 software (originlab 2021). valverde et al. new zealand journal of forestry science (2022) 52:15 page 4 taper equations the methodology proposed by scolforo et al. (2008) was used to estimate coefficients of the taper and volume equations using the nlme package version 3.1-153 developed by pinheiro et al. (2016) and implemented for linear and non-linear mixed effects models. in addition, the first-order continuous autocorrelation function (car1) and the power variance function were used to eliminate the total within-bole correlation and heteroskedasticity effects. the compatibility between volume and taper equations (it was carried out with the best taper equation) used a system of independent equations with simultaneous estimation of parameters by seemingly unrelated regression (sur), according to the methodology of diéguez-aranda et al. (2006) and zhao et al. (2019), the analysis used systemfit package version 1.1-24 (henningsen et al. 2019). all analyses were conducted in r. for all analyses, four non-linear taper equations were evaluated considering previous studies by son et al. (2009), hall et al. (2020) and hirigoyen et al. (2021) for eucalyptus (table 2) that used the single equation of ormerod (1973) and kozak et al. (1969), the segmented equations of max and burkhart (1976) and the variable form equation of kozak (2004). for total and merchantable volume estimation, the schumacher and hall (1933) equation was implemented (eq. 1), given that this equation has been widely used with eucalyptus in previous studies (trincado & reference equation kozak (2004) kozak et al. (1969) max & burkhart (1976) ormerod (1973) where: d is the diameter to be estimated (cm); h is the reference height (m); h is the total height of the tree (m); dbh is the diameter at 1.3 m above the ground (cm); hst is the stump height (m); α0, α1, β0, β1, β2, β3 are the parameters to be estimated. table 2: evaluated taper equations. burkhart 2006; de souza vismara et al. 2016). according to scolforo et al. (2019), a minimum merchantable diameter of 6 cm is adequate for merchantable volume estimation. (1) where: v is the total or merchantable volume per individual tree (m3 tree-1), dbh is the diameter at 1.3 m (cm), h is the total height or merchantable height (m) and β0, β1 and β2 are the parameters to estimate. after model adjustment, a 1:1 relationship between predicted and observed data was observed to evaluate potential under or overestimate individual tree volume (total and merchantable) estimates. the distribution of residuals was also analysed to evaluate their uniformity and bias. evaluation of genotype, irrigation regime and interaction effects indicator variables were used in each taper and volume equation to evaluate whether there was an effect of the genotype x irrigation regime interaction (scenario 1), genotype effect (scenario 2) or irrigation regime effect (scenario 3) and each scenario was analysed independently. indicator variables were implemented according to quiñonez-barraza et al. (2014) where ij=1 if factor=j; 0 otherwise, and where ij represents each factor analysed. for scenario 1 (genotype x irrigation regime interaction); j=2 for engh-high, j=3 for egl-high, j=4 for engl-high, j=5 for en-high, j= 6 for eb-high and j= 7 for es-high, j=8 for egh-low, j=9 for engh-low, j=10 for egl-low, j=11 for engl-low, j=11 for en-low, j= 12 for eb-low and j= 13 for es-low (egh-high was the reference). for scenario 2 (genotype); j=2 for engh, j=3 for egl, j=4 for engl, j=5 for en, j= 6 for eb and j= 7 for es (enh was the reference). finally, scenario 3 (irrigation regime); j= 2 for low (high was the reference). the model parameters were rewritten based on indicator variables, so that αi and βi could be represented as αi = αi1 + αi2i2 +… + αinin and βi = βi1 + βi2i2 +… + binin. each full model with indicator variables only comprised the significant parameters different from zero at a significance level of 5% (α = 0.05). in order to assess the genotype x irrigation regime interaction, the genotype or irrigation regime effect significantly affect the taper and volume equations; the likelihood ratio test (lrt) (eq. 2) was used to test the full versus reduced equation. (2) where: lrt is likelihood ratio test, la is maximum likelihood of la (equation of each scenario) and lb is likelihood of lb (reduced equation). the test was performed using a mixed chi-square distribution of (n-1), where n is genotype x irrigation regime x repetitions. the null hypothesis analysed that there are no differences between the reduced equation model and the equation model with the indicator variable (each evaluated scenario). selection of best taper equation to select the best taper equation, we used the approach proposed by scolforo et al. (2018) and hirigoyen et al. (2021), considering: the adjusted coefficient of determination (adj-r2) (eq. 3), the root mean squared error (rmse) (eq. 4), the akaike information criterion (aic) (eq. 5) and the bayesian information criterion (bic) (eq. 6). additionally, final equations were ranked using the methodology proposed by hirigoyen et al. (2021) in which the results of the variables adj-r2, rmse, aic and bic were used. (3) (4) (5) (6) where, n denotes the number of observations; p is the number of independent regressors; r2 is the coefficient of determination; yi is an observed value of diameter. ŷi is a predicted value of diameter; m is the maximum likelihood; k is the number of independently adjusted parameters within the equation and w is the free parameters to be estimated. validation of taper and volume equations since it was impossible to obtain an independent validation data set, the leave-one-out jackknife method was used to test equations (yang & kung 1983; rodríguez et al. 2013). the following criteria were evaluated: mean bias error (bias) (eq. 7), percentage mean bias error (bias%) (eq. 8), standard error of the estimate (see) (eq.9) and percentage standard error of the estimate (see%) (eq.10). also, the bias variation at different heights was analysed in the best taper equation; 10 relative height classes were created (ej. 0-10, 10-20, ... 90-10) that grouped all the trees. finally, with the total and merchantable volume equations, the bias variation analysis was performed according to dbh, for which 10 dbh classes were used (ej. 8, 10, … 26), that aggregated all the observations. (7) valverde et al. new zealand journal of forestry science (2022) 52:15 page 5 (8) (9) (10) where, yi is an observed value of diameter or bole volume. ŷi is a predicted value of diameter or bole volume. ȳ is the mean of the observed values of diameter or volume. n is the number of observations, k is the number of parameters in the equation. bias is the mean bias error and see is the standard error of the estimate . results individual tree characteristics for selected genotypes under irrigation treatments mean cumulative growth estimates of selected trees for each genotype are presented in table 3. the dbh data ranged from 11.97 to 22.17 cm and five genotypes (eb, en, engh, egh and engl) did not show statistical dbh differences by irrigation regime. contrastingly, es, ecg and ecg genotypes showed smaller dbh at the low irrigation regime. for h, values ranged from 14.15 to 20.70 m, and only es, ecg, egl and engl genotypes showed a significant decrease in height for the low irrigation regime treatment. finally, total and commercial volume values ranged from 0.074 to 0.334 m3 tree-1 for the high irrigation and from 0.029 to 0.223 m3 tree-1, for the low irrigation regime. all genotypes showed decreased individual tree volume under low irrigation, being engl and egh those that showed the largest response to irrigation. regarding the bole form (figure 3), the same inverse relationship between d/d and h/h was determined for the eight genotypes and irrigation regime treatments. these relationships showed wide variation in-ground line diameter of the evaluated trees (h/h close to zero), and d/d variability decreased as h/h variability increased in both irrigation regimes for the four evaluated equations. effects of genotype and irrigation on taper and volume equations no significant effects of genotype x irrigation regimes interaction, genotypes and irrigation regimes on taper equations were found (table 4). the lrt test showed p-values greater than 0.22 for the four evaluated equations, and simplified reduced equations independent of genotype and/or irrigation regime are viable. the same results were observed for total and commercial volume equations (table 4). therefore, using a generalised equation is optimal since the variables analysed did not generate a gain in accuracy. valverde et al. new zealand journal of forestry science (2022) 52:15 page 6 irrigation genotype dbh (cm) h (m) tv (m3 tree-1) mv (m3 tree-1) high eb 19.35ab (1.95) 18.95a (0.28) 0.282b (0.018) 0.185b (0.024) ecg 16.05b(1.84) 15.84b (0.13) 0.139d (0.082) 0.082d (0.068) egh 16.85b(1.45) 19.45a (0.21) 0.210bc (0.012) 0.132bc (0.073) egl 12.82c(1.40) 15.40b (0.09) 0.091e (0.044) 0.044e (0.022) en 22.17a (2.70) 20.70a (0.33) 0.334a (0.022) 0.223a (0.012) engh 19.45ab (1.57) 18.57a (0.24) 0.245b (0.015) 0.179b (0.089) engl 17.50b (1.52) 16.52b (0.17) 0.179b (0.012) 0.142b (0.092) es 19.27ab (1.97) 17.97ab (0.24) 0.241b (0.015) 0.155b (0.084) average 17.93 (1.92) 17.92 (0.21) 0.215 (0.013) 0.137 (0.067) low eb 18.55ab (1.06) 18.06a (0.20) 0.201bc (0.033) 0.133bc (0.097) ecg 14.70c(1.80) 15.80c (0.11) 0.117d (0.063) 0.063d (0.050) egh 16.97b(1.77) 16.77ab (0.16) 0.164c (0.060) 0.105c (0.045) egl 11.97c(1.15) 14.15c (0.07) 0.074e (0.089) 0.029e (0.064) en 21.37a (1.38) 19.38a (0.33) 0.337a (0.095) 0.225a (0.077) engh 21.55a (1.81) 18.81a (0.32) 0.326a (0.069) 0.219a (0.056) engl 15.62c(1.85) 15.85c (0.14) 0.141d (0.085) 0.085d (0.063) es 17.6b(1.93) 15.93c (0.19) 0.195b (0.012) 0.125 b (0.056) average 17.29 (1.84) 16.84 (0.19) 0.194 (0.012) 0.123 (0.080) table 3: mean of diameter (dbh), total height (h), total volume (tv) and merchantable volume (mt) characterisation of the eight selected eucalyptus genotypes in contrasting irrigation regimes (standard deviation for each parameter in parenthesis; different letters indicate significant differences at 0.05). taper equation selection all adjusted parameters of the four equations analysed were significant (table 5). the equation with the best fit was kozak’s (2004) model with the lowest rmse, aic and bic in comparison to other models, and presented the best accuracy and smallest residual distribution range (±1.4%) (figure 4a). the second-best equation was kozak et al.’s (1969) model, that showed a good adj-r2 (0.960) and rmse, aic, and bic values slightly higher than those of kozak’s (2004) model. however, its residual distribution (figure 4b) showed a larger dispersion (±4%) but still maintained a uniform distribution. the ormerod (1973) equation showed intermediate rmse, aic and bic values suggesting a lower quality fit compared to kozak’s (2004) and kozak et al.’s (1969) models and a broader but uniform residual distribution range compared to kozak (2004) or kozak et al. (1969) models (figure 4c). finally, the max and burkhart (1976) equation presented the poorest fit from all models with the lowest adj-r2 values and the highest rmse, aic, bic, and mad estimates. also, its residual distributions showed the highest heteroscedasticity of all four equations (figure 4d). valverde et al. new zealand journal of forestry science (2022) 52:15 page 7 figure 3: relationship between relative diameter ratio (d/d) and relative height ratio (h/h) for all evaluated eucalyptus genotypes under both irrigated treatments. equation variable genotype x irrigation genotype irrigation lrt p lrt p lrt p taper kozak (2004) 2.02 0.09 ns 1.33 0.12 ns 0.44 0.33 ns kozak et al. (1969) 1.33 0.12 ns 0.45 0.33 ns 0.32 0.25 ns max and burkhart (1976) 0.34 0.25 ns 0.22 0.50 ns 0.24 0.52 ns sharmar and oderwald (2001) 0.45 0.33 ns 0.22 0.50 ns 0.23 0.52 ns ormerod (1973) 1.46 0.10 ns 0.43 0.35 ns 0.30 0.27 ns volume total 1.94 0.10 ns 1.10 0.20 ns 0.40 0.41 ns merchantable 1.90 0.11 ns 0.99 0.36 ns 0.38 0.40 ns table 4: likelihood ratio test (lrt) and their respective p-value (p) for genotype, irrigation regime and genotype x irrigation regime interaction effects on taper and volume equations. note: ns not significant, * significant at 0.05. volume equations for total and merchantable volume equations all coefficients of the general model equations were significant (p-value<0.001) and an estimated error (se) less than 0.023 (table 6). the equations showed good fits, adj-r2 estimates greater than 0.98, and low rmse, aic and bic values. when analyzing our generalised equation against smalian estimates for total volume (figure 5a), an underestimation of 3 to 8% was observed in trees with individual volumes ranging from 0.25 to 0.32 m3 tree-1, and underestimation increased as the volume of the tree increased. residuals distribution (figure 5b) showed uniformity and its variation was less than 0.3%, indicating a good accuracy level. similar to total volume, the merchantable volume equation (figure 5c) showed a slight tendency to underestimate volume in trees with a merchantable volume greater than 0.22 m3 tree-1 and the same trend to increase underestimation as the size of the tree increased was observed but reached a maximum of 5% of valverde et al. new zealand journal of forestry science (2022) 52:15 page 8 equation parameter se p-value adj-r2 rmse aic bic ranking kozak (2004) α0 α1 α2 β0 β1 β2 β3 β4 β5 1.032 0.980 0.010 0.312 -0.722 0.745 2.533 0.052 -0.589 0.033 0.009 0.001 0.002 0.010 0.012 0.040 0.003 0.004 0.001* 0.010* 0.003* 0.007* 0.001* 0.006* 0.001* 0.009* 0.001* 0.986 0.877 109.500 90.322 1 kozak et al. (1969) β0 β1 -2.092 0.820 0.250 0.076 0.005* 0.003* 0.960 1.040 120.437 109.101 2 ormerod (1973) β0 0.582 0.045 0.001* 0.936 1.149 123.809 112.233 3 max and burkhart (1976) α1 α2 β0 β1 β2 β3 0.822 0.285 -0.793 -1.066 1.540 0.361 0.012 0.034 0.089 0.233 0.345 0.098 0.001* 0.002* 0.003* 0.004* 0.002* 0.001* 0.913 2.800 130.711 133.010 4 table 5: adjusted coefficients and statistical criteria values for selected taper equations considering all eucalyptus genotypes and irrigation regimes. note: ns not significant, * significant at 0.05 figure 4: residual plots for the generalised model equations adjusted across genotypes and irrigation regime treatments. underestimation. on the other hand, the distribution of the residuals (figure 5d) was homogeneous and showed less than 0.15% variation indicating high accuracy for this equation. validation of equations validation with the leave-one-out jackknife method (table 7) showed that the selected general equations were valid for predicting the taper and volume of eucalyptus boles. kozak’s (2004) taper equation showed high flexibility with a negative bias of -0.081 cm (bias% of -0.692%), the see% was less than 4.5% (see of 0.509 cm), showing good precision in data estimation. regarding bias variation in relative height classes (figure 6a), showed a uniform bias in classes from 0 to 70%, with a mean value of -0.007 cm. in classes above 70% (relative height), the bias increased with a negative trend, with an average value of 0.210 cm. similar results were obtained regarding the volume equations (total and commercial) (table 7). again, the bias showed negative values, less than 5.0 x10-3 m3 tree-1, with an average bias% of -3.27%, showing good accuracy for volume estimation. on the other hand, see was less than 8.0 x10-3 m3 tree-1 (average see% 5.23%), showing good precision for estimating volume per tree. according to dbh classes, the bias variation analysis had a uniform bias between 8 and 22 cm (average 1.57 x10-3 m3 tree-1) for total volume (figure 6b), and trees with a dbh ≥22 cm showed a negative increase in bias on average of -1.83 x10-2 m3 tree-1. finally, the merchantable volume equation showed excellent accuracy in dbh classes (figure 6c); with an average bias of -1.25 x 10-3 m3 tree-1 in classes from 8 to 18 cm, and it was in trees with dbh ≥20 cm, that bias increased with an average of -1.50 x 10-2 m3 tree-1. valverde et al. new zealand journal of forestry science (2022) 52:15 page 9 volume equation parameter se p-value adj-r2 rmse aic bic total β0 β1 β2 2.750 x10-5 2.082 0.974 >0.001 0.023 0.009 >0.001* 0.002* 0.001* 0.980 0.020 132.789 138.901 merchantable β0 β1 β2 3.912x10-5 1.71 1.164 >0.001 0.019 0.008 >0.001* 0.002* 0.001* 0.982 0.017 129.444 132.562 table 6: adjusted coefficients and statistical criteria values for total and merchantable volume equations considering all eucalyptus genotypes and irrigation regimes. note: ns not significant, * significant at 0.05 figure 5: (a) (c) predicted (ve) versus observed (vo) individual tree total (t) and merchantable (m) volume. (b) (d) plot of residues distribution against individual tree total (t) and merchantable (m) volume predicted values considering a generalised model considering all genotypes and irrigation regime treatments. valverde et al. new zealand journal of forestry science (2022) 52:15 page 10 discussion effects of irrigation regime and genotype on taper and volume equations developing individual tree taper, total and merchantable volume equations is essential to estimate and make productivity projections (li et al. 2017) and optimise a forest crop’s growth ((scolforo et al. 2019). in our study, a generalised taper and volume equations was obtained in which the effect of the genotype, water regime and interaction of both variables was considered. our results indicated that none of these effects affected taper and individual tree volume equations (table 4), similar to what has been reported before by gomat et al. (2011) and scolforo et al. (2019) in which a single generalised model equation could be used for eucalyptus regardless of genotype and climatic environment. gomat et al. (2011) and scolforo et al. (2019) highlighted that irrigation regime may affect tree growth, but they found no evidence of these effects on individual tree bole shape. binkley et al. (2017) showed that temperature and precipitation variations directly affected growth rate and transpiration but not bole profile shape for several eucalyptus clones across a large climatic gradient. the plasticity of eucalyptus growing in different water availability environments affects their productivity but does not change their individual tree shape (hill & hollender 2019). studies developed by souza et al. (2016) and cerqueira et al. (2021) determined that taper is mainly affected by variables such as competition for space, severe water stress, or aspects associated with the spatial location of cultivation, but broad climatic factors are not significant. however, as scolforo et al. (2018) suggested, excluding climatic variables, such as water regime, does not mean that it may not add precision to taper-volume equations for different species. in the case of genotypes from advanced tree improvement programs, clonal material is selected to maximise productivity and other desirable characteristics, whereas taper variability is usually deployed by selecting cylindrical trees of maximum individual tree volume to optimise final harvest (vallejos et al. 2010). this may explain why the genotype effect was not significant in our study and in practice may be omitted from taper and volume equations for a broad range of genotypes for a single species (scolforo et al. 2019). interestingly, our study found similar results, even for a broad range of taxa tested at this site. equation statistical criteria value taper kozak (2004) bias (cm) -0.081 bias% -0.689 see (cm) 0.509 see% 4.351 total volume bias (10-3 m3 tree-1) -4.402 bias% -2.268 see (10-3 m3 tree-1) 8.371 see% 4.31 merchantable volume bias (10-3 m3 tree-1) -5.250 bias% -4.268 see (10-3 m3 tree-1) 7.557 see% 6.144 table 7: statistical criteria (bias, bias%, see and see%) obtained to validate taper and volume (total and merchantable) equations with the leaveone-out jackknife method. figure 6: bias variation obtained in each class analysed (relative height and dbh) in the validation of: (a) taper; (b) total; and (c) merchantable volume equations. the bars represent the 95% confidence intervals. valverde et al. new zealand journal of forestry science (2022) 52:15 page 11 selected taper equation kozak’s (2004) equation showed the best fit in our study and many authors have considered it the optimal equation. this result is similar to previous studies carried out in different eucalyptus species such as son et al. (2009), scolforo et al. (2018) and scolforo et al. (2019), in which it was suggested that the kozak’s (2004) equation showed the best response to adaptation to the shape of eucalyptus boles. kozak’s (2004) equation has the advantage of describing the lower part of the shape of the bole as a neiloid, the middle as a paraboloid and the upper part as a cone. the equation considers a bole shape transition that responds to most species that have been analysed, providing flexibility and reduced error (rojo et al. 2005), aspects that have allowed its use on a large number of coniferous and broadleaf species (li & weiskittel 2010). the model allows a simple adjustment and may provide representative generalised models for taxa or geographic regions (son et al. 2009). kozak et al.’s (1969) and ormerod’s (1973) equations, which are considered simple models, showed lower quality of fit compared to kozak’s (2004) equation. in addition, the simplicity of these equations does not provide a better representation of the shape of the bole generating over or underestimates (souza et al. 2018). muhairware (1999) indicated that simple models provide ease of estimation, are algebraically integrable but provide serious limitations for species with irregular shapes or that show transitions along the bole (de andrade 2014), an aspect that caused the adjustment to be lower, increased variation of residuals (figure 4), and final equations were not considered in our study. finally, max & burkhart’s (1976) equation showed the poorest adjustment and most considerable bias of all the evaluated equations. although it is a segmented model that shows biologically consistent behavior since it makes the transition from a neiloid to a paraboloid, in addition to having an algebraic simplicity of use (mctague & weiskittel 2021); however, it had significant deficiencies in the explanation of transition in the bole shape by generating a non-continuous model (mctague & weiskittel 2021; salekin et al. 2021). our results suggest that it is less functional than other models such as kozak’s (2004) model, providing the poorest adaptation to model bole shape, and it was not considered for providing a valuable final model. study considerations and limitations the generalisation of equations that accurately describe the shapes of trees in different irrigation regimes and with different genotypes simplifies forest management, productivity projection, and decision-making (miguel et al. 2011; da silva menezes et al. 2020), and our model contributes to this. however, two elements must be considered: firstly, the plantation experiment comprised of middle-aged trees, comparable to studies such as gomat et al. (2011) and campos et al. (2014), where eucalyptus after three years of age had the same bole shape although the canopy had not completely closed. changes in stocking may affect bole shape and proposed equations need to be used with caution for more advanced stages of plantation with or without silvicultural treatments applied. the second aspect to consider is the validation of our equations. unfortunately, due to logistical and budget constraints and the availability of the same genotypes in a single experiment, it was impossible to validate our equations with an independent data set. therefore, it was decided to use the leave-one-out jackknife validation method, ideal for analyzing small-sized samples and avoiding overestimating the bias and standard error in the equations (pal 2017). furthermore, it is a safe validation method in volume and taper equations (yang & kung 1983; rodríguez et al. 2013), showing a greater accuracy gain if used in equations that have been fitted with mixed models that reduce the error of estimating the coefficients of the equation compared to other methods (trincado & burkhart 2006). very few studies evaluate taper and volume in large sets of genotypes grown in different water conditions, so having this information provides a first step for modeling the species in the study region. in addition, previous studies by benbrahim and gavaland (2003) have shown that taper studies without independent validation data sets are viable when seeking to understand new silvicultural conditions and species, such as was the case of our study. conclusions no statistically significant effects of irrigation regime, genotype, and interactions of genotype with irrigation regime were found for any of the individual tree taper, total and merchantable volume equations evaluated; therefore, the use of a generalised equation regardless of the taxa or water regime may provide reliable estimates across the evaluated genotypes under contrasting water availability conditions. the kozak (2004) equation showed the best performance of all evaluated models and equations of total and commercial volume showed a slight underestimation of individual tree volume for larger trees. the use of a generalised equation for taper and total and commercial volume, regardless of taxa or water regime, would simplify forest modelling, management and estimates of eucalyptus plantation productivity. competing interests the authors have no competing interests. acknowledgements we gratefully acknowledge the support of many professionals from cmpc s.a. (forestal mininco) who provided invaluable support on field installations, maintenance and establishment of irrigation syboles. funding this work was funded by the government of chile via conicyt fondecyt regular project 1190835, conicyt fondef project it16i10087 and anid basal fb210015, also funding for maintenance of this trials was provided by cmpc forestal mininco s.a., the forest productivity cooperative at universidad de concepción chile and support of national agency for research and development (anid), scholarship program, doctorado becas chile/2020-21202023. author contributions all the authors contributed to all aspects of the study and manuscript preparation. references arias-aguilar, d., valverde, j.c., & campos, r. (2020). effect of planting density and tree species selection on forest bioenergy systems: tree growth, nutrient storage and wood chemical properties. greenhouse gases: science and technology, 10(6), 1165-1175. https://doi.org/10.1002/ghg.2008 benbrahim, m., & gavaland, a. (2003). a new stem taper function for short-rotation poplar. scandinavian journal of forest research, 18(4), 377-383. https:// doi.org/10.1080/02827580310005171 binkley, d., campoe, o.c., alvares, c., carneiro, r.l., cegatta, í., & stape, j.l. (2017). the interactions of climate, spacing and genetics on clonal eucalyptus plantations across brazil and uruguay. forest ecology and management, 405, 271-283. https:// doi.org/10.1016/j.foreco.2017.09.050 booth, t.h. (2013). eucalypt plantations and climate change. forest ecology and management, 301, 2834. https://doi.org/10.1016/j.foreco.2012.04.004 brennan, e.b., hrusa, g.f., weinbaum, s.a., & levison jr, w. (2001). resistance of eucalyptus species to glycaspis brimblecombei (homoptera: psyllidae) in the san francisco bay area. pan-pacific entomologist, 77(4), 249-253. campos, b., binoti, d., silva, m., leite, h , & binoti, m. (2014). efeito do modelo de afilamento utilizado sobre a conversão de fustes de árvores em multiprodutos. scientia forestalis, 42(104), 513-520. cerqueira, c.l., môra, r., tonini, h., arce, j.e., carvalho, s., & vendruscolo, d.g.s. (2021). modeling of eucalyptus tree stem taper in mixed production systems. embrapa pecuária sul-artigo em periódico indexado (alice). https://doi.org/10.18671/ scifor.v49n130.22 ciren. (1999). estudio agrológico viii región. descripciones de suelos, materiales y símbolos. santiago, chile. da silva menezes, l., figueredo, j.b.l., costa, l.s., castro, r.v.o., & júnior, c.a.a. (2020). taper modeling and economic evaluation of multi-products obtained from wood of short-rotation eucalyptus stands. floresta, 50(3), 1439-1448. https://doi. org/10.5380/rf.v50i3.60513 de andrade, v.c.l. (2014). modelos de taper do tipo expoente-forma para descrever o perfil do fuste de árvores. pesquisa florestal brasileira, 34(80), 271283. https://doi.org/10.4336/2014.pfb.34.80.614 de souza vismara, e., mehtätalo, l., & batista, j.l.f. (2016). linear mixed-effects models and calibration applied to volume models in two rotations of eucalyptus grandis plantations. canadian journal of forest research, 46(1), 132-141. https://doi. org/10.1139/cjfr-2014-0435 diéguez-aranda, u., castedo-dorado, f., álvarezgonzález, j.g., & rojo, a. (2006). compatible taper function for scots pine plantations in northwestern spain. canadian journal of forest research, 36(5), 1190-1205. https://doi.org/10.1139/x06-008 garber, s.m., & maguire, d.a. (2003). modeling stem taper of three central oregon species using nonlinear mixed effects models and autoregressive error structures. forest ecology and management, 179(13), 507-522. https://doi.org/10.1016/s03781127(02)00528-5 gomat, h.y., deleporte, p., moukini, r., mialounguila, g., ognouabi, n., saya, a.r., vigneron, p., & saint-andre, l. (2011). what factors influence the stem taper of eucalyptus: growth, environmental conditions, or genetics? annals of forest science, 68(1), 109-120. https://doi.org/10.1007/s13595-011-0012-3 goodwin, a.n. (2009). a cubic tree taper model. australian forestry, 72(2), 87-98. https://doi.org/ 10.1080/00049158.2009.10676294 hall, k.b., stape, j., bullock, b.p., frederick, d., wright, j., scolforo, h.f., & cook, r. (2020). a growth and yield model for eucalyptus benthamii in the southeastern united states. forest science, 66(1), 25-37. https:// doi.org/10.1093/forsci/fxz061 henningsen, a., hamann, j.d., & henningsen, m.a. (2019). package ‘systemfit’. hill, j.l., & hollender, c.a. (2019). branching out: new insights into the genetic regulation of shoot architecture in trees. current opinion in plant biology, 47, 73-80. https://doi.org/10.1016/j. pbi.2018.09.010 hirigoyen, a., navarro-cerrillo, r., bagnara, m., franco, j., requin, f., & rachid-casnati, c. (2021). modelling taper and stem volume considering stand density in eucalyptus grandis and eucalyptus dunnii. iforest-biogeosciences and forestry, 14(2), 127. https://doi.org/10.3832/ifor3604-014 hung, t.d., brawner, j.t., meder, r., lee, d.j., southerton, s., thinh, h.h., & dieters, m.j. (2015). estimates of genetic parameters for growth and wood properties in eucalyptus pellita f. muell. to support tree breeding in vietnam. annals of forest science, 72(2), 205-217. https://doi.org/10.1007/s13595014-0426-9 valverde et al. new zealand journal of forestry science (2022) 52:15 page 12 https://doi.org/10.1002/ghg.2008 https://doi.org/10.1080/02827580310005171 https://doi.org/10.1080/02827580310005171 https://doi.org/10.1016/j.foreco.2017.09.050 https://doi.org/10.1016/j.foreco.2017.09.050 https://doi.org/10.1016/j.foreco.2012.04.004 https://doi.org/10.18671/scifor.v49n130.22 https://doi.org/10.18671/scifor.v49n130.22 https://doi.org/10.5380/rf.v50i3.60513 https://doi.org/10.5380/rf.v50i3.60513 https://doi.org/10.4336/2014.pfb.34.80.614 https://doi.org/10.1139/cjfr-2014-0435 https://doi.org/10.1139/cjfr-2014-0435 https://doi.org/10.1139/x06-008 https://doi.org/10.1016/s0378-1127(02)00528-5 https://doi.org/10.1016/s0378-1127(02)00528-5 https://doi.org/10.1007/s13595-011-0012-3 https://doi.org/10.1080/00049158.2009.10676294 https://doi.org/10.1080/00049158.2009.10676294 https://doi.org/10.1093/forsci/fxz061 https://doi.org/10.1093/forsci/fxz061 https://doi.org/10.1016/j.pbi.2018.09.010 https://doi.org/10.1016/j.pbi.2018.09.010 https://doi.org/10.3832/ifor3604-014 https://doi.org/10.1007/s13595-014-0426-9 https://doi.org/10.1007/s13595-014-0426-9 husch, b., miller, c., & beers, t. (1993). forest mensuration (3 ed.): malabar: krieger publishing company. kozak, a. (2004). my last words on taper equations. the forestry chronicle, 80(4), 507-515. https://doi. org/10.5558/tfc80507-4 kozak, a., munro, d., & smith, j. (1969). taper functions and their application in forest inventory. the forestry chronicle, 45(4), 278-283. https://doi. org/10.5558/tfc45278-4 li, r., & weiskittel, a.r. (2010). comparison of model forms for estimating stem taper and volume in the primary conifer species of the north american acadian region. annals of forest science, 67(3), 302. https://doi.org/10.1051/forest/2009109 li, y., suontama, m., burdon, r.d., & dungey, h.s. (2017). genotype by environment interactions in forest tree breeding: review of methodology and perspectives on research and application. tree genetics & genomes, 13(3), 60. https://doi. org/10.1007/s11295-017-1144-x lizarralde, i., broto, m., rodríguez, f., & bravo, f. (2008). taper equations and wood products: assessing the carbon flow of the forest through its products managing forest ecosystems: the challenge of climate change (pp. 165-177): springer. https:// doi.org/10.1007/978-1-4020-8343-3_10 lu, k., bi, h., watt, d., strandgard, m., & li, y. (2018). reconstructing the size of individual trees using log data from cut-to-length harvesters in pinus radiata plantations: a case study in nsw, australia. journal of forestry research, 29(1), 13-33. https:// doi.org/10.1007/s11676-017-0517-1 max, t.a., & burkhart, h.e. (1976). segmented polynomial regression applied to taper equations. forest science, 22(3), 283-289. mctague, j.p., & weiskittel, a. (2021). evolution, history, and use of stem taper equations: a review of their development, application, and implementation. canadian journal of forest research, 51(2), 210235. https://doi.org/10.1139/cjfr-2020-0326 miguel, e.p., do amaral machado, s., figueiredo filho, a., & arce, j.e. (2011). modelos polinomiais para representar o perfil e o volume do fuste de eucalyptus urophylla na região norte do estado de goiás. floresta, 41(2). https://doi.org/10.5380/ rf.v41i2.21883 muhairwe, c.k. (1999). taper equations for eucalyptus pilularis and eucalyptus grandis for the north coast in new south wales, australia. forest ecology and management, 113(2-3), 251-269. https://doi. org/10.1016/s0378-1127(98)00431-9 nogueira, g.s., leite, h.g., reis, g.g., & moreira, a.m. (2008). influência do espaçamento inicial sobre a forma do fuste de árvores de pinus taeda l. revista árvore, 32, 855-860. https://doi.org/10.1590/ s0100-67622008000500010 ormerod, d. (1973). a simple bole model. the forestry chronicle, 49(3), 136-138. https://doi. org/10.5558/tfc49136-3 osler, g., west, p., & downes, g. (1996). effects of bending stress on taper and growth of stems of young eucalyptus regnans trees. trees, 10(4), 239-246. https://doi.org/10.1007/bf02185675 pal, r. (2017). chapter 4 validation methodologies. in r. pal (ed.), predictive modeling of drug sensitivity (pp. 83-107): academic press. https://doi. org/10.1016/b978-0-12-805274-7.00004-x pinheiro, j., bates, d., debroy, s., & sarkar, d. (2016). nlme: linear and non-linear mixed effects models. r package. quiñonez-barraza, g., los santos-posadas, d., héctor, m., álvarez-gonzález, j.g., & velázquez-martínez, a. (2014). sistema compatible de ahusamiento y volumen comercial para las principales especies de pinus en durango, méxico. agrociencia, 48(5), 553-567. rodríguez, f., lizarralde, i., & bravo, f. (2013). additivity on nonlinear stem taper functions: a case for corsican pine in northern spain. forest science, 59(4), 464-471. https://doi.org/10.5849/ forsci.12-023 rojo, a., perales, x., sánchez-rodríguez, f., álvarezgonzález, j., & von gadow, k. (2005). stem taper functions for maritime pine (pinus pinaster ait.) in galicia (northwestern spain). european journal of forest research, 124(3), 177-186. https://doi. org/10.1007/s10342-005-0066-6 rubilar, r., hubbard, r., emhart, v., mardones, o., quiroga, j.j., medina, a., valenzuela, h., espinoza, j., burgos, y., & bozo, d. (2020). climate and water availability impacts on early growth and growth efficiency of eucalyptus genotypes: the importance of gxe interactions. forest ecology and management, 458. doi: 10.1016/j.foreco.2019.117763 https://doi. org/10.1016/j.foreco.2019.117763 salekin, s., catalán, c.h., boczniewicz, d., phiri, d., morgenroth, j., meason, d.f., & mason, e.g. (2021). global tree taper modelling: a review of applications, methods, functions, and their parameters. forests, 12(7), 913. https://doi. org/10.3390/f12070913 schumacher, f., & hall, f. (1933). logarithmic expression of timber-tree volume. journal of agricultural research, 47, 719-734. scolforo, h.f., mctague, j.p., burkhart, h., roise, j., carneiro, r.l., & stape, j.l. (2019). generalized stem taper and tree volume equations applied to eucalyptus of varying genetics in brazil. canadian journal of forest research, 49(5), 447-462. https:// doi.org/10.1139/cjfr-2018-0276 valverde et al. new zealand journal of forestry science (2022) 52:15 page 13 https://doi.org/10.5558/tfc80507-4 https://doi.org/10.5558/tfc80507-4 https://doi.org/10.5558/tfc45278-4 https://doi.org/10.5558/tfc45278-4 https://doi.org/10.1051/forest/2009109 https://doi.org/10.1007/s11295-017-1144-x https://doi.org/10.1007/s11295-017-1144-x https://doi.org/10.1007/978-1-4020-8343-3_10 https://doi.org/10.1007/978-1-4020-8343-3_10 https://doi.org/10.1007/s11676-017-0517-1 https://doi.org/10.1007/s11676-017-0517-1 https://doi.org/10.1139/cjfr-2020-0326 https://doi.org/10.5380/rf.v41i2.21883 https://doi.org/10.5380/rf.v41i2.21883 https://doi.org/10.1016/s0378-1127(98)00431-9 https://doi.org/10.1016/s0378-1127(98)00431-9 https://doi.org/10.1590/s0100-67622008000500010 https://doi.org/10.1590/s0100-67622008000500010 https://doi.org/10.5558/tfc49136-3 https://doi.org/10.5558/tfc49136-3 https://doi.org/10.1007/bf02185675 https://doi.org/10.1016/b978-0-12-805274-7.00004-x https://doi.org/10.1016/b978-0-12-805274-7.00004-x https://doi.org/10.5849/forsci.12-023 https://doi.org/10.5849/forsci.12-023 https://doi.org/10.1007/s10342-005-0066-6 https://doi.org/10.1007/s10342-005-0066-6 https://doi.org/10.1016/j.foreco.2019.117763 https://doi.org/10.1016/j.foreco.2019.117763 https://doi.org/10.3390/f12070913 https://doi.org/10.3390/f12070913 https://doi.org/10.1139/cjfr-2018-0276 https://doi.org/10.1139/cjfr-2018-0276 scolforo, h.f., mctague, j.p., raimundo, m.r., weiskittel, a., carrero, o., & scolforo, j.r.s. (2018). comparison of taper functions applied to eucalypts of varying genetics in brazil: application and evaluation of the penalized mixed spline approach. canadian journal of forest research, 48(5), 568-580. https:// doi.org/10.1139/cjfr-2017-0366 shiver, b., & brister, g. (1992). tree and stand volume functions for eucalyptus saligna. forest ecology and management, 47(1-4), 211-223. https://doi. org/10.1016/0378-1127(92)90275-e son, y.m., kim, h., lee, h.y., kim, c.m., kim, c.s., kim, j.w., joo, r.w., & lee, k.h. (2009). taper equations and stem volume table of eucalyptus pellita and acacia mangium plantations in indonesia. journal of korean society of forest science, 98(6), 633-638. souza, g.s.a.d., cosenza, d.n., araújo, a.c.d.s.c., pimenta, l.v.a., souza, r.b., almeida, f.m., & leite, h.g. (2018). evaluation of non-linear taper equations for predicting the diameter of eucalyptus trees. revista árvore, 42. https://doi.org/10.1590/180690882018000100002 souza, r.r., nogueira, g.s., júnior, l.s.m., pelli, e., de oliveira, m.l.r., abrahão, c.p., & leite, h.g. (2016). forma de fuste de árvores de eucalyptus em plantios com diferentes densidades iniciais stem form of eucalyptus trees in plantations under different initial densities. scientia forestalis, 44(109), 33-40. https://doi.org/10.18671/scifor. v44n109.03 team, r.c. (2021). r: a language and environment for statistical computing. retrieved from: https:// www.r-project.org trincado, g., & burkhart, h.e. (2006). a generalized approach for modeling and localizing stem profile curves. forest science, 52(6), 670-682. vallejos, j., badilla, y., picado, f., & murillo, o. (2010). metodología para la selección e incorporación de árboles plus en programas de mejoramiento genético forestal. agronomía costarricense, 34(1), 105-119. https://doi.org/10.15517/rac. v34i1.6704 yang, y.c., & kung, f.h. (1983). method for estimating bole volume. journal of forestry, 81(4), 224-227. zhao, d., lynch, t.b., westfall, j., coulston, j., kane, m., & adams, d.e. (2019). compatibility, development, and estimation of taper and volume equation systems. forest science, 65(1), 1-13. https://doi. org/10.1093/forsci/fxy036 valverde et al. new zealand journal of forestry science (2022) 52:15 page 14 https://doi.org/10.1139/cjfr-2017-0366 https://doi.org/10.1139/cjfr-2017-0366 https://doi.org/10.1016/0378-1127(92)90275-e https://doi.org/10.1016/0378-1127(92)90275-e https://doi.org/10.1590/1806-90882018000100002 https://doi.org/10.1590/1806-90882018000100002 https://doi.org/10.18671/scifor.v44n109.03 https://doi.org/10.18671/scifor.v44n109.03 https://www.r-project.org https://www.r-project.org https://doi.org/10.15517/rac.v34i1.6704 https://doi.org/10.15517/rac.v34i1.6704 https://doi.org/10.1093/forsci/fxy036 https://doi.org/10.1093/forsci/fxy036 evaluating basic density calibrations based on nir spectra recorded on the three wood faces and subject to different mathematical treatments evelize a. amaral*, luana m. dos santos, paulo r. g. hein, emylle v. s. costa, sebastião carlos s. rosado and paulo f. trugilho department of forestry sciences–wood science and technology, lavras university, lavras, minas gerais, 372000-900, brazil *corresponding author: evelizeamaral@yahoo.com.br (received for publication 19 february 2020; accepted in revised form 4 february 2021) abstract background: near infrared (nir) spectroscopy has been successfully applied to estimate the chemical, physical and mechanical properties of various biological materials, including wood. this study aimed to evaluate basic density calibrations based on nir spectra collected from three wood faces and subject to different mathematical treatments. methods: diffuse reflectance nir spectra were recorded using an integrating sphere on the transverse, radial and tangential surfaces of 278 wood specimens of eucalyptus urophylla x eucalyptus grandis. basic density of the wood specimens was determined in the laboratory by the immersion method and correlated with nir spectra by partial least squares regression. different statistical treatments were then applied to the data, including standard normal variate, multiplicative scatter correction, first and second derivatives, normalization, autoscale and meancenter transformations. results: the predictive model based on nir spectra measured on the transverse surface performed the best (r²cv = 0.85 and rmse = 25.5 kg/m³) while the model developed from the nir spectra measured on the tangential surface had the poorest performance (r²cv = 0.53 and rmse = 46.8 kg/m³). the difference in performance between models based on original (untreated) and mathematically-treated spectra was minimal. conclusions: multivariate models fitted to nir spectra were found to be efficient for predicting the basic density of eucalyptus wood, especially when based on spectra measured on the transversal surface. for this data set, models based on the original spectra and mathematically treated spectra had similar performance. the reported findings show that mathematical transformations are not always able to extract more information from the spectra in the nir. new zealand journal of forestry science amaral et al. new zealand journal of forestry science (2021) 51:2 https://doi.org/10.33494/nzjfs512021x100x e-issn: 1179-5395 published on-line: 17/02/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access process impractical at large scale. near infrared (nir) spectroscopy is one of the non-destructive methods that allows the determination of many wood characteristics, such as, chemical, anatomical and physio-mechanical properties (tsuchikawa & schwanninger 2013; tsuchikawa & kobori 2015). the technique consists of exposing the samples to electromagnetic radiation in the region of the spectra between wavelengths of 12500 cm-1 to 3600 cm-1 to obtain the absorbance spectra (williams and norris 2001). the method has a number of advantages, including: speed (one minute or less introduction the determination of density and its variation within the tree, both in the radial and longitudinal direction is fundamental for understanding wood quality (silva et al. 2015). thus, determining the basic density of wood has become a crucial step in the management routine of forest-based companies. to assess the quality of wood in forest plantations, forestry companies need rapid and cost-effective techniques, since analysing it using conventional methods can consume time and money, which makes the keywords: nir signature; multivariate statistics; wood properties; eucalyptus timber, hardwood. mailto:evelizeamaral@yahoo.com.br http://creativecommons.org/licenses/by/4.0/), amaral et al. new zealand journal of forestry science (2021) 51:2 page 2 per sample); being non-invasive; suitable for use in the production line, allowing online or real-time analysis; requiring minimum sample preparation; and uses relatively simple instruments that can be transported over long distances (muñiz 2012). nir spectra are correlated with material composition or properties determined by standardised methods using multivariate tools in order to generate a predictive model (meder et al. 2010). multivariate regression models developed from nir spectra have been successfully used to estimate wood density across a range of species (schimleck et al. 1999; gindl et al. 2001; schimleck et al. 2005; jiang et al. 2006; jones et al. 2006; mora et al. 2008; hein et al. 2009; hein 2010, arriel et al. 2019). distortions and errors in the predictive models developed from nir spectra can occur, since part of the spectral information may not be correlated with the investigated property. control over experimental procedures and post-processing of the data is required to remove, reduce or standardise irrelevant information and consequently, improve the quality of the signal in the calibration, remove the imperfections present in the original spectra, without changing the information contained in it (naes et al. 2002). therefore, research is required to define the most appropriate way to apply the technique and generate reliable predictions. for example, hein et al. (2010) demonstrated that the chemical properties of eucalyptus urophylla wood are better estimated from the spectra measured in milled wood than in whole (unprocessed) wood. costa et al. (2018) investigated parameter settings of the nir spectrometer and which wood surface is most suitable for measuring spectra and generating models to estimate wood density in eucalyptus. according to sandak et al. (2016), there is still a need to better understand the fundamental issues and the impact that the most commonly utilised methods for pre-processing spectral data have on predictive models for wood properties (and those of other ligno-cellulosic materials). in short, it is not still clear how mathematical transformations affect the spectral variation on different wood surfaces. once nir spectra have been obtained, the aim of this study, therefore, was to evaluate basic density calibrations based on nir spectra recorded on transverse, radial and tangential wood surfaces and subject to different mathematical treatments. common mathematical treatments were applied to the nir spectra collected from eucalyptus wood samples and partial least squares (pls) regressions for estimating wood density were developed and compared. methods origin and sample preparation the specimens used in this study were obtained from a progeny test of eucalyptus urophylla x eucalyptus grandis located in minas gerais state in brazil. trees were felled at 6 and 6.5 years of age. central boards produced from 10 trees were air-dried and clapboards were cut so that the radial and tangential planes were well aligned with the surface of each wood piece. a total of 278 wood specimens were produced with nominal size of 50 mm x 25 mm x 25 mm (l x r x t) in length, width and thickness, respectively. only defect-free specimens (without cracks or knots) were considered for nir spectroscopic analysis. nir spectra acquisition and basic density determination nir spectra were recorded from 12,500 cm-1 to 3,600 cm-1 with a spectral resolution of 8 cm-1 in diffuse reflection mode using a fourier transform nir spectrometer (model vector 22/n, mpa, brukeroptik gmbh, ettlingen, germany). an integrating sphere was used to obtain the nir spectra used in this study. the integration sphere is a lead sulphide detection system, which receives the incident ray after reflection in the sample. nir spectra were recorded on the radial, tangential and transverse surfaces of each wood specimen (figure 1). nir spectra readings were performed in an acclimatised room (temperature of 20°c and a relative humidity of 65%). under these conditions, the moisture content of wood specimens stabilised at 12%. after spectra acquisition, the basic density of the wood specimens was determined as the ratio between the mass of the oven-dried specimens and their saturated volume (measured by the immersion method) according to standard nbr 11941 (nbr 2003). multivariate statistics partial least squares (pls-r) regressions were developed to describe the relationship between basic density of wood and nir spectra for each specimen surface using the unscrambler software (camo as, norway, v.9.7). for calibration and validations, only the spectral range from 9,000 cm-1 to 4,000 cm-1 was considered as indicated by costa et al. (2018). the number of latent variables used in these regressions was automatically suggested by figure 1: transverse, radial and tangential surface the wood the software. in order to suppress part of the noise and improve the signal quality, the following pre-treatments were applied: standard normal variation (snv), multiplicative scatter correction (msc), savitzky golay derivatives (13-point filter and first and second-order polynomials 1d or 2d), normalisation (n), autoscale (as) and centre and scale (cs). anomalous samples were detected from studentised residues and leverage plot and excluded from the models. snv and the first derivative are two pre-treatments commonly used to remove distortions and errors in nir spectra. the snv method is applied to every spectrum individually. the average and standard deviation of all the data points for that spectrum is calculated. the average value is subtracted from the absorbance for every data point and the result is divided by the standard deviation (reis et al. 2013). the first derivative is widely used in original spectra obtained from wood and consists of better defining overlapping peaks in the same region and making the baseline correction in wood spectra as a result of the particle morphology (costa et al. 2018). pls-r models were evaluated by cross-validation. the data were divided into six subsets, each containing 46 or 47 specimens. preliminary models were developed using data from five of the subsets and validated using the remaining subset that was not used to develop the model. thus, each preliminary model was calibrated with 232 samples and validated using 46 samples. in each preliminary model, the samples were selected at random. this procedure was repeated six times, so that all subsets were used for validation. the final model for each wood surface (and for each mathematical treatment) had its regression coefficients calculated from the average of the six preliminary models. the ranking of models was based on the following criteria: (1) coefficient of determination of the model in the cross validation (r²cv); (2) standard cross-validation error (rmsecv); (3) number of latent variables (vl) used in the calibration and (4) ratio performance to deviation (rpd). the rmsecv measures the efficiency of the calibration model in predicting the property of interest in a batch of unknown samples. the rpd was first used by williams and sobering (1993) and is conceptualised as the relationship between the standard error of the measured and predicted values. according to williams and sobering (1993) calibrations with rpd between 2 and 3 are classified as “sufficient for approximate predictions” and rpd between 3 and 5 are considered “satisfactory for prediction”. results effect of wood surfaces on nir model performance the mean basic density of wood specimens was 462 kg/m3 with a standard deviation of 68 kg/m3. the basic density was higher (494 kg/m3) in the wood specimens from 6.5-year-old trees than from the specimens taken from 6-year-old trees (416 kg/m3). the goodness of fitness statistics associated with the pls-r models for estimating the wood density from untreated nir spectra collected from different surfaces of the specimen are given in table 1. the best model for predicting wood density was developed using untreated nir spectra recorded on the transverse wood surface. this model (model 1) had r²c = 0.86 and rmsec = 24.3 kg/m3 in calibration (table 1). cross-validation using 6 subsets of 46–47 wood specimens and yielded an r²cv of 0.85 and rmsecv of 25.5 kg/m3. model 2 developed from spectra collected from the radial surface had an r²cv = 0.70 and rmsecv= 37.3 kg/m3 (table 1). both these models are considered satisfactory. on the other hand, model 3, which is based on nir spectra recorded on the tangential surface, had the poorest performance (r²cv = 0.53 and rmsecv =46.8 kg/m3). the relationship between wood density values obtained from the gravimetric method and those estimated from nir-based models 1, 2 and 3 is shown in figure 2. effect of mathematical treatment on nir models a comparison of the mean untreated spectra recorded on transverse, radial and tangential wood surfaces with those treated using the first derivative and standard normal variate methods are presented in figure 3. these two mathematical treatments enhanced the quality of the signal and improved the goodness of fit statistics associated with the pls-r models (table 2). the mathematical treatment that resulted in the best fit statistics was the second derivative (model 7) which yielded an r²c of 0.90 and rmsec of 21.5 kg/m3. however, for cross-validations, the best fit statistics were obtained for models developed from nir spectra treated using the first derivative (model 6) and standard normal variate methods (snv, model 4) which yielded the higher r²cv (0.86) and lower rmsecv (25.4 kg/m3). amaral et al. new zealand journal of forestry science (2021) 51:2 page 3 model surface r²c rmsec (kg/m3) r²cv rmsecv (kg/m3) rpd vl 1 transverse 0.87 24.3 0.85 25.5 2.68 6 2 radial 0.72 35.8 0.70 37.3 1.83 6 3 tangential 0.58 44.0 0.53 46.8 1.46 6 r²c coefficient of determination of the calibration; rmsec root mean standard calibration error; r²cv coefficient of determination of the cross validation; rmsecv root mean standard error of cross-validation; rpd performance ratio of standard deviation; vl – latent variable. table 1: calibrations and cross-validations for estimating wood density based on nir spectra recorded from transverse, radial and tangential surface of wood specimens. amaral et al. new zealand journal of forestry science (2021) 51:2 page 4 discussion nir models for wood density numerous studies have been undertaken with the objective of developing nir spectroscopic models for estimating wood density (tsuchikawa & schwanninger 2013, tsuchikawa & kobori 2015). in eucalyptus, viana et al. (2010) studied six different clones at 6 years of age and found that models fitted to nir spectra were efficient at predicting basic density, chemical and anatomical properties. the goodness of statistics obtained for the predictive models for wood density reported in the present study (r² = 0.53 to 0.85 and rpd = 1.46 to 2.68) are similar to those from other studies that have used nir spectroscopy to estimate wood density. for example, the models developed by schimleck et al. (1999) to predict the basic density of eucalyptus globulus wood had r² values between 0.62 and 0.80. in larix decidua mill, gindl et al. (2001) developed models based on nir spectra that had r² values of 0.98-0.99 in calibrations and 0.95-0.97 in cross-validations. jones et al. (2006) evaluated the basic density of pinus taeda l. wood samples taken from trees ranging in age from 21 to 26 years across three different regions of georgia, usa. their models to predict the basic density from untreated spectra had an r² of 0.90 and rpd of 2.28 using six latent variables. anisotropic effect on nir-based models in the present study, the most robust models were developed using nir spectra recorded on the transverse and radial surfaces (models 1 and 2 of table 1) while figure 2: wood basic density values determined by immersion method and estimated by models based on nir spectra recorded on transverse (model 1), radial (model 2) and tangential (model 3) wood surfaces. figure 3: averaged untreated nir spectra (a); first derivative nir spectra (b); and nir spectra after standard normal variate (c); recorded on transverse, radial and tangential wood surfaces. those developed using spectra collected on the tangential surface had poorer performance (model 3, table 1). similar results were obtained in other studies carried out with across a range of different species. for example, jiang et al. (2006) also compared the accuracy of estimation of basic density of chinese fir (cunninghamia lanceolata (lamb.) hook.) wood from nir spectra in the tangential, radial, and transverse surfaces. they reported that the best predictive model was generated from the transverse surface. hein et al. (2009) evaluated the robustness of models for predicting wood density in 14-year-old eucalyptus urophylla using spectra taken from the three wood surfaces. they concluded that the best pls-r models for wood density predictions were derived from radial surface nir spectra and that models developed from tangential surface spectra had the poorest performance. schimleck et al. (2005) compared models to predict wood properties based on radial and transverse faces of strips from pinus taeda. they found that differences between the two sets of calibrations were small, indicating that either face could be used for nir analysis. these differences in performance of models developed using spectra obtained from transverse, radial or tangential surfaces can be explained by the variation in the wood anatomical structure in the different planes (costa et al. 2018). variation in the exposition of the anatomical features in the different planes affects the absorbance and reflectance of nir radiation. nir spectra obtained from radial and transverse surfaces represent wood formed during a period ranging from several months through to multiple years, whereas the nir spectra taken from the tangential surface represents wood produced over a short period of time. therefore, they are less representative of the range of variation in wood properties of the entire specimen. moreover, the nir spectra obtained in this study represent only a few millimetres of material and were used to predict the wood density of specimens with dimensions of 50 x 25 x 25 mm. in this study, the effect of applying mathematical treatments to the nir spectra was negligible. according to sandak et al. (2016), the goal of the signal (spectra) pre-processing is to eliminate or minimise variability within spectra that is not related to the investigated property of interest. our study showed that there were no differences in goodness-of-fit statistics for models based on original (untreated) and mathematically treated spectra. therefore, mathematical transformations are not always able to extract more information from the spectra in the nir. conclusions overall, the key finding from this study was that nir spectroscopy in conjunction with multivariate analysis could generate efficient models to predict density in eucalyptus wood. models for estimating wood density based on nir spectra recorded on transverse or radial wood surfaces had better predictive performance than those developed from nir spectra collected from tangential wood surfaces. mathematical transformations that have improved the performance of models in previous studies did not improve the fit statistics for the models developed in the present study. despite this, we recommend that these mathematical transformations be applied and tested in future studies as there may be situations where they can significantly improve model performance. competing interests the authors declare that they have no competing interests. authors’ contributions eaa and prgh designed the study. eaa, lms, evsc and prgh analysed the data and wrote the manuscript. eaa, lms and evsc participated in the experimental part and analysed the data. prgh and pft reviewed the manuscript. all authors read and approved the final manuscript. amaral et al. new zealand journal of forestry science (2021) 51:2 page 5 model math. treat. r²c rmsec (kg/m3) r²cv rmsecv (kg/m3) rpd vl 4 snv 0.87 24.2 0.86 25.4 2.69 6 5 msc 0.87 25.1 0.86 26.3 2.60 6 6 1d 0.88 22.8 0.86 25.4 2.69 6 7 2d 0.90 21.5 0.81 29.5 2.32 6 8 normaliz. 0.87 24.2 0.86 25.4 2.69 6 9 autoscale 0.86 24.6 0.85 25.9 2.64 6 10 m cent 0.87 24.3 0.85 25.5 2.68 6 treat treatment; r²c coefficient of determination of the calibration; rmsec root mean standard calibration error; r²cv coefficient of determination of the cross validation; rmsecv root mean standard error of cross-validation; rpd performance ratio of standard deviation; vl – latent variable. table 2. calibrations and cross-validations for estimating wood density based on nir spectra recorded from transverse surface of wood specimens after different treatments. acknowledgements the authors thank the wood science and technology graduation program (dcf/ufla, brazil) for all the support for this study. thanks to carlos henrique da silva and heber dutra for technical support. this study was financed in part by the coordenação de aperfeiçoamento de pessoal de nível superior brasil (capes) finance code 001, by the conselho nacional de desenvolvimento científico e tecnológico (cnpq: grants n. 405085/20168) and by fundação de amparo à pesquisa do estado de minas gerais (fapemig). p.r.g. hein was supported by cnpq grants (process no. 303675/2017-9). references arriel, t.g., ramalho, f.m.g., lima, r.a.b., souza, k.i.r., hein, p.r.g., & trugilho, p.f. (2019). developing near infrared spectroscopic models for predicting density of eucalyptus wood based on indirect measurement. cerne, 25(3), 294-300. https://doi. org/10.1590/01047760201925032646 associação brasileira de normas técnicas (abnt). (2003). nbr11941. madeira determinação da densidade básica. 6 p. rio de janeiro, brazil. h t t p s : / / w w w. n o r m a s . c o m . b r / v i s u a l i z a r / abnt-nbr-nm/2789/abnt-nbr11941-madeiradeterminacao-da-densidade-basica camo software inc. (2007). the unscrambler, version 9.7. camo software as. accessed: november 2016 from: http://www.camo.com costa, e.v.s., rocha, m.f.v., hein, r.g., amaral, e.a., santos, l.m.; brandão, l.e.v.s., & trugilho, p.f. (2018). influence of spectral acquisition technique and wood anisotropy on the statistics of predictive near infrared-based models for wood density. journal of near infrared spectroscopy, 26(2), 106-116. https://doi.org/10.1177/0967033518757070 gindl, w., teischinger, a., schwanninger, m., & hinterstoisser, b. (2001). the relationship between near infrared spectra of radial wood surfaces and wood mechanical properties. journal of near infrared spectroscopy, 9(1), 255-261. https://doi. org/10.1255/jnirs.311 hein, p.r.g., chaix, g., & lima, j.t. (2010). effects of sample preparation on nir spectroscopic estimation of chemical properties of eucalyptus urophylla s.t. holzforschung, 64(1), 45-54. https:// doi.org/10.1515/hf.2010.011 hein, p.r.g., lima, j.t., & chaix, g. (2009). robustness of models based on near infrared spectra to predict the basic density in eucalyptus urophylla wood. journal of near infrared spectroscopy, 17(3), 141150. https://doi.org/10.1255/jnirs.833 jiang, z.h., huang, a.m., & wang, b. (2006). near infrared spectroscopy of wood sections and rapid density prediction. spectroscopy and spectral analysis 2006, 26(6), 1034-1037. jones, p.d., schimleck, l.r., peter, g.f., daniels, r.f., & clark, a. (2006). nondestructive estimation of wood chemical composition of sections of radial wood strips by diffuse reflectance near infrared. wood science and technology, 40(8), 709-720. https://doi.org/10.1007/s00226-006-0085-6 meder, r., trung, t., & schimleck, l. (2010). guest editorial: seeing the wood in the trees: unleashing the secrets of wood via near infrared spectroscopy. journal of near infrared spectroscopy, 18(6), 5-7. https://doi.org/10.1255/jnirs.913 mora, c.r., schimleck, l.r., & isik, f. (2008). near infrared calibration models for the estimation of wood density in pinus taeda using repeated sample measurement. journal of near infrared spectroscopy, 16(6), 517-528. https://doi. org/10.1255/jnirs.816 muñiz, g.i.b., magalhães, w.l.e., carneiro, m. e., & viana, l.c. (2012). fundamentos e estado da arte da espectroscopia no infravermelho próximo no setor de base florestal. ciência florestal, 22(4), 865-875. https://doi.org/10.5902/198050987567 naes, t., isaksson, t., fearn, t., & davies, t. (2002). userfriendly guide to multivariate calibration and classification. journal of chemometrics, 17(10), 571-572. reis, n., franca, a.s., & oliveira, l.s. (2013). quantitative evaluation of multiple adulterants in roasted coffee by diffuse reflectance infrared fourier transform spectroscopy (drifts) and chemometrics. talanta, 115(1), 569-576. https://doi.org/10.1016/j. talanta.2013.06.004 sandak, j., sandak, a., & meder, r. (2016). tutorial: assessing trees, wood and derived products with near infrared spectroscopy: hints and tips. journal of near infrared spectroscopy. 24(1), 485-505. https://doi.org/10.1255/jnirs.1255 schimleck, l.r., michell, a.j., raymond, c.a., & muneri, a. (1999). estimation of basic density of eucalyptus globulus using near-infrared spectroscopy. canadian journal of forest research, 29(1), 194201. https://doi.org/10.1139/x98-204 schimleck, l.r., sturzenbecher, r., mora, c., & jones, p.d. (2005). comparison of pinus taeda l. wood property calibrations based on nir spectra from the radial-longitudinal and radial-transversal faces of wooden strips. holzforschung, 59(2), 214-218. https://doi.org/10.1515/hf.2005.034 schultz, t.p., & burns, d.a. (1990). rapid secondary analysis of lignocellulose: comparison of near infrared (nir) and fourier transform infrared (ftir). tappi journal, 73(5), 209-212. silva da, c.j., vale do, a.t., & miguel, e.p. (2015). densidade básica da madeira de espécies de cerradão no estado de tocantins. pesquisa florestal brasileira, 35(82), 63-75. https://doi.org/10.4336/2015.pfb.35.82.822 amaral et al. new zealand journal of forestry science (2021) 51:2 page 6 https://doi.org/10.1590/01047760201925032646 https://doi.org/10.1590/01047760201925032646 https://www.normas.com.br/visualizar/abnt-nbr-nm/2789/abnt-nbr11941-madeira-determinacao-da-densidade-basica https://www.normas.com.br/visualizar/abnt-nbr-nm/2789/abnt-nbr11941-madeira-determinacao-da-densidade-basica https://www.normas.com.br/visualizar/abnt-nbr-nm/2789/abnt-nbr11941-madeira-determinacao-da-densidade-basica http://www.camo.com https://doi.org/10.1177/0967033518757070 https://doi.org/10.1255/jnirs.311 https://doi.org/10.1255/jnirs.311 https://doi.org/10.1515/hf.2010.011 https://doi.org/10.1515/hf.2010.011 https://doi.org/10.1255/jnirs.833 https://doi.org/10.1007/s00226-006-0085-6 https://doi.org/10.1255/jnirs.913 https://doi.org/10.1255/jnirs.816 https://doi.org/10.1255/jnirs.816 https://doi.org/10.5902/198050987567 https://doi.org/10.1016/j.talanta.2013.06.004 https://doi.org/10.1016/j.talanta.2013.06.004 https://doi.org/10.1255/jnirs.1255 https://doi.org/10.1139/x98-204 https://doi.org/10.1515/hf.2005.034 https://doi.org/10.4336/2015.pfb.35.82.822 tsuchikawa, s., & schwanninger, m. (2013). a review of recent near-infrared research for wood and paper. applied spectroscopy reviews, 48(1), 43-71. https://doi.org/10.1080/05704920601036707 tsuchikawa, s., & kobori, h. (2015). a review of recent application of near infrared spectroscopy to wood science and technology. journal wood science 61(3):213-220. https://doi.org/10.1007/s10086015-1467-x viana, l.c.,trugilho, p.f., hein, p.r.g.; lima, j.t., & silva, j.r.m. (2010). modelos de calibração e a espectroscopia no infravermelho próximo para predição das propriedades químicas e da densidade básica da madeira de eucalyptus. ciência florestal, 20(1), 367-376. https://doi. org/10.5902/198050981859 williams, p.c, & norris, k. (2001). near-infrared technology in the agricultural and food industries. 2nd edition. st. paul, mn, usa: american association of cereal chemists. 330 p. williams, p.c., & sobering, d.c. (1993). comparison of commercial near infrared transmittance and reflectance instruments for analysis of whole grains and seeds. journal of near infrared spectroscopy, 1(1), 25-33. https://doi.org/10.1255/jnirs.3 amaral et al. new zealand journal of forestry science (2021) 51:2 page 7 https://doi.org/10.1080/05704920601036707 https://doi.org/10.1007/s10086-015-1467-x https://doi.org/10.1007/s10086-015-1467-x https://doi.org/10.5902/198050981859 https://doi.org/10.5902/198050981859 https://doi.org/10.1255/jnirs.3 new zealand journal of forestry science mason new zealand journal of forestry science (2019) 49:1 https://doi.org/10.33494/nzjfs492019x24x e-issn: 1179-5395 published on-line: 15 february 2019 influences of mean top height definition and sampling method on errors of estimates in new zealand’s forest plantations e. g. mason* school of forestry, college of engineering, university of canterbury, private bag 4800, christchurch, new zealand *author: euan.mason@canterbury.ac.nz (received for publication 5 september 2018; accepted in revised form 9 january 2019) abstract background: a study was undertaken of 51 stand inventories to compare two alternative mean top height (mth) calculation methods prevalent in new zealand, and to evaluate the consequences of creating height versus diameter at breast height (h-d) curves at a stand-level during inventories as opposed to fitting h-d curves at a plot-level. methods: the dataset was separated into two groups; one with plots having less than 6 heights measured and one with more than 5 heights measured. mth was calculated using all possible combinations of the two calculation methods and with h-d curves either at a stand-level or a plot-level. graphs were prepared to compare the 4 alternative mth estimation techniques for all plots. in addition standard deviations of mth between plots were calculated within stands, and then these were compared for different mth calculation methods using interleaved histograms and with a mixed effects analysis of variance. results: results showed that the two mth calculation methods were almost identical so long as h-d curves were fitted at a plot-level, but they differed substantially when curves were fitted at a stand-level. in addition, fitting h-d curves at a stand-level reduced independence of samples, resulting in substantial decreases in estimated standard deviations in mth within samples, thereby artificially reducing confidence intervals around sample estimates. conclusions: inventory estimates of mth were found to depend on calculation method, and so a standard definition is required. in addition, h-d curves fitted at a stand level undermined the assumption that sampling units were independent, and thereby reduced estimated variation between plots by up to 69%, depending on mth calculation method. forest inventory procedures in new zealand’s forest plantations should be redesigned to enable accurate definition of confidence intervals around sample estimates, and to facilitate the use of inventories for estimating variation in productivity across landscapes. keywords: inventory; mean top height; sampling © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access introduction mean top height (mth) is an important variable estimated in inventories and permanent sample plots (psps) in new zealand (goulding 2005), and it is a critical variable for the estimation of site productivity through site index. the relationship between height growth and productivity has long been recognised (bauer 1881), and mth at age 20 is commonly used as an index of site productivity of pinus radiata d.don. in new zealand (goulding 2005). site index has been used as a final test calibration for eco-physiological models of site productivity in sweden (mason et al. 2017), and new zealand plantation owners are keen to access similar eco-physiological estimates for their forests calibrated with site index. calculations of site index using mth and age estimates from psps and mason new zealand journal of forestry science (2019) 49:1 page 2 forest inventory plots have been proposed for calibrating site productivity models at high resolution across their estates, but inventory estimates and psp estimates differ in two potentially important respects. firstly, mths in inventories and psps are typically calculated in different ways; the forest research institute (fri) method (goulding 2005), and the carter holt harvey (chh) method (woollons 2003). the fri method involves fitting a height-dbhob (h-d) curve using a näslund equation (equation 1) (näslund 1936) to all trees with heights and diameters at breast height outside bark (dbhobs, generally measured at 1.4 m above ground level in new zealand), and then dbhobs of trees representing the largest 100 dbhobs/ha are averaged to determine “mean top diameter” (mtd). finally, the mtd is inserted into the näslund equation to determine mth. several new zealand references refer to the näslund equation as the “petterson equation”, but although petterson (1955) reported on it, näslund, who worked with petterson, invented it and so it should always bear his name. the chh method also involves fitting a näslund equation to all trees with heights and dbhobs measured, but then the equation is used to estimate heights of unmeasured trees among those representing the largest 100 dbhobs/ha, and the heights, both measured and estimated, of trees representing the largest 100 dbhobs/ ha are averaged to determine mth. h=1.4+ (b+ a/dbhob)(-2.5) (1) where h = estimated height, and dbhob = diameter at breast height outside bark. coefficients a and b can be estimated using simple linear regression techniques following a transformation (equation 2). y=a+b* dbhob (2) where y = dbhob/(h−1.4)0.4 and other variables are as described previously. woollons (2003) compared the two mth estimation techniques in experimental plots with all trees’ heights and diameters measured and found that under those circumstances they were almost identical. however, if only a few heights were measured in each plot and h-d curves were fitted at a stand-level rather than a plotlevel, the two techniques might yield different estimates of mth, and this has not been tested previously. secondly, mths of plots estimated during inventories can lack independence, while those in psps do not. inventory estimates in new zealand generally lack independence because only a very few heights (often 2–3 and in some plots none) are measured in each plot, and so h-d curves are fitted at a stand-level rather than one curve per plot. psps, on the other hand, generally contain a minimum of 12 trees with heights measured, chosen to cover the range of dbhobs in a plot (ellis & hayes 1997), and so h-d curves can be fitted independently to each plot. preliminary investigations in new zealand have suggested that psp estimates of mth might yield more diverse and somewhat different calibrations of ecophysiological estimates of productivity than calibrations with inventory mth estimates. in addition, when inventory mth estimates were plotted on maps, there often appeared to be very little variation within stands but large variation between adjacent stands (figure 1). inventory plots are generally much more numerous than psps, however, and so it would be good to be able to use inventory estimates for calibrations. this led to the question of how much these two sources of mth estimates might differ, and also to whether or not forest managers might be under-estimating variation between plots in their inventory samples. rayonier (nz) ltd. conducted 51 inventories in 2017 with more heights measured per plot than usual, and the large inventory dataset enabled a study to answer the following three questions: 1. is there a substantial difference between inventory estimates of mth when h-d curves are fitted within plots compared to when they are fitted within stands? 2. when only a few heights are measured in each plot and h-d curves are fitted at a stand level, does the fri mth calculation technique differ substantially from the chh technique? 3. does the lack of independence in inventory estimates of mth result in an underestimate of variation between plots within stands? figure 1: site indices (mth at age 20) for pinus radiata d.don estimated by standard inventory procedures across part of a forest estate, coloured by site index. each point represents a plot location, and those within the same grid and at the same spacing are in the same inventory. the lines are forest boundaries. typically 1–3 heights were measured per plot, and all h-d curves were estimated at a stand level. note how different inventories show markedly different site indices for almost the same points, and how little variation in mth estimates exist within inventories. methods fifty-one inventories were conducted in stands of pinus radiata d.don in southern new zealand in a forest estate of slightly less than 44,000 ha planted with approximately 20% open-pollinated and 80% controlled-pollinated nursery stock. in total there were 865 plots distributed on randomly located grids throughout the stands. this arrangement of plots has been recommended as one that provides unbiased estimators of crop condition (gordon & pont 2015). features of the plots are shown in table 1. table 1: features of plots in the inventories in each plot all the dbhobs were measured with a diameter tape, and a subset of trees’ heights (as indicated in the table) were measured with a vertex hypsometer. the distribution of numbers of height measurements per plot was bimodal, with peaks at 3 and 7 heights per plot (figure 2). figure 2: frequencies of height measurements per plot. the dataset was separated into two groups; one with plots having less than 6 heights measured and one with more than 5 heights measured. mth was calculated in the following four ways (where numbers of height measurements allowed): 0. 1. h-d curves fitted at a stand level and mth calculated using the chh method 2. h-d curves fitted at a stand level and mth calculated using the fri method 3. h-d curves fitted at a plot level and mth calculated using the chh method 4. h-d curves fitted at a plot level and mth calculated using the fri method graphs were prepared to compare the four alternative mth estimation techniques for all plots with more than 5 heights measured per plot. in some cases plots were also examined using plots with <6 and >2 heights measured. in 15 cases, where heights measured within stands were few, mth calculations at a plot level were unreliable owing to illogical curve coefficients, and so these estimates were excluded from the study. standard deviations of mth were calculated within stands, and then these were compared for different mth calculation methods using interleaved histograms. comparing standard deviations between methods using within-plot h-d curves and within-stand h-d curves involved comparisons using h-d curves from differently sized samples. a further calculation of mth and standard deviation was conducted using height and dbhob measurements from within stands that had a similar median and distribution to those within plots where numbers of heights measured was greater than five. this latter step was accomplished by firstly randomly selecting only one measurement per plot and then randomly adjusting the numbers in each stand to be within an appropriate range. the median in each case was 7 heights measured. comparing these standard deviations with those from mth calculation methods involving h-d curves would therefore indicate the extent to which using a common h-d curve for all plots on stand would influence variation between mth estimates between plots in a stand. standard deviations were transformed (λ=0.53) using a box-cox (box & cox 1964) method (equation 3) to make their frequency distribution as normal as possible. a mixed effects model was then fitted to examine the joint effects of calculation method and level of the h-d curve (equation 4). the level of h-d curve had three categories: plot-level, standlevel using all heights, and stand-level using a random selection of heights with a median of seven per stand. stand was a random effect in this analysis. (y λ −1)/λ, if λ ≠ 0 y(λ) = log y, if λ = 0 (3) sdijkl = pi + mj + (pm)ij + sk + ϵijkl (4) where sd = transformed standard deviation of mth between plots in stand, p = level of h-d calculation, m = method of mth calculation, s = random effect of stand, and ϵ = random error. in order to further explore reasons why variation of mth estimates within stands might be influenced by a common h-d curve, mth and stems/ha estimated from each plot were divided by stand average mth and stems/ha respectively calculated from all plots, to mason new zealand journal of forestry science (2019) 49:1 page 3 feature min. median mean max. years since planting 15.5 23.8 23.2 36.8 plot size (ha) 0.01 0.022 0.028 0.12 stems/ha 50 600 718 2700 trees/plot 1 17 17.7 69 heights/plot 0 4 4.64 24 mean dbhob (cm) 17.2 37.0 37.5 68.8 mth (fri, stand h-d, m) 18.4 29.2 28.7 38.1 � produce relative mth and relative stems/ha, and then relative mths were plotted against relative stems/ha for different methods. the dataset used for the study contained a range of plot sizes (table 1), and this can cause a bias in estimation of mean top height (garcia 1998; garcia & batho 2005; magnussen 1999; ochal et al. 2017). garcia (1998) suggested that if mth was defined as the mean height of the largest dbh tree in each 0.01 ha of a stand as advocated by rennolls (1978) then in plots where tree positions have not been defined a u-statistic estimator can be employed to provide a nearly unbiased estimate of mean top height defined in this way. the u-statistic estimator is the height of all trees weighted by the number of times they appear as the largest tree in all possible mean top-sized groups of trees in a plot. this procedure has since been employed successfully (garcia & batho 2005; ochal et al. 2017). to check whether or not the conclusions of the study reported here were affected by bias due to varying plot size, two alternative definitions were developed that were u-statistic estimator analogues of the fri and chh definitions used in new zealand, and the analysis was repeated using these two definitions. the definitions used either all näslund curve estimates of height (the fri analogue), or measured estimates when possible and otherwise curve estimates (the chh analogue). neither of these definitions is currently employed in new zealand, and so they were used simply to check on the validity of the study’s conclusions. all calculations, analyses, and geographical information system operations were done in r software (r development core team 2004). results comparing the chh and fri calculations with h-d curves at a plot level, in the same manner as woollons (2003), replicated woollons’ result showing that two methods appeared to be equivalent for practical purposes (figure 3). figure 3: a comparison of the fri and chh mth calculation methods using plots where at least six heights were measured per plot with plot-level h-d curves. the line shows where two estimates would be exactly equivalent. however, the two methods gave different results when compared with h-d curves estimated at a stand level, and the range of estimates mths was noticeably smaller (figure 4). figure 4: a comparison of the fri and chh mth calculation methods using plots where at least six heights were measured per plot with stand-level h-d curves. the line shows where two estimates would be exactly equivalent. there was also a hint that estimates for at least one of the methods might be biased for smaller mths. when fewer than six heights were measured in each plot, the differences between the two methods became even more pronounced (figure 5). figure 5: a comparison of the fri and chh mth calculation methods using plots where less than six heights were measured per plot with stand-level h-d curves. the line shows where two estimates would be exactly equivalent. for both calculation methods, the stand-level and plot-level estimates differed substantially (figures 6 and 7), but the divergence was greater for the fri method. in addition, there was evidence of bias at the extremes of these plots. figure 6: a comparison of mth calculation methods using stand-level and plot-level h-d curves in plots where at least six heights were measured per plot, using the fri calculation method. the line shows where two estimates would be exactly equivalent. mason new zealand journal of forestry science (2019) 49:1 page 4 figure 7: a comparison of mth calculation methods using stand-level and plot-level h-d curves in plots where at least six heights were measured per plot, using the chh calculation method. the line shows where two estimates would be exactly equivalent. histograms of standard deviations by mth calculation method and level of h-d equation showed that there was a tendency for stand-level equations to have smaller standard deviations than plot-level equations and that mth calculation method influenced standard deviation when stand-level methods were applied (figure 8). the mixed effects analysis of standard deviations with stand as a random effect showed clearly that the interaction between h-d curve level and mth calculation method was statistically significant (p<0.0001). furthermore, standard deviations of stand level curves calculated with only 7 heights did not differ significantly from those calculated using all heights in a stand (figure 8). relative mth tended to show a negative slope (p=2.2e-16) when plotted against relative stems/ha using within-stand h-d curves, and this trend more evident when using the fri calculation method (figure 9) than when using the chh one. by contrast, using plot-level h-d curves exhibited a slightly positive trend (p=2.358e-06) (figure 10). figure 8: mean standard deviations of plot mth estimates within stands versus level of h-d curve (“plot” = h-d at a plot level, “stand” = using all heights in a stand, and “stand7” = using a median of 7 heights/stand), and mth calculation method (chh or fri). bars with the same letter are not significantly different. figure 9: relative mth versus relative stems/ha, using stand-level h-d curves in plots where at least six heights were measured per plot, using the fri calculation method. the line shows a fitted linear regression (p=2.2e-16). repeating the analysis with contrasting definitions of mean top height employing the definition of “mean top” suggested by rennolls (1978) and u-statistic estimators developed by garcia (1998) yielded very similar results to those obtained with the new zealand definitions. the magnitudes of differences in standard deviation shown in figure 8 were slightly less when using the u-statistic estimator mth definitions. for instance instead of the 69% reduction in standard deviation observed between plot-level and stand level näslund curves for the fri technique, the reduction was 60%. all patterns and trends shown in the figures for the new zealand definitions of mth were the same when using the two u-statistic estimator definitions. figure 10: relative mth versus relative stems/ha, using plot-level h-d curves in plots where at least six heights were measured per plot, using the chh calculation method. the line shows a fitted linear regression (p=2.358e-06). discussion and conclusions independence of sampling units is a fundamental assumption for the use of statistics such as t to calculate confidence limits around sample estimates (‘student’ 1908; fisher & mackenzie 1923). “independence of sampling units” means that sampling units are not mason new zealand journal of forestry science (2019) 49:1 page 5 related; that the value obtained from one sampling unit is unaffected by the value obtained from another. using heights and dbhobs from other sampling units to formulate an h-d relation for mth calculation clearly violates this assumption. if sampling units lack independence, then standard errors may be underestimated, leading to overestimates of precisions of sample estimates. in the study reported here, using stand-level h-d curves reduced the width of estimated confidence intervals by 69% using the fri mth calculation method, and by 29% using the chh method, with a median of seven height measurements per plot. the reason for this is clear, h-d curves obtained independently from plots have different levels on an h-d graph, and using only one “average” h-d curve across a stand constrains the range of possible estimates of mth (figure 11). the chh method gives greater weighting to local height measurements, thereby reducing the extent of lack of independence compared to the fri method. measuring a median of only three height measurements/plot, as is common in new zealand’s plantation inventories, would cause the chh method to underestimate precision of estimated mth to a greater extent than outlined in the study reported here. 300 index calculations (kimberley et al. 2005) would also be affected by lack of independence, because site index is estimated before solving a basal area/ha function to establish a 300 index estimate. figure 11: h-d curves at a plot-level (blue), and stand-level (red) for one example stand. points show mth estimates, using the fri calculation method. an alternative explanation for greater variation among plot-level h-d estimates was that the larger number of measured heights and dbhobs used to make a stand-level h-d curve might provide a more accurate h-d equation, but this possibility was shown to be of little consequence by the comparison between standlevel curves using a median of 7 heights and dbhobs and plot-level h-d curves constructed with the same number of measured heights and dbhobs (figure 8). however, stochastic simulations of samples from a large population with linear correlations similar to those between height and dbhob suggest that a median of seven heights is barely adequate (results not shown), and that ellis & hayes (1997) are right in recommending at least 12 height measurements in psps to get more precise estimates of mth. more height measurements per plot is likely to reduce variation in mth between plots, but not to the extent that the variation is as small as variation between mths estimated from a stand-level h-d curve (figure 11). a future study should examine the extent of variation in stands with all heights measured, perhaps by lidar (saremi, kumar, et al. 2014; saremi, lalit, et al. 2014) and compare alternative sampling strategies when all heights are known. stems/ha varied enormously within stands (figure 9), to the point where locally high stocking led to lower estimates of mth with stand-level h-d curves compared to estimates of height in areas with locally low stocking. a stand-level h-d curve would tend to wrongly indicate that trees should be shorter where locally high stocking led to smaller diameters, but when plot-level h-d curves were used the reverse was true, albeit to a lesser extent (figure 10). the latter result may reflect a small increase in mean and mean top heights with stems/ha, as has been reported in previous studies (maclaren et al. 1995; mason 1992). in addition, having more trees to sample from may cause a small increase in the heights of the largest 100 stems/ha and hence mth when a local h-d curve is employed. stems/ha estimates from each plot may be useful as a covariate in estimating mth if this effect is characterised across a wide range of stands. the effect of stand-level h-d curves in reducing estimates of mth variability very likely explains the bias observed at extremes in figures 6 and 7. this challenges our definition of the word “stand” as an area of land with more or less homogeneous site quality, forest species composition, stand structure and age structure, unless we give a great deal of latitude to the phrase “more or less”. the results of this study confirmed the findings of woollons (2003) that when h-d equations were estimated at a plot level, the chh and fri mth calculation methods differed little in their respective estimates (figure 3). however clearly the two methods differed substantially when h-d curves were formed at a stand level (figure 4). this latter finding can be explained by the fact that using a stand-level h-d curve causes estimates of mth to be less independent with the fri calculation method than with the chh method. implications of these findings for forest managers are that: estimates of stand-level mth and volume will be much less precise than statistical calculations indicate if h-d equations are formed at a stand-level. when using the chh mth calculation method this effect will become larger as numbers of heights measured per plot is reduced. studies designed to evaluate site productivity within stands for the purposes of precision forestry will be misled by site indices derived from mths where h-d curves were formed using stand-level data, especially if a median of only 3 height measurements are obtained per plot. consideration should be given to increasing the numbers of heights measured in each inventory plot in order to allow plot-level h-d curves to be calculated. in mason new zealand journal of forestry science (2019) 49:1 page 6 mason new zealand journal of forestry science (2019) 49:1 page 7 new zealand psps are established with at least 12 height measurements/plot (ellis & hayes 1997), which may seem too great an expense for inventories, except that measuring larger plots may reduce overall inventory costs because fewer plots may be required for any given level of inventory precision (when precision is calculated properly, from independent sampling units). ellis & hayes recommend that psps should be large enough to accommodate at least 20 trees that will be in the final crop, and median sizes of psps are generally around 0.1 ha, compared to a median of 0.022 ha for the inventory plots used for this study. optimising inventory design should be a priority, given the new findings described here. trees for height measurement should be chosen across the range of dbhobs in a plot, including the largest and smallest dbhobs, but with more weighting to larger trees if mth estimates are the objective. forest managers should consider creating a consistent definition of mth. as it is more robust when sampling units lack independence, the chh method is recommended. however, sampling with truly independent sampling units should be the preferred option. previous studies have found that the “mean top” criterion is best established as an absolute number of trees per hectare rather than a proportion (ritchie et al. 2012), and in order to reduce dependence of mth estimates on plot size, the british columbian definition based on the mean of largest trees in 0.01 ha plots (garcia & batho 2005), estimated with u-statistic estimators in larger plots (garcia 1998), should also be considered. the practicality of these alternatives in new zealand conditions will be the subject of future studies. ethics approval none required. consent for publication owners of the data have provided consent for publication. availability of data the data used in this study are confidential to rayonier (nz) ltd. competing interests the author has no competing interests. funding funding this study has been provided by rayonier (nz) ltd., nelson forests ltd., kaingaroa timberlands ltd., wenita ltd., global forest partners ltd., and the new zealand forest growers levy trust. author’s contributions the author conceived of the study, set the objectives, did all analyses and wrote the entire paper. acknowledgements i am very grateful for the financial support of rayonier (nz) ltd., nelson forests ltd., kaingaroa timberlands ltd., wenita ltd., global forest partners ltd., and the new zealand forest growers levy trust that made this work possible. rayonier (nz) ltd. undertook to measure more heights per plot and made raw inventory data available, and both alex tolan and charles hosking offered valuable advice about data structures and company policies. references ‘student’. (1908). the probable error of a mean. biometrika, 6(1), 1-25. bauer, f. (1881). die rotbuche in bezug auf ertrag, zuwachs und form. berlin: verlap von paul parey. box, g.e.p., & cox, d.r. (1964). an analysis of transformations. journal of the royal statistical society. series b (methodological), 26(2), 211-252. ellis, j.c., & hayes, j.d. (1997). field guide for sample plots in new zealand forests. fri bulletin 186. rotorua, new zealand: forest research institute. fisher, r.a., & mackenzie, w.a. (1923). studies in crop variation: the manurial response of different potato varieties. journal of agricultural science, 13, 311-320. garcia, o. (1998). estimating top height with variable plot sizes. canadian journal of forest research, 28(10), 1509-1517. garcia, o., & batho, a. (2005). top height estimation in lodgepole pine sample plots. western journal of applied forestry, 20(1), 64-68. gordon, a.d., & pont, d. (2015). inventory estimates of stem volume using nine sampling methods in thinned pinus radiata stands, new zealand. new zealand journal of forestry science, 45:8. goulding, c.j. (2005). measurement of trees. in m. colley (ed.), forestry handbook (pp. 145-148). christchurch, new zealand: new zealand institute of forestry (inc.). kimberley, m., west, g., dean, m., & knowles, l. (2005). the 300 index a volume productivity index for radiata pine. new zealand journal of forestry, 50(2), 13-18. maclaren, j.p., grace, j.c., kimberley, m.o., knowles, r.l., & west, g.g. (1995). height growth of pinus radiata as affected by stocking. new zealand journal of forestry science, 25(1), 73-90. magnussen, s. (1999). effect of plot size on estimates of top height in douglas-fir. western journal of applied forestry, 14(1), 17-27. mason, e.g. (1992). decision-support systems for establishing radiata pine plantations in the central north island of new zealand. (phd), university of canterbury, christchurch. mason, e.g., holmström, e., & nilsson, u. (2017). using hybrid physiological/mensurational modelling to predict site index of pinus sylvestris l. in sweden: a pilot study. scandinavian journal of forest research, 8. doi: 10.1080/02827581.2017.1348539 näslund, m. (1936). skogsförsöksanstalens gallringsförsök i tallskog primärbearbetning (vol. 29 (1)). stockholm, sweden: centraltryckeriet. mason new zealand journal of forestry science (2019) 49:1 page 8 ochal, w., socha, j., & pierzchalski, m. (2017). the effect of the calculation method, plot size, and stand density on the accuracy of top height estimation in norway spruce stands. iforest, 10, 498-505. petterson, h. (1955). yield of coniferous forests. medd. stat. skogsförsöksanct, (1b, 45). sweden. r development core team. (2004). r: a language and environment for statistical computing. vienna, austria: r foundation for statistical computing. rennolls, k. (1978). ‘top height’; its definition and estimation. commonwealth forestry review, 57(3), 215-219. ritchie, m., zhang, j., & hamilton, t. (2012). effects of stand density on top height estimation for ponderosa pine. western journal of applied forestry, 27(1), 18-24. saremi, h., kumar, l., stone, c., melville, g., & turner, r. (2014). sub-compartment variation in tree height, stem diameter and stocking in a pinus radiata d.don plantation examined using airborne lidar data. remote sensing, 6(8), 7592-7609. saremi, h., lalit, k., turner, r., & stone, c. (2014). airborne lidar derived canopy height model reveals a significant difference in radiata pine (pinus radiata d.don) heights based on slope and aspect of sites. trees: structure and function, 28(3), 733-744. woollons, r.c. (2003). examination of mean top height definitions and height estimation equations for pinus radiata in new zealand. new zealand journal of forestry, 48(3), 15-18. list of abbreviations chh: carter holt harvey (describing a mth calculation method) dbhob: diameter at breast height (1.4 m) over bark fri: forest research institute (describing a mth calculation method) h: estimated height of an individual tree h-d: height versus diameter (describing a curve) lidar: light detection and ranging, a method used to examine the earth’s surface mtd: mean top diameter (mean dbhob of the largest 100 trees/ha defined by dbhob) mth: mean top height psp: permanent sample plot a comparison between traditional ordinary least-squares regression and three methods for enforcing additivity in biomass equations using a sample of pinus radiata trees mohan kc1,2*, euan g. mason2, horacio e. bown3, grace jones2,4 1ministry of industry, tourism, forests and environment, bagmati province, division forest office, lalitpur, 44700, nepal 2school of forestry, university of canterbury, private bag 4800, christchurch, new zealand 3faculty of forestry, university of chile, po box 9206, santiago, chile 4department of forestry and wood technology, linnaeus university, pg vejdes väg, 351 95 växjö, sweden *corresponding author: mohankc.forestry@gmail.com (received for publication 14 january 2020; accepted in revised form 6 august 2020) abstract background: additivity has long been recognised as a desirable property of systems of equations to predict the biomass of components and the whole tree. however, most tree biomass studies report biomass equations fitted using traditional ordinary least-squares regression. therefore, we aimed to develop models to estimate components, subtotals and aboveground total biomass for a pinus radiata d.don biomass dataset using traditional linear and nonlinear ordinary leastsquares regressions, and to contrast these equations with the additive procedures of biomass estimation. methods: a total of 24 ten-year-old trees were felled to assess above-ground biomass. two broad procedures were implemented for biomass modelling: (a) independent; and (b) additive. for the independent procedure, traditional linear models (linols) with scaled power transformations and y-intercepts and nonlinear power models (nlinols) without y-intercepts were compared. the best linear (transformed) models from the independent procedure were further tested in three different additive structures (linadd1, linadd2, and linadd3). all models were evaluated using goodness-of-fit statistics, standard errors of estimates, and residual plots. results: the linols with scaled power transformations and y-intercepts performed better for all components, subtotals and total above-ground biomass in contrast to nlinols that lacked y-intercepts. the additive model (linadd3) in a joint generalised linear least-squares regression, also called seemingly unrelated regression (sur), provided the best goodnessof-fit statistics and residual plots for four out of six components (stem, branch, new foliage and old foliage), two out of three subtotals (foliage and crown), and above-ground total biomass compared to other methods. however, bark, cone and bole biomass were better predicted by the linols method. conclusions: sur was the best method to predict biomass for the 24-tree dataset because it provided the best goodnessof-fit statistics with unbiased estimates for 7 out of 10 biomass components. this study may assist silviculturists and forest managers to overcome one of the main problems when using biomass equations fitted independently for each tree component, which is that the sum of the biomasses of the predicted tree components does not necessarily add to the total biomass, as the additive biomass models do. new zealand journal of forestry science kc et al. new zealand journal of forestry science (2020) 50:7 https://doi.org/10.33494/nzjfs502020x90x e-issn: 1179-5395 published on-line: 25/10/2020 © the author(s). 2020 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access pinus radiata d.don, native to california, is a widely planted commercial tree species in the southern hemisphere, including new zealand, australia, chile, spain and south africa (lavery & mead 2000; mead introduction forests play a vital role in the carbon cycle to mitigate climate change by accumulating and sequestering atmospheric carbon dioxide (co2) (houghton 1991). keywords: above-ground; additive; biomass; linear; nonlinear; radiata pine; seemingly unrelated regression http://creativecommons.org/licenses/by/4.0/), kc et al. new zealand journal of forestry science (2020) 50:7 page 2 2013). it is grown primarily for timber production, as this species is versatile, fast-growing, and has a wide range of end uses (lavery & mead 2000; lewis et al. 1993; rogers 2002; sutton 1999; toro & gessel 1999). the global plantation area of p. radiata is now more than 4.2 million hectares (mead 2013). in new zealand, it is the predominant planted species, and accounts for about 90% of the total 1.7 million hectares of forest plantations (nixon et al. 2017). plantation forests in new zealand have not only been recognised as providing financial returns from traditional wood products, but also as providing environmental services by accumulating biomass and sequestering a substantial amount of carbon. to quantify such benefits, a precise biomass model with a required level of accuracy is essential. tree biomass estimation is required by scientists and practitioners alike as a surrogate of ecosystem production, product outturn and carbon accounting, among others. biomass modelling is important for estimating carbon sequestration of forest ecosystems, as individual tree biomass or its components is aggregated to yield the stand biomass (zheng et al. 2015). allometric models are commonly used to assess the biomass accumulated in forests. allometric relationships can be developed from destructive sampling by using several forms of regression equations. generally, biomass equations are fitted in linear form using logarithmic transformation of b and d of the form, b = adb, where b is the biomass of the tree, or its components, and d is the diameter of the tree (baskerville 1972; beauchamp & olson 1973; canadell et al. 1988; santa regina et al. 1997; sprugel 1983). clutter et al. (1983) explained various linear and nonlinear additive regression models to estimate the biomass of an individual tree or its components. the additivity of biomass equations has long been recognised as a desirable property, so that predictions of tree components added together equals predictions of total tree biomass (cunia & briggs 1984, 1985; parresol 1999, 2001). three procedures for forcing additivity have been proposed (cunia & briggs 1985; parresol 1999): (a) adding the best regression functions of the components’ biomass to determine the total biomass regression function; (b) using the same independent variables for each component; and (c) using joint generalised least-squares regression, also known as seemingly unrelated regression (sur), in which statistical dependencies among sample data are accounted for by forcing constraints on the regression coefficients. these three procedures have been extensively applied for estimating tree biomass around the world (canga et al. 2013; návar, gonzález et al. 2004). the additive procedure in sur has been mostly used for biomass modelling of single species (cunia & briggs 1984, 1985; green & reed 1985; parresol 1999; zheng et al. 2015). over the last 50 years, a substantial number of biomass studies for p. radiata have been undertaken in new zealand (beets & madgwick 1988; beets et al. 2007; beets & pollock 1987; cromer et al. 1985; madgwick 1983, 1985, 1994; madgwick et al. 1977; mead et al. 1984; moore 2010; webber & madgwick 1983; will 1964). previous studies for p. radiata in new zealand aimed to find the best biomass equations using various functional linear and nonlinear forms, with models generally fitted separately for each individual biomass component and for the whole tree. separately calculated biomass equations ignore inherent correlation among the component equations measured on the same tree (kozak 1970; parresol 1999). simultaneous fits with related equations using additive procedures have greater statistical efficiency, as they take into account statistical dependencies among biomass components in parameter estimation recorded from the same tree (bi et al. 2010; bi et al. 2004; carvalho & parresol 2003; parresol 1999, 2001). two country specific systems of additive biomass equations were developed for p. radiata using routinely measured stand variables from australia and new zealand (bi et al. 2010). it has been noted that prediction accuracy varies across methodological differences and uncertainties associated over a range of stand variables (bi et al. 2010; moore 2010). as there is uncertainty about how to better meet additivity requirements, this study was undertaken to compare traditional linear and nonlinear ordinary least-squares regressions, and additive procedures in the estimation of tree component and total biomass for a dataset composed of 24 trees of p. radiata. methods study site and experiment this study was carried out in the canterbury region of new zealand, planted with p. radiata in 2005, in a forestry trial designed to test the effect of stocking, genetics, fertiliser application, and follow-up weed control treatment on productivity and wood quality (mason 2008). the site is located at latitude 43° 37.2′ s and longitude 172° 20.4′ e, and about 45 m above sea level on a flat landscape. the site has a mean annual air temperature between 11 and 13 °c with a monthly minimum (july) of −2 to +4 °c and a monthly maximum (january) of 20 to 23 °c (macara 2016). annual rainfall is about 618 mm with a monthly range between 38 and 68 mm (macara 2016). the experiment consisted of 48 permanent plots with a randomised complete block split-split design, with the arrangement of factors within four complete blocks (mason 2008). during the summer of 2015 to 2016, 24 ten-year-old trees of p. radiata were harvested and measured from six plots of the trial, and within each plot four trees were felled. these plots consisted of three levels of stocking (625, 1250 and 2500 stems ha−1), two levels of follow-up weed control treatment (herbicide and no chemical treatment) and two clones (1 and 2). biomass data trees were felled at ground level. the over-bark diameter of each tree at breast height was recorded at 1.4 m. total tree height was measured from ground level to the tip of the tree bole. for each tree, the components were separated into stem, branch, bark, foliage, and cones. needles and twigs less than 1 cm in diameter were considered foliage, and this was separated into “new” and “old” foliage. the total fresh mass of all components including subsamples were measured immediately after felling, using a portable balance. all the cones and small branches were weighed separately. the logs were separated into small pieces and weighed fresh in the field. a subsample of stem discs with bark (cut at the 1.4 m section and every 2 m upwards in the stem) and subsamples of all other components, were weighed to determine fresh weight in the field. subsamples were dried in an oven at 70 °c until constant mass was achieved, and then this weight was recorded. dry mass of each component was calculated as the fresh mass recorded in the field for that component multiplied by the ratio of subsample dry to fresh mass (eq. 1): (1) where y is the total dry mass (kg), dw and fw refers to the sub sampled dry and fresh mass (kg) respectively, tfw is the total fresh mass (kg), and i is the tree component such as stem, bark, branch, new foliage, old foliage and cones. descriptive statistics of the trees including components, sub-total and above-ground total biomass are shown in table 1. the notations and definitions used in this manuscript are explained in the abbreviations section. variance stabilisation biomass data generally exhibit non-constant variance in model residuals (parresol 1993, 2001). when developing predictive equations, variance can be stabilised either by providing a weight function or by using transformations (parresol 1993, 2001). curvilinearity and heterogeneity in variance of all linear models were reduced by transforming the response as well as explanatory variables using scaled-power transformations (eq. 2), widely known as box-cox transformations (box & cox 1964). the predicted values of these models were back transformed to the original form using eq. 3. a similar variance stabilisation technique was implemented by zheng (2015) while using the additive procedure of biomass modelling for quercus variabilis in northern china. (2) (3) where y(λ) is the transformed variable, and λ is a coefficient of the transformed variable that varies normally between −3 and +5 (cook & weisberg 2009), y' is the back-transformed variable. a λ term is chosen to make the frequency distribution of each variable as close to normal as possible, thus promoting linear relationships and stabilising variance. model assessment and evaluation in this study, a dataset consisting of 24 trees was used to evaluate the fitting bias, precision, and validity of models using the following goodness-of-fit statistics: root mean square error (rmse), mean absolute bias (mab), mean prediction error (mpe), residual standard error (rse), coefficient of variation (cv), coefficient of determination (r2), index of agreement (ioa), and akaike information criterion (aic). models were considered better with small aic, rmse, mab, mpe, rse, and cv of the residuals, kc et al. new zealand journal of forestry science (2020) 50:7 page 3 1description max min mean 2sd 3ci (p= 95%) tree variables dbh (cm) 28 8.2 18.68 5.46 2.30 h (m) 13.77 8.85 11.66 1.19 0.50 crl (m) 6 0.2 3.35 2.07 0.87 components stem (kg tree-1) 118.46 9.29 60.46 32.24 13.62 branch (kg tree-1) 62.68 0.28 17.42 20.92 8.83 of (kg tree-1) 34.84 0.94 13.18 11.63 4.91 bark (kg tree-1) 11.92 0.65 5.18 3.19 1.35 cone (kg tree-1) 16.96 0.05 3.65 3.87 1.63 nf (kg tree-1) 9.79 0.29 3.37 2.86 1.21 subtotals bole (kg tree-1) 128.69 9.94 65.64 35.33 14.92 crown (kg tree-1) 123.63 3.12 37.60 37.29 15.75 foliage (kg tree-1) 61.59 2.84 20.19 17.45 7.37 total agt (kg tree-1) 241.65 13.07 103.24 71.45 30.17 1 abbreviation details provided at the end of the text; 2sd = standard deviation, 3ci = confidence interval table 1: descriptive statistics for the 24-felled trees used for developing regression models, and components, sub-total and above-ground total biomass. and large r2 and ioa. the interpretation of these fitting statistics can be found in von gadow and hui (2001) and goicoa et al. (2011). in addition, model performance was assessed by residual plots and histograms of residuals. modelling procedure in this research, two procedures were implemented to estimate components, subtotals and above-ground total biomass: (1) independent; and (2) additive. all models were fitted to estimate biomass in terms of kg tree−1. independent procedure for biomass estimation in this procedure, biomass equations were fitted independently using traditional linear ordinary leastsquares regressions with scaled power transformations and y-intercepts (denoted as, linols; eq. 4) and nonlinear ordinary least-squares power equations that lacked y-intercepts (denoted as, nlinols; eq. 5). the mathematical specifications of these models are as follows (parresol 1999; zeng 2011; zianis et al. 2005). (4) (5) where fl(xl, βl) is the regression function for the above ground biomass or one of its components, xl are tree dimension variables such as d, h and crl (l = 1, 2, . . . . , p) while βl denote the regression coefficients. each component equation contained its own independent variables. all components, subtotals and agt biomass equations were fitted separately using the lm and nls function of r statistical software (r core team, 2018), for linear and nonlinear regressions, respectively. additive procedure of biomass estimation in this procedure, biomass equations were fitted based on three additive procedures, described and compared by parresol (1999, 2001). the additivity requirement to estimate total tree biomass is ensured by (a) adding the separately calculated best regression functions of each component, (b) using the same independent variables for each component, and (c) using joint generalised leastsquares methods, also known as seemingly unrelated regression (sur). in sur, statistical dependencies among components are forced by constraining regression coefficients (cunia & briggs 1985; parresol 1999). in kc et al. new zealand journal of forestry science (2020) 50:7 page 4 this study, four restrictions were provided for the sur model: (1) foliage; (2) crown; (3) bole; and (4) agt, as illustrated in figure 1. for example, foliage biomass is the sum of nf, of and cone biomass (eq. 6). mathematically, the additive system of biomass equations in additive error terms with cross-equation correlation is specified in eq. 6 where ŷi represents the predicted biomass of a given component and fi(xi, βi) is a regression function for the biomass component, (i = cone, new foliage, old foliage, branch, bark and stem, foliage, crown, bole and agt biomass). the residual is εij for the i th equation and j is an index for component. all additive biomass equations were fitted in the r statistical software (r core team 2018) using the systemfit package (henningsen & hamann 2007). in the first additive procedure, the additivity was ensured by adding individually calculated best regression functions of each component to give a total biomass regression function (cunia & briggs 1985; parresol 1999). the best regression functions obtained from the independent procedure of biomass modelling that were fitted separately for each component given in table a.1 were used. the additive structure of this model, denoted as linadd1, is specified in eq. 7. in the second additive procedure, additivity was implemented by using the same explanatory variables for each component. for this, the most frequent independent variable (d) was selected from the best linear regression function as it was best fitted for stem, bark, foliage, bole, and agt (table a.1). using d as an independent variable figure 1: a statistical framework showing model structure with four restrictions (foliage, crown, bole and agt) for biomass additivity. (6) kc et al. new zealand journal of forestry science (2020) 50:7 page 5 (7) (8) for all components, the additive structure of the model, denoted as linadd2, is specified in eq. 8. in the third additive procedure, we used different explanatory variables in a joint generalised linear leastsquares regression, known as sur (cunia & briggs 1985; parresol 1999). for this, best-fitted explanatory variables from the independent procedure of biomass modelling were used for stem, cone, branch, nf, and of (table a.1). we used the second-best regression d2h as an independent variable for bark (data not shown). the additive structure of the model, denoted as linadd3, is specified in eq. 9. results comparison of fitted equations for components, subtotals and agt tested linols and nlinols equations with their bestfit results are given in table a.1, and fitted statistics with their regression estimates are presented in table a.2. we attempted to take into account follow-up herbicide, stocking, and clone factors into all models as dummy variables. these were found to be non-significant (p>0.05) so were discarded from all subsequent modelling. in comparison, linols provided relatively higher r2 values than nlinols for all, except for branch and cone biomass (table a.2). however, plotting residuals with predicted values and with other variables demonstrated that nlinols regression was unsuitable for these two components (data not shown). therefore, overall, the best fitted linols model according to goodness-of-fit statistics and residual plots were eq. (i) for stem, bark, foliage, bole and agt biomass, eq. (ii) for cone biomass, eq. (iii) for branch biomass, eq. (viii) for nf and crown biomass, and eq. (ix) for of biomass (table a.1). finally, these selected linols models were further tested in the additive process of biomass estimations. the estimated coefficients for six components, three subtotals, and agt using four methods (linols, linadd1, linadd2 and linadd3) are presented in table 2 and their goodness of fit statistics are given in table 3. the distribution of residuals with predicted values of the fitted best models for six components, three subtotals, and agt are given in fig. 2. the linadd3 fitted in sur was considered best to predict stem (eq. 10), branch (eq. 11), nf (eq. 13), of (eq. 14) biomass as it provided the better-fitting statistics when compared to the other three equations (table 3). for stem, linadd3 simultaneously decreased the rmse, rse, and cv by 0.1%, mpe by 0.2% while r2 increased by 0.005%, compared to the other three equations. for branches, linadd3 provided a marginal decrease in the goodness-of-fit statistics (e.g. rmse, rse, and cv by 1%) in contrast to linols and linadd1. for nf, linadd3 model recorded a decrease in fitting statistics (e.g. rmse by 3.7%, rse by 1.43%), in contrast to the other three kc et al. new zealand journal of forestry science (2020) 50:7 page 6 (9) kc et al. new zealand journal of forestry science (2020) 50:7 page 7 components parameter estimates methods λ value linols linadd1 linadd2 linadd3 cone β10 −2.301 ** (0.769) −2.225 ** (0.767) −5.240 ** (1.483) −2.229 ** (0.767) λd = 0.34 λco = 0.27 β11 0.132 *** (0.029) 0.129 *** (0.029) 1.262 *** (0.296) 0.129 *** (0.029) nf β20 −4.959 *** (0.375) −4.959 *** (0.375) −4.959 *** (0.375) −4.213 *** (0.663) λd = 0.34 λcrl = 1.45 λnf = 0.07β21 1.184 *** (0.075) 1.184 *** (0.075) 1.184 ns (0.075) 1.059 *** (0.117) β22 −0.035 ns (0.026) of β30 −4.089 *** (0.331) −3.209 *** (0.336) −4.089 *** (0.331) −3.105 *** (0.350) λd = 0.34 λcrl = 1.45 λof = 0.01β31 1.266 *** (0.066) 1.119 *** (0.063) 1.266 *** (0.066) 1.101 *** (0.066) β32 −0.006 *** (0.001) −0.007 *** (0.002) branch β40 −5.803 *** (0.666) −5.786 *** (0.619) −6.486 *** (0.681) −5.758 *** (0.619) λd = 0.34 λcrl = 1.67 λbr = 0.04β41 1.981 *** (0.149) 1.987 *** (0.128) (1.760) *** (0.136) 1.973 *** (0.128) β42 −0.049 * (0.019) −0.051 *** (0.009) −0.049 *** (0.009) bark β50 −4.983 *** (0.373) −4.983 *** (0.373) −4.983 *** (0.373) −3.340 *** (0.279) λd2h = 0.3 λba = 0.36 β51 1.416 *** (0.074) 1.416 *** (0.074) 0.373 *** (0.074) 0.146 *** (0.007) stem β60 −6.691 *** (0.493) −6.691 *** (0.493) −6.691 *** (0.493) −6.655 *** (0.492) λd = 0.34 λst = 0.38 β61 3.264 *** (0.099) 3.264 *** (0.099) 3.264 *** (0.099) 3.257 *** (0.098) foliage β70 −3.259 *** (0.393) λd = 0.34 λfol = 0.03 β71 1.213 *** (0.079) crown β80 −0.731 ns (0.510) λd = 0.34 λcrl = 1.45 λcr = −0.13β81 0.693 *** (0.089) β82 −0.046 * (0.021) bole β90 −7.002 *** (0.510) λd = 0.34 λbol = 0.38 β91 3.403 *** (0.102) agt β100 −1.182 *** (0.244) λd = 0.34 λagt = 0.08 β101 1.309 *** (0.049) table 2: regression model for each biomass component across modelling techniques. table shows parameter estimates, their standard error between parentheses and significance indicated as: ns, non-significant, *, p<0.05; **, p<0.01; ***, p<0.001. box-cox transformation values (λ) are also presented. note: the λ value shown in the table indicates that the variables were subjected to a scaled power transformation. the estimated parameter values for each technique are presented in power-transformed scale. table 2: confusion matrix kc et al. new zealand journal of forestry science (2020) 50:7 page 8 biomass m od el r m se m a b m p e r se cv r 2 io a r a n k stem linols 4.864 3.391 23.660 5.080 8.046 0.976 0.994 3 linadd1 4.864 3.391 23.660 5.080 8.046 0.976 0.994 4 linadd2 4.864 3.391 23.659 5.080 8.046 0.976 0.994 2 linadd3 4.859 3.406 23.611 5.075 8.038 0.976 0.994 1 branch linols 8.034 4.935 64.538 8.588 46.124 0.846 0.962 2 linadd1 8.046 4.97 64.733 8.601 46.194 0.846 0.962 3 linadd2 11.168 6.658 124.713 11.664 64.118 0.703 0.909 4 linadd3 7.958 4.848 63.322 8.507 45.688 0.849 0.962 1 bark linols 0.880 0.604 0.774 0.919 16.987 0.921 0.979 1 linadd1 0.879 0.604 0.774 0.919 16.987 0.921 0.979 3 linadd2 0.879 0.604 0.774 0.919 16.987 0.921 0.979 1 linadd3 0.899 0.613 0.809 0.939 17.374 0.917 0.979 4 nf linols 0.896 0.609 0.803 0.936 26.627 0.898 0.973 2 linadd1 0.896 0.609 0.803 0.936 26.627 0.898 0.973 4 linadd2 0.896 0.609 0.803 0.936 26.627 0.898 0.973 3 linadd3 0.863 0.599 0.745 0.923 25.642 0.905 0.975 1 of linols 2.879 2.109 8.289 3.007 21.852 0.936 0.985 3 linadd1 2.486 1.758 6.181 2.658 18.869 0.952 0.989 2 linadd2 2.879 2.109 8.289 3.007 21.852 0.936 0.985 3 linadd3 2.436 1.727 5.935 2.604 18.490 0.954 0.989 1 cone linols 2.743 1.777 7.522 2.865 75.229 0.475 0.781 1 linadd1 2.760 1.783 7.619 2.883 75.710 0.468 0.774 3 linadd2 2.811 1.785 7.904 2.936 77.113 0.449 0.753 4 linadd3 2.759 1.783 7.613 2.882 75.681 0.469 0.775 2 foliage linols 4.561 3.179 20.800 4.764 22.592 0.929 0.981 2 linadd1 4.509 3.194 20.327 4.939 22.334 0.930 0.982 2 linadd2 4.648 3.307 21.606 4.855 23.026 0.926 0.981 4 linadd3 4.478 3.173 20.048 4.905 22.180 0.931 0.982 1 crown linols 15.359 9.019 235.915 16.043 40.845 0.823 0.955 4 linadd1 10.146 6.633 102.932 11.403 26.979 0.923 0.981 2 linadd2 14.044 8.408 197.219 14.668 37.346 0.852 0.959 3 linadd3 10.023 6.570 100.466 11.265 26.655 0.925 0.981 1 bole linols 5.287 3.765 27.951 5.522 8.055 0.977 0.994 1 linadd1 5.295 3.762 28.035 5.530 8.067 0.987 0.994 4 linadd2 5.295 3.762 28.035 5.530 8.067 0.987 0.994 2 linadd3 5.293 3.799 28.014 5.658 8.064 0.987 0.994 3 agt linols 17.131 11.419 293.497 17.894 16.594 0.940 0.985 3 linadd1 14.704 9.702 216.201 16.526 14.242 0.956 0.989 2 linadd2 17.405 11.171 302.933 18.179 16.859 0.938 0.984 4 linadd3 14.667 9.708 215.132 16.485 14.207 0.956 0.989 1 table 3: goodness-of-fit statistics for given biomass components using four methods of modelling. note: rank indicates the model’s performance in comparison. for example, a model in rank 1 is a best and rank 4 is a worst in terms of goodness-of-fit statistics, and residual plots. kc et al. new zealand journal of forestry science (2020) 50:7 page 9 (a) stem (eq.10) (b) branch (eq.11) (c) bark (eq. 12) (d) nf (eq. 13) (e) of (eq. 14) (f) cone (eq. 15) 20 40 60 80 100 12010 -5 0 5 10 15 predicted stem biomass (kg) r es di du al ( kg ) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 2122 23 24 0 10 20 30 40 50 60 70 -1 0 0 10 20 predicted branch biomass (kg) r es di du al ( kg ) 1 2 3 45 6 7 8 910 11 12 1314 15 16 17 18 19 20 21 22 23 24 2 4 6 8 10 12 -2 -1 0 1 2 3 predicted bark biomass (kg) r es di du al ( kg ) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 0 2 4 6 8 -2 -1 0 1 2 predicted nf biomass (kg) r es di du al ( kg ) 1 23 4 56 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 0 10 20 30 -6 -4 -2 0 2 4 predicted of biomass (kg) r es di du al ( kg ) 1 2 3 4 567 8 9 10 11 12 13 1415 16 17 18 19 20 21 22 23 24 2 4 6 -4 -2 0 2 4 6 8 10 predicted cone biomass (kg) r es di du al ( kg ) 1 2 3 4 5 6 7 8 9 10 1112 13 14 15 16 17 18 19 20 21 22 23 24 equations. for of, the linadd3 provided a decrease in rmse (by 10.9%), mab (by 12.7%), mpe (by 20.3%), rse (by 9.6%) and cv (by 10.9%), and an increase in r2 (by 1.2%) and ioa (by 0.3%), in contrast to the other three equations (table 3). the linols was considered the best to predict bark (eq. 12), cone (eq. 15) and bole (eq. 12) biomass exhibiting relatively good fit statistics as compared to sur models (table 3). for bark, linols simultaneously decreased rmse, mab, mpe, rse and cv by 0.7%, 0.5%, 1.5%, 0.7%, and 0.7%, respectively, and it increased r2 by 0.1% in contrast to the other three additive sur equations (table 3). for cone, the linols equation provided a compatible decrease in rmse, rse, and cv by 1.2%; mab decreased by 0.4%, mpe decreased by 2.4%, r2 increased by 3% and ioa increased by 1.9%, in contrast to the other three additive equations tested. for bole, the linols indicated there was a marginal decrease in rmse (0.1%), mpe (0.2%), rse (0.1%) and cv (0.1%), compared to three additive equations (table 3). the additive equation fitted in sur was considered the best to predict foliage (eq. 16) and crown (eq. 17) biomass as linadd3 provided better fit statistics (table 3). for foliage, linadd3 provided a decrease in rmse, mab, mpe and cv by 2.1%, 1.6%, 4.1% and 2.1%, respectively, and r2 by 0.3% and ioa by 0.1% compared to other three equations (table 3). for crown, using linadd3, average fitting statistics decreased by 21.5% for rmse, 16.7% for mab, 36.3% for mpe, 18.1% for rse, and 21.5% for cv; and increased r2 by 7% and ioa by 1.7%, in contrast to other three equations. the linadd3 fitted in sur (eq. 19) was considered the best to predict agt biomass as it provided a decrease in rmse by 10.1%, mab by 9.3%, mpe by 18.7, rse by 5.8% and cv by 10.1%; and an increase in r2 by 1.2% and ioa by 0.3%, in contrast to the other three methods. discussion in this study, follow-up herbicide, stocking, and clone factors fitted into models as dummy variables were all non-significant. the possible reason for this insignificance could be that the site or silvicultural effects kc et al. new zealand journal of forestry science (2020) 50:7 page 10 (g) foliage (eq. 16) (h) crown (eq. 17) (i) bole (eq. 18) (j) agt (eq. 19) figure 2: residuals vs predicted values of biomass for the selected best models. the solid black horizontal line across zero represent baseline and the dotted red line is loess curve 0 10 20 30 40 50 -5 0 5 10 15 predicted foliage biomass (kg) r es di du al ( kg ) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 0 20 40 60 80 100 120 -1 0 0 10 20 predicted crown biomass (kg) r es di du al ( kg ) 1 2 3 4 5 67 8 9 10 11 1213 1415 16 17 18 19 20 21 22 23 24 20 40 60 80 100 120 -1 0 -5 0 5 10 15 predicted bole biomass (kg) r es di du al ( kg ) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 50 100 150 200 -2 0 -1 0 0 10 20 30 40 predicted agt biomass (kg) r es di du al ( kg ) 1 2 3 4 5 6 7 8 910 11 12 13 14 15 16 1718 19 20 21 22 23 24 figure 2: residuals vs predicted values of biomass for the selected best model for each biomass compponent (a–j). the solid black horizontal line across zero represent baseline and the dotted red line is loess curve. were explained by the tree size variables because the silvicultural treatments may simply speed up/slow the growth of the trees (beets & pollock 1987; moore 2010), as opposed to changing their allometry. the biomass of p. radiata plantation changes with a wide range of climatic, edaphic, silvicultural and genetic factors (beets & pollock 1987; bi et al. 2010; moore 2010). another reason of this lack of significance could be the small sample size (duncanson et al. 2015) as the parameterisation of allometric equations depends significantly on the size of sample. in this study, the best model of each component selected from independent procedures provided the logical base equations for further tests in the additivity of biomass equations. the same approach was also used by other researchers (magalhães & seifert 2015; návar et al. 2002) to utilise additive properties in biomass modelling. biomass additivity reduces the discrepancy between the sum of predicted values for components and those for a total tree (kozak 1970), and it has long been documented as a desirable property of systems of equations to predict total tree biomass (bi et al. 2004; parresol 2001). three procedures were implemented for the additivity in the biomass model (parresol 1999, 2001): (1) using a separately calculated best linear function of the biomass of the components (best linear functions were d, d and h, d, d and crl, d and crl2, and d2 for stem, branch, bark, nf, of, and cone biomass, respectively); (2) using the most frequently observed predictor (d) as the same independent variable for all components; and (3) using different independent variables for each component by forcing four linear restrictions on the regression coefficients, the sur technique. the additivity of biomass equations to predict biomass of components, subtotal, and agt has been explained in some other studies (carvalho & kc et al. new zealand journal of forestry science (2020) 50:7 page 11 10 11 12 13 14 15 16 17 18 19 (beets & madgwick 1988; beets & pollock 1987), and nutrients and silvicultural practices (beets & madgwick 1988; madgwick 1985; mead et al. 1984) may influence the biomass allometry. conclusions two procedures for biomass modelling were compared, namely, independent and additive. for the independent procedure of biomass modelling, linols models with scaled power transformations and y-intercepts and nlinols power models that lacked y-intercepts were compared for six components, three subtotals, and agt biomass. the linols models with scaled power transformations and y-intercepts provided superior results in contrast to nlinols power models without y-intercepts. the best-fitted component equations from linols models were further tested in an additive procedure. a system of additive equations in sur with different independent variables for each component (linadd3) showed better performance than linols, linadd1, and linadd2. besides, the linear sur model provided comparatively unbiased estimates for stem (eq. 10), branch (eq. 11), nf (eq. 13), of (eq. 14), foliage (eq. 16), crown (eq. 17), and agt (eq. 19), while linols showed comparatively better fitting statistics for bark (eq. 12), cone (eq. 15) and bole biomass (eq. 18) for the dataset of this study. since seven out of ten biomass components were well fitted with sur that provided lower variance by taking account of the existence of correlations among residuals of the component equations, we suggest that sur could be a superior method for fitting biomass equations. list of terms and abbreviations competing interests the authors declare that they have no competing interests. kc et al. new zealand journal of forestry science (2020) 50:7 page 12 parresol 2003; magalhães & seifert 2015; návar et al. 2002; návar, méndez et al. 2004; parresol 1999, 2001; zhao et al. 2015). a linear sur model (linadd3) of this study provided better results than linols, linadd1, and linadd2, in terms of goodness-of-fit statistics, standard error of estimates and residual plots. linadd3 fitted in sur was superior to the other two additive models since it considered the correlation between each component equation, and provided greater statistical efficiencies (carvalho & parresol 2003). in contrast to our results, a study reported that the additive model (denoted as con) that used the same independent variable for all components, similar to our linadd2 model, was statistically superior to the linear and nonlinear sur model with the different independent variables in parameter restriction (magalhães & seifert 2015). however, the authors (magalhães & seifert 2015) indicated that the con method had the limitation that it did not take into account the correlations among plant parts. applying sur to the system of additive equations with the same explanatory variables for each component does not provide precise estimation of biomass (srivastava & giles 1987). therefore, the linadd3, a sur model that consisted of different explanatory variables for each component is consistent with that of srivastava and giles (1987), which was more effective than the other two additive models. model linadd1 also consisted of the same explanatory variables for two-component equations such as stem and bark, and linadd2 consisted of the same independent variable for all component equations. therefore, linadd1 and linadd2 were not effective compared to linadd3. in addition, the individually calculated best equations for each component (linols) provided the least efficient biomass estimates for all components except for bark, cone and bole biomass, compared with the linear sur model (linadd3). researchers recommended using sur to estimate biomass as it provides greater statistical efficiency than separately calculated equations for each component (bi et al. 2010; bi et al. 2004; kozak 1970; návar et al. 2002; návar, méndez et al. 2004; parresol 2001). although this study focused on testing different procedures to fit biomass equations and highlighted the importance of sur, there could be concerns over the applicability of the resulting models given the size of the dataset from the trial. while applying these models, it is advisable to consider that the dataset used was relatively small with only 24 trees of the same age class, with d (8.2 to 28 cm), h (8.85 to 13.77 m), and crl (0.2 to 6 m). small sample sizes may provide biased estimates, as the allometric parameters are sample size dependent (duncanson et al. 2015). in addition, the models developed were based on only two out of five clones, selected across a range of treatment plots. extrapolation should be avoided as uncertain prediction errors are expected from the selected models. the site characteristics (duncanson et al. 2015; madgwick 1994; mason 2008), growth inputs (dong et al. 2015), tree age d diameter at breast height (cm) h total tree height (m) d2h product of h and d2 (cm2.m) crl crown length (m) nf new foliage of old foliage foliage sum of cone, nf and of biomass (kg) crown sum of foliage and branch biomass (kg) bole sum of stem and bark biomass (kg) components cone, nf, of, bark, branch, stem subtotal foliage, crown and bole agt sum of all components (above-ground total biomass) in kg λ variable-specific transformation coefficient β variable-specific parameter estimate ŷco, ŷnf, ŷof, ŷbr, ŷba, ŷst, ŷfol, ŷcr, ŷbol, ŷagt predicted biomass in kg for cone, nf, of, branch, bark, stem, foliage, crown, bole and agt, respectively. authors’ contributions egm, heb and mkc developed the concept for this manuscript. mkc performed the statistical analysis and drafted the manuscript. egm and heb contributed to the design of the study and with interpreting the results. egm, heb and gj participated in biomass sampling, sample processing and contributed to subsequent writing. all authors read and approved the final manuscript. acknowledgements we are particularly grateful to prof. john walker for his assistance in designing field and laboratory protocols. we would like to thank numerous undergraduate students at the school of forestry, university of canterbury (uc), new zealand, who contributed collecting the samples in the field. we want to thank school of forestry, university of canterbury (uc), new zealand for providing funding to undertake field work. we thank anonymous reviewers, and editors for their insightful comments and suggestions for improving this manuscript. the corresponding author thanks the ministry of foreign affairs and trade and the nzaid programme that provided an m.sc. research scholarship at the university of canterbury (uc), new zealand. references baskerville, g. (1972). use of logarithmic regression in the estimation of plant biomass. canadian journal of forest research, 2(1), 49-53. https://doi. org/10.1139/x72-009 beauchamp, j.j., & olson, j.s. (1973). corrections for bias in regression estimates after logarithmic transformation. ecology, 54(6), 1403-1407. https:// doi.org/10.2307/1934208 beets, p., & madgwick, h. (1988). above-ground dry matter and nutrient content of pinus radiata as affected by lupin, fertiliser, thinning, and stand age. new zealand journal of forestry science, 18(1), 43-64. beets, p., pearce, s., oliver, g., & clinton, p. (2007). root/ shoot ratios for deriving below-ground biomass of pinus radiata stands. new zealand journal of forestry science, 37(2), 267. beets, p., & pollock, d. (1987). accumulation and partitioning of dry matter in pinus radiata as related to stand age and thinning. new zealand journal of forestry science, 17(2), 246-271. bi, h., long, y., turner, j., lei, y., snowdon, p., li, y., harpur, r., zerihun, a., & ximenes, f. (2010). additive prediction of aboveground biomass for pinus radiata d. don plantations. forest ecology and management, 259(12), 2301-2314. https://doi.org/10.1016/j. foreco.2010.03.003 bi, h., turner, j., & lambert, m.j. (2004). additive biomass equations for native eucalypt forest trees of temperate australia. trees, 18(4), 467-479. https:// doi.org/10.1007/s00468-004-0333-z box, g.e., & cox, d.r. (1964). an analysis of transformations. journal of the royal statistical society. series b (methodological), 211-252. https:// doi.org/10.1111/j.2517-6161.1964.tb00553.x canadell, j., riba, m., & andres, p. (1988). biomass equations for quercus ilex l. in the montseny massif, northeastern spain. forestry: an international journal of forest research, 61(2), 137-147. https:// doi.org/10.1093/forestry/61.2.137 canga, l., aranda, u., khouri, a., & obregón, c. (2013). above-ground biomass equations for pinus radiata d. don in asturias. forest systems, 22(3), 408-415. https://doi.org/10.5424/fs/2013223-04143 carvalho, j.p., & parresol, b.r. (2003). additivity in tree biomass components of pyrenean oak (quercus pyrenaica willd.). forest ecology and management, 179(1-3), 269-276. https://doi.org/10.1016/ s0378-1127(02)00549-2 clutter, j.l., fortson, j.c., pienaar, l.v., brister, g.h., & bailey, r.l. (1983). timber management: a quantitative approach. new york: john wiley & sons, inc. cook, r., & weisberg, s. (2009). applied regression including computing and graphics. new york: john wiley & sons. cromer, r., barr, n., williams, e., & mcnaught, a. (1985). response to fertiliser in a pinus radiata plantation. 1. above-ground biomass and wood density. new zealand journal of forestry science, 15(1), 59-70. cunia, t., & briggs, r. (1984). forcing additivity of biomass tables: some empirical results. canadian journal of forest research, 14(3), 376-384. https://doi. org/10.1139/x84-067 cunia, t., & briggs, r. (1985). forcing additivity of biomass tables: use of the generalized least squares method. canadian journal of forest research, 15(1), 23-28. https://doi.org/10.1139/x85-006 dong, l., zhang, l., & li, f. (2015). developing additive systems of biomass equations for nine hardwood species in northeast china. trees, 29(4), 1149-1163. https://doi.org/10.1007/s00468-015-1196-1 duncanson, l., rourke, o., & dubayah, r. (2015). small sample sizes yield biased allometric equations in temperate forests. scientific reports, 5, 17153. https://doi.org/10.1038/srep17153 goicoa, t., militino, a., & ugarte, m. (2011). modelling aboveground tree biomass while achieving the additivity property. environmental and ecological statistics, 18(2), 367-384. https://doi.org/10.1007/ s10651-010-0137-9 green, e.j., & reed, d.d. (1985). compatible tree volume and taper functions for pitch pine. northern journal of applied forestry, 2(1), 14-16. https://doi. org/10.1093/njaf/2.1.14 henningsen, a., & hamann, j. (2007). systemfit: a package for estimating systems of simultaneous equations in r. journal of statistical software, 23(4), 1-40. https://doi.org/10.18637/jss.v023.i04 houghton, r.a. (1991). tropical deforestation and atmospheric carbon dioxide. climatic change, 19(1), 99-118. https://doi.org/10.1007/bf00142217 kozak, a. (1970). methods for ensuring additivity of biomass components by regression analysis. the forestry chronicle, 46(5), 402-405. https://doi. org/10.5558/tfc46402-5 lavery, p.b., & mead, d.j. (2000). pinus radiata: a narrow endemic from north america takes on the world. kc et al. new zealand journal of forestry science (2020) 50:7 page 13 in: d.m. richardson (ed.), ecology and biogeography of pinus (pp. 432-449). cambridge uk: cambridge university press. lewis, n.b., ferguson, i.s., sutton, w., donald, d., & lisboa, h. (1993). management of radiata pine. melbourne: inkata press pty ltd/butterworth-heinemann. macara, g.r. (2016). the climate and weather of canterbury. niwa science and technology series, taihoro nukurangi. retrieved from https://www. niwa.co.nz/static/web/canterbury_climatology_ second_ed_niwa.pdf. madgwick, h. (1983). estimation of the oven-dry weight of stems, needles, and branches of individual pinus radiata trees. new zealand journal of forestry science, 13(1), 108-109. madgwick, h. (1985). dry matter and nutrient relationships in stands of pinus radiata. new zealand journal of forestry science, 15(3), 324-336. madgwick, h. (1994). pinus radiata: biomass, form and growth. rotorua nz: hai madgwick. madgwick, h., jackson, d.s., & knight, p. (1977). aboveground dry matter, energy, and nutrient contents of trees in an age series of pinus radiata plantations. new zealand journal of forestry science, 7(3), 445468. magalhães, t.m., & seifert, t. (2015). biomass modelling of androstachys johnsonii prain: a comparison of three methods to enforce additivity. international journal of forestry research: 878402. https://doi. org/10.1155/2015/878402 mason, e. (2008). influences of silviculture, genetics and environment on radiata pine corewood properties: results from recent studies and a future direction. new zealand journal of forestry, 53(2), 26-31. mead, d. (2013). sustainable management of pinus radiata plantations. fao forestry paper no. 170. rome: fao. mead, d., draper, d., & madgwick, h. (1984). dry matter production of a young stand of pinus radiata: some effects of nitrogen fertiliser and thinning. new zealand journal of forestry science, 14(1), 97-108. moore, j. (2010). allometric equations to predict the total above-ground biomass of radiata pine trees. annals of forest science, 67(8), 806. https://doi. org/10.1051/forest/2010042 návar, j., gonzález, n., graciano, l., dale, v., & parresol, b. (2004). additive biomass equations for pine species of forest plantations of durango, mexico. madera y bosques, 10(2). návar, j., méndez, e., & dale, v. (2002). estimating stand biomass in the tamaulipan thornscrub of northeastern mexico. annals of forest science, 59(8), 813-821. https://doi.org/10.1051/forest:2002079 návar, j., méndez, e., nájera, a., graciano, j., dale, v., & parresol, b. (2004). biomass equations for shrub species of tamaulipan thornscrub of north-eastern mexico. journal of arid environments, 59(4), 657-674. https://doi.org/10.1016/j.jaridenv.2004.02.010 nixon, c., gamperle, d., pambudi, d., & clough, p. (2017). plantation forestry statistics: contribution of forestry to new zealand. wellington, nz. parresol, b. (1993). modeling multiplicative error kc et al. new zealand journal of forestry science (2020) 50:7 page 14 variance: an example predicting tree diameter from stump dimensions in bald cypress. forest science, 39(4), 670-679. parresol, b. (1999). assessing tree and stand biomass: a review with examples and critical comparisons. forest science, 45(4), 573-593. parresol, b. (2001). additivity of nonlinear biomass equations. canadian journal of forest research, 31(5), 865-878. https://doi.org/10.1139/x00-202 r core team. (2018). r: a language and environment for statistical computing. r foundation for statistical computing, vienna, austria. retrieved from https:// www.r-project.org/. rogers, d.l. (2002). in situ genetic conservation of monterey pine (pinus radiata d. don): information and recommendations. new zealand journal of forestry science, 32(23), 395-398. santa regina, i., tarazona, t., & calvo, r. (1997). aboveground biomass in a beech forest and a scots pine plantation in the sierra de la demanda area of northern spain. annals of forest science, 54(3), 261269. https://doi.org/10.1051/forest:19970304 sprugel, d. (1983). correcting for bias in log-transformed allometric equations. ecology, 64(1), 209-210. https://doi.org/10.2307/1937343 srivastava, v.k., & giles, d.e. (1987). seemingly unrelated regression equations models: estimation and inference (vol. 80): crc press. sutton, w.r.j. (1999). the need for planted forests and the example of radiata pine. new forests, 17(1), 95-110. https://doi.org/10.1023/a:1006567221005 toro, j., & gessel, s. (1999). radiata pine plantations in chile. new forests, 18(1), 33-44. https://doi. org/10.1023/a:1006597823190 von gadow, k., & hui, g. (2001). modelling forest development (vol. 57): springer science & business media. webber, b., & madgwick, h. (1983). biomass and nutrient content of 29-year-old pinus radiata stand. new zealand journal of forestry science, 13(2), 222-228. will, g. (1964). dry matter production and nutrient uptake by pinus radiata in new zealand. the commonwealth forestry review, 57-70. zeng, w. (2011). methodology on modeling of single-tree biomass equations for national biomass estimation in china. doctoral dissertation. chinese academy of forestry, beijing. zhao, d., kane, m., markewitz, d., teskey, r., & clutter, m. (2015). additive tree biomass equations for midrotation loblolly pine plantations. forest science, 61(4), 613-623. https://doi.org/10.5849/ forsci.14-193 zheng, c., mason, e., jia, l., wei, s., sun, c., & duan, j. (2015). a single-tree additive biomass model of quercus variabilis blume forests in north china. trees, 29(3), 705-716. https://doi.org/10.1007/s00468-0141148-1 zianis, d., muukkonen, p., mäkipää, r., & mencuccini, m. (2005). biomass and stem volume equations for tree species in europe. silva fennica monographs no. 4, 63 p. kc et al. new zealand journal of forestry science (2020) 50:7 page 15 linols equation tested models fitted best with i stem, bark, foliage, bole, agt ii cone iii branch iv v vi vii viii nf,crown ix of nlinols equation tested models fitted best with i stem, nf, of, cone, foliage, bole, agt ii iii iv bark v branch, crown vi vii table a.1: tested linear and nonlinear ordinary least-squares equations with their best-fit results. appendix kc et al. new zealand journal of forestry science (2020) 50:7 page 16 component m od el fit statistics parameter estimates r m se m a b m pe r se cv % r 2 io a β0 β1 β2 linols best models stem i 4.86 3.39 23.66 5.08 8.05 0.98 0.99 −6.691*** (0.493) 3.264*** (0.0986) branch iii 8.03 4.93 64.54 8.59 46.12 0.85 0.96 −5.803*** (0.666) 1.981*** (0.149) −0.049* (0.0196) bark i 0.88 0.6 0.77 0.92 16.99 0.94 0.98 −4.983*** (0.373) 1.416*** (0.074) nf viii 0.9 0.61 0.8 0.94 26.63 0.92 0.97 −3.803*** (0.792) 0.992*** (0.137) −0.054* (0.033) of ix 2.88 2.11 8.29 3.01 21.85 0.96 0.98 −2.87 *** (0.441) 1.064*** (0.079) −0.009** (0.0024) cone ii 2.74 1.78 7.52 2.86 75.23 0.48 0.78 −2.301** (0.769) 0.132*** (0.029) foliage i 4.56 3.18 20.8 4.76 22.59 0.94 0.98 −3.259*** (0.393) 1.213*** (0.079) crown viii 15.36 9.02 235.92 16.04 40.85 0.82 0.96 −0.731 ns (0.51) 0.693*** (0.089) −0.046* (0.021) bole i 5.29 3.77 27.95 5.52 8.05 0.98 0.99 −7.003*** (0.51) 3.403*** (0.102) agb i 17.13 11.42 293.5 17.89 16.59 0.96 0.98 −1.182*** (0.244) 1.309*** (0.049) nlinols best models stem i 5.27 3.88 27.72 5.5 8.71 0.97 0.99 0.292*** (0.069) 1.803*** (0.075) branch v 6.87 4.3 47.18 7.34 39.44 0.89 0.97 0.517 ns (1.126) 3.791*** (0.474) −3.199*** (0.645) bark iv 0.9 0.6 0.8 0.94 17.31 0.92 0.98 0.0021 ns (0.001) 0.931*** (0.071) nf i 0.88 0.61 0.77 0.92 26.14 0.9 0.97 0.001 ns (0.001) 2.868*** (0.269) of i 2.42 1.79 5.86 2.53 18.37 0.95 0.99 0.001 ns (0.001) 3.044*** (0.195) cone i 2.64 1.71 6.98 2.76 72.46 0.51 0.82 0.001 ns (0.002) 2.861** (0.762) foliage i 4.41 3.03 19.42 4.6 21.83 0.93 0.98 0.003 ns (0.002) 2.985*** (0.23) crown v 8.56 5.73 73.19 9.15 22.75 0.95 0.99 0.156 ns (0.19) 3.456*** (0.264) −1.987*** (0.379) bole i 5.76 4.25 33.15 6.01 8.77 0.97 0.99 0.302*** (0.072) 1.819*** (0.076) agb i 16.34 11.24 266.87 17.06 15.82 0.95 0.99 0.095* (0.045) 2.347*** (0.149) table a.2: goodness-of-fit statistics across best linols and nlinols regression models for each biomass component. table shows parameter estimates, their standard error between parentheses and significance indicated as: ns, non significant, *, p<0.05; **, p<0.01; ***, p<0.001. note: the estimated parameter values are presented in power-transformed scale. model i, ii, iii and so on refers to the best-fitted equations given in table a.2. trends in discount rates used for forest valuation in new zealand bruce manley new zealand school of forestry, university of canterbury, christchurch 8140, new zealand bruce.manley@canterbury.ac.nz (received for publication 5 april 2019; accepted in revised form 6 may 2019) abstract background: the discount rate is a key input for estimating the market value of a forest. data collected in surveys of forest valuers from 1997 to 2017 indicate a reduction in implied discount rate (idr) over time with lower idrs for larger forests. the purpose of this study was to formally analyse these trends. methods: there are three steps to the analysis: 1. relationships were developed for the idr data from 1997 to 2017; 2. further relationships were developed for idr data from 2009 to 2017 for which forest size (i.e. net stocked area) rather than just size class is available; and 3. detailed forest transaction data from 2011 to 2017 were used to develop a model to estimate average crop value from key variables including discount rate. this process allowed an analysis to confirm whether or not trends in discount rate with time and forest size were significant. results: analysis of the implied discount rate (idr) revealed that the reduction over time is significant and that the discount rate for large forests (>10,000 ha) has declined more than for smaller forests. analysis of data from 2009, for which forest size rather than size class is available, showed that forest area has a significant effect on idr. finally, the discount rate within the crop-value model, developed using transaction data collected since 2011, was found to vary with time and forest size; i.e. discount rate decreased as time or forest size increased. conclusions: overall, it can be concluded that the discount rates implicit in new zealand forest transactions have declined over time, with the scale of the reduction depending on forest size. new zealand journal of forestry science manley. new zealand journal of forestry science (2019) 49:4 https://doi.org/10.33494/nzjfs492019x48x e-issn: 1179-5395 published on-line: 20 june 2019 © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access value) of the forecast cashflows of a forest crop equal to the transaction value. in analysing forest transactions (i.e. forest sales), average value is sometimes used as a comparative measure to benchmark the market evidence (manley 2016a). however, average value is influenced by many factors. the influence of average stand age on average crop valuei for new zealand forest transactions from 2011 to 2017 is shown in figure 1. the variation at any age indicates that other factors are also important introduction discounted cashflow analysis has been widely used for forest valuation since faustmann (1849) used it to estimate both the value of land and the value of the tree crop growing on that land. a key input to discounted cashflow analysis is the choice of discount rate. a common approach in new zealand is to use the discount rate implied by recent transactions (manley 2018). the implied discount rate (idr) is estimated for a transaction as the discount rate that makes the npv (net present keywords: crop value; discount rate; plantation forestry manley. new zealand journal of forestry science (2019) 49:4 page 2 determinants of value. manley and bell (1992) showed that crop value is a function of age, species composition, site productivity, past silvicultural investment, terrain, and distance to market. given that no two forests are the same, average crop value of one forest only provides limited ability to estimate the value of another forest unless adjustments are made for the underlying determinants of forest value. in new zealand, idr provides a better way to summarise transaction evidence in order to estimate the market value of forests. keating (1990) reported the discount rates implicit in the sales price achieved in the sale of state plantations. since then, the idr rather than average value ($ ha-1) has been the key statistic extracted from transaction information. forest valuers in new zealand have been surveyed every two years about the discount rate they use to determine the market value of a tree crop (manley 1998, 1999, 2001, 2003, 2005, 2007, 2010, 2012, 2014, 2016b, 2018). as part of these surveys, valuers were also asked for transaction information; in particular, their estimate of the discount rate implicit in the transaction price of recent forest salesii. there is a large variation in reported idrs (figure 2). however, some trends in idr are apparent: • a reduction in idr over time; and • lower idrs for larger forests manley (2016a) analysed data from 27 new zealand forest transactions between 2011 and 2013 and developed a model to predict the market value of the tree crop from the discounted stumpage revenue associated with the average stand in each forest that was sold. inputs to the model (model 1) are average age (adjusted for species mix), total recoverable volume, harvest cost, distance to port and the proportion of pruned area. parameters in the model represent log price, pruned log price differential, unit transport cost and discount rate. the model has a residual standard error (rse) of $2154 ha-1 (table 1). value = [(a + b* pruned * 0.25 – c * distance to port – harvest cost)*volume30] / (1+d) 30-age (model 1) where: • value is average tree crop value ($ ha-1). in cases where the transaction involved land as well as the tree crop, land market value was deducted. • pruned is proportion of forest area that has been pruned. it is assumed that pruned logs make up 25% of the volume from a pruned crop. • distance to port is distance to nearest port (km). • harvest cost is average harvesting cost ($ m-3) including logging, loading, roading and overheads). • volume30 is total recoverable volume (m 3 ha-1) at age 30 years. • age is average forest age adjusted so that age is reduced by 15 years for the proportion of nonradiata pine area. this adjustment allows for the longer rotation age of these other species which are mostly douglas-fir. figure 1: average crop value for each of 78 new zealand forest transactions between 2011 and 2017 (i.e. dataset 3) plotted against average stand age. values are real $2017. average stand age is adjusted for species mix. figure 2: idrs (applied to current rotation pre-tax cashflows) for transactions reported in each of the discount rate surveys. forests are identified by size class (small <1000 ha; medium 1000 to 10,000 ha; large >10,000 ha). [figure 3 of manley (2018)] parameter estimate p-value at-wharf price ($ m-3) a 68.15 <<0.001 pruned log price premium ($ m-3) b 75.38 0.002 transport cost ($ m-3 km-1) c 0.103 0.019 discount rate d 0.086 <<0.001 table 1. parameter estimates for model 1 developed by manley (2016a) for 2011 to 2013 transactions. rse = $2154 ha-1 the coefficients (table 1) can be considered to represent: a = at-wharf price ($ m-3) b = premium paid for pruned logs ($ m -3) c = unit transport cost ($ m-3 km-1) d = discount rate with an extended data set available for 2011 to 2017, this model provides an alternative framework for analysing trends in discount rate evident from transaction information. the purpose of this paper was to formally analyse trends in idr. this was done using two different approaches: direct analysis of idrs and analysis of crop value with idr embedded within the valuation model. there were three steps to the analysis: 1. relationships were developed for the idr data from 1997 to 2017; 2. further relationships were developed for idr data from 2009 to 2017 for which net stocked area (rather than just size class) is available; and 3. detailed forest transaction data from 2011 to 2017 were used to develop a model to estimate average crop value from key variables including discount rate. this provides a complementary approach to confirm whether trends in the discount rate with time and forest size are statistically significant. it is important to differentiate this analysis from previous work that postulated that the discount rate used to value an investment should reduce over time; i.e. that the cashflows for an investment should be discounted at a variable rate, with a successively lower rate used for distant cashflows. arguments for this, as summarised by price (2011), include: (i) this is the way in which humans actually discount (henderson & bateman 1995); (ii) present generations are entitled to discount their own consumption, but not that of future generations (kula 1981); and (iii) when discount rate is uncertain, future cashflows should be discounted at lower rates (newell & pizer 2003). in contrast, in the analysis reported here a constant discount rate is used in valuing the future cashflows for a specific tree crop with only the current rotation considered. trends in the discount rate recognised by the market over time are evaluated. methods three sets of data were used, one for each of the three steps to the analysis. the first dataset was the full set of idr data from 1997 to 2017. the second dataset was the subset for which net stocked area (rather than just size class) was available. the third dataset was the subset for which detailed forest data is available. data came from the 11 surveys of forest valuers carried out between 1997 and 2017. in each survey, all known valuers active in the valuation of new zealand plantations were surveyed. although the response rate was always high (near 100%) the idr dataset did not include all transactions as some forests, particularly smaller forests, had been sold without subsequent analysis of idr by any of the respondents. 1. idr data from 1997 to 2017 data were available from the surveys of forest valuers for 119 transactions: 63 small forests, 33 medium forest and 23 large forests. the available data consisted of: • year of transaction. • valuers’ estimates of idr (applied to pretax cashflows for the current rotation) for each transaction. the average idr for each transaction was used in the analysis. • forest size class (small/medium/large) for each transaction. (small <1000 ha; medium 1000 to 10,000 ha; large >10,000 ha). data were analysed (using the r non-linear least squares function nls) for trends in discount rate over time allowing for potential differences with forest size class. the general structure of the model initially used was: r = e – [exp(f *t) – 1] (model 2) where: • r is implied discount rate applied to pre-tax cashflows • t is time (years since 1997) in this model, e is the estimated discount rate at t=0. if there is no trend with time, f=0 and r=e for all values of t. this model form was chosen because it allowed for the general trends evident in figure 2 and it subsequently allowed for a simple comparison between different size classes with only a single coefficient (f ) changing. 2. idr data from 2009 to 2017 data were available for 70 transactions: 43 small forests, 19 medium forest and 8 large forests. the available data consisted of: • year of transaction. • valuers’ estimates of idr (applied to pre-tax cashflows for the current rotation) for each transaction. again, the average idr for each transaction was used in the analysis. • forest size (net stocked area in hectares (nsa)) for each transaction. this dataset is a subset of the 1997 to 2017 dataset. initially the same analysis was conducted as for the 1997 to 2017 dataset. subsequently, forest size was included in the model as a continuous variable. manley. new zealand journal of forestry science (2019) 49:4 page 3 3. transaction data from 2011 to 2017 data were available for 78 transactions. the available data consisted of: • year of transaction. • crop value ($ ha-1). • forest size (net stocked area – nsa in hectares) for each transaction. • proportion of area in radiata pine. • proportion of area pruned. • average age. • average volume at age 28 years (m3 ha-1). • average harvesting cost ($ m-3 for logging, loading, roading and overheads). • distance to nearest port (km). both the crop value and harvesting cost were converted to real $june 2017 using ppi (producer price index, outputs level 1, all industries). the model developed by manley (2016a) for 2011 to 2013 data was re-estimated using all data. the main change was to assume a harvest age of 28 years rather than 30 years. this was done to better reflect current practice. subsequent steps allowed for: • changes in real price over time. • trends in discount rate with time and forest size. of the 78 transactions, 58 were included in the idr dataset for 2009 to 2017; i.e. those transactions from 2011 to 2017 for which an idr was provided for current rotation pre-tax cashflows. some survey respondents provided transaction details but idr data only for posttax cashflows or multiple rotations. other respondents provided details for some transactions but not idrs. results 1. idr data from 1997 to 2017 the initial model evaluated was model 2 and the initial discount rate level was estimated as 10.3% (table 2). the trend for a reducing discount rate over time was significant. the residual standard error (rse) was 1.65%. model 3 was evaluated to determine whether there are different trends over time for the three different forest size groups. in this model, the parameter f corresponds to the medium forest size. parameter g is not significant (table 3) indicating that there is no significant difference in trends in discount rate over time between small and medium forests. the difference in discount rate trends between medium and large forests (parameter h) is only marginally significant; i.e. not significant at the 5% probability level but significant at the 10% probability level. rse is 1.60%. r = e – [exp((f + g * small + h * large) * t) – 1] (model 3) where: small and large are dummy variables that take on the value of 1 if the forest is in that size class, and 0 otherwise. making the small forest size the base case in model 3 (with dummy variables for medium and large) confirmed that there was no significant difference in trend between small and medium forests but that the difference between small and large forests was significant (p=0.002). consequently model 4 was developed in which small and medium forests were pooled. the pooled small/medium group was significantly different from the large group (p=0.003). rse is 1.61% (table 4). r = e – [exp((f + h * large) * t) – 1] (model 4) where: large is a dummy variable that takes on the value of 1 if the forest is in that size class, and 0 otherwise. a variation of model 4 was analysed in which large forests were allowed to have a different level of initial discount rate. this was done by replacing e by the expression e + i*large. however, the coefficient i was not significant (p=0.65). consequently, model 4 was chosen as the preferred model. estimates from model 4 show the trends in discount rate over time with a lower discount rate for large forests (figure 3). manley. new zealand journal of forestry science (2019) 49:4 page 4 parameter estimate p-value e 10.311 <<0.001 f 0.0596 <<0.001 table 2. parameter estimates for model 2 fitted to data from 1997 to 2017. rse = 1.65% parameter estimate p-value e 10.453 <<0.001 f 0.0657 <<0.001 g -0.00963 0.226 h 0.0177 0.059 table 3. parameter estimates for model 3 fitted to data from 1997 to 2017. rse = 1.60% parameter estimate p-value e 10.453 <<0.001 f 0.0657 <<0.001 h 0.0243 0.003 table 4. parameter estimates for model 4 fitted to data from 1997 to 2017. rse = 1.61% 2. idr data from 2009 to 2017 when the 2009 to 2017 dataset was used, the time variable (t) represented the number of years since 2009. when fitted to this dataset, model 2 had an rse of 1.79% (table 5). the trend with time was significant. model 3 allowed for different trends for each size class. this resulted in a model with an rse of 1.71% (table 6). there was no significant difference in trends between medium and large forests (p=0.22) but the difference between small and medium forests was marginally significant (p=0.08). using small rather than medium as the base in model 3 showed that the difference between small and large forests was significant (p=0.008). again, there were indications of different trends in discount rate over time for different forest size classes. as net stocked area (nsa) was available for transactions in the 2009 to 2017 dataset, it was possible to treat forest area as a continuous variable rather than using discrete size classes. here, the natural logarithm of nsa was used to scale the effect of time (model 5). this model reduced rse to 1.67% (table 7). both model coefficients e and f were significant although residuals are large. there was no strong pattern when model residuals were plotted against ln(nsa) (figure 4). model estimates show trends over time for forests of different size (figure 5). forest area varied from 30 ha to 61,000 ha in the dataset used. r = e – [exp(f *ln(nsa)*t)-1] (model 5) a variation of model 5 allowing the initial level of the discount rate to vary with forest size was tested. this was done by replacing e by the expression e + i*ln(nsa). however, the coefficient i was not significant (p=0.88). manley. new zealand journal of forestry science (2019) 49:4 page 5 figure 3: estimates of idr over time for different forest size classes using model 4 fitted to data from 1997 to 2017. parameter estimate p-value e 9.648 <<0.001 f 0.177 <<0.001 g -0.0684 0.081 h 0.0462 0.217 parameter estimate p-value e 9.690 <<0.001 f 0.150 <<0.001 table 5. parameter estimates for model 2 fitted to data from 2009 to 2017. rse = 1.79% rse = 1.71% table 6. parameter estimates for model 3 fitted to data from 2009 to 2017. parameter estimate p-value e 9.617 <<0.001 f 0.0212 <<0.001 table 7. parameter estimates for model 5 fit to data from 2009 to 2017. rse = 1.67% figure 4: residuals (actual – predicted) for model 5 (fitted to data from 2009 to 2017) plotted against ln(nsa). figure 5: estimates of idr over time for five different forest sizes using model 5 fitted to data from 2009 to 2017. an alternative approach was to model the effects of forest size and time as additive rather than multiplicative (model 6). the result for this model was not as good as for model 5 with an rse of 1.71% (table 8). nevertheless the coefficients for both ln(nsa) and t were significant. r = e – [exp(f *ln(nsa)+g*t)-1] (model 6) 3. transaction data from 2011 to 2017 the initial step was to re-estimate the coefficients of the model developed by manley (2016a) to predict the market value of the tree crop from the discounted stumpage revenue associated with the average stand in each forest that was sold. the model (table 9) has an rse of $4269 ha-1. all four coefficients are significant. the a and b coefficients of model 7 are estimates of log prices over the period 2011 to 2017. however, over this period log prices increased (figure 6). consequently, the pf olsen log price index (converted to real $june 2017 and rebased so that june 2017 equals 1) was included in the model (model 7). this addition resulted in a better model with the rse reducing to $3518 ha-1 (table 10). value = [(a*lpi + b* lpi*pruned * 0.25 – c * distance to port – harvest cost)*volume28] / (1+d) 28-age (model 7) where lpi is the pf olsen log price index in real $june 2017 rebased so that $june 2017 equals 1. the final step was to allow for the discount rate to vary with time and forest size. based on the findings of the previous section, the parameter d was replaced by the expression used in model 7: e – [exp(f *ln(nsa)*t)-1] (used in model 8 to replace parameter d in model 7) where t is years since 2009. all coefficients in this model (model 8) are significant and the rse is further reduced to $3069 ha-1 (table 11). the residuals do not exhibit any strong patterns when plotted against ln(nsa), year or age (figures 7, 8 and 9). model estimates show the trends over time for forests of different sizes (figure 10). forest area varied from 10 ha to 132,000 ha in the dataset used. manley. new zealand journal of forestry science (2019) 49:4 page 6 table 8. parameter estimates for model 6 fitted to data from 1997 to 2017. rse = 1.71% parameter estimate p-value e 10.585 <<0.001 f 0.110 <<0.001 g 0.084 <<0.001 parameter estimate p-value at-wharf price ($ m-3) a 96.90 <<0.001 pruned log price premium ($ m-3) b 87.04 <<0.001 transport cost ($ m-3 km-1) c 0.209 <<0.001 discount rate d 0.104 <<0.001 table 9. regression coefficients for model 1 fitted using data from 2011 to 2017. the model form is identical to manley (2016a) except that a rotation age of 28 years rather than 30 years is assumed. rse = $4269 ha-1 figure 6: pf olsen log price index (in nominal $ and real $june 2017) from march 2010 to december 2018. this index is based on prices for log grades “weighted in proportions that represent a broad average of log grades produced from a typical pruned forest with an approximate mix of 40% domestic and 60% export supply” [source pf olsen]. conversion to real $june 2017 was done using ppi (producer price index, outputs level 1, all industries). table 10. regression coefficients for model 7 fitted using data from 2011 to 2017. parameter estimate p-value at-wharf price ($ m-3) a 95.53 <<0.001 pruned log price premium ($ m-3) b 82.39 <<0.001 transport cost ($ m-3 km-1) c 0.155 <<0.001 discount rate d 0.085 <<0.001 rse = $ 3518 ha-1 manley. new zealand journal of forestry science (2019) 49:4 page 7 discussion idr model analysis using data from 1997 to 2017 revealed that trends in discount rate over time were significant and provided a strong indication of a forest size effect. the forest size effect was clear once area was introduced as a continuous variable using the 2009 to 2017 dataset. the effects of size and time (i.e. years since 2009) can be modelled as being either multiplicative or additive. the former approach produced a better model and was adopted here. valuation model applying model 1 to 2011 to 2017 transaction data increased rse to $4269 ha-1 compared to $2154 ha-1 when only 2011 to 2013 data were used in the earlier manley (2016a) study. however, improvements were made by: • including the log price index – rse reduces to $3518 ha-1. • allowing discount rate to vary with time and forest size – rse further reduces to $3069 ha-1. the model structure determines the discounted stumpage revenue of the average stand in each forest allowing for changes in real log price and discount rate over time. a comparison of the model parameters with typical industry values revealed that: • the value for at-wharf price (parameter a) was $95.53 m-3 in model 7 and $97.14 m-3 in model 8. average at-wharf prices for unpruned logs in 2017 were $110–120 m-3. including lower priced pulplogs or chiplogs reduced this to a level closer to the estimated model parameters. • pruned log prices were typically $50–60 m-3 higher than average unpruned log prices. the pruned log price premium implicit in transaction information ($82.39 m-3 in model 7 and $97.34 m-3 in model 8) was greater than this. • the unit transport cost ($0.15 m-3 km-1 in parameter estimate p-value at-wharf price ($ m-3) a 97.14 <<0.001 pruned log price premium ($ m-3) b 97.34 <<0.001 transport cost ($ m-3 km-1) c 0.1764 <<0.001 base discount rate e 0.121 <<0.001 discount rate adjuster f 0.0009051 <<0.001 table 11. regression coefficients for model 8 fitted using data from 2011 to 2017. this model allows for discount rate to vary with forest size over time. rse = $ 3069 ha-1 figure 7: residuals for model 8 (fitted to data from 2011 to 2017) plotted against ln(nsa). figure 8: residuals for model 8 (fitted to data from 2011 to 2017) plotted against year. figure 9: residuals for model 8 (fitted to data from 2011 to 2017) plotted against adjusted age (i.e. average forest age adjusted for species mix). figure 10: estimates of idr over time for five different forest sizes using model 8 fitted to data from 2011 to 2017. model 7 and 0.18 m-3 km-1 in model 8) aligned well with the value of $0.15 t-1 km-1 used in the agrihq june 2017 report. • the discount rate of 8.5% estimated for model 7 is higher than the average reported idr of 7.6% for 2016–17 in manley (2018). the general conclusion is that the parameter estimates are realistic given that models 7 and 8 are simplifications. for example, they consider only discounted stumpage revenue and ignore annual overhead costs and land rents. an attempt was made to include annual costs in the model but the associated parameter was not significant. one interpretation of the pruned log price coefficient is that the market for forests recognises a greater premium for pruned logs than is evident in the current log market. the coefficient was calculated assuming that 25% of the volume of a pruned stand will be extracted as pruned logs. the pruned log price coefficient will vary inversely with the assumed proportion decreasing or increasing. comparison of two different approaches initially idr is the dependent variable used for modelling in models 2 to 6. at the final stage, in models 7 and 8, crop value is the dependent variable with the model structure including a term to represent the discount rate. these two distinct approaches are discussed below. in calculating idrs, valuers make assumptions about future volumes, prices and costs. the idr is the balancing variable in making the npv of future cashflows equate to the known transaction value. consequently, the estimated idr depends on the assumptions made about future volumes (harvesting strategy and yields), prices and costs. if different valuers have different views about these factors, they will generate different idrs for the same transaction. this variation is evident in the published survey results. the alternative approach uses average crop value as the dependent variable and predicts it based on: • maturity (current age relative to a target rotation age of 28 years). • species composition (proportion of area in species other than radiata pine). • site productivity (volume at age 28). • past silvicultural investment (proportion of area pruned). • terrain/harvest difficulty (average harvesting cost). • distance to market. model inputs were provided by valuers. again, there was some variation in these inputs between valuers. however, this approach does not rely on valuers’ inputs on two key variables: • log price. this is an output from the model. • harvesting strategy. harvest age is fixed at age 28 years. table 2: confusion matrix there are differences between approaches in the coefficients estimated for discount rate. the e coefficient was 9.6 (table 7) for model 5 (idr model) and 12.1 (table 11) for model 8 (average crop value model). the latter model estimated a greater impact of forest size on discount rate (figure 10) compared to the idr model (figure 5). trends in discount rates the consistent findings from the three elements of this study are that the reduction in discount rate over time is statistically significant and that the impact of forest size is also significant. the effect of forest size on discount rate revealed by analysis of the 1997 to 2017 idr data (figure 3) was muted because only size class data were available. differences became greater once forest area was included as a continuous variable (see figures 5 and 10). the greatest differentiation with forest size was exhibited in the analysis of transaction data for 2011 to 2017 (figure 10). the purpose of the models was to document past trends. although the models contain significant coefficients they only partially explain the variation in idr or crop value model residuals are large. the models should not be used to forecast future discount rates as past trends are unlikely to be a good predictor of the future. limitations the exponential functional form was chosen to model reductions in discount rate because it allows for the pattern of reduction apparent in figure 2. the variant used (model 2) is simple with only two coefficients, one for level and another that, when multiplied by time, gives the rate of change. in the extension to model 5, the rate of change is also determined by ln(nsa). an inherent limitation is that for a given forest size, the rate of change is set and the reduction in discount rate increases with time. this is evident in figure 5 where, for example, the reduction between years 7 and 8 is greater than the reduction between years 6 and 7. conclusions must also be tempered by the limited number of large-forest transactions in the datasets. for example, the dataset of 2011 to 2017 transactions includes only eight large forests over 10,000 ha: four 2011–13 transactions, three 2014–15 transactions and one 2016–17 transaction. the four transactions since 2014 have considerable leverage over the specific results obtained. implications for forest valuers given the limitations of model form and data, together with the large model residuals, forest valuers should not use the models developed here to set discount rates. the ultimate determinant of discount rates for new zealand plantations will continue to be the market. forest valuers are attempting to mimic the market when they act for buyers and sellers in a transaction or when they estimate the fair value of a tree crop for company reporting. the key message for forest valuers is that the market is recognising lower discount rates for larger forests. manley. new zealand journal of forestry science (2019) 49:4 page 8 discount rates declared in financial reporting discount rates being used for financial reporting have tended to decline since 2013 (figure 11). the companies shown are all large (>10,000 ha) apart from invercargill city forests, sunchang forestry nz and te waihou plantations, all of which are of medium size (1000 to 10,000 ha). the discount rates shown are those used by the independent forest valuer in determining crop value. there have been different trends for individual companies with discount rates not declining for some but markedly for others. the discount rates used in 2010 to 2012 are lower than those indicated by the idr model while those in recent years are higher than shown in figure 5 or figure 10. overall the reductions have not been at the same rate indicated by the reductions in idrs for medium/large companies. comparison to sewall survey us forest valuation company james w. sewall company regularly carries out its own survey of discount rates. in the sewall investor survey, active investors are asked “what is the ‘base’ discount rate (real, pre-tax, before timo fees & expenses) required for successful bids on generic timberland investments in the u.s. now?” respondents are subsequently asked to “provide the discount rate premium over the u.s. base rate” for a range of international forest investments including new zealand pine. since 2011, the average discount rate has declined for both us timberland and new zealand pine (figure 12). again the reduction is not as steep as that indicated by the analysis of idrs. why the decline in discount rate over time? the reduction in idr over time has been driven by the supply and demand for plantations by the international investment community. medium and large new zealand plantations have been actively sought when put up for sale. in many cases purchasers have been pension funds, often investing via timos (timber investment management organisations). since 2010, the total timo forest area has plateaued (figure 13). as noted by new forests (2017), “a significant proportion of the high-quality timberland estates in us, australia and new zealand is already in institutional ownership....”. it is also stated that “with rising allocations to real asset investments and a finite pool of such investments, institutional investors will increasingly seek to hold high-quality assets in their portfolio.” purchasers in new zealand have not only been pension funds or timos. wood processors and log traders have also purchased forests to guarantee their supply chain. as a consequence, there has been an increasing demand for a limited area of investment-grade plantations. in recent years, medium as well as large forests have had multiple parties seeking to buy them. the demand for medium to large plantations in new zealand has led to higher prices being paid for them with a consequent reduction in idr. conclusions the three steps in the analysis all produce results with a similar pattern. initial analysis using idr data from 1997 to 2017 reveals that the reduction in discount rates over time is significant and that the discount rate for large forests (>10,000 ha) has reduced more than for smaller forests. analysis of data from 2009, for which forest area rather than area class is available, shows that forest area has a significant effect on idrs. finally, detailed manley. new zealand journal of forestry science (2019) 49:4 page 9 figure 12: average discount rates for usa and new zealand from james w. sewall surveys of forest investors. source: james w. sewall company figure 11: discount rates declared in financial reporting for some new zealand-registered companies with annual reports in the public domain. all rates are applied to current rotation pre-tax cashflows. figure 13:timo forest area under management. source: timberlink llc. transaction data collected since 2011 are used to develop a model to predict crop value from the average stand in each forest. discount rates estimated by this model are found to vary with time and forest size; i.e. the discount rate decreases as time or forest size increases. overall, it can be concluded that the discount rates implicit in new zealand forest transactions have reduced, with the scale of the reduction depending on forest size. ethics approval not applicable. consent for publication not applicable. availability of data please contact the author for further information. competing interests the author declares that he has no competing interests. acknowledgements forest valuers are thanked for contributing to discount rate surveys since 1997. scott downs provided data on the pf olsen log price index, james w. sewall company provided data on the sewall investor survey while timberlink llc provided data on the forest area under management by timos. two anonymous referees provided very useful feedback on a draft of the manuscript. references faustmann, m. (1849). “on the determination of the value which forest land and immature stands possess for forestry.” english edition edited by m. cane, oxford institute commonwealth forestry paper 42, 1968, entitled “martin faustmann and the evolution of discounted cash flow”. henderson, n., & bateman, i. (1995). empirical and public choice evidence for hyperbolic social discount rates and the implications for intergenerational discounting. environmental resource economics, 5, 413–423. keating, j.e. (1990, december). lessons from the forest marketplace. nz forest industries, 16–17. kula, e. (1981). future generations and discounting rules in public sector investment appraisal. environment and planning a 13, 899–910. manley, b. (1998). discount rates used for forest valuation results of a pilot survey. new zealand forestry, 42(4), 47. manley, b. (1999). discount rates used for forest valuation results of 1999 survey. new zealand journal of forestry, 44(3), 39–40. manley, b. (2001). discount rates used for forest valuation results of 2001 survey. new zealand journal of forestry, 46(3), 14–15. manley, b. (2003). discount rates used for forest valuation results of 2003 survey. new zealand journal of forestry, 48(3), 29–31. manley, b. (2005). discount rates used for forest valuation results of 2005 survey. new zealand journal of forestry, 50(3), 7–11. manley, b. (2007). discount rates used for forest valuation results of 2007 survey. new zealand journal of forestry, 52(3): 21–27. manley, b. (2010). discount rates used for forest valuation results of 2009 survey. new zealand journal of forestry, 54(4): 19–23. manley, b. (2012). discount rates used for forest valuation results of 2011 survey. new zealand journal of forestry, 56(4), 21–28. manley, b. (2014). discount rates used for forest valuation results of 2013 survey. new zealand journal of forestry, 59(2), 29–36. manley, b. (2016a). analysis of new zealand forest transactions 2011–2013. new zealand journal of forestry, 60(4), 29–32. manley, b. (2016b). discount rates used for forest valuation results of 2015 survey. new zealand journal of forestry, 61(2), 28–35. manley, b. (2018). discount rates used for forest valuation results of 2017 survey. new zealand journal of forestry, 63(2), 35–43. manley, b., & bell, a. (1992). analysis of the value of the state plantations sold in 1990. new zealand journal of forestry, 37(3), 22–27. new forests. (2017). 2017 timberland investment outlook. https://newforests.com.au/wp-content/ uploads/2017/09/2017-timberland-investmentoutlook-web-1.pdf newell, r.g., & pizer, w.a. (2003). discounting the distant future: how much do uncertain rates increase valuation? journal of environmental economics and management, 46, 52–71. price, c. (2011). optimal rotation with declining discount rate. journal of forest economics, 17, 307–318. endnotes i crop value is the value of the tree crop and excludes the value of the land. ii in the surveys valuers were asked for implied discount rates for both pre-tax and post-tax cashflows. only the former was considered in this analysis. manley. new zealand journal of forestry science (2019) 49:4 page 10 selection of pinus spp. progenies in lavras (minas gerais, brazil) at 36 months of age érick martins nieri1,*, antônio carlos porto2, rodolfo soares de almeida3, lucas amaral de melo3, eduardo willian resende3, generci assis neves4, luana maria dos santos3 and júlio cézar tannure faria3 1 department of forestry sciences, federal university of the south and southeast of pará, são félix do xingu, pará, 68380-000, brazil 2 department of biology, federal university of lavras, cp 3037, lavras, minas gerais,, 37200-900, brazil 3 department of forestry sciences, federal university of lavras, street address, lavras, minas gerais, 37200-900, brazil 4 empresa resineves, itapeva, são paulo, 18400-000, brazil *corresponding author: ericknieri@unifesspa.edu.br (received for publication 6 june 2020; accepted in revised form 31 january 2022) abstract background: the selection of superior genotypes and adapted to the edaphoclimatic conditions of the region of the introduction produces gains in productivity for forest stands. the objective of this study was to select progenies of pinus spp. planted in lavras, minas gerais (mg), brazil. methods: the experimental site was located on dystrophic haplic cambisol. the progeny test was designed as a randomised complete block with 30 replicates and single plot. the treatments corresponded to one progeny of pinus massoniana, three pinus caribaea var. bahamensis and 33 pinus caribaea var. hondurensis arranged with a 3 x 3 m spacing. the traits height (h), diameter at breast height (dbh) and crown projection area (cpa) were measured at 36 months of age. results: the results showed that heritability in the narrow sense was 0.24 for dbh, 0.27 for h and 0.50 for cpa. the dbh and h traits showed a high-magnitude positive correlation. the p7, p15, p27, p31 and p33 progenies showed better performance than the other progenies for the evaluated traits. direct and indirect selection showed similar gains, which favors the use of indirect selection; i.e., when selecting progenies for dbh, progenies with better performance in h are also selected. additionally, dbh may be used at advanced ages given the difficulty of measuring height. the progeny of pinus caribaea var. hondurensis showed superior performance compared with pinus caribaea var. bahamensis and pinus massoniana for the region of lavras, mg. conclusions: this study suggests the possibility of expanding the production of pinus caribaea var. hondurensis in the region of lavras with progenies p7, p15, p27, p31 and p33, because in the initial assessments they showed greater adaptability to the edaphoclimatic conditions. nevertheless, performing a future selection with the aim of evaluating resin production is recommended. new zealand journal of forestry science nieri et al. new zealand journal of forestry science (2022) 52:4 https://doi.org/10.33494/nzjfs522022x116x e-issn: 1179-5395 published on-line: 25/02/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access conditions of the planted region (ibá 2017). the genus pinus occurs naturally throughout north america, central america, asia, europe and north africa and is considered a group of fast-growing exotic plants in brazil. when well-managed, this genus has supplied the market with the raw material that was historically supplied by native forest species such as araucaria spp. (shimizu 2008, corrêa et al. 2012). introduction the demand for timber products in brazil and worldwide tends to increase with increasing population growth. globally, it is estimated that approximately 250 million additional hectares of planted forests will be needed by 2050, and to meet this demand, it will be necessary to use breeding to select for genotypes of fast-growing, productive species that are adapted to the edaphoclimatic keywords: genotypes; genetic parameters; behaviour; adaptability. mailto:ericknieri@unifesspa.edu.br http://creativecommons.org/licenses/by/4.0/), nieri et al. new zealand journal of forestry science (2022) 52:4 page 2 the genus pinus includes pinus caribaea, a fastgrowing species that is tropical in origin and is used for wood production and, in particular, for resin extraction. this species is among those most frequently used for homogeneous reforestation in various parts of the world due to its adaptability to diverse climates and the broad application/destination of its products, which have undergone genetic improvement over recent decades. these improvements include an increase in volumetric yield and in the yield of trees with straight trunks, fewer whorls and thicker branches, which maximise the industrial usefulness of the wood (missio et al. 2004, shimizu 2008, silva et al. 2011). genetic breeding is a key tool that enables gains in productivity because the selection of superior genotypes involves the assessment of traits of interest by estimating parameters, correlations and genetic gains that allow the selection of superior individuals between and within pinus progenies (manfio et al. 2012). however, genetic breeding can be very expensive, because selection is performed through evaluation stages and with adult individuals, and requires several selection cycles. to minimise this expense, several studies have attempted to select genotypes at younger ages, as this requires less time for selection and makes it possible to change the objectives of selection (xavier et al. 2009). early selection enables breeders of forest species to identify the traits of juvenile trees that are of economic interest at the end of a rotation, i.e., to predict at the juvenile stage the performance of an adult individual, which reduces the time required to complete a selection cycle. by reducing the test time, more rapidly identifying inferior genotypes and more promptly recommending new genotypes for commercial cultivation, i.e., by shortening the breeding cycle, the program of obtaining more productive genotypes is accelerated and the time between generations is decreased (gonçalves et al. 1998, massaro et al. 2010). selection at younger ages has been successful for the traits of height (h) and diameter at breast height (dbh) in eucalyptus spp. at two years of age (massaro et al. 2010). these traits at young and old ages were highly positively correlated, indicating that this type of selection can generate significant gains (moraes et al. 2014). there are studies in the literature on pinus caribaea and pinus massoniana in other countries, such as those by sanchez et al. (2014), liu et al. (2013) and hodge & dvorak (2001). however, there are no studies involving the selection of genotypes in juvenile pinus caribaea and pinus massoniana in the region of lavras, minas gerais (mg), brazil. thus, the aim of the present study was to select pinus spp. progenies grown in lavras at 36 months of age. methods the experimental site was located on the campus of the federal university of lavras (universidade federal de lavras, ufla) at 21°14’19.6”s latitude and 44°58’28.5”w longitude and at 905 metres above sea level. according to the köppen classification, the climate is cwb (highland tropical) with mild summers and a mean annual temperature of 19.6 °c, a mean annual rainfall of 1511 mm, a mean annual relative humidity of 76.2% and a total annual evaporation of 901.1 mm (alvares et al. 2013). the progeny test was conducted at the ecotone between the cerrado (brazilian savanna) and seasonal semideciduous forest on soil categorized as dystrophic haplic cambisol using the brazilian soil classification system (embrapa 2013). the test consisted of 37 progenies, including 33 progenies of pinus caribaea var. hondurensis (p1–p33), three progenies of pinus caribaea var. bahamensis (ft1– ft3) and one progeny of pinus massoniana (fec) obtained from open-pollinated and phenotypically selected maternal trees located in commercial plantations of resineves agroflorestal ltda., itapeva, são paulo, brazil. the seedlings were produced by the company in 55 cm³ tubes filled with commercial substrate. a soil chemical analysis was performed on the 0-20 cm soil layer (table 1) using random zigzag sampling. a total of 15 points were collected and a composite was then created. soil preparation included cleaning the entire area with the aid of a tractor (4 x 2 front-wheel drive), ant control (sulfluramid baits) and weed control (broad-spectrum herbicide with postemergence action). in the month of planting, the soil was prepared by heavy harrowing of the entire area. planting was performed in mid-september 2015. the spacing used was 3 x 3 m, and the holes were opened at the time of planting with the aid of a dibble. in each hole, along with the pinus seedling, spores of mycorrhizal fungi that had been collected from fruiting bodies found within pinus spp. stands were applied, as well as 200 ml of a solution of water-retaining polymer at a concentration of 1 g of polymer per plant. after planting, a solution of termiticide was applied to the stem of each seedling for the control of subterranean termites. the experiment was arranged in a randomised complete block design with 30 replicates and single-tree plots. to control for edge effects, the entire experiment ph k p ca mg al h + al v zn fe mn cu b s mg dm-3 cmol dm-3 % mg dm-3 6 32 1.7 1.2 0.3 0.1 1.6 48.8 2.2 59.9 3.3 2.1 0.3 7.0 table 1: soil chemical analysis of the 0-20 cm layer of the experimental site. v = base saturation index. was surrounded by two rows of plants of the same species. the treatments consisted of three progenies of pinus caribaea var. bahamensis, 33 progenies of pinus caribaea var. hondurensis and one progeny of pinus massoniana. during the weeks following planting, it was necessary to irrigate the seedlings using a truck with a water reservoir and irrigators to ensure the initial survival of the seedlings until the beginning of the rainy season. during the entire period, leaf-cutter ants were monitored, and weeds were controlled by mechanised mowing (4 x 2 front-wheel drive) and the use of broadspectrum postemergence herbicides. at 36 months after planting, the survival, h, dbh and crown projection area (cpa) were evaluated. the percent survival was quantified as the total number of initial individuals for each species divided by the number of surviving individuals at 36 months and multiplied by 100. the cpa was obtained by measuring the crown projection between plants within the planting row (cpbp) and between the planting rows (cpbr). after these measurements, the cpa was calculated using the formula presented by nieri et al. (2018) (equation 1). cpa = [(cpbr) × (cpbp) × π] / 4 (1) genetic parameters were estimated for each measured trait via the restricted maximum likelihood/ best linear unbiased prediction (reml/blup) method using a mixed models analysis, which was performed with the r package asreml using the following model (equation 2): y = xb + za + e (2) where y is the vector of the observations, b is the vector of the replicate effects (fixed), a is the vector of the individual additive genetic effects (random) and e is the vector of the residuals (random). the variables x and z represent the incidence matrices for the cited effects. the solution of the fixed effects (b) and random effects (a) of the model was obtained by solving the following equation (henderson et al., 1959): where a-1 is the inverse matrix of the additive relationship based on the pedigree using the inbreeding index. the following distributions and mean and variance structures were assumed: the progenies of pinus caribaea var. bahamensis and pinus massoniana were considered fixed, obtaining the best linear unbiased estimator (blues). the estimated genetic parameters were individual heritability (hi 2), individual heritability in the narrow sense (hr 2) (equation 3), mean heritability of the progenies (hm 2) (equation 4), selective accuracy at the progeny level (rgg) (equation 5), the coefficient of genetic variation (cvg) (equation 6) and the coefficient of residual variation (cve) (equation 7). (3) (4) (5) (6) (7) where σi 2 is additive variance, σg 2 is genetic variance, σe 2 is residual variance, σf 2 is phenotypic variance, pev is the prediction error variance of genotypic values, and m is the mean of the analysed traits. the genetic (rg), (eequation 8), environmental (re) (equation 9) and phenotypic (rf) (equation 10) correlations were estimated between the h, dbh and cpa traits evaluated at 36 months after planting. (8) (9) (10) where rgxy , rfxy and rexy are the genetic, phenotypic and environmental correlation coefficients, respectively, and σgxgy 2, σfxfy 2and σexey 2 are the mean genetic, phenotypic and environmental covariances, respectively, between the traits the in pairs extracted by the model using the sum of pairs of traits as response variable (kempthorne 1957). the gain from direct selection (equation 11) and indirect selection for h, dbh and cpa were calculated using the rank-summation index (imm) proposed by mulamba & mock (1978) for the evaluated traits (equation 12). to calculate gain, 30.3% was used as the selection intensity. (11) (12) nieri et al. new zealand journal of forestry science (2022) 52:4 page 3 table 1: description of the study sites ̭ ̭ ̭ ̭ ̭ ̭ ̭ ̭ ̭̭̭ ̭ where sg is the selection gain, n is the number of selected progenies, i(mm)j is the index value associated with individual j, ri is the ranking of an individual relative to the jth trait and n is the number of traits considered in the index. to test the significance of the random effects of the model, an analysis of deviance was performed using the likelihood ratio test (lrt) for the three traits (dbh, h and cpa). results the mean percent survival of the progenies of pinus caribaea var. bahamensis, pinus caribaea var. hondurensis and pinus massoniana was 99% (table 2). these results were initial variables used to verify the adaptability of species to environmental conditions. the analysis of deviance for the genetic and statistical parameters of the progenies of pinus caribaea var. hondurensis revealed significant differences among the progenies tested for the variables dbh, h and cpa in the region of lavras at 36 months of age (table 3). the random progenies of pinus caribaea var. hondurensis had a mean (μa) dbh of 6.74 cm, a mean h of 4.34 m and a mean cpa of 4.18 m2 plant-1, which were considered satisfactory for the lavras region. the estimation of heritability in the narrow sense (hr 2) can be considered intermediate for dbh (0.24) and h (0.22) and high for the cpa (0.50). the individual heritability and the mean heritability of progenies for dbh and h are considered intermediate and high, respectively, and for cpa, both are considered high. these parameters are fundamental and reflect that the tested genotypes have dbh, h and cpa traits that will be transmitted to their offspring and will consequently gain from the selection of superior genotypes. when evaluating experimental precision and accuracy, the selective accuracy was 0.84 for dbh, 0.83 for h and 0.91 for cpa. the coefficient of genetic variation expresses the genetic variability found among the progenies, and the higher this coefficient, the greater the variability among the progenies. the coefficients were 19.52 for dbh, 15.80 for h and 26.44 for cpa. when considering the individual genetic variation coefficient, i.e., within the progenies, the values obtained were 9.76 for dbh, 7.90 for h and 13.22 for cpa. figure 1 shows that the p7, p15, p27, p31 and p33 progenies of pinus caribaea var. hondurensis were among the 11 best progenies with regard to h, dbh and cpa. nieri et al. new zealand journal of forestry science (2022) 52:4 page 4 however, the progenies derived from pinus caribaea var. bahamensis (ft1, ft2 and ft3) and pinus massoniana (fec) were ranked in the middle and lower parts of the plots. genetic correlation involves a mechanism that allows the explanation of the joint variation in two variables/ traits. table 4 shows a strong positive genetic correlation between dbh and h, dbh and cpa, and h and cpa. table 4 shows that the variables dbh and height had a strong genetic, phenotypic, and environmental correlation. the dbh and cpa had a strong genetic correlation. however, for environmental and phenotypic correlation, an average correlation was observed. when correlating h and cpa, there was a strong genetic correlation and average environmental and phenotypic correlation, as observed for dbh in correlation with the cpa. parameter diameter at breast height (dbh) height (h) crown projection area (cpa) μ 5.871 4.082 3.758 μa 6.740 4.349 4.188 μf 5.002 3.814 3.327 σi 2 1.732** 0.472** 1.226** σe 2 4.200 1.267 0.922 σf 2 5.932 1.739 2.148 hi 2 0.29 0.27 0.57 hr 2 0.24 0.22 0.50 hm 2 0.70 0.69 0.83 rgg 0.84 0.83 0.91 cvg 19.52 15.80 26.44 cve 34.79 29.28 32.40 μ fec¹ 4.03 3.73 3.46 μ ft1² 6.41 4.44 3.77 μ ft2² 3.75 2.93 2.33 μ ft3² 5.77 4.13 3.72 species survival (%) pinus caribaea var. bahamensis 99 pinus caribaea var. hondurensis 98 pinus massoniana 100 mean 99 table 2: percent survival of the progenies of pinus caribaea var. bahamensis, pinus caribaea var. hondurensis measured at 36 months in lavras, mg. table 3: genetic parameters and deviance analysis for dbh, h and cpa of 33 p. caribaea var. hondurensis progenies at 36 months after planting in lavras, mg. overall mean of progenies (μ), mean of the random progenies (μa), mean of the fixed progenies (μf), additive variance (σi 2), environmental variance (σe 2), phenotypic variance (σf 2), individual heritability (hi 2), individual heritability in the narrow sense (hr 2), mean heritability of the progenies (hm 2), selective accuracy at the progeny level (rgg), genetic coefficient of variation (cvg), residual coefficient of variation (cve). ** significant at 1% probability of error by the likelihood ratio test (lrt). ¹ mean of the pinus massoniana progeny. ² mean of each pinus caribaea var. bahamensis progeny. ̭ ̭ ̭ ̭ ̭ ̭ ̭ when using and comparing different selection strategies and methods at the progeny level, it was observed that, as expected, the gain with direct selection was better than for indirect for all variables. however, for dbh and h, the difference was approximately 1.3% (figure 2). the direct selection showed greater gains for the cpa than selection by the method of the mulamba e mock (1978). the results of indirect selection indicate that for dbh and h, gains similar to direct selection were obtained, which reflects the high degree of correlation between the traits. discussion the observed results demonstrate the potential of these species planted in lavras, given that macedo et al. (2018) emphasises that survival is the initial parameter to determine the species potential for a region. the progenies of pinus caribaea var. hondurensis had results similar to those reported by durigan et al. (2004) in assis, são paulo, at three years after planting; by donadoni et al. (2010) in vilhena, rondônia, at three years of age and by tambarussi et al. (2018) in canoinhas, santa catarina, at four years after planting. the latter observed an average of 5 cm dbh, 4 m h and 1.63 m² plant-1 cpa, which demonstrates the potential of the tested progenies for the region of lavras. the growth in dbh and h for pinus massoniana and pinus caribaea var. bahamensis were similar in comparison with studies by liu et al. (2013) and romanelli & sebbenn (2004), respectively. according to the classification of resende (2002), heritabilities below 0.15 were classified as low magnitude, heritabilities between 0.15 and 0.50 are classified as intermediate magnitude, and heritabilities above 0.50 are classified as high magnitude. the heritabilities observed were similar to those obtained by tambarussi et al. (2010), souza et al. (2017) and tambarussi et al. (2018) and are higher than those reported by silva et al. (2011). according to ramalho et al. (2012), heritability can be conceptualised as the proportion of genetic nieri et al. new zealand journal of forestry science (2022) 52:4 page 5 figure 1: ranking of progenies of pinus caribaea var. bahamensis, pinus caribaea var. hondurensis and pinus massoniana for dbh, h and cpa in lavras, mg, at 36 months of age. table 4: genetic, phenotypic and environmental correlations among dbh, h and cpa for progenies of pinus caribaea var. hondurensis in the region of lavras, mg, at 36 months after planting. treatment height crown projection area dbh rg 0.830** 0.874** re 0.723** 0.657** rf 0.822** 0.561** height rg 0.861** re 0.658** rf 0.638** **significant by the t test at 1% probability of error. variation present in the total phenotypic variance, which according to souza et al. (2017), estimates the reliability of the phenotypic value as an indicator of reproductive value; for this reason, the mentioned heritabilities are frequently included in expressions aimed at gain prediction, which in turn assists breeders in making decisions in genetic improvement programs. resende & duarte (2007) reported that, to promote selection and recombination for plant breeding, recommended selection accuracy values are above 0.70. the selective accuracy values were similar to those obtained by tambarussi et al. (2018), who selected pinus elliottii x pinus caribaea hybrids at four years after planting and found an accuracy of 0.91 for dbh and 0.93 for h. souza et al. (2017), working with pinus caribaea var. hondurensis at five years old, obtained values of 0.53 for dbh and 0.70 for h. the results obtained for the evaluated traits were higher than those reported by silva et al. (2011), souza et al. (2017) and pires et al. (2013). coefficients above 7% are considered high, and the higher this coefficient is, the greater the genetic variability among progenies (sebbenn et al. 1998). consistent with the present study, sebben et al. (2010) conducted a survey at an altitude of 562 m with a mean annual rainfall of 1400 mm on soils classified as typic allic dystrophic red latosol (according to the brazilian soil classification system) and found that pinus caribaea var. hondurensis exhibited growth in dbh and h of 2.6% and 4%, respectively, which were higher values than those observed in pinus caribaea var. bahamensis. for dbh and h, strongly positive genetic, phenotypic and environmental correlations were observed. these results favored the use of indirect selection, where progenies ranked with high dbh values consequently showed high h values (sebbenn et al. 2010). consistent with these results, tambarussi et al. (2018) found a strong genetic correlation between dbh and h (0.84). for the correlation between cpa and dbh or h, intermediate phenotypic and environmental correlations were observed, which indicates that these variables are similarly influenced by the environment, as the correlation corresponding to environmental causes is considered the total effect of all the variable factors in the environment that caused a positive correlation (falconer & mackay 1996). a highly correlated response to selection is expected when it is performed on variables with high-magnitude positive correlations; i.e., selecting for one trait will result in a similar response for the other trait (macedo et al. 2013). based on these results, it is recommended to use and measure only the dbh to select the best progenies in stands with advanced ages, because determining plant height requires increased time and labor, and can therefore increase the cost of the evaluation. direct selection performed on one trait can affect gains in other traits (missio et al. 2004). when selecting pinus elliottii x pinus caribaea hybrids using dbh, tambarussi et al. (2018) noted that the genetic correlations indicated probable positive selection effects for both dbh and h as well as for tree cpa (falconer & mackay 1996). these results suggest that selection using the nieri et al. new zealand journal of forestry science (2022) 52:4 page 6 figure 2: gains from selection for dbh, h and cpa using the direct method and the method of mulamba & mock (1978) for progenies of pinus caribaea var. hondurensis in lavras, mg, at 36 months. method of mulamba & mock (1978)–indirect selection– enables gains because the traits have a strong genetic correlation (missio et al. 2004, bhering et al. 2012). the gains obtained under direct and indirect selection were higher than those reported by tambarussi et al. (2010), who found gains of 5.99% for dbh and 5.82% for h for pinus caribaea var. hondurensis at 14 years of age. according to macedo et al. (2018) and nieri et al. (2018), the dbh and h measurements enable evaluation of the adaptability of genotypes introduced to exotic environments and their potential for establishment, as they express the adaptability and vigor of the seedlings under ecological conditions observed in the field after planting, given the magnitude of genotype x environment interactions in loco. to obtain progenies with greater adaptability for the lavras region, and because the genetic breeding program is in its early stages, it is recommended to select 30.3% of the progenies for future crossings to obtain greater gains. however, for this to occur, genetically different progenies must be selected. the progenies of pinus caribaea var. hondurensis showed greater genetic variability and higher performance and consequently better adaptability, thus, it is recommended to continue genetic improvement with the progenies of pinus caribaea var. hondurensis. tests with more families are recommended to certify the competitiveness and suitability of the progenies of pinus caribaea var. bahamensis and pinus massoniana in lavras. however, the superiority of pinus caribaea var. hondurensis is indicative of its better performance in the region. most studies claim that there is a positive correlation between resin yield and growth in dbh and cpa (liu et al. 2013). according to these same authors, dbh has a higher correlation with resin production (0.70) and therefore should be considered when selecting superior progenies for resin production. it is essential to select the progenies that showed the best adaptability to the lavras region, because the breeding program begins with the selection of promising progenies with adaptability using basic traits such as dbh and h. therefore, an initial selection using dbh may include progenies with high resin yield, as the progenies were initially selected by resin production. therefore, selection at a younger age is recommended to determine the resin yield among the best progenies selected. conclusions the progenies of pinus caribaea var. hondurensis showed higher performance than pinus caribaea var. bahamensis and pinus massoniana for the region of lavras, minas gerais, brazil, at 36 months after planting. the progenies of pinus caribaea var. hondurensis have significant genetic variation, which allows for the continuity of the genetic breeding programme. indirect selection was adequate for the dbh and h traits, mainly due to the high genetic correlation between these traits. based on this study, the expansion of pinus caribaea var. hondurensis in the region of lavras, mg, is suggested using the progenies p7, p15, p27, p31 and p33, because in the initial assessments, these progenies showed greater adaptability to the local edaphoclimatic conditions. future selection should be performed at a more advanced age to include the evaluation of resin production. competing interests the authors declare that they have no competing interests. authors' contributions emn, lam and gan designed the study. emn, lam, rsa, ewr, lms and jctf participated in the experimental part. emn, acp, lam, lms and rsa analysed the data and wrote the manuscript. emn and lam reviewed the manuscript. all authors read and approved the final manuscript. references alvares, c.a., stape, j.l., sentelhas, p.c., de moraes, g.j.l., sparovek, g. (2013). köppen’s climate classification map for brazil. meteorologische zeitschrift, 22(6), 711-728. https://doi.org/10.1127/09412948/2013/0507 bhering, l.l., laviola, b.g., salgado, c.c., sanchez, c.f.b., rosado, t.b., alves, a.a. (2012). genetic gains in physic nut using selection indexes. pesquisa agropecuária brasileira, 47(3), 402-408. https:// doi.org/10.1590/s0100-204x2012000300012 corrêa, p.r.r., auer, c.g., dos santos, a.f., higa, a.r. (2012). seleção precoce de progênies de pinus radiata a sphaeropsis sapinea. ciência florestal, 22(2), 275281. https://doi.org/10.5902/198050985734 donadoni, a.x., pelissari, a.l., drescher, r., rosa, g.d. (2010). relação hipsométrica para pinus caribaea var. hondurensis e pinus tecunumanii em povoamento homogêneo no estado de rondônia. ciência rural, 40(12), 2499-2504. https://doi. org/10.1590/s0103-84782010001200010 durigan g, contieri w.a, de melo a.c.g., nakata h. (2004). crescimento e sobrevivência de espécies arbóreas plantadas em terreno permanentemente úmido em região de cerrado. in: vilas bôas o., durigan g. (orgs.) pesquisas em conservação e recuperação ambiental no oeste paulista: resultados da cooperação brasil-japão. páginas e letras editora e gráfica, são paulo, br, pp. 447-456. embrapa (2013). sistema brasileiro de classificação de solos. 3rd ed. brasilia, brazil: embrapa. falconer, d.s., mackay, t.f. (1996). introduction to quantitative genetics. 4th ed. london: longman group. gonçalves, p.s., bortoletto, n., fonseca, f.s., bataglia, o.c., ortolani, a.a. (1998). early selection for growth nieri et al. new zealand journal of forestry science (2022) 52:4 page 7 https://doi.org/10.1127/0941-2948/2013/0507 https://doi.org/10.1127/0941-2948/2013/0507 https://doi.org/10.1590/s0100-204x2012000300012 https://doi.org/10.1590/s0100-204x2012000300012 https://doi.org/10.5902/198050985734 https://doi.org/10.1590/s0103-84782010001200010 https://doi.org/10.1590/s0103-84782010001200010 vigor in rubber tree genotypes in northwestern são paulo state (brazil). genetics and molecular biology, 21(4), 620-630. https://doi.org/10.1590/ s1415-47571998000400018 henderson, c.r., kempthorne, o., searle, s.r., von krosigk, c.m. (1959). the estimation of environmental and genetic trends from records subject to culling. biometrics, 15, 192-218. https:// doi.org/10.2307/2527669 hodge, g.r., dvorak, w.s. (2001). genetic parameters and provenance variation of pinus caribaea var. hondurensis in 48 international trials. canadian journal of forest research, 31(3), 496-511. https:// doi.org/10.1139/x00-189 industria brasileira de árvores. relatório 2017. http://iba.org/images/shared/biblioteca/iba_ relatorioanual2017.pdf accessed 23 october 2018. kempthorne, o. (1957). an introduction to genetic statistics. new york: john wiley & sons inc. 545 p. liu, q., zhou, z., fan, h., liu, y. (2013). genetic variation and correlation among resin yield, growth, and morphologic traits of pinus massoniana. silvae genetica, 62(1), 38-43. https://doi.org/10.1515/ sg-2013-0005 macedo, r.l.g., vale, a.b., carvalho, f., venturin, n., nieri, e.m. (2018). eucalipto em sistemas agroflorestais. 2nd ed. lavras, brazil: ufla. macedo, h.r., freitas, m.l.m., moraes, m.l.t., zanata m., sebben, a.m. (2013). variação, herdabilidade e ganhos genéticos em progênies de eucalyptus tereticornis aos 25 anos de idade em batatais-sp. scientia forestalis, 41(100), 533-540. manfio, c.e., motoike, s.y., resende, m.d.v., santos, c.e.m., sato, a.y. (2012). avaliação de progênies de macaúba na fase juvenil e estimativas de parâmetros genéticos e diversidade genética. pesquisa florestal brasileira, 32(69), 63-68. https://doi. org/10.4336/2012.pfb.32.69.63 massaro, r.a.m., bonine, c.a.v., scarpinati, e.a., paula, r.c. (2010). viabilidade de aplicação da seleção precoce em testes clonais de eucalyptus spp [early selection viability in eucalyptus spp clonal tests]. ciência florestal, 20(4), 597-609. https://doi. org/10.5902/198050982418 missio, r.f., cambuim, j., moraes, m.l.t., paula, r.c. (2004). seleção simultânea de caracteres em progênies de pinus caribaea morelet var. bahamensis [seleção simultânea de caracteres em progênies de pinus caribaea morelet var. bahamensis]. scientia forestalis, 66(1), 161-168. moraes, c.b, freitas, t.c.m, pieroni, g.b., resende, m.d.v., zimback, l., mori, e.s. (2014). estimativa de parâmetros genéticos para seleção precoce de clones de eucalyptus para a região de ocorrência de geadas. scientia forestalis, 42(104), 219-227. mulamba, n.n., mock, j.j. (1978). improvement of yield potential of the eto blanco maize (zea mays l.) population by breeding for plant traits. egyptian journal of genetics and cytology, 7(1), 40-51. nieri, e.m., macedo, r.l.g., martins, t.g.v., melo, l.a., venturin, r.p., venturin, n. (2018). comportamento silvicultural de espécies florestais em arranjo para integração pecuária floresta. floresta, 48(2), 195202. https://doi.org/10.5380/rf.v48i2.54744 pires, v.c.m., martins, k., bôas, o.v., freitas, m.l.m., sebbenn, a.m. (2013). variabilidade genética de caracteres silviculturais em progênies de polinização aberta de pinus caribaea var. bahamensis. scientia forestalis, 41(1), 113-119. ramalho, m.a.p, dos santos, j.b., pinto, c.a.b.p., souza, e.a., gonçalves, f.m.a, souza, j.c. (2012). genética na agropecuária. lavras, brazil: editora ufla. resende m.d.v. (2002). software selegen reml/blup. colombo, brazil: embrapa. resende, m.d.v., duarte, j.b. (2007). precisão e controle de qualidade em experimentos de avaliação de cultivares. pesquisa agropecuária tropical, 37(3), 182-194. romanelli, r.c., sebbenn, a.m. (2004). parâmetros genéticos e ganhos na seleção para a produção de resina em pinus elliottii var. elliottii no sul do estado de são paulo. instituto florestal, 16(1), 11-23. sanchez, m., ingrouille, m.j., cowan, r.s., hamilton, m.a., fay, m.f. (2014). spatial structure and genetic diversity of natural populations of the caribbean pine, pinus caribaea var. bahamensis (pinaceae), in the bahamas archipelago. botanical journal of the linnean society, 174(3), 359-383. https://doi. org/10.1111/boj.12146 sebbenn, a.m., siqueira, a.c.m.f., kageyama, p.y., machado, j.a.r. (1998). parâmetros genéticos na conservação da cabreúva myroxylon peruiferum l.f. allemão. scientia forestalis, 53(1), 31-38. sebbenn, a.m., boas o.v., max, j.c.m., freitas, m.l.m. (2010). estimativa de parâmetros genéticos e ganhos na seleção para caracteres de crescimento em teste de progênies de pinus caribaea var. hondurensis e var. bahamensis, em assis-sp. revista instituto florestal, 22(2), 279-288. shimizu, j.y. (2008). pinus na silvicultura brasileira. colombo, brazil: embrapa florestas. silva, j.m., aguiar, a.v., mori, e.s., moraes, m.l.t. (2011). variação genética e ganho esperado na seleção de progênies de pinus caribaea var. caribaea em selvíria, ms. scientia forestalis, 39(90), 241-252. souza t.s., dos santos w., deniz l.d., alves, a.p.o., shimizu j.y., sousa v.a, aguiar a.v. (2017). variação genética em caracteres quantitativos em pinus caribaea var. nieri et al. new zealand journal of forestry science (2022) 52:4 page 8 https://doi.org/10.1590/s1415-47571998000400018 https://doi.org/10.1590/s1415-47571998000400018 https://doi.org/10.2307/2527669 https://doi.org/10.2307/2527669 http://iba.org/images/shared/biblioteca/iba_relatorioanual2017.pdf http://iba.org/images/shared/biblioteca/iba_relatorioanual2017.pdf https://doi.org/10.1515/sg-2013-0005 https://doi.org/10.1515/sg-2013-0005 https://doi.org/10.4336/2012.pfb.32.69.63 https://doi.org/10.4336/2012.pfb.32.69.63 https://doi.org/10.5902/198050982418 https://doi.org/10.5902/198050982418 https://doi.org/10.5380/rf.v48i2.54744 https://doi.org/10.1111/boj.12146 https://doi.org/10.1111/boj.12146 hondurensis. scientia forestalis, 45(113), 177-185. https://doi.org/10.18671/scifor.v45n113.18 tambarussi, e.v., marques, e.g., andrejow, g.m.p., peres, f.s.b., pereira, f.b. (2018). análise dialélica na avaliação do potencial de híbridos de pinus elliotti x pinus caribaea para a formação de populações de melhoramento. scientia forestalis, 46(119), 395403. https://doi.org/10.18671/scifor.v46n119.07 tambarussi, e.v., sebbenn, a.m., moraes, m.l.t., zimback, l., palomino, e.c., mori, e.s. (2010). estimative of genetic parameters in progeny test of pinus caribaea morelet var. hondurensis barret & golfari by quantitative traits and microsatellite markers. bragantia, 69(1), 39-47. https://doi.org/10.1590/ s0006-87052010000100006 xavier, a., wendling i., silva, r.l. (2009). silvicultura clonal. viçosa, brazil: editora ufv. nieri et al. new zealand journal of forestry science (2022) 52:4 page 9 https://doi.org/10.18671/scifor.v45n113.18 https://doi.org/10.18671/scifor.v46n119.07 https://doi.org/10.1590/s0006-87052010000100006 https://doi.org/10.1590/s0006-87052010000100006 bioefficacy of euphorbia peplus latex as an antifeedant and insecticide against gonipterus platensis larvae on eucalyptus globulus amanda huerta1*, ítalo chiffelle2, carolina arias1, tomislav curkovic3 and jaime e. araya3 1 university of chile, faculty of forestry and nature conservation sciences, department of silviculture and nature conservation, p.o. box 9206, santiago, chile. 2 university of chile, faculty of agronomic sciences, department of agroindustry and enology, p.o. box 1004, santiago, chile. 3 university of chile, faculty of agronomic sciences, department of crop protection, p.o. box 1004, santiago, chile. *corresponding author: ahuerta@uchile.cl (received for publication 25 october 2021; accepted in revised form 22 october 2022) abstract background: gonipterus platensis (marelli) is part of the gonipterus scutellatus species complex which consists of three species that have spread beyond their natural ranges. due to its high reproductive potential and a capacity for intense defoliation by both larvae and adults, g. platensis causes tree growth loss and stem deformities. the antifeeding effect and insecticide efficacy of latex from petty spurge, euphorbia peplus l. (euphorbiaceae), on larvae of g. platensis, were evaluated through bioassays, with a view to its integrated management. methods: eucalyptus globulus leaves treated by immersion in euphorbia peplus latex solutions were infested with five third-instar larvae, and the area consumed was determined at 24 h. the antifeeding effect of five latex solutions applied with a brush on eucalyptus globulus leaves, using a random experimental design of six treatments and five replications, was also evaluated. larval mortality was recorded daily and analysed by an anova and tukey’s test. the lc50 (lethal concentration to kill 50% of the individuals) was calculated by probit analysis and chi2 tests were performed. results: the ethanolic solutions of the latex caused strong antifeeding effect, with total inhibition of larval feeding at all solutions. larval mortality increased significantly over time up to 76%, due to the effect of the ethanol latex solutions, in all solutions by day 6. the lower lc50 values were 0.049 and 0.012% w/v on days 5 and 6, respectively. conclusions: these results indicate that euphorbia peplus latex is a promising bioinsecticide and a possible alternative for integrated pest management. however, further tests should be carried out. new zealand journal of forestry science huerta et al. new zealand journal of forestry science (2023) 53:2 https://doi.org/10.33494/nzjfs532023x195x e-issn: 1179-5395 published on-line: 17/11/2022 © the author(s). 2023 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access for intense defoliation by both larvae and adults, g. platensis causes tree growth loss and stem deformities (dos santos bobadilha et al. 2019), which are strongly related to climatic conditions (adame et al. 2022). in chile, g. platensis was detected in 1998 and since then it has affected eucalypt plantations economically (servicio agrícola & ganadero 2010). in chile, eucalypts are highly valued for their rapid growth and the quality of the wood for pulp (rua et al. 2020), and eucalypt plantations currently cover ~860 thousand ha (instituto forestal 2020). introduction the eucalyptus snout beetle, gonipterus platensis (marelli) (coleoptera: curculionidae), is part of the gonipterus scutellatus gyllenhal species complex which consists of three species that have spread beyond their natural ranges. as a group, the species complex is a global pest of eucalyptus, and g. platensis has the largest distribution outside of the three species in the complex (mapondera et al. 2012; schröder et al. 2020). due to its high reproductive potential and a capacity keywords: botanical insecticide; eucalyptus snout beetle; eucalyptus globulus; gonipterus scutellatus; petty spurge; pichoga. mailto:ahuerta@uchile.cl http://creativecommons.org/licenses/by/4.0/), huerta et al. new zealand journal of forestry science (2023) 53:2 page 2 in south africa, control trials with pyrethroids were carried out, which were quickly abandoned due to the high cost of the treatments and because of the successful introduction of a biological control agent (romanyk and cadahía 2002). in chile, two biological control agents have been introduced, the egg endoparasitoids anaphes nitens huber and a. tasmaniae huber & prinsloo (hymenoptera: mymaridae), which are native to australia. these species have been released in several areas, and although they have reduced the g. platensis populations, the problem persists probably due to the diversity of habitats and climate (corporación nacional forestal 2012, 2017). in portugal, a. nitens was found to be ineffective at altitudes above 400 m (ceia et al. 2021). in brazil, entomopathogenic fungi (beauveria spp. and metarhizium anisoplliae) have been evaluated against g. platensis adults, identifying fungal strains with superior lethality than existing commercialized strains (jordan et al. 2021). pest management could be improved further by controlling other life stages (larvae, pupae, and adults), using other techniques such as increasing the diversity of biological control agents, exploring environmentally friendly biopesticides, selecting and/ or developing eucalyptus genotypes with tolerance to infestation, and silvicultural control (schöroder et al. 2020). despite growing evidence of environmental damage and human health concerns, the global use of synthetic insecticides has continued to grow substantially over the past 50 years (isman 2020). thus, it is necessary to find alternatives replacing synthetic pesticides to control of pest insects. to reduce their negative effects, new natural botanical insecticides have been developed, based on extracts of leaves, fruits, or other plant structures, with diverse results. these compounds are biodegradable, reduce pest resistance appearance, and have a lesser impact on flora and fauna, among other properties (amri et al. 2013). the latex from plants in the family euphorbiaceae, especially those in the genus euphorbia, has toxic, irritant, and medicinal effects (bittner et al. 2001; docampoa et al. 2010), but it also has insecticide properties attributed mainly to triterpenes, flavonoids, alkaloids, coumarins, cyanogenetic glycosides and tannins and others secondary metabolites (mendivelso et al. 2003; ogbourne et al. 2004). petty spurge, euphorbia peplus l., is a 15 to 40 cm tall herbaceous toxic plant, typical of gardens, sidewalks, orchards, and ruderal sites in europe, temperate asia, north africa, north and south america and oceania, in places with a temperate climate, shady, humid, and fertile soils (mendivelso et al. 2003). this species has been tested for antimicrobial, analgesic, antipyretic, anti-inflammatory (ali et al. 2013), antifeeding (hua et al. 2017) and insecticide activities (ghramh et al. 2019). this work evaluated the antifeeding effect and insecticide efficacy of euphorbia peplus latex on g. platensis larvae, through laboratory bioassays, with a view to integrated pest management (ipm). gonipterus platensis intensely defoliates eucalypt plantations worldwide, without adequate management in most regions. therefore, an antifeedant and insecticide based on euphorbia peplus latex could be valuable as an alternative control method with little environmental impact. methods sampling and latex preparation the chemical composition of euphorbia peplus latex was determined by hua et al. (2017), with 13 terpenoid compounds, including 12 diterpenoids and an acyclic triterpene alcohol. for the present study, latex was extracted from fresh stems (approx. 2.5 kg) of euphorbia peplus plants collected during spring at the antumapu campus of the university of chile in santiago (33° 34’s, 70° 37’w). approximately 5 ml of latex (density 0.958 g/ml) were extracted manually from euphorbia peplus plants collected at random, to avoid effects of individual plants. fresh stems were cut, and the latex was collected by gravity in the chemistry laboratory, department of agroindustry and enology, faculty of agronomic sciences, university of chile, in santiago at room temperature (19±2°c) and held in a refrigerator at -4°c until use in bioassays. sampling and rearing of insects young larvae of g. platensis were collected in the summer from eucalyptus globulus labill trees in san bernardo, metropolitan region (33° 58’s, 70° 70’w), and were taken in cloth bags to the forestry entomology laboratory, faculty of forestry and nature conservation sciences, university of chile, santiago. these larvae were placed in petri dishes (10 cm diameter) lined with whatman no. 1 filter paper humidified with distilled water. larvae were kept in a bioclimatic incubator (model jspc-420c, jsr research inc., chungehungnan-do, korea) at 20±3°c, 60±6% rh, and a photoperiod of 14:10 (day:night) until they reached the third instar. these conditions were maintained in the following bioassays. evaluation of the antifeeding effect the antifeeding effect of the latex ethanolic solutions was evaluated following the method described by defagó et al. (2006). petri dishes (10 cm diameter) containing five third instar larvae and two eucalyptus globulus leaves of the same size, one an untreated control (only ethanol at 96% v/v) and the other one treated by immersion (1 min and after air drying) in the solutions of latex (10, 30, 50, 70, and 100% w/v), with five replicates, and kept in the climatic chamber. after 24 h, the percentage of leaf area eaten (either treated or control ones) by larvae was determined by the imagej programme (schneider et al. 2012). the percentage of antifeeding effect was calculated as (1 [t / c]) x 100, where t and c were the consumed levels of the treated and control leaves, respectively. foliar area consumed was analysed through the wilcoxon test (p < 0.05), and the antifeeding effect was analysed using anova followed by tukey tests between solutions (p < 0.05). evaluation of the insecticide efficacy the insecticide efficacy bioassay on g. platensis was conducted using third instar larvae placed on fresh eucalypt leaves. five third instar larvae were placed on fresh eucalypt leaves in 10 cm diameter petri dishes lined with slightly moistened filter paper at the bottom and fresh washed eucalypt leaves, with the petiole wrapped in wet cotton. six treatments (five latex solutions [10, 30, 50, 70, and 100% w/v] plus a control) were compared; the control was treated only with ethanol at 96% v/v, using a simple random design with five replicates. the treatments were applied with a hairbrush on both sides of washed eucalypt leaves. dead and live larvae were counted daily in six times during the test. daily and total mortality (% ± standard error) were determined. mortality results were normalized using bliss degrees to stabilize the variance error. after checking the assumptions of normality and homoscedasticity, the data were subjected to anova (6 x 6). when significant differences occurred between treatments or days, tukey tests were run (p ≤ 0.05). for the statistical analysis, the first factor was set to be the treatment and the second the days after the application. data were adjusted mathematically to obtain the best-fitting function to obtain the lc50 (lethal concentration to kill 50% of the individuals) using a probit model described previously (robertson et al. 1984). data fit to the probit model was confirmed with a chi2 test. all statistical analysis were done using infostat (2009) software. results antifeeding effect the ethanol latex extracts had a strong antifeeding effect, with almost total inhibition of feeding by g. platensis in all treatments (table 1). insecticide efficacy during the test, mortality of g. platensis larvae increased due to the effect of the ethanol euphorbia peplus latex solutions. statistical analyses indicated significant differences between the cumulative mean mortality (%) of g. platensis larvae resulting from the treatments and the control [f5,25 (0.05) = 16.7] and among the number of days after the initial exposure [f5,25 (0.05) = 12.4]). on day 6, with the lowest solution (10% w/v) the minimum mortality was 76%. the greatest mortality occurred with the two highest solutions at 96% and 100%, which indicates the substantial insecticide potential of euphorbia peplus latex on these larvae (figure 1). lc50 the results from the probit analysis indicate an lc50 decreasing over time. the lowest and most promising lc50s were 0.012 and 0.049% w/v on day 6 and 5, respectively (table 2). thus, the latex from euphorbia peplus can be considered an effective insecticide for g. platensis since more than 50% mortality was obtained after 4 days exposure even to a low-concentration solution of the extract (<0.15% w/v). discussion antifeeding effects while there are several trials with botanical insecticides in the euphobia genus, there are very few with euphorbia peplus or extracts tested on larvae g. platensis. antifeeding results were almost 100% effective at all the solutions evaluated, thus the latex of euphorbia peplus can protect against feeding by larvae of g. platensis on leaves of euphorbia globulus. similarly, the latex of euphorbia peplus showed potent antifeedant activity against the larvae of cotton bollworm (helicoverpa armigera (hübner) (lepidoptera: noctuidae) (hua et al. 2017). a somewhat slower feeding inhibition was observed in larvae of pieris brassicae and spodoptera littoralis on day 2, with 100% inhibition in s. littoralis and a 30% solution of euphorbia peplus latex (chaieb et al. 2001). our results show a more potent activity of euphorbia peplus latex on g. platensis larvae since an almost 100% antifeeding effect was obtained at 10% w/v after only 24 h. however, a 100% antifeedant effect on third-instar larvae of g. platensis was obtained with even lower concentrations of 2.4% w/v aqueous extracts of cestrum parqui l’heritier (solanaceae) leaves, a south american shrub (huerta et al. 2021). larval mortality the ethanolic solutions of the latex obtained from euphorbia peplus caused >76% larval mortality on g. platensis in all treatments (starting at 10% w/v) six days after exposure. a consistent dose-dependent insecticide effect of the euphorbia peplus latex was observed with >90% larval mortality at >70% w/v. by comparison, the insecticide effect of extracts from azadirachta indica a.juss. (meliaceae) neem oil on g. platensis larvae, reached only a maximum of 40% mortality at 3% w/v concentration on day 8 after exposure (pérez otero huerta et al. new zealand journal of forestry science (2023) 53:2 page 3 table 1: description of the study sites latex (% w/v) foliar area consumed antifeeding effect (%)1 untreated (control) treated 10 2.48 0.03* 98.9 ± 1 a 30 2.55 0.01* 99.3 ± 1 a 50 2.38 0.00* 99.8 ± 1 a 70 2.91 0.02* 100.0 ± 0 a 100 1.73 0.00* 100.0 ± 0 a table 1: mean antifeeding effect (%) of ethanolic solutions of euphorbia peplus latex on feeding by gonipterus platensis third instar larvae on eucalyptus globulus leaves in a free choice test, at several latex solutions (% w/v) after 24 h. 1 means sharing a letter do not differ significantly according to tukey’s test (p <0.05%). * significant differences between the consumption of treated and untreated leaves in a wilcoxon’s test (p <0.05%). et al. 2003). in another study, ethanol extracts from new and mature leaves of the peruvian pepper tree schinus molle l. (anacardiaceae) were evaluated at 3.4% and 4.8% w/v, respectively, leading to 100% and 94.7% cumulative mortality 10 days after application (chiffelle et al. 2017). the insecticide effect of leaf, seed, and root extracts of antelaea azadiracha l. (meliaceae) was also evaluated on larvae of the eucalypt weevil, providing 72.5% mortality on day 7 (santolamazza & fernández 2004), a considerably lower effect than in our tests with euphorbia peplus latex. finally, 2.4% w/v aqueous extracts from cestrum parqui leaves provided 52% mortality of third-instar larvae of g. platensis on day 6, being more effective than low concentrations of euphorbia peplus latex tested here (huerta et al. 2021). huerta et al. new zealand journal of forestry science (2023) 53:2 page 4 the lower lc50 was 0.012 % w/v for the ethanolic latex extract on day 6 in our study. in comparison, the lc50 of ethanol extracts from new and mature leaves of s. molle against larvae of g. platensis on day 6 was 0.79 and 0.63% w/v, respectively (chiffelle et al. 2017), therefore less effective than those achieved in our current study. euphorbia peplus ethanol leaf extract was also lethal against fourth instar larvae of culex pipiens linnaeus (diptera, culicidae), with an lc50 of 0.14 % w/v after 24 h exposure (ghramh et al. 2019). although aqueous extracts of leaves of cestrum parqui were shown to be effective as an antifeedant and an insecticide against g. platensis larvae, the lc50 on day 6 was 1.84% w/v (huerta et al. 2021), higher than that obtained in the present study. this means less plant material of euphorbia peplus latex is required than of cestrum parqui leaves to achieve a similar control effect. 1 means with different small letters, and capital letters, indicate significant differences between solutions and days after initial exposure, respectively, according to tukey tests (p≤0.05). figure 1: cumulative mean mortality (% ± standard error) of g. platensis third instar larvae resulting from ethanolic solutions of latex from euphorbia peplus at one to six days from initial exposure. time (days) slope (mean ± error standard) lc50 (% w/v) (95 % ci 1) chi-square2 4 14.27 ± 6 0.125 (0.061-0.135) 8.52 5 13.34 ± 4 0.049 (0.012-0.053) 2.54 6 10.43 ± 3 0.012 (0.009-0.015) 2.52 table 2: mean lethal concentration (lc50) and parameters of probit regression for effects of ethanol solutions of euphorbia peplus latex on g. platensis larvae at different evaluation times. 1 ci: confidence interval. 2 goodness of fit for probit model, critical chi-square value 9.49 (df=4; p≤0.05). conclusions the latex of euphorbia peplus had an almost 100% antifeeding effect on g. platensis larvae. in addition, the ethanolic solutions of the latex caused >76% larval mortality in all treatments, with the least lc50 of 0.012% w/v on day 6. these results indicate that euphorbia peplus is a possible source of botanical insecticide compounds that could be used in ipm of g. platensis. however, further studies on adults as well as field tests are required to determine the effectiveness of euphorbia peplus latex application under field conditions. competing interests the authors have no competing interests to declare. abbreviations anova: analysis of variance ci: confidence interval lc50: lethal concentration to kill 50% of the individuals % w/v: percentage weight/volume funding this research was supported by laboratory of forest entomology (department of silviculture and nature conservation) and laboratory chemistry (department of agroindustry and enology), both of the universidad of chile. references adame, p., alberdi, i., cañellas, i., hernández, l., aguirre, a., ruano, a, moreno-fernández, d., gonzález, a.i.,torres, m.b., & montes, f. (2022). drivers and spread of non-native pests in forests: the case of gonipterus platensis in spanish eucalyptus plantations. forest ecology and management, 510, 120104. https://doi.org/10.1016/j. foreco.2022.120104 ali, a.a., sayed, h.m., ibrahim, s.r., & zaher, a.m. (2013). chemical constituents, antimicrobial, analgesic, antipyretic, and anti-inflammatory activities of euphorbia peplus l. phytopharmacology, 4, 69–80. amri, i., hamrouni, l., hanana, m., & jamoussi, b. (2013). reviews on phytotoxic effects of essential oils and their individual components: news approach for weeds management. international journal applied biology and pharmaceutical technology, 4, 96-114. bittner, m., alarcón, j., aqueveque, p., becerra, j., hernández, v., hoeneisen, m., & silva, m. (2001). estudio químico de especies de la familia euphorbiaceae en chile. boletín de la sociedad chilena de química, 46(4),1-16. https://doi. org/10.4067/s0366-16442001000400006 ceia, r.s., faria, n., lopes, p.b., alves, j., da silva a.a., valente, c., gonçalves, c.i., mata, v.a., santos s.a.p., azevedo-pereira h.m.v.s., sousa j.p., & da silva, l.p. (2021). local and landscape effects on the occurrence and abundance of the eucalyptus weevil gonipterus platensis (coleoptera: curculionidae). forest ecology and management, 500: 119618. https://doi.org/10.1016/j.foreco.2021.119618 chaieb, i., ben halima, m., & ben hamouda, m. (2001). the effect of food containing cestrum paquii (solanaceae) extract on various damaging lepidoptera. mededelingen (rijksuniversiteit te gent. fakulteit van de landbouwkundige en toegepaste biologische wetenschappen), 66, 479490. chiffelle, i., huerta, a., sandoval, c.a., & araya, j.e. (2017). efecto insecticida de extractos de hojas de schinus molle en larvas de gonipterus platensis. revista de la facultad nacional de agronomía, medellín, colombia, 70(3), 8263-8270. https://doi. org/10.15446/rfna.v70n3.58272 corporación nacional forestal (2012). el manejo sanitario de las plantaciones forestal. chile forestal, 360, 35-37. corporación nacional forestal (2017). finaliza liberación de controladores biológicos en región metropolitana. santiago, chile: conaf. https:// www.conaf.cl/finaliza-ultima-etapa-de-liberacionde-controladores-biologicos-en-la-rm/ accessed 6 may 2022. defagó, m., valladares, g., banchio, e., carpinella, c., & palacios, s. (2006). insecticide and antifeedant activity of different plant parts of melia azedarach on xanthogaleruca luteola. fitoterapia, 77, 500505. https://doi.org/10.1016/j.fitote.2006.05.027 docampoa, p., cabrerizoa, s., paladinob, n., parrañob, m., ruffolob, v., & muttia, o. (2010). eritrodermia secundaria a planta productora de látex (synadenium grantii). archivos argentinos de pediatría, 108(6), 126-129. dos santos bobadilha, g., baptista vidaurre, g., câmara, a.p., fantuzzi neto, h., da silva oliviera, j.t., pirez soliman, e., verly lopes, d.j., & zanuncio, j.c. (2019). effect of defoliator insect on growth and wood properties of eucalypt trees. european journal of wood products, 77, 861–868. https:// doi.org/10.1007/s00107-019-01435-6 ghramh, h.a., khan, k.a., & ibrahim, e.h. (2019). biological activities of euphorbia peplus leaves ethanolic extract and the extract fabricated gold nanoparticles (aunps). molecules, 24, 1431. https://doi.org/10.3390/molecules24071431 hua, j., liu, y., xiao, c.j., jing, s.x., luo, s.h., & li, s.h. (2017). chemical profile and defensive function of the latex of euphorbia peplus. phytochemistry, 136, 56-64. https://doi.org/10.1016/j. phytochem.2016.12.021 huerta et al. new zealand journal of forestry science (2023) 53:2 page 5 https://doi.org/10.1016/j.foreco.2022.120104 https://doi.org/10.1016/j.foreco.2022.120104 https://doi.org/10.4067/s0366-16442001000400006 https://doi.org/10.4067/s0366-16442001000400006 https://doi.org/10.1016/j.foreco.2021.119618 https://doi.org/10.15446/rfna.v70n3.58272 https://doi.org/10.15446/rfna.v70n3.58272 https://www.conaf.cl/finaliza-ultima-etapa-de-liberacion-de-controladores-biologicos-en-la-rm/ https://www.conaf.cl/finaliza-ultima-etapa-de-liberacion-de-controladores-biologicos-en-la-rm/ https://www.conaf.cl/finaliza-ultima-etapa-de-liberacion-de-controladores-biologicos-en-la-rm/ https://doi.org/10.1016/j.fitote.2006.05.027 https://doi.org/10.1007/s00107-019-01435-6 https://doi.org/10.1007/s00107-019-01435-6 https://doi.org/10.3390/molecules24071431 https://doi.org/10.1016/j.phytochem.2016.12.021 https://doi.org/10.1016/j.phytochem.2016.12.021 huerta, a., chiffelle, i., passalaqua, c., curkovic, t., & araya, j.e. (2021). antifeeding and insecticide effects of leaf extracts from cestrum parqui (solanaceae) on larvae of gonipterus platensis (coleoptera: curculionidae). agriculture & food journal of international scientific publications, 9, 100-107. infostat (2009). infostat versión 2009, grupo infostat, fca. córdoba, argentina: universidad nacional de córdoba. instituto forestal (2020). anuario forestal 2020, chilean statistical yearbook of forestry. boletín estadístico nº174. santiago, chile: ministerio de agricultura, gobierno de chile. isman, m. (2020). botanical insecticides in the twentyfirst century-fulfilling their promise? annual review of entomology, 65, 233-249. https://doi. org/10.1146/annurev-ento-011019-025010 jordan, c., dos santos, p.l., oliveira, l., domingues, m.m., gêa, b., ribeiro, m.f., mascarin, g.m., & wilcken, c.f. (2021). entomopathogenic fungi as the microbial frontline against the alien eucalyptus pest gonipterus platensis in brazil. scientific reports, 11, 7233. https://doi.org/10.1038/s41598-02186638-9 mapondera, t.s., burgess, t., matsuki, m., & oberprieler, r. (2012). identification and molecular phylogenetics of the cryptic species of the gonipterus scutellatus complex (coleoptera: curculionidae: gonipterini). australian journal of entomology, 51, 175–188. https://doi.org/10.1111/j.14406055.2011.00853.x mendivelso, d., palacios, c., pinzón, a., & victoria, m. 2003. estudio morfológico y anatómico de euphorbia peplus linneo (euphorbiaceae). acta biológica colombiana, 8(2), 99-103. ogbourne, s., suhrbier, a., jones, b., cozzi, s., boyle, g., morris, m., mcalpine, d., johns, j., scott, t., sutherland, k., gardner, j., le, t., lenarczyk, a., aylward j., & parsons, p. (2004). antitumor activity of 3-ingenyl angelate: plasma membrane and mitochondrial disruption, and necrotic cell death. cancer research, 64, 2833-2839. https://doi. org/10.1158/0008-5472.can-03-2837 pérez otero, r., mansilla vásquez, p., & rodríguez iglesias, j. (2003). eficacia y efectos en laboratorio de diferentes insecticidas en el control del defoliador del eucalipto gonipterus scutellatus y de su parasitoide anaphes nitens. boletín de sanidad vegetal plagas, 29, 649-658. robertson, j., smith, k., savin, n., & lavigne, r. (1984). effects of dose selection and sample size on the precision of lethal dose estimates in dose-mortality regression. journal of economic entomology, 77(4), 833-837. https://doi.org/10.1093/jee/77.4.833 romanyk, n., & cadahía, d. (2002). plagas de insectos en las masas forestales. madrid, spain: sociedad española de ciencias forestales. ediciones mundiprensa, 336 p. rua, j.c.p., barreiro, s., branco, m., & tomé, m. (2020). estimating defoliation impact of gonipterus platensis on eucalyptus globulus stands productivity using a forest simulator based on 3-pg. forest ecology and management, 478: 118495. https:// doi.org/10.1016/j.foreco.2020.118495 santolamazza, s., & fernández, f. (2004). efectos de dos insecticidas de síntesis y de dos bio-insecticidas sobre el defoliador del eucalipto gonipterus scutellatus gyllenhal y su agente de control biológico anaphes nitens girault. boletín de sanidad vegetal plagas, 30, 265-277. schneider, c.a., rasband, w.s., & eliceiri, k.w. (2012). nih image to imagej: 25 years of image analysis. nature methods, 9, 671-675. https://doi.org/10.1038/ nmeth.2089 schröder, m., slippers b., wingfield, m., & hurley, b. (2020). invasion history and management of eucalyptus snout beetles in the gonipterus scutellatus species complex. journal of pest science, 93, 11-25. https://doi.org/10.1007/s10340-01901156-y servicio agrícola & ganadero. 2010. informativo fitosanitario forestal no 4 (3). unidad de vigilancia y control de plagas forestales y exóticas invasoras. santiago, chile: ministerio de agricultura. 8 p. huerta et al. new zealand journal of forestry science (2023) 53:2 page 6 https://doi.org/10.1146/annurev-ento-011019-025010 https://doi.org/10.1146/annurev-ento-011019-025010 https://doi.org/10.1038/s41598-021-86638-9 https://doi.org/10.1038/s41598-021-86638-9 https://doi.org/10.1111/j.1440-6055.2011.00853.x https://doi.org/10.1111/j.1440-6055.2011.00853.x https://doi.org/10.1158/0008-5472.can-03-2837 https://doi.org/10.1158/0008-5472.can-03-2837 https://doi.org/10.1093/jee/77.4.833 https://doi.org/10.1016/j.foreco.2020.118495 https://doi.org/10.1016/j.foreco.2020.118495 https://doi.org/10.1038/nmeth.2089 https://doi.org/10.1038/nmeth.2089 https://doi.org/10.1007/s10340-019-01156-y https://doi.org/10.1007/s10340-019-01156-y implications of ectomycorrhizal inoculation for drought stress tolerance of atlas cedar (cedrus atlantica (endl.) carrière) seedlings hamida gaba chahboub1*, mohamed sghir lamhamedi2 and ouzna abrous-belbachir3 1national forest research institute po 37, division of forestry and plant breeding, laboratory of forest ecophysiology. cheraga, algiers, algeria 2forest study center, laval university, 2325 university street, quebec, qc g1v qa6, canada 3university of science and technology houari boumedienne (usthb) po 32, faculty of biology, laboratory of biology and physiology of organisms (lbpo), el alia algeria *corresponding author: chahboubhamida@yahoo.fr (received for publication 7 september 2020; accepted in revised form 23 june 2021) abstract background: ectomycorrhizal inoculation is a promising strategy to minimise the initial transplant shock and increase plant survival and growth during the first years of out-planting in the field. the aim of this research was to investigate the effect of sporal inoculum of three ectomycorrhizal fungi: cortinarius cedretorum, amanita vaginata and inocybe geophylla on tolerance levels of atlas cedar (cedrus atlantica (endl.) carrière) seedlings subjected to applied drought stress in nursery conditions. methods: carpophores, seeds and organic forest soil were collected under pure stands of atlas cedar. after fifteen months of growth, seedlings were subjected to drought stress by withholding water for thirty days; we assessed morphological and physiological variables of all seedling batches (inoculated and uninoculated, controlled and stressed seedlings) results: all roots of inoculated stressed seedlings were mycorrhizal. the mycorrhization rates were 67%, 64.6% and 53.6% for stressed seedlings inoculated with cortinarius cedretorum, amanita vaginata, inocybe geophylla, respectively. this root mycorrhization was accompanied by a significant improvement in seedling growth, especially height and length of the main root (10.2 cm, 52 cm) reached in stressed seedlings inoculated with cortinarius cedretorum. there was a significant increase in relative water content, total chlorophyll, carotenoids, soluble sugars and starch, superoxide dismutase and ascorbate peroxydase enzyme activities in inoculated stressed seedlings compared with uninoculated seedlings. conclusions: inoculation of atlas cedar seedlings with spores of ectomycorrhizal fungi remains a very effective alternative for improving growth and the morphological and physiological status of seedlings under drought conditions. cortinarius cedretorum appears to be consistently advantageous followed by amanita vaginata and inocybe geophylla. new zealand journal of forestry science chahboub et al. new zealand journal of forestry science (2021) 51:7 https://doi.org/10.33494/nzjfs512021x131x e-issn: 1179-5395 published on-line: 09/07/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access episodic and recurrent droughts combined with high temperatures are among the main factors responsible for the physiological weakening of trees and increased vulnerability to other physiological decay processes (ettobi et al. 2009, kherchouche et al. 2013). in north africa, the success of reforestation programmes has become a major challenge because, after planting, seedlings are subjected to extreme environmental stresses (drought, high temperature, introduction atlas cedar (cedrus atlantica (endl.) carrière) (pinaceae) is an endemic species of the north african mountains in algeria and morocco. in algeria, the cedar forest is found in both the saharan and tellian mountains where it represents 1.3% of the total forest area of the country (belloula & beghami 2018). it has great ecological, floristic, socio-economic and heritage value to the local population (m’hirit & benzyane 2006). prolonged keywords: cedrus atlantica, drought stress, sporal inocula, forest soil. http://creativecommons.org/licenses/by/4.0/), chahboub et al. new zealand journal of forestry science (2021) 51:7 page 2 low soil fertility, etc.) during the establishment phase (lamhamedi et al. 2000). the use of biological methods as a practical way to alleviate soil stresses, including drought, on plant growth has received increased attention (miransari et al. 2009). ectomycorrhizal inoculation is a promising strategy to minimise the initial transplant shock and increase plant survival and growth during the first years of out-planting in the field. (boukcim et al. 2002). ectomycorrhizal fungi can improve growth and physiological status of seedlings by enhancing their photosynthetic capacity and by increasing the uptake of water and mineral nutrients (read et al. 2004), promote soil aggregation, and improve plant health by increasing protection against biotic (pathogen attacks) and abiotic (drought, salinity and heavy metals) stresses (rillig & mummey 2006). indeed, ectomycorrhizal fungi enhance water uptake under dry conditions by exploring a larger volume of soil than non-mycorrhizal roots, and hyphae can enter small soil pores that are not accessible to the short roots of a plant (lehto & zwiazek 2011) thus improving plant water relations under low water conditions (dunabeitia et al. 2004). in natural conditions, atlas cedar trees develop symbiotic associations with many different ectomycorrhizal species (hocine et al. 1994). however, it is difficult to form mycorrhizae even after artificial inoculations. therefore, the application of the inoculum to atlas cedar has been limited by the difficulty in obtaining a widespread and reproducible colonisation of root systems after their inoculation by mycelia of different fungal species (nezzar-hocine 1998). abourouh (2000) and lamhamedi et al. (2009) reported that no comprehensive studies have yet been conducted to assess the mycorrhization of atlas cedar seedlings with spores collected from the same site and their possible behaviour in nursery. there have been few studies associated with the benefit of ectomycorrhizal symbiosis to cedrus atlantica under water stress and it is not yet known how the mycorrhizae play an important role in the growth and water relations of forest trees. inoculation of atlas cedar plants using ectomycorrhizal spores collected from the same seed collection site and soil is a more efficient technique for reaching a satisfactory mycorrhization rate in the nursery. in this way, the spore inoculation may ultimately provide an inexpensive and viable strategy for nursery growers that they can integrate easily into nursery operations for forest trees. in addition, unlike solid and liquid inoculum where only one strain (i.e. a genotype) is used, the spore inoculum allows the introduction of a variety of fungal genotypes and increase of the probability of obtaining compatible genotypes (fungus-cedar) that can promote good colonisation of the roots. the inoculation of atlas cedar seedlings with fungal spore inoculum of cortinarius cedretorum (r. maire), amanita vaginata (fries) vittidini and inocybe geophylla (pers.) had the best mycorrhization rate and improved atlas cedar seedling growth (gaba-chahboub et al. 2016). although different studies have shown that ectomycorrhizal fungi confer drought tolerance for several forest species (lamhamedi et al. 1992; lehto & zwiazek 2011), studies concerning the effects of inoculation by spore-based atlas cedar seedlings on growth parameters, as well as physiological and biochemical parameters in response to drought stress are limited, and little is known about the effect of ectomycorrhizal inoculation on reactive oxygen species metabolism. therefore, the aim of this research was to determine whether mycorrhizal symbiosis minimises the damage caused by oxidative stress via the regulation of antioxidant enzymes. we assume that inoculation of atlas cedar seedlings with ectomycorrhizal spores enhances their growth when subjected to drought stress. the purpose of this study is to test the controlled mycorrhization of atlas cedar seedlings by ectomycorrhizal spore inoculation under water limited conditions and to demonstrate their effectiveness on morphological, physiological and biochemical variables in cedrus atlantica seedlings as compared to uninoculated stressed seedlings. materials and methods study site the experiment was carried out in january 2013. both soil and atlas cedar seeds were collected from the same stand located in the chrea national park (36 ° 25 ‘n, 2° 52’ e, elevation: 1600 m, rainfall: 1400 mm/year) located in the north-western tell atlas mountains of algeria where prevailing winds are from the northwest and tend to be wet. the rainy period is about ten months (september to june) and the dry period typically occurs over two months, july and august. the mean maximum temperatures of hottest month oscillate between 26.3 °c and 33.6 °c, and the mean minimum temperatures of the coldest month can vary between 0.4 °c and 7.3 °c (halimi 1980). organisms and soil used in the study the forest soil was collected from the a1 horizon (located at depths varying between 25 to 30 cm in the main zone prospected by the roots) of lower cretaceous schists that were classified as more or less clayey (zaidi 2002). granulometric analysis of the collected soil showed the following composition: 13.9% clay, 20.3% fine silt, 25.2% coarse silt, 2.3% fine sand and 38.2% coarse sand (aubert 1978).the soil is slightly acid, rich in organic matter and little limestone (1.2%) with a total carbon of 5.2%, total nitrogen of 0.08% (john 1970), assimilable phosphorus 0.9 mg/kg soil (olsen et al. 1954), organic matter of 9.1% (anne 1945); a ph of 5.9 and an electrical conductivity of 2.9 ds/m. the atlas cedar seeds were collected from ten trees located within a distance of 20 m from each other. the seeds were surface sterilised with 30% hydrogen peroxide (h2o2) for 5 min, rinsed three times with distilled water and then stratified at 4 °c for two weeks in the dark, under semi-axenic conditions. seeds were transferred to petri dishes lined with wet filter paper before they were germinated in a growth chamber at 25 °c day / night and humidity of 65%. three ectomycorrhizal fungi were collected from the same seed collection site. these are cortinarius cedretorum (cc), amanita vaginata (av) and inocybe geophylla (ig). the fresh carpophores were gently cleaned by a small brush to remove root fragments and soil particles still adhering to them. they were dried firstly in the open air and then in an oven at 30 °c for 24 hours to remove all traces of moisture. subsequently, they were ground using a grinder for a very short time (2 min). the spore inoculums were screened using a sieve with a mesh size of 200 μm. these spore suspension inoculums were kept in the refrigerator (4 °c) until use. the forest soil was sterilised by autoclaving at 120 °c for 20 min twice over 48 hours period. experimental design the experiment was conducted in a greenhouse at the national forest research institute (bainem arboretum) located at 15 km west of algiers, 36° 45’ n, 3° 25’e). the randomised experimental design included four treatments side by side: three treatments that consisted of atlas cedar seedlings each inoculated with cc, av and ig, and a control treatment that consisted of uninoculated control seedlings. twenty seedlings were randomly allocated to each treatment. after one month of growth in riedacker containers (height: 17 cm, diameter: 10 cm) with a “w” or “m” shaped section (thermoflan, molières-cavaillac, le vigan, france), each seedling was inoculated with 0.1 g of ground spore inoculums mixed with 250 ml of sterile distilled water. using a pipette, the spore inoculum was placed at a depth of 4 cm in direct contact with the roots of each seedling (abourouh 1994) for the three treated groups and seedlings of the control group were left uninoculated. all the seedlings were watered three times per week with sterile distilled water in order to maintain soil moisture levels near saturation to promote plant and fungus growth. the experiment was conducted in the greenhouse with a 16-hour day length. day and night-time temperatures were 25 and 18 °c, respectively and relative humidity was 65%. after fifteen months of growth, while all seedlings were still actively growing and kept in the same position, a drought stress treatment was applied by withholding water for thirty days. each batch of inoculated and uninoculated seedlings was further subdivided into two batches: stressed and control seedlings. overall, there were eight batches of seedlings: uninoculated control seedlings (ucs); uninoculated stressed seedlings (uss); inoculated control seedlings (ics ,ccc, avc and igc); and inoculated stressed seedlings (iss, ccs, avs and igs). mycorrhizal colonisation and growth measurements the extent of mycorrhizal colonisation was assessed based on the number of mycorrhizal apices, number of fine roots, and the rate of mycorrhization of the roots. these assessments were made on root systems that had been thoroughly and carefully washed to ensure that they were free of soil. growth data (stem height, length of main root and whole plant dry mass) were collected on ten samples from each batch that had been oven dried at 80 °c for 48 hours. measurement of physiological and biochemical variables relative water content (rwc)was determined according to (clark & mccaig 1982). needles were sampled randomly from the last rosette of each seedling (five replicate from each batch). fresh weight (fw) was determined from the needles sampled in the upper part of seedlings; turgid weight (tw) was obtained by floating the needles in tubes with deionised water for at least 48 h in the dark before being weighed again. dry weight (dw) was determined after drying needles at 80°c for 48 h. rwc was calculated from the following equation: rwc = [(fw – dw) / (tw – dw)] × 100. biochemical analysis was done on samples of fresh needles that were collected from the upper part of five seedlings from each batch (five replicates from five seedlings for each biochemical variable). chlorophyll and carotenoid contents were determined in 80% acetone extract. the absorbance readings were obtained at 663 nm (chlorophyll a), 647 (chlorophyll b) and 470 nm (carotenoids) by an optizen pop spectrophotometer and contents were calculated according to lichtenthaler (1987). the soluble sugars and starch contents were determined using anthrone reagent (mccready et al. 1950). proline content was estimated by the method described in rasio et al. (1987); absorbance was read at 528 nm. proline content was expressed as µmole proline/g of dry matter (dm). lipid peroxidation was determined by measuring the amount of malondialdehyde (mda) produced by the thiobarbituric acid reaction (alia et al. 1995). antioxidant enzyme activity assay catalase (cat, ec 1.11.1.6) activity was assayed by measuring the rate of disappearance of h2o2 according to the method described in aebi et al. (1983). 150 mg of fresh vegetal material was ground with 1 ml of 0.1m trishcl buffer (ph = 8.1) and the mixture was centrifuged at 12,000 rpm. for the assay of the enzymatic activity. 100 μl of the supernatant was added to 2 ml of potassium phosphate buffer (0.1 m, ph = 7) and the optical density was read by adding 20 μl h2o2 at 6% every 30 seconds over the course of 3 to 4 minutes. the enzyme activity was expressed in u mg-1 protein (u=1 µm of h2o2 reduction min-1 mg-1 protein). total ascorbate peroxydase (apx, ec 1.11.1.1) activity was assayed according to nakano & asada (1981). the reaction was started with the addition of h2o2, and ascorbate oxidation measured at 290 nm for one minute; a decline in optical density is proportional to enzyme activity. total superoxide dismutase (sod, ec1.15.1.1) assay was performed according to marklund (1974) with some modifications. the evaluation of the autooxidation of pyrogallol was carried out by differential measurement between a control and a test at 420 nm every thirty seconds for four minutes. statistical analysis the data from these measurements and assays were subjected to a descriptive statistical analysis and analysis of variance (anova). the two-way anova included two chahboub et al. new zealand journal of forestry science (2021) 51:7 page 3 table 1: description of the study sites fixed factors (inoculum and water stress), and means were compared according to the method of newman and keuls (dagnelie 1999). all analyses were undertaken using the statistica software (tibco software inc, palo alto, usa). results effect of mycorrhization on atlas cedar seedlings growing under drought stress conditions for uninoculated seedlings, significant reductions in seedling height (57%), length of the main root (20.8%), dry weight of whole seedling (42%) and number of fine roots (53.8%) were observed in stressed seedlings (uss) compared with control seedlings (ucs). relative to uninoculated stressed seedlings, stressed seedlings that had been inoculated all had significantly higher values of the size and morphological characteristics that were measured. for example, values of seedling height in the ccs, avs and igs were 191%, 171% and 125% greater, respectively compared with uss. the mean length of the main root in inoculated stressed seedlings was also significantly higher compared to uninoculated stressed seedlings. in the ccs, avs and igs treatments, mean root lengths were 108%, 68% and 52.8% greater, respectively compared with uss. we found highly significant increases in the average number of fine roots between uss and iss. these enhancement rates were:130%, 100% and 52.3%, respectively in ccs, avs and igs compared to uss. there was no reduction in mycorrhization rate or the number of mycorrhizal apices number in iss compared to ics (table. 1). effect of drought stress on relative water content (rwc) in ucs, the relative water content was 83% while in uss it was reduced significantly to 60% reflecting the installation of physiological drought. in ics, rwc was higher than in ucs; they were 88%, 87.7% and 86.8%, respectively in ccc, avc and igc. moreover, iss maintained their water status and their rwc were near to that of ics; we registered 86%, 85.9%, and 84.7%, respectively, for ccs, avs and igs (fig. 1). total chlorophyll content (chla+b) in the needles of ucs, the total chlorophyll content was 225 ± 0.007 µg/g dw. this content increased significantly and reached 278 ± 0.002 µg/g dw, 263 ± 0.004 µg/g dw and 253 ± 0.004 µg/g dw, respectively, in ccc, avc and igc. under drought stress, the total chlorophyll content was reduced significantly in uss (161 ± 0.006 µg/g dw) was 28.4% lower compared to ucs. in ccs and avs, the total chlorophyll content was 255 ± 0,003 µg/g dw and 243 ± 0.006 µg/g dw, respectively, and were significantly higher than values recorded in uss (fig. 1). carotenoid content needle carotenoid content in ccc, avc, igc were 50.4 ± 0.09 µg/g dw, 49.6 ± 0,09 µg/g dw and 49 ± 0,10 µg/g dw, respectively, and were significantly higher than in chahboub et al. new zealand journal of forestry science (2021) 51:7 page 4 ucs 7.2%, 4.8%, 3.6% increase, respectively in ccc, avc and igc. in uss, carotenoid content reduced significantly to 40.6 ± 0.05 µg/dw. otherwise, carotenoid contents were significantly higher in iss; we recorded mean values of 48.6 ± 0.03 µg/g dw, 48 ± 0.08 µg/g dw and 47.6 ± 0.05 µg/g dw respectively in ccs, avs and igs (fig. 1). malondialdehyde content the average mda content in needles of ucs was 4.1 ± 0.027 nmole/mg dw. this content decreased in ics, where it was of 1.9 ± 0.003 nmole/mg dw, 2 ± 0.03 nmole/mg dw and 2 ± 0.004 nmole/mg dw, respectively, in ccc, avc and igc. under drought stress, we found that the average mda content increased to 5 ± 0.088 nmole/mg dw (20.6% increase) in uss compared to ucs indicating that drought stress aroused lipid peroxidation. in iss it was significantly lower than in uss with mean values of 3.5 ± 0.025 nmole/ mg dw, 3.6 ± 0.075 nmole/mg dw; 3.6 ± 0.02 nmole/mg dw, respectively, in ccs, avs and igs reductions of 28.8%, 27% and 26.4%, respectively) (fig. 1). soluble sugars the soluble sugars content in the needles of ucs was 1.9 ± 0.007 mg/g dw. this content increased significantly in ics where mean values of 2.2 ± 0.012 mg/g dw, 2.1 ± 0.006 mg/g dw and 2.1 ± 0.006 mg/g dw, respectively, were obtained for ccc, avc and igc (fig. 1). these corresponded to increases of 15.9%, 13% and 11%, respectively. for the uss, the soluble sugars content was significantly lower than in iss. mean values of 2.1 ± 0.14 mg/g dw, 2 ± 0.057 mg/g dw and 1.9 ± 0.006 mg/g dw, respectively were obtained for ccs, avs and igs. compared with uss, these contents corresponded to decreases of 41.8%, 39.5% and 38%, respectively. starch content there was no significant difference in starch content between ics and iss. the following mean values were recorded for ccs, avs and igs: 1.8 ± 0.006 mg/g dw, 1.7 ± 0.005 mg/g dw, and 1.7 ± 0.006 mg/g dw, respectively. small differences in starch content seemed were observed between ucs and uss (1.5 ± 0.01 mg/g dw and 1.3 ± 0.008 mg/g dw, respectively) (fig. 1). free proline content the drought stress induced a significant increase in proline content in uss, where a mean value of 8.9 ± 0.039 µmole/g dw was recorded. this content was lower in iss where values of 8 ± 0.117 µmole/g dw, 8.1 ± 0.085 µmole/g dw and 8.4 ± 0.045 µmole/g dw, respectively, were recorded in ccs, avs and igs corresponding to decreases of 11%, 9.3%, 5.1%, respectively), but still higher than values observed in ics (fig.1). cat activity there were no significant differences between the cat activity of needles in inoculated and uninoculated seedlings control and stressed seedlings. in iss, the cat activity was 95 ± 0.02 µmole h2o2 min -1 mg-1 protein, 95 ± 0.01µmole h2o2 min -1 mg-1 protein and 96 ± 0.01 µmole h2o2 min -1 mg-1 protein, respectively, in ccs, avs and igs (fig. 1). chahboub et al. new zealand journal of forestry science (2021) 51:7 page 5 g ro w th a nd m yc or rh iz at io n va ri ab le s m yc or rh iz al in oc ul at io n an d st re ss s it ua ti on o f s ee dl in gs u ni no cu la te d co nt ro l se ed lin gs (u cs ) u ni no cu la te d st re ss ed se ed lin gs (u ss ) in oc ul at ed co nt ro l se ed lin gs (c cc ) in oc ul at ed st re ss ed se ed lin gs (c cs ) in oc ul at ed co nt ro l se ed lin gs (a vc ) in oc ul at ed st re ss ed se ed lin gs (a vs ) in oc ul at ed co nt ro l se ed lin gs (i g c) in oc ul at ed st re ss ed se ed lin gs (i g s) se ed lin g he ig ht ( cm ) 5. 5 ± 0. 49 3. 5 ± 0. 56 a 10 .5 ± 0 .5 4 10 .2 ± 0 .3 2b 10 .0 ± 0 .3 5 9. 5 ± 0. 50 b 8. 7 ± 0. 47 7. 9± 0 .6 1c le ng th o f t he m ai n ro ot (c m ) 30 .2 ± 0 .7 25 .0 ± 0 .5 7 a 56 .0 ± 1 .2 8 52 .0 ± 2 .3 4b 47 .0 ± 0 .8 2 42 .0 ± 0 .7 9 b 43 .0 ± 2 .6 5 38 .2 ± 3 .9 6 c w ho le -s ee dl in g dr y w ei gh t ( g) 0. 3 ± 0. 01 0. 21 ± 0 .0 2a 3. 5 ± 0. 83 2. 5 ± 0. 57 b 2. 1 ± 0. 30 2. 3 ± 0. 11 b 2. 0 ±0 .3 7 1. 7 ± 0. 25 c m yc or hi za ti on r at e (% ) 70 .0 ± 1 .6 4 67 .0 ± 1 .9 5 68 .0 ± 1 .5 7 64 .6 ± 1 .9 6b 57 .0 ± 1 .6 1 53 .6 ± 1. 51 c n um be r of m yc or rh iz al sp ec ie s 13 0. 0 ± 5. 72 12 4. 0 ± 3. 53 12 4. 0 ± 8. 36 11 6. 4 ± 9. 20 90 .0 ± 8 .2 5 82 .2 ± 8 .7 9 n um be r of fi ne r oo ts 10 .0 ± 1 .1 5 6. 5 ± 1. 20 a 22 .0 ± 2 .4 3 15 .0 ± 3 .9 1b 21 .0 ± 3 .2 0 13 .0 ± 2 .8 6 b 17 .0 ± 0 .3 8 10 .0 ± 0 .5 4c ta b le 1 : i m pa ct o f m yc or rh iz at io n on g ro w th tr ai ts o f 1 5m on th -o ld c ed ru s at la nt ic a se ed lin gs s ub je ct ed to d ro ug ht s tr es s. e ac h va ri ab le w as th e av er ag e of 1 0 in de pe nd en t m ea su re m en ts . f or e ac h va ri ab le , t he d iff er en t l et te rs s ho w ed s ig ni fic at iv e di ffe re nc es b et w ee n un in oc ul at ed s tr es se d se ed lin gs ( u ss ) an d in oc ul at ed s tr es se d se ed lin gs ( is s) b as ed o n th e st ud en tn ew m an -k eu ls te st ( p < 0. 05 ). chahboub et al. new zealand journal of forestry science (2021) 51:7 page 6 figure 1: impact of mycorrhization on physiological and biochemical variables of cedrus atlantica seedlings under drought stress (ni: uninoculated seedlings, cc.: seedlings inoculated with cortinarius cedretorum, av: seedlings inoculated with amanita vaginata, ig: seedlings inoculated with inocybe geophylla).(five replicates for each parameter) letters indicate the significant differences between batches of seedlings, homogeneous groups have approximatively the same mean values and the same letter following the newmankeuls test (p <0.05). total apx activity results showed that the highest needle apx activity was observed in igs and igc. however, no significant differences were observed between these two treatment groups. the increase of apx activity was approximately 13.5% in igs compared to uss (fig. 1). total sod activity among the uninoculated and inoculated seedlings, sod activity of needles in iss was significantly higher than in ics. mean values of sod activity of 35.4 u mg-1 protein min-1, 34.5 u mg-1 protein min-1 and 32.3 u mg-1 protein min-1, respectively, were observed in ccs, avs and igs. no significant differences in sod activity were observed between ucs and ics (fig.1). discussion our results showed that inoculated stressed seedlings had better growth than the uninoculated stressed seedling. this enhancement occurred for all growth variables recorded in this study: seedling height, length of the main root, total dry matter and number of fine roots. an increase of the ratio of the length of the main root to the height of seedlings was also observed. the adaptation of atlas cedar to drought is due to the extensive root growth, its ability to explore the soil thoroughly and the alternating waves of aerial growth with waves of growth root (ducrey 1994). thus, it has been showed in some forest species as cedrus atlantica (aussenac 1985), pinus ponderosa (mcmillin 1995) and fagus sylvatica (van hees 1997) that water constraint induces preferentially high biomass allocation to the roots when the stress intensity is important. the enhancement of the root part at the expense of the aerial part is considered as a criterion for drought resistance (pallardy 1993; liu et al. 2014). in another study, plant tolerance to drought could be primarily due to a large volume of soil explored by roots and the extra-radical hyphae of the fungi (songsri et al. 2008, kambiranda et al. 2011, zhang et al. 2016). cedar trees can be considered as an anisohydric species because they are adapted to short episodic drought (mcdowell et al. 2008). atlas cedar is characterised by a very high dynamic potential of water absorption by roots, which colonise mainly the upper layers of soil. in addition, cedar plants respond quickly to rainfall via the efficiency of its ectomycorrhizal root system at 20 to 50 cm of depth. under water stress, atlas cedar maintains its physiological activity up to very pronounced drought levels (bouahmed et al. 2019). relative water content is a key indicator of cell and tissue hydration, and it is important for optimal physiological functioning and the growth processes (altinkut et al .2001). in our study, the rwc of iss is higher than that of uss. this confirms the effectiveness of the three-spore inoculum on the upkeep of a high rwc in the needles of inoculated seedlings. our results also showed that the total chlorophyll content (a + b) and carotenoids content decreased significantly in uss and that total chlorophyll and carotenoids content in the needles of iss are significantly higher than those of uss. this suggests that there is a beneficial effect of mycorrhizal symbiosis on the induction chlorophyll and carotenoids accumulation (baslam et al. 2011). similar results were obtained in six-month-old pseudotsuga menziesii plants inoculated with the ectomycorrhizal fungus rhizopogon vinicolor and subjected to water stress conditions, which had higher chlorophyll levels than the stressed uninoculated plants (dixon & hiolhiol, 1992). the iss also reacted by significantly increasing soluble sugars and starch content compared to uss. indeed, the relative increase in soluble sugars content could contribute to an osmotic adjustment that would allow the plant to balance its water content and to maintain a high cellular integrity in plant tissues during water deficit periods (rai 2002; chaves et al. 2003). proline accumulation under stress conditions is due to the induction of the gene encoding the enzyme p5cs (pyrroline-5-carboxylatesynthetase) and inhibition of the proline dehydrogenase (prodh) gene (szabados & savouré 2010). the accumulation of proline is concomitant with soluble sugars in some plant species (clifford et al. 1998) and the ability of some genotypes to accumulate the osmotica (proline and sugars) is used as a selection criterion of drought tolerant genotypes (nouri 2002). mycorrhizal symbiosis protects plants against a variety of abiotic stresses using various processes such as improved photosynthetic rate, uptake and accumulation of mineral nutrients, accumulation of osmo-protectants and changes in the rhizosphere ecosystem (yin et al. 2016). it is known that drought stress induces oxidative stress with the generation of reactive oxygen species (ros). they attack the membrane polyunsaturated fatty acids (pufa) resulting in their breakdown into small hydrocarbon fragments like ketones or malondialdehyde (mda) or other related products (zeriri et al. 2012). in our study, mda content of leaves in uss was significantly higher than that of iss. in previous studies, it was reported that mycorrhizal plants had lower mda content than non-mycorrhizal plants, suggesting that mycorrhization decreased lipids oxidative damages during stress conditions (wu et al. 2006). our experiments on antioxidant enzymatic activities cat, apx and sod indicated that applied drought stress to atlas cedar seedlings did not significantly modify the cat enzyme activity and that the highest needle apx activity was observed in igs and igc, while the highest needle sod activities were observed in inoculated stressed seedlings relative to uss. it is known that when plants are subjected to stress, the first reactive oxygen species (ros) scavenging enzyme active in the enzymatic mechanism is sod which plays a key role in cellular defences against ros (scandalios 1993). the increase of sod activity in leaves is closely related to a higher ability to scavenge active oxygen radicals under water stress (abedi & pakniyat 2010). moreover, there is a correlation between the sod and cat activities because the sod is responsible for the formation of hydrogen peroxide h2o2 and the increase of this last is correlated to cat activity (khalifa et al. 2011). chahboub et al. new zealand journal of forestry science (2021) 51:7 page 7 the importance of root antioxidant enzymes in the regulation of cellular metabolism under water stress conditions has been demonstrated by shvaleva et al. (2006). the balance between apx and cat activities is critical for suppressing toxic levels of ros in cells (apel & hirt 2004). in most of the previous studies, mycorrhizal application enhanced the antioxidant system (lambais et al. 2003). increased sod activity in mycorrhizal plants is not related to their nutritional status but is the direct effect of mycorrhizal association in response to drought treatment of the host plant (ruiz-lozano & azcon 1995). ectomycorrhizal plants show high resistance to drought stress by preventing oxidative stress and effectively removing ros (bartels 2001). coordinated activation of sod and cat can prevent cellular damage to the host plant during ectomycorrhizal formation between castanea sativa mill and pisolithus tinctorius (baptista et al. 2007). the increase in antioxidant enzymatic activities induced by mycorrhizae is associated with an increase in photosynthetic activity, biomass and plant nutrition. however, the stimulation of antioxidant molecules and antioxidant enzymatic activities depends on the host plant and mycorrhizal fungi (roldan et al. 2008). it seems that mycorrhizal inoculation notably influences the activity of antioxidant enzymes in citrus leaves under water stress and an increase in the activity of antioxidant enzymes alleviates water stress (roldán et al. 2008). conclusions this study showed that inoculating atlas cedar seedlings with ectomycorrhizal spores were effective in improving host morphological and physiological status of atlas cedar seedlings under drought conditions. among the three mycorrhizal fungi inoculants, cortinarius cedretorum appears to be consistently advantageous for cedrus atlantica seedlings. it exhibited the greatest impact on the cedar seedlings growth under water drought, followed by av and ig; this last had the least effect on cedar seedlings, probably reflecting its poor symbiotic efficiency under these drought conditions. this study provides interesting perspectives and a very valuable framework for further studies focusing on integrated analysis of the ectomycorrhizal symbiosis effects under drought stress. it also showed the services delivered by ectomycorrhizal symbiosis, which has important implications to forest ecosystems of atlas cedar in water-limited environments and could potentially lead to development of an efficient conservation strategy in order to reach sustainable objectives in forest ecosystem productivity of this endemic species. competing interests the authors declare that they have no competing interests. authors’ contributions ch undertook the conception and design, acquisition of data, drafting of the manuscript, analysis and interpretation of data, and final approval of the version to be published. lms critically revised the manuscript for important intellectual content. abo interpreted data, drafted the manuscript and undertook analysis and interpretation of data. references abedi, t., & pakniyat, h. (2010). antioxidant enzyme changes in response to drought stress in ten cultivars of oil seed rape (brassica napus l.). czech journal of genetics and plant breeding, 46(1), 2734. https://doi.org/10.17221/67/2009-cjgpb abourouh m. (1994). les ectomycorhlzes du cedre de l’atlas: etat des connaissances et perspectives. annales de la recherche forestière au maroc. le cèdre de l’atlas (cedrus atlantica, manetti). silva mediterranea. actes du séminaire international sur le cèdre de l’atlas ifrane (maroc), 7 11 juin 1993. annales de la recherche forestiere au maroc, 27(spécial), 337-348 http://www.fao.org/3/ ah002f/ah002f00.htm abourouh, m. (2000). ectomycorrhization of the main forest species in morocco. doctoral thesis, mohammed university, rabat, morocco. aebi, h. (1983). catalase in vitro. oxygen radicals in biological systems. methods in enzymology, 105, 121-126. https://doi.org/10.1016/s00766879(84)05016-3 alia, prasad, k.v.s.k. & pardha saradhi, p. (1995). effect of zinc on free radicals and proline in brassica and cajanus. phytochemistry, 39(1), 45-47. https://doi. org/10.1016/0031-9422(94)00919-k altinkut, a., kazan, k., ipekci, z., & gozukirmizi, n. (2001). tolerance to paraquat is correlated with the traits associated with water stress tolerance in segregating f2 populations of barley and wheat. euphytica, 121, 81-86. https://doi. org/10.1023/a:1012067711200 anne, p. (1945). rapid determination of organic carbon in soils. agronomic annals, 15(2), 161-172. apel, k., & hirt, h. (2004). reactive oxygen species: metabolism, oxidative stress, and signal transduction. annual review of plant biology, 55, 373-399. https://doi.org/10.1146/annurev. arplant.55.031903.141701 aubert, g. (1978). soil analysis methods, 2nd ed. marseille, france: regional center for educational documentation, crdp. aussenac, g. (1985). le potentiel hydrique de l’arbre: une donnée essentielle pour la compréhension de chahboub et al. new zealand journal of forestry science (2021) 51:7 page 8 https://www.researchgate.net/journal/1212-1975_czech_journal_of_genetics_and_plant_breeding https://www.researchgate.net/journal/1212-1975_czech_journal_of_genetics_and_plant_breeding https://doi.org/10.17221/67/2009-cjgpb http://www.fao.org/3/ah002f/ah002f00.htm http://www.fao.org/3/ah002f/ah002f00.htm https://doi.org/10.1016/s0076-6879(84)05016-3 https://doi.org/10.1016/s0076-6879(84)05016-3 https://doi.org/10.1016/0031-9422(94)00919-k https://doi.org/10.1016/0031-9422(94)00919-k https://doi.org/10.1023/a:1012067711200 https://doi.org/10.1023/a:1012067711200 https://doi.org/10.1146/annurev.arplant.55.031903.141701 https://doi.org/10.1146/annurev.arplant.55.031903.141701 l’écophysiologie des essences forestières. science du sol, 4, 217-226. baptista, p., martins, a., salomé pais, m.s., tavares, r.m., & lino-neto, t. (2007). involvement of reactive oxygen species during early stages of ectomycorrhiza establishment between castanea and pisolithus tinctorius. mycorrhiza, 17, 185-193. https://doi.org/10.1007/s00572-006-0091-4 bartels, d. (2001). targeting detoxification pathways: an efficient approach to obtain plants with multiple stress tolerance? trends in plant science, 6(7), 284-286. https://doi.org/10.1016/s13601385(01)01983-5 baslam, m., pascual, i., sanchez-diaz, m., erro, j., garcia mina, j.m., & goicoechea, n. (2011). improvement of nutritional quality of greenhouse-grown lettuce by arbuscular mycorrhizal fungi is conditioned by the source of phosphorus nutrition. journal of agriculture and food chemistry, 59(20), 1112911140. https://doi.org/10.1021/jf202445y belloula, s., & beghami, y. (2018). assessment of the dynamics of atlas cedar decline (cedrus atlantica manetti) by remote sensing in the aures area, algeria. arab world geographer, 21(2-3), 154-167. bouahmed, a., vessella, f., schirone, b., krouchi, f., & derridj, a. (2019). modeling cedrus atlantica potential distribution in north africa across time: new putative glacial refugia and future range shifts under climate change. regional environmental change, 19, 1667-1682. https://doi.org/10.1007/ s10113-019-01503-w boukcim, h., conventi, s., & mousain, d. (2002). ectomycorrhization of cedrus atlantica under controlled conditions: efficiency of two forms of mycelial inoculum. annals of forest sciences, 59, 839-846. https://doi.org/10.1051/ forest:2002082 chaves, m.m., maroco, j.p., & pereira, j.s. (2003). understanding plant responses to drought: from genes to the whole plant. functional plant biology, 30, 239-264. https://doi.org/10.1071/fp02076 clarke, j.m., & mccaig, t.n. (1982). excised leaf water relation capability as an indicator of drought resistance of triticum genotypes. canadian journal of plant science, 62, 571-578. https://doi. org/10.4141/cjps82-086 clifford, s.c., arndt, s.k., corlett, j.e., joshi, s., sankhla, n., popp, m., & jones, h.g. (1998). the role of solute accumulation, osmotic adjustment and changes in cell wall elasticity in drought tolerance in ziziphus mauritiana (lamk.). journal of experimental botany, 49(323), 967-977. https://doi.org/10.1093/ jxb/49.323.967 dagnelie, p. (1999). theoretical and applied statistics. vol 2. brussels, belgium: oneand two-dimensional statistical inference. dixon, r.k., & hiol hiol, f. (1992). mineral nutrition of pinus caribea and eucalyptus camaldulensis inoculated with pisolithus tinctorius and thelephora terrestris. communications in soil science and plant analysis, 23, 13-14. https://doi. org/10.1080/00103629209368674 ducrey, m. (1994). adaptation of the atlas cedar (cedrus atlantica manetti) to the mediterranean climate: ecophysiological aspects of its response to drought. annals of forest research of morocco, 27, 140-152. dunabeitia, m.k., hormilla, s., garcia-plazaola j.i., txarterina, k., arteche, u., becerril, j.m. ( 2004). differential response of three fungal species to environmental factors and their role in the mycorhization of pinus radiata d. don. mycorrhiza, 14, 11-18. et-tobi, m. (2008). dendrometric and phytosanitary inventory of azrou and ait youssi lamekla cedar forests. study of the causes of the dieback of the cedar grove of the middle atlas (ifrane). morocco. convention fao/ utf/ mor/028hce flcd, morocco, 77 p. gaba-chahboub, h., lamhamedi, m.s., & abrousbelbachir, o. (2016). effect of ectomycorrhizal inoculation in the nursery on the growth and nutrition of atlas cedar plants in algeria. tropical woods and forests, 330(4), 57-67. https://doi. org/10.19182/bft2016.330.a31319 halimi, a. (1980). the blideen atlas, climate and plant layers. vol 1. algeria: university publication office, 523 p. hocine, h., belarbi, h., & perrin, r. (1994). possibilities of mycorrhization of cedrus atlantica manetti. annals of forest research in morocco, 27 (special), 349361. john, m.k. (1970). colorimetric determination in soil and plant material with ascorbic acid. soil science, 68, 171-177. https://doi.org/10.1097/00010694197004000-00002 kambiranda, d.m., vasanthaiah, h.k.n., katam, r., ananga, a., basha s.m., & naik, k. (2011). impact of drought stress on peanut (arachis hypogaea l.) productivity and food safety. in hemanth vasanthaiah, (ed.), plants and environment (pp. 249-272). london: intechopen. khalifa, f.k., khalil, f.a., barakat, h.a., & hassan, m.m. (2011). protective role of wheat germ and grape seed oils in chlorpyrifos-induced oxidative stress, biochemical and histological alterations in liver of rats. australian journal of basic and applied sciences, 5(10), 54-66. kherchouche, d., kalla, m., gutierrez, e.m., briki, a., & hamchi, a. (2013). drought and decline of the atlas cedar (cedrus atlantica manetti) in the belezma massif (algeria). drought, 24(2), 129-137. https:// doi.org/10.1684/sec.2013.0384 chahboub et al. new zealand journal of forestry science (2021) 51:7 page 9 https://doi.org/10.1007/s00572-006-0091-4 https://doi.org/10.1016/s1360-1385(01)01983-5 https://doi.org/10.1016/s1360-1385(01)01983-5 https://doi.org/10.1021/jf202445y https://doi.org/10.1007/s10113-019-01503-w https://doi.org/10.1007/s10113-019-01503-w https://doi.org/10.1051/forest:2002082 https://doi.org/10.1051/forest:2002082 https://doi.org/10.1071/fp02076 https://doi.org/10.4141/cjps82-086 https://doi.org/10.4141/cjps82-086 https://doi.org/10.1093/jxb/49.323.967 https://doi.org/10.1093/jxb/49.323.967 https://doi.org/10.1080/00103629209368674 https://doi.org/10.1080/00103629209368674 https://doi.org/10.19182/bft2016.330.a31319 https://doi.org/10.19182/bft2016.330.a31319 https://doi.org/10.1097/00010694-197004000-00002 https://doi.org/10.1097/00010694-197004000-00002 https://doi.org/10.1684/sec.2013.0384 https://doi.org/10.1684/sec.2013.0384 chahboub et al. new zealand journal of forestry science (2021) 51:7 page 10 lambais, m.r., ríos-ruiz, w.f., & andrade, r.m. (2003). antioxidant responses in bean (phaseolus vulgaris) roots colonized by arbuscular mycorrhizal fungi. new phytologist, 160(2), 421-428. https://doi. org/10.1046/j.1469-8137.2003.00881.x lamhamedi, m.s., bernier, p.y., & fortin, j.a. (1992). hydraulic conductance and soil water potential at the soil-root interface of pinus pinaster seedlings inoculated with different dikaryons of pisolithus sp. tree physiology, 10, 231-244. https://doi. org/10.1093/treephys/10.3.231 lamhamedi, m.s., ammari, y., fecteau, b., fortin, j.a., & margolis, h. (2000). problem of forest nurseries in north africa and orientation strategies. agriculture notebooks, 9(5), 369-380. lamhamedi, m.s., abourouh, m., & fortin, j.a. (2009). technological transfer: the use of ectomycorrhizal fungi in conventional and modern forest tree nursery in northern africa. in: d. khasa, piché, y. & coughlan, a.p. (eds) advances in mycorrhizal science and technology (pp.139-152) ottawa: nrc research press. lehto, t., & zwiazek, j.j. (2011). ectomycorrhizas and water relations of trees. mycorrhiza, 21, 71-90. https://doi.org/10.1007/s00572-010-0348-9 lichtenthaler, h.k. (1987). chlorophylls and carotenoids: pigments of photosynthetic biomembranes. methods in enzymology, 148, 350-382. https://doi. org/10.1016/0076-6879(87)48036-1 liu, p., yin, l., deng, x., wang, s., tanaka, k., & zhang, s. (2014). aquaporin mediated increase in root hydraulic conductance is involved in siliconinduced improved root water uptake under osmotic stress in sorghum bicolor l. journal of experimental botany, 65(17), 4747-4756. https:// doi.org/10.1093/jxb/eru220 m’hirit, o., & benzyane, m. (2006). le cèdre de l’atlas: mémoire du temps sprimont, belgium: mardaga. marklund, s., & marklund, g. (1974). involvement of the superoxide anion radical in the autoxidation of pyrogallol and a convenient assay for superoxide dismutase. european journal of biochemistry, 47(3), 469-474. https://doi. org/10.1111/j.1432-1033.1974.tb03714.x mcdowell, n., pockman, w.t., allen, c.d., breshears, d.d., cobb, n., kolb, t., plaut, j., sperry, j., west, a., williams d.g., & yepez, e.a. (2008). mechanisms of plant survival and mortality during drought: why do some plants survive while others succumb to drought? new phytologist, 178(4), 719-739. https:// doi.org/10.1111/j.1469-8137.2008.02436.x mcmillin, j.d., & wagner, m.r. (1995). effects of water stress on biomass partitioning of ponderosa pine seedlings during primary root growth and shoot growth periods. forest science, 41(3), 594-610. miransari, m. (2009). contribution of arbuscular mycorrhizal symbiosis to plant growth under different types of soil stress. plant biology, 12(4), 563-569. https://doi.org/10.1111/j.14388677.2009.00308.x mccready, r.m., guggoz, j., silviera v., & owens, h.s. (1950). determination of starch and amylose in vegetables. analytical chemistry, 22(9), 1156-1158. https://doi.org/10.1021/ac60045a016 nezzar hocine, h. (1998). natural mycorrhizal associations of cedrus atlantica in the djurdjura massif (algeria) and controlled mycorrhization. doctoral thesis, blaise pascal university, clermontferrand ii, france. nouri, l. (2002). osmotic adjustment and maintenance of photosynthetic activity in durum wheat (triticum durum desf.) in conditions of water deficit. magister thesis. university of constantine. algeria. nakano, y., & asada, k. (1981). hydrogen peroxide is scavenged by ascorbate-specific peroxidase in spinach chloroplasts. plant and cell physiology, 22, 867-880. olsen, s.r., cole, c.v., watanabe, f.s., & dean, l.a. (1954). estimation of available phosphorus in soils by extraction with sodium bicarbonate. [circular 939]. washington: usda, u.s. government printing office, 1-19. pallardy, s.g., & rhoads, j.l. (1993). morphological adaptations to drought in seedlings of deciduous angiosperms. canadian journal of forest research, 23, 1766-1774. https://doi.org/10.1139/x93-223 rai, v.k. (2002). role of amino acids in plant responses to stresses. biologia plantarum, 45, 481-487. https:// doi.org/10.1023/a:1022308229759 rasio, a. (1987). several mechanisms of water stress adaptation in durum wheat. genetica agraria, 42, 90 p. read, d.j., leake, j.r., & perez-morenco, j. (2004). mycorrhizal fungi as drivers of ecosystem processes in heathland and boreal forest biomes. canadian journal of botany, 82(8), 1243-1263. https://doi.org/10.1139/b04-123 rillig, m.c., & mummy, d.l. (2006). mycorrhizae and soil structure. new phytologist, 171, 41-53. https://doi. org/10.1111/j.1469-8137.2006.01750.x roldán, a., díáz-vivancos, p., herníndez, j.a., carrasco, l., & caravaca, f. (2008). superoxide dismutase and total peroxidase activities in relation to drought recovery performance of mycorrhizal shrub seedlings grown in an amended semiarid soil. journal of plant physiology, 165, 715-722. https:// doi.org/10.1016/j.jplph.2007.02.007 ruiz lozano, j.m., & azcon, r. (1995). hyphal contribution to water uptake in mycorrhizal plants https://nph.onlinelibrary.wiley.com/action/dosearch?contribauthorstored=r%c3%ados-ruiz%2c+w+f https://nph.onlinelibrary.wiley.com/action/dosearch?contribauthorstored=andrade%2c+r+m https://doi.org/10.1046/j.1469-8137.2003.00881.x https://doi.org/10.1046/j.1469-8137.2003.00881.x https://doi.org/10.1093/treephys/10.3.231 https://doi.org/10.1093/treephys/10.3.231 https://doi.org/10.1007/s00572-010-0348-9 https://doi.org/10.1016/0076-6879(87)48036-1 https://doi.org/10.1016/0076-6879(87)48036-1 https://doi.org/10.1093/jxb/eru220 https://doi.org/10.1093/jxb/eru220 https://doi.org/10.1111/j.1432-1033.1974.tb03714.x https://doi.org/10.1111/j.1432-1033.1974.tb03714.x https://doi.org/10.1111/j.1469-8137.2008.02436.x https://doi.org/10.1111/j.1469-8137.2008.02436.x https://doi.org/10.1111/j.1438-8677.2009.00308.x https://doi.org/10.1111/j.1438-8677.2009.00308.x https://doi.org/10.1021/ac60045a016 https://doi.org/10.1139/x93-223 https://doi.org/10.1023/a:1022308229759 https://doi.org/10.1023/a:1022308229759 https://doi.org/10.1139/b04-123 https://doi.org/10.1111/j.1469-8137.2006.01750.x https://doi.org/10.1111/j.1469-8137.2006.01750.x https://doi.org/10.1016/j.jplph.2007.02.007 https://doi.org/10.1016/j.jplph.2007.02.007 chahboub et al. new zealand journal of forestry science (2021) 51:7 page 11 as affected by the fungal species and water status. physiologia plantarum, 95, 472-478. https://doi. org/10.1111/j.1399-3054.1995.tb00865.x scandalios, l.m. (1993). oxygen stress and superoxide dismutase. plant physiology, 101(1), 7-12. https:// doi.org/10.1104/pp.101.1.7 shvaleva, a.l., f costae e, silva., breia, e., jouve, l., hausman,j.f., almeida, m.h., maroco, j.p., rodrigues, m.l., pereira, j.s., & chaves, m.m. (2006). metabolic responses to water deficit in two eucalyptus globulus clones with contrasting drought sensitivity. tree physiology, 26 (2), 239248. https://doi.org/10.1093/treephys/26.2.239 songsri, p., jogloy,s., vorasoot, n., akkasaeng,c., patanothai, a., & holbrook, c.c. (2008). root distribution of drought resistance peanut genotypes in response to drought. journal of agronomy and crop science, 194(2), 92-103. https://doi.org/10.1111/j.1439037x.2008.00296.x szabados, l., & savouré, a. (2010). proline: a multifunctional amino acid. trends in plant sciences, 15(2), 89-97. https://doi.org/10.1016/j. tplants.2009.11.009 van hees, a.f.m. (1997). growth and morphology of pedunculate oak (quercus robur l.) and beech (fagus sylvatica l.) seedlings in relation to shading and drought. annals of forest science, 54, 9-18. https://doi.org/10.1051/forest:19970102 wu, q.s., xia, r.x., & zou, y.n. (2006). reactive oxygen metabolism in mycorrhizal and non-mycorrhizal citrus (poncirus trifoliata) seedlings subjected to water stress. journal of plant physiology, 163(11), 1101-1110. https://doi.org/10.1016/j. jplph.2005.09.001 yin, n., zhang, z., wang, l., & qian, k. (2016). variations in organic carbon aggregation, and enzyme activities of gangue-fly ash-reconstructed soils with sludge and arbuscular mycorrhizal fungi during 6-year reclamation. environment science pollution research institute, 23(17), 17840-17849. https:// doi.org/10.1007/s11356-016-6941-5 zaidi, s. (2002). natural regeneration of cedar (cedrus atlantica manetti) and its forest dynamics in the blidean atlas. magister thesis, vegetal ecology, university of science and technology houari boumediene. algeria. zeriri, i., tadjine, a., grara, n., belhaouchet, n., berrebbah, h., & djebar, m.r. (2012). potential toxicity of an insecticide of the family of carbamates on a bioindicator model of the pollution the earthworm octodrilus complanatus (oligochaeta, lumbricidae). annals of biological research, 3(11), 5367-5373. zhang, x., li, w., fang, m., jixian, y., meng, s. ( 2016). effects of arbuscular mycorrhizal fungi inoculation on carbon and nitrogen distribution and grain yield and nutritional quality in rice (oryza sativa l). journal of the science of food and agriculture, 97, 2919-1925. https://doi.org/10.1111/j.1399-3054.1995.tb00865.x https://doi.org/10.1111/j.1399-3054.1995.tb00865.x https://doi.org/10.1104/pp.101.1.7 https://doi.org/10.1104/pp.101.1.7 https://www.ncbi.nlm.nih.gov/pubmed/?term=shvaleva%20al%5bauthor%5d&cauthor=true&cauthor_uid=16356921 https://www.ncbi.nlm.nih.gov/pubmed/?term=costa%20e%20silva%20f%5bauthor%5d&cauthor=true&cauthor_uid=16356921 https://www.ncbi.nlm.nih.gov/pubmed/?term=breia%20e%5bauthor%5d&cauthor=true&cauthor_uid=16356921 https://www.ncbi.nlm.nih.gov/pubmed/?term=hausman%20jf%5bauthor%5d&cauthor=true&cauthor_uid=16356921 https://www.ncbi.nlm.nih.gov/pubmed/16356921 https://doi.org/10.1093/treephys/26.2.239 https://doi.org/10.1111/j.1439-037x.2008.00296.x https://doi.org/10.1111/j.1439-037x.2008.00296.x http://www.sciencedirect.com/science/article/pii/s1360138509002982 http://www.sciencedirect.com/science/article/pii/s1360138509002982 http://www.sciencedirect.com/science/journal/13601385/15/2 https://doi.org/10.1016/j.tplants.2009.11.009 https://doi.org/10.1016/j.tplants.2009.11.009 https://doi.org/10.1051/forest:19970102 https://doi.org/10.1016/j.jplph.2005.09.001 https://doi.org/10.1016/j.jplph.2005.09.001 https://doi.org/10.1007/s11356-016-6941-5 https://doi.org/10.1007/s11356-016-6941-5 development of an ssr-based dna fingerprinting method for black wattle (acacia mearnsii de wild) michael w. bairu1,2, willem g. coetzer1,3, assefa b. amelework2* 1institute for commercial forestry research, p. o. box 100281, scottsville 3209, south africa 2agricultural research council, vegetable and ornamental plants, private bag x293 pretoria 0001, south africa (current address) 3 department of genetics, university of the free state, p. o. box 339, bloemfontein 9300, south africa (current address) *corresponding author: assefaa@arc.agric.za (received for publication 28 may 2019; accepted in revised form 20 july 2020) abstract background: the most commonly used method for extracting dna from plant leaf tissue involves cetyl trimethylammonium bromide but some species, such as acacia mearnsii, contain high levels of secondary metabolites and polysaccharides that interfere with this process. various modifications have been proposed for effective removal of these biomolecules but these methods can be time consuming. therefore, this study was initiated to optimise the cetyl-trimethylammonium bromide protocol for the extraction of high-quality genomic dna and to develop a fingerprinting tool using cross species transferable simple sequence repeat markers for genetic diversity studies in a. mearnsii. methods: five ctab-based modification were examined and 49 cross-species microsatellite markers, developed for several acacia species, were tested in four multiplex panels of a. mearnsii populations. results: the modified protocol yields high quantity and quality dna from a. mearnsii leaves using high concentration of nacl to remove polysaccharides and polyvinylpolypyrrolidone (pvpp) to eliminate polyphenols during dna purification. in addition, omitting the selective precipitation and nacl gradient steps in the extraction protocol, enabled us to extract dna 10–20 min faster than the normal protocol. of the tested microsatellite loci, 11 were successful in amplifying sharp and high-intensity bands in all the four multiplex panels and were polymorphic. the level of polymorphism ranged from 0.115 to 0.794, with a mean 0.50 and mean number of alleles varied from 2 to 10, with overall mean of 6 alleles per locus. the mean observed and expected heterozygosity ranged from 0.058 to 0.970 and 0.102 to 0.796, respectively. the 11 microsatellite loci that were effectively amplified from a. mearnsii dna were adequate in detecting genetic variation among the tested populations. conclusions: these pcr-based, multi-allelic, co-dominant microsatellite markers provide a powerful tool for genetic, breeding and conservation studies in a. mearnsii. new zealand journal of forestry science bairu et al. new zealand journal of forestry science (2020) 50:6 https://doi.org/10.33494/nzjfs502020x56x e-issn: 1179-5395 published on-line: 14/08/2020 © the author(s). 2020 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access 2015). acacia mearnsii was first introduced to south africa in 1864 (de beer 1986). south africa is the third largest, and one of the oldest, plantation resource areas in the southern hemisphere (owen & van der zel 2000). acacia mearnsii is grown in approximately 130 000 hectares, that stretches from south-eastern mpumalanga (27.0245° s, 30.7925° e) to kwazulu-natal (29.7285° s, 30.5319° e) in the north (the department of forestry introduction acacia mearnsii is a well-known commercially grown tree species in south africa. this species is cultivated mainly for its timber, timber products, pulp and its tannin-rich bark for the leather tanning industry. acacia mearnsii is indigenous to australia but is currently grown intensively in india, japan, south africa, kenya, tanzania, uganda, brazil, uruguay and argentina (ildis keywords: acacia mearnsii, cross-species microsatellite markers, dna extraction, pcr amplification https://doi.org/10.33494/nzjfs502020x56x http://creativecommons.org/licenses/by/4.0/), bairu et al. new zealand journal of forestry science (2020) 50:6 page 2 and water affair 2003). acacia mearnsii was first introduced into south africa via a bag of seeds with no prior information on the genetic variability. consequently, a. mearnsii plantations were presumed to have a narrow genetic base. the high out crossing nature and the low self-incompatibility of acacia species and their broad geographic adaptability enabled the species to maintain high genetic diversity (duminil et al. 2009). however, the out crossing rate is highly dependent on flower fecundity, pattern and synchronization (butcher et al. 2000). these factors are highly influenced by the environment. intensive breeding was carried out over the years and populations were established in geographically isolated regions depending on the objective of breeding (frost tolerance, disease resistance and timber and bark quality). these populations were exposed to specific environmental factors that led to populations with distinct phenotypes. it was suggested that fragmentation of populations in a given region might result in an increase in the level of inbreeding and a decrease in allelic richness (butcher et al. 2000). natural selection, mutation, genetic drift and mating systems all affect the patterns of genetic variation among populations (millar et al. 2008). to investigate the influence of the above genetic forces on the genetic diversity and patterns of variation, it is helpful to develop appropriate genetic analysis tools. population genetic diversity and conservation studies using molecular techniques are important for population or species survival; and the extraction of high quality genomic dna and polymerase chain reaction (pcr) amplification protocols are essential pre-requisites (bonin et al. 2004; tan & yiap 2009). extraction of highquality dna from plant tissue is challenging because plants have variable levels of metabolites and structural biomolecules that interfere with existing dna extraction protocols (salblok et al. 2009; sahu et al. 2012). secondary metabolites, polysaccharides and polyphenols, are plant biomolecules that often interfere with dna isolation, enzymatic digestion and pcr (weising et al. 2005). in the case where species are characterized by their high secondary metabolite content, such as a. mearnsii, dna extraction procedures require intensive testing and adaptation of different protocols. the most commonly used basic dna extraction methods from plant leaf tissue are derived from the original cetyl trimethylammonium bromide (ctab) based protocol (doyle & doyle 1990). for species like a. mearnsii, which contain high levels of secondary metabolites and polysaccharides, ensuring extraction of high-quality dna is important for downstream genetic analysis (sahu et al. 2012; healey et al. 2014). to contend with the problems associated with secondary metabolites various modifications have been proposed for effective removal of these biomolecules (weising et al. 2005). the modified protocols include the addition of polyvinylpyrrolidone (pvp), insoluble pvpp, bovine serum albumin (bsa), reducing agents like β-mercaptoethanol or high salt solutions to the extraction buffer (allen et al. 2006; varma et al. 2007; lade et al. 2014; arruda et al. 2017). however, these methods can be time consuming, relying on long incubation steps, or requiring multiple dna washes and precipitations that decrease overall yield (healey et al. 2014). therefore, this study was initiated to optimise the ctab protocol for the extraction of high-quality genomic dna and to develop a fingerprinting tool using cross species ssr markers for genetic diversity studies in a. mearnsii. methods sample preparation the development of the dna extraction and ssr-pcr amplification protocols were part of the genetic diversity study to support the institute for commercial forestry research’s a. mearnsii breeding and seed production programmes (https://www.icfr.ukzn.ac.za). the four breeding populations used in this study were derived from a base population constituted with 1081 genotypes collected from different progeny trials resulting from controlled crosses and targeted selfing (dunlop et al. 2003). the four breeding population were reconstituted based on similarity in disease resistance, bark yield and quality, stem form and vigour. two hundred and twenty-eight individuals were selected from the four breeding populations (tp1 = 78, tp2 = 34, tp3 = 64, and tp4 = 52). at sampling time, three of these breeding populations were seven years old and one was a 15-yearold population (tp1). young, fresh, and healthy leaves free from visible contamination were sampled, wrapped in moist paper and placed in a cooler box. upon arrival to the laboratory, each sample was ground into fine powder with mortar and pestle following the addition of liquid nitrogen. the powder was transferred to duplicate 2 ml reaction tubes and immediately stored in a -80 oc freezer. one set of samples were used for dna extraction and the duplicates were stored for future reference. dna extraction various ctab dna extraction protocols were tested for their suitability with a. mearnsii leaf material. the initial protocol used here was developed by centro internacional de mejoramiento de maíz y trigo (cimmyt 2005) and modified by bairu et al. (2006) and moyo et al. (2008). the protocol was amended to increase dna quantity and quality and to shorten the time spent on extractions. briefly, some of the modifications included an altered nacl concentration, to adding pvpp to the ctab extraction buffer, and the collapse of the chloroform and ctab/nacl steps into a single step as described below and summarised in table 1. the final protocol contained the following steps: about 150 mg ground leaf material was added to 800 µl preheated at 65 °c ctab extraction buffer (2% w/v ctab, 100 mm tris-hcl (ph 8.0) 20 mm edta, 1.4 m nacl), with 8 µl 2-mercaptoethanol and 2% pvpp in a clean 2 ml tube. the mixture was incubated in a water bath at 65 °c for 1 hour, while mixing every 10 min by inversion. immediately following incubation 800 µl chloroform: isoamyl alcohol (24:1) and 160 µl preheated at 65 °c ctab/nacl (10% ctab, https://www.icfr.ukzn.ac.za 0.7 m nacl) was added and mixed for 10 min by inversion. the mixture was centrifuged for 10 min at 10000 rpm followed by recovery of the top aqueous layer into a 1.5 ml clean tube. precipitation of dna was achieved by adding 800 µl ice-cold isopropanol to the supernatant and mixed gently by inversion. the mixture was then centrifuged for 15 min at 10 000 rpm followed by careful decanting of the isopropanol, leaving the pellet in the tube. the pellet was washed in two steps, first with 70% ice cold ethanol and then 95% ethanol by inversion. after removing the ethanol by decanting, the pellet was allowed to air dry in a laminar flow bench. finally, the dried pellet was re-suspended in 100 µl te buffer (10 mm tris-hcl, 1 mm edta, ph 8.0) and stored at -20 °c. following dna extraction, the concentration and purity of dna was estimated using the jenway genova nano spectrophotometer (bibby scientific ltd., staffordshire, uk). ssr marker screening due to the lack of species specific ssr markers for a. mearnsii, 49 simple sequence repeat (ssr) markers were used from other acacia species (additonal file). the markers were chosen based on their ability to amplify across species (adamski et al. 2013; aggarwal et al. 2011; miller 2009; ng et al. 2005; butcher et al. 2000). initially, unlabelled primer sets were tested using pcr amplification and the pcr products were visualization on a 2% metaphor™ agarose gel (lonza rockland inc., rockland, usa) with tae buffer. only those markers with visible bands were selected for further evaluation. the forward primer of each selected primer set was then labelled with a fluorescent dye and used for downstream amplification. the pcr products were sent to the central analytical facility at stellenbosch university, south africa for fragment analysis. the software program, gene marker® v2.4.0 (soft genetics) was used to score the electropherograms. after this step, only polymorphic markers that amplified more than two alleles across eight a. mearnsii samples of known genetic variation in the breeding programme were selected for further analyses. pcr amplification the initial pcr reactions were performed using the dreamtaq master mix, ready-to-use solution containing dreamtaq dna polymerase, optimized dreamtaq buffer, mgcl2, and dntps, from thermo scientific. each reaction mixture contained; ~150 ng dna, 5 µl dreamtaq pcr master mix (thermo scientific), 0.4 µm of each primer, 0.8 µl bsa (1mg/ml) and dh2o to make up the final volume of 10 µl. a gradient pcr was performed to determine the optimal annealing temperature (ta) for each primer set tested. the temperature range for the gradient pcr was set from 48–60 °c, while the remaining parameters set were according to the manufacturer’s protocols. the pcr products were viewed on a 2% metaphor™ agarose gel. the optimum ta was established for each primer set and the touchdown pcr method was used for all subsequent pcr amplifications using the dreamtaq master mix. the touchdown protocol had the following conditions: initial denaturation step at 95 °c for 3 min, 10 cycles of 95 °c for 20 seconds, ta 10 °c higher than the optimum and decreasing every cycle by 1 °c for 20 seconds and an elongation step of 72 °c for 30 seconds, followed by 25 cycles of 95 °c for 20 seconds, optimum ta for 20 seconds, 72 °c for 30 seconds, with a final elongation step of 72 °c for 5 minutes. additional magnesium chloride (mgcl2; 1 mm) was added to pcr mixtures for primer sets with weak amplicons. the various annealing temperatures for each primer pair (ta = 48–52 °c) and the different levels of amplicons across the primer sets, made multiplexing difficult. each primer was, therefore, amplified in single-plex and added to a single tube for fragment analysis as a ‘multiplex’. four multiplex sets (multiplex a, b, c and d) were established (table 2). all amplified products were sent to the central analytical facility at stellenbosch university, south africa for fragment analysis. gene marker® v2.4.0 was used for scoring all genotypes. this process was time consuming and an alternative was needed. the kapa2g fast multiplex kit (kapa biosystems, cape town, south africa) was tested for its multiplexing utility, using the selected acacia ssr primer sets. the kapa2g fast multiplex kit contains kapa2g fast hotstart dna polymerase, a buffer optimised for multiplex pcr, with 0.2 mm of each dntp and 3 mm mgcl2 (at 1x). this kit is pre-optimized for multiplex pcr reactions to use with primer interactions, primer concentrations, dna quality and quantity, with marginal changes to annealing temperatures. each pcr reaction mixture contained: ~150 µl dna, 5 μl kapa2g fast multiplex mix, 0.6 µl bsa (1mg/ml), 0.1–0.4 μm of each primer and dh2o to make up the final volume of 10 µl. the pcr cycle bairu et al. new zealand journal of forestry science (2020) 50:6 page 3 ctab protocol mean dna concentration (ng/µl) mean 260/280 ratio mean 260/230 ratio 1 (0.7 m nacl in ctab buffer) 44.217# 2.089# 2 (with ctab/nacl step) 15.981 1.787 3 (1.4 m nacl in ctab buffer) 113.222 1.908 0.966 4 (combined ctab/nacl step) 131.214 1.915 1.309 5 (final protocol) (pvpp added) 159.515 1.887 1.754 table 1: comparison between the different modified ctab protocols tested for dna extraction from acacia mearnsii leaf material. #the mean taken from four samples parameters were set up following the manufacturer’s protocols. the pcr cycle parameters were as follows: initial denaturation at 95 °c for 3 minutes, 30 cycles of 95 °c for 15 seconds, 60 °c for 30 seconds, and 72 °c for 20 seconds, with a final elongation step of 72 °c for 1 minute. the above mentioned multiplex sets were bairu et al. new zealand journal of forestry science (2020) 50:6 page 4 used for amplification and the amplified products were sent to the central analytical facility at stellenbosch university, south africa for fragment analysis. all loci were successfully amplified using the kapa2g fast multiplex kit. all subsequent amplifications were done using this kit. multiplex dye primer sequence (5’-3’) repeat motifs type repeat ta size range (bp) na source a ah2-1 fam f: gacagagggagcattttgta (ct)12 di60 146-160 10 aggarwal et al. 2011 r: cagacaagaccagagaatgac ah3-18 fam f: tgagacaattaatggtggtg (taa)5 tri 60 209-221 4 aggarwal et al. 2011 r: tttacaagggaaaagctgag am465 cy 3.5 f: tgggtatcacttccaccatt (ac)23 di60 113-131 5 butcher et al. 2000 r: aggctgcttctttgtgcagg b ah3-1 pet f: ctaaggcacttggatcattc (tct)5 tri60 214-217 2 aggarwal et al. 2011 r: agagagagagaggcacactg ah3-10 fam f: agggatatcggatgcttact (gat)7 tri60 178-202 10 aggarwal et al. 2011 r: aaagatgcagcagacctatc ak15 vic f: cacccccacgttatcttaca (tat)5 tri60 297-309 4 adamski et al. 2013 r: gactggcgaaagagtcgaa c ah16 hex f: gagggtaatgcttcaagtagac (ga)16 di60 86-88 2 ng et al. 2005 r: tgcgtgtctccccactactc ah56 cy3.5 f: gatagctcatagaaacaccatacc (ga)9 di60 123-129 4 ng et al. 2005 r: ggcgaagctctctctctctctctctct ak89 fam f: aggggaaggacgaaagttgt (ac)7 di60 160-174 5 adamski et al. 2013 r: gcaagaggagcttcaagtgg d ah01 fam f: ttgaggttgagggtgatgaa (ga)6 di60 # 106-116 5 ng et al. 2005 r: ggcaagcctctctctctct ah2-13 ned f: gaagaagcaggaggaggtag (ag)7 di60 # 143-151 7 aggarwal et al. 2011 r: tgttttccacttctcacaca table 2: detail of the eleven selected ssr markers with bands amplified in a. mearnsii. # 0.5 mm additional mgcl2 added; ta= annealing temperature; na= number of observed alleles. only the annealing temperature used for the kapa2g fast multiplex kit is reported. statistical analyses estimates of null allele frequencies were performed using the software program freena (chapuis and estoup 2007) using the expectation maximization algorithm (em) (dempster et al. 1977). the influence of null alleles on genetic diversity estimates was assessed with a wilcoxon signed rank test using r (r core team 2015) and fst per locus values for corrected uncorrected null alleles were made using the excluding null alleles (ena) method (chapuis & estoup 2007). genetic diversity per locus was assessed using the mean number of alleles (na), observed (ho) and unbiased expected (uhe) heterozygosity using genalex v6.5 (peakall & smouse 2012), and the inbreeding coefficient (fis) was determined using genepop v4.3 (rousset 2008). polymorphic information content (pic) was estimated using cervus v3.0 (kalinowski et al. 2007), probability of identity (pid) for unrelated individuals and probability of identity for full siblings (pidsibs) were estimated using genealex v6.5, and allelic richness (ar), using the rarefaction method, as implemented in fstat (goudet 2001). deviations from hardy-weinberg equilibrium were calculated using genepop v4.3. pid and pidsibs estimate the probability that two randomly chosen full-sibs within a given population that have the same genotype on a set of makers. the pid assumes that there is no linkage disequilibrium and population substructure. when such assumptions do not hold, the pidsibs is often used as a conservative upper bound of the “real” probability. results dna extraction, quality and quantity in this study, the ctab dna isolation technique was employed to extract dna from acacia mearnsii leaf samples with slight modifications. the mean concentration and purity of dna samples extracted from the leaves of acacia mearnsii based on the five protocols tested are presented in table 1. the mean dna concentration varied significantly ranging from 44 to 159 ng/µl, at low concentration of nacl and at high concentration of polyvinyl polypyrrolidone (pvpp), respectively. the addition of a high concentration nacl and pvpp provided the best results, with high dna concentrations (>100 ng/µl) and near optimum wavelength ratios (mean 260/280 = 1.887; mean 260/230 = 1.754). efficient dna extraction was achieved with the optimised ctab protocol, with less than 5% reextraction. however, the 260/230 ratio values from the spectrophotometer assessments indicated that some residual phenols and/or carbohydrates might still be present in the dna extracts. subsequently, bovine serum albumin (bsa) was used during pcr amplification to eliminate the effect of the co-extracted compounds. ssr screening and summary statistics of the 49 assessed microsatellite makers developed for other acacia species, only 11 (22.5%) loci amplified clear reproducible high-quality dna bands in a. mearnsii (table 2). these markers were polymorphic and consistent amplification was achieved among the four populations of a. mearnsii tested. the proportion of missing data ranged from 0% to 3.5% per locus. missing data of up to ~4% of the scored genotypes is, however, acceptable for population genetic studies (putman and carbone 2014) and no genotype was eliminated from the analysis. the proportion of null alleles calculated per locus over all samples ranged from 0.00 (ak89) to 25.8 (ah3-10), with a mean of 10.8 over all loci. the null allele frequencies (naf) for locus ah2-13 and ah310 were higher than 20% (table 3). the observed null allele frequencies per population for ah2-13 and ah310 ranged from 15.5–20.6% and from 19.4–32.1% per population, respectively (data not shown). although the values are high, they are still within the range of values often reported in other studies using ssr loci develop in other species (dakin and avise 2004). there was no significant difference detected between excluding null allele (ena) corrected and uncorrected fst values (p-value > 0.05; bonferroni corrected) and therefore, it was decided to keep all loci for the subsequent analyses. the genotypes showed a wide range of allelic diversity from 2 to 10 alleles per locus. the highest allele number (na = 10) was observed at markers ah2-1 and ah3-10 and the lowest was for ah16, with an overall mean na of 5.3 (table 2). based on alleles detected among four multiplex panels in all four a. mearnsii populations, dinucleotide repeat ssrs were relatively more polymorphic than those with tri-nucleotide repeats. the unbiased expected heterozygosity (uhe) ranged from 0.103 to 0.796 per locus, while the observed heterozygosity (ho) varied from 0.058 to 0.97 (table 3). only three loci (ah3-1, ah3-18 and am465) did not show significant deviation from hardy-weinberg equilibrium (table 3). a high level of variation was observed for the inbreeding coefficient (fis) estimates among loci (fis = -0.567 to 0.653). negative fis values were only observed at two loci (ah3-18, ak89), indicating a heterozygote excess for ak89. two loci were found to be highly polymorphic (ah2-1 and ah3-10). both loci had pic values greater than 0.7, with high allelic richness (ar) values. more than 50% of the markers had pic values greater than 0.50. the combined pid value for the ssr panel was 7.2 x 10-8. therefore 1 in about 14 million trees will have the same genotype. the pidsibs value of 1 x 10-3 provides a lower bound for the number of loci required for the successful identification of individuals, with 1 in about 1000 individuals sharing the same genotype if all individuals are full siblings. this indicates that these markers had a high discriminatory power and were found to be highly suitable for genetic diversity analysis. the allelic richness (ar) estimates were based on a minimum sample size of 30 individuals, the values ranged from 2 (ah16 and ah3-1) to 9 (ah3-1), with a mean of 4.8 alleles per locus. discussion exploiting the differences in solubility of polysaccharides and dna in the ctab buffer by adjusting the concentration of sodium chloride can aid the removal of polysaccharides (weising et al. 2005). it has been bairu et al. new zealand journal of forestry science (2020) 50:6 page 5 described that a high salt concentration in the extraction buffer assists in eliminating polysaccharides by increasing their solubility in ethanol (fang et al. 1992; lodhi et al. 1994; varma et al. 2007). the addition of nacl at concentrations higher than 0.5 m, along with ctab, successfully removes polysaccharides during dna extraction (moreira & oliveira 2011; lucas et al. 2019). in the present study, a concentration of 1.4 m nacl was used in the extraction buffer that further improved the quality of the extracted dna. endogenous dnases can degrade the extracted dna unless edta is added (weising et al. 2005). edta has an inhibitory effect on magnesium-dependant dnases by binding to magnesium ions through chelation (weising et al. 2005). in this study, the addition of pvpp, which have strong h-receptor for binding and removal of polyphenolics into the ctab buffer was helpful in removing the polyphenols and polysaccharides from leaf samples of acacia mearnsii (kolosova et al. 2004). this protocol resulted in a mean total dna amount of 15.95 μg extracted from 150 mg leaf tissue. similar results were obtained with modified ctab methods in cotton (15–30 μg from 100 mg plant tissue; ali et al. 2019), as well as arabidopsis thaliana, zea mays and nicotiana sp. (5–30 μg from 200 mg plant tissue; allen et al. 2006). however, the values obtained in this study are much lower than those obtained from other species characterized by high levels of secondary metabolites (sahu et al. 2012; arruda et al. 2017). tiwari et al. (2012), obtained a dna concentration ranged from 179 to 833 ng/µl using the modified ctab protocol in selected medicinal plants. similarly, sahu et al. (2012) achieved dna concentration ranged from 8.8 to 9.9 μg/μl that was amenable of rapd markers analysis. the difference in dna concentration could be attributed to the various modifications such as high salt concentration, pvp, pvpp and avoiding the use of liquid nitrogen and selective precipitation and washing steps in addition to species differences. the 260/280nm and 260/230nm wavelength ratios are well-known measures of nucleic acid quality. the recommended values for the 260/280 ratio ranged from 1.8 to 2.0 and the optimised ctab protocol in this study resulted in an absorbance value of 1.89, which is within the accepted range indicating the protocol is efficient in obtaining high-quality dna samples. the quality of the dna obtained using the optimized protocol is sufficient since ratios in the range of 1.6–1.8 are acceptable for pcr reactions. similarly, in a pure dna sample, the reference interval for 260/230 ratio is 2.0–2.2, however, the ratio obtained in this study was 1.75 which falls below the recommended level. this value was similar to those reported by murray and thompson (1980) but lower than values reported by arruda et al. (2017). this indicates that the quality of the dna is low due to possible contamination by residual polyphenolic compounds and/or carbohydrates in the extract (moncada et al. 2013). the high quality dna obtained by arruda et al. (2017) might be attributed to the higher concentration of ctab (3%) and nacl (2.5 m) compared to ctab (2%) and nacl (1.4 m) used in this study. this could aid to more efficient elimination of polysaccharides since the composition and concentration of reagents can interfere with the quality and quantity of extracted dna (borges et al. 2012). in addition, the low 260/280 and 260/230 ratios obtained in this study might be attributed to high quantities of tannin in a. mearnsii leaves (elgailani & ishak 2014). the systematic exploration of microsatellite markers across species should be the first step in developing ssrs bairu et al. new zealand journal of forestry science (2020) 50:6 page 6 locus na ar ho uhe fis pid pidsibs pic naf hwd p-value ah2-1 8.25 9.96 0.514 0.779 0.393 0.057 0.36 0.792 0.18 0.00 ah2-13 4.00 7.96 0.216 0.446 0.608 0.340 0.60 0.410 0.21 0.00 ah3-18 3.75 4.97 0.407 0.431 0.001 0.360 0.62 0.395 0.00 0.19 am465 4.25 5 0.634 0.663 0.042 0.170 0.46 0.608 0.02 0.37 ah3-10 9.00 10 0.336 0.796 0.575 0.054 0.36 0.794 0.26 0.00 ah3-1 2.00 2 0.095 0.102 0.080 0.780 0.88 0.115 0.02 0.23 ak15 3.25 4 0.361 0.563 0.403 0.270 0.53 0.485 0.16 0.00 ah16 2.00 2 0.058 0.164 0.653 0.710 0.85 0.151 0.14 0.00 ah56 4.50 6 0.337 0.598 0.445 0.210 0.49 0.557 0.17 0.00 ak89 5.50 6.97 0.970 0.608 -0.567 0.210 0.50 0.551 0.00 0.00 ah01 6.25 7 0.622 0.639 0.005 0.170 0.48 0.594 0.04 0.00 mean 4.80 5.99 0.416 0.524 0.240 0.496 0.11 se 0.36 0.83 0.041 0.035 0.11 0.067 0.03 table 3: genetic diversity parameter values measured per locus over all samples ar= allelic richness; ho= observed heterozygosity; he= unbiased expected heterozygosity; fis= inbreeding coefficient; pid= probability of identity per locus; pidsibs= probability of identity for full siblings per locus; pic= polymorphic information content; naf(%)= null allele frequency as percentage; hwd= deviation from hardy-weinberg; se= standard error in under-studied species. developing new ssr primers specific to the species of interest is both time consuming and costly (yosodha et al. 2005; ravishankar et al. 2015). it is, therefore, more feasible to source ssr primers developed for other related species. the use of ssr markers across species from the same genus (omondi et al. 2010; aggarwal et al. 2011; adamski et al. 2013; le roux et al. 2013; roberts et al. 2013) and across different genera (peakall et al. 1998) have been reported for leguminosae. in this study, the possibility of crossspecies transferability of 49 microsatellites, derived from a. mangium, a. koa and a. auriculiformis, into a. mearnsii was investigated. of the tested 49 cross-species markers, only 11 were effective in amplifying sharp and high intensity bands. the transfer rate was 22.5%, which is moderately low. this result was in agreement with butcher et al. (2000), who reported low levels of crossspecies ssr amplification within acacia species. the low transferability might be attributed to the genetic divergence among the acacia species due to exposure to different bottlenecks and geographic isolation (varshney et al. 2005; barbará et al. 2007). similarly, several studies indicated the low transferability rate of ssrs among plant species (luro et al. 2008; koppolu et al. 2010; lee et al. 2011). although transferability was in general very low, the mean number of alleles amplified per locus was relatively higher (6.0) in this study compared to the 2.9 allele per locus reported by (adamski et al. 2013) using 16 acacia koa ssr markers on seven acacia species. the high levels of polymorphisms of ssr loci indicated that the application of these markers for genetic diversity studies in a. mearnsii was useful and cost-effective compared to developing new ssr markers specific to the species. this study provided an important insight in the development of microsatellite markers suitable for genetic studies in a. mearnsii. in the current study, we only used ssrs to study their cross-species transferability and on the genetic diversity of four a. mearnsii breeding population. however, their genomic distribution, biological function and other uses should be further investigated (vieira et al. 2016). various factors such as genotyping errors of loci should be considered in choosing a sufficient ssr panel for population genetic and parentage studies. a wellknown problem with ssr is the occurrence of null alleles and over estimation of alleles due to polymerase strand-slippage in dna replication (vieira et al. 2016). the analysis of null allele frequencies per genotype conducted using the maximum likelihood estimates in this study confirmed the presence of null alleles in all loci but ak89. for these loci, null-allele frequencies estimates ranged between 0% and 25.8%, with 64% of the loci showing null-allele frequencies above 5%. in this case, the null alleles occur when an allele fails to amplify via pcr and heterozygotes might falsely be scored as homozygote. null alleles can possible be either caused by mutations in the flanking regions of a microsatellite sequence (chapuis & estoup 2007; dakin & avise 2004), or associated with shorter alleles outperforming the longer alleles, usually due to lower dna quality or quantity (gagneux et al. 1997; wattier et al. 1998). bairu et al. new zealand journal of forestry science (2020) 50:6 page 7 consequently, null alleles may inflate genetic diversity parameter estimates (chapuis & estoup 2007; carlsson 2008) and result in inaccurate parentage assessments (dakin & avise 2004). carlsson (2008) reported that in datasets containing null alleles, the genetic differentiation estimate fst were slightly overestimated. if the frequency of a null allele is <0.1, some estimators can be used directly without adjustment; if it is >0.5, the potency of estimation is too low and such a locus should be excluded (huang et al. 2016). in this study, the estimated null allele frequencies were generally below the threshold levels. all ssr loci were used for the downstream diversity analyses. it was reported by carlsson (2008) that loci affected by null alleles can still be useful for population level studies. however, breeders should interpret the results with caution as the effect of null alleles on parentage analysis is more substantial, as it could lead to false parentage assignments (pemberton et al. 1995). since these breeding populations were derived from the same base population, it is therefore, advisable to use parentage assignment methods that consider the presence of null alleles in the dataset (dakin & avise 2004) during further selection and breeding planning. the mean genetic diversity value observed in a. mearnsii (he = 0.524) in this study was much higher than the study conducted on a. longifolia in portugal (he = 0.190) (vicente et al. 2018) using ssrs. however, the value is more comparable with a study of a. senegal (he = 0.519) (djibo et al. 2017), a. auriculiformis (he = 0.60) (son et al. 2016) and a. mangium (he = 0.56) (son et al. 2016) using ssrs. the inbreeding coefficient (fis) measures the deviation of heterozygosity from expected values under the assumptions of hardy weinberg. a negative and positive fis value indicates an excess or deficiency of heterozygosity in a population relative to the ratio under hwe. all the markers except ak89 revealed a positive fis value suggesting a deficiency of heterozygotes. the deficiency of heterozygotes can be attributed to either inbreeding or existence of null alleles or the presence of subpopulations within populations (wahlund effect) (jordana 2003). of the 11 microsatellite loci analysed, 9 loci deviated from hwe in a. mearnsii (p < 0.001). this might be due to the small sample size used or the existence of null alleles (wang 2008). linkage disequilibrium observed among pairs of loci in this study appeared to be associated with the presence of null alleles in some of the analysed loci. all the microsatellite loci, except ak89 and ah01, which were found to depart from the hardy-weinberg equilibrium (hwe), had null allele frequencies exceeding 10%. the probability of identity (pid) is also a widely used individual identification estimator (peakall & sydes 1996; reed et al. 1997; waits et al. 2001; ferrie et al. 2013), which provides the probability that two randomly drawn individuals within a given population will have identical genotypes at multiple loci. for pid, it is assumed that there is neither linkage disequilibrium nor population substructure. in the case of such assumptions not holding true, the pid might drastically underestimate the above-mentioned probability. hence, probability of identity for siblings (pidsibling) is often used as a conservative upper bound of the “real” probability (waits et al. 2001). low pid values are required to accurately estimate individual identity and not falsely classify individuals as identical (waits et al. 2001; vidya & sukumar 2005). locus pid values in the range of 0.01 – 0.0001 are generally used in wildlife forensics (waits et al. 2001), suggesting the values in the current study are slightly higher than ideal. it is important to take the population size and genetic history of the species under study into account in choosing an appropriate marker panel using pid estimates (schwartz & monfort 2008). for example, if the study population comprise of 10000 individuals, a pid < 0.0001 (1 in about 10000 individuals) should be sufficient for individual identification. conclusions in this study, we managed to get high quantity and quality dna from a. mearnsii leaves by modifying the ctab-based dna extraction procedure using high concentration of nacl to remove polysaccharides and pvpp to eliminate polyphenols during dna purification. in addition, by combining the ctab and nacl steps in the extraction protocol, we succeeded in the extraction of dna 10–20 min faster than the normal protocol. according to our knowledge, this is the first assessment of cross-species microsatellite marker transferability and the use of these markers to study the genetic variation in a. mearnsi. however, it was also demonstrated that the majority of cross-species microsatellite markers were not transferable across species in the genus acacia. considering the high polymorphism demonstrated by the studied microsatellite markers, the cross-species amplification is an interesting alternative to the development of new microsatellites in a. mearnsii. these multi-allelic, pcr-based, co-dominant microsatellite loci provides a powerful tool for genetic, breeding and conservation genetic studies of a. mearnsii. competing interests the authors declare no competing interest. the institution where the research was done, and the funding body are duly acknowledged. authors’ contributions mb designed, initiated and managed the project, and contributed to laboratory work and manuscript preparation. wc undertook most of the laboratory work, and contributed to data analysis and manuscript preparation. aa undertook data analysis and manuscript preparation. additional files additional file 1: list of 49 cross-species ssr markers used in this study acknowledgements the institute for commercial forestry research (icfr), south africa for affording the authors the opportunity to do this research and the south african wattle growers union (through icfr) for making the funding available. table 2: confusion matrix references adamski, d.j., dudley, n.s., morden, c.w. & borthakur, d. (2013). cross-amplification of nonnative acacia species in the hawaiian islands using microsatellite markers from acacia koa. plant biosystems 147, 1088-1091. https://doi.org/10.1080/11263504.2 012.749958 aggarwal, r.k., allainguillaume, j., bajay, m.m., barthwal, s., bertolino, p., chauhan, p., et al. (2011). permanent genetic resources added to molecular ecology resources database 1 august 2010 – 30 september 2010. molecular ecology resources 11, 219-222. https://doi.org/10.1111/j.17550998.2010.02944.x ali, q. salisu, i.b., raza, a., shahid, a.a., rao, a.q., & husnain, t. (2019). a modified protocol for rapid dna isolation from cotton (gossypium spp.). methodsx 6, 259-264. https://doi.org/10.1016/j. mex.2019.01.010 allen, g.c., flores-vergara, m.a., krasynanski, s., kumar, s., & thompson. w.f. (2006). a modified protocol for rapid dna isolation from plant tissues using cetyltrimethylammonium bromide. nature protocols 1, 2320-2325. https://doi.org/10.1038/ nprot.2006.384 arruda, s.r., pereira, d.g., silva-castro, m.m., brito, m.g., & waldschmidt, a.m. (2017). an optimized protocol for dna extraction in plants with a high content of secondary metabolites, based on leaves of mimosa tenuiflora (willd.) poir. (leguminosae). genetics and molecular research 16, 1-9. https:// doi.org/10.4238/gmr16039063 bairu, m.w., fennell, c.w., & van staden, j. (2006). the effect of plant growth regulators on somaclonal variation in cavendish banana (musa aaa cv. ‘zelig’). scientia horticulturae 108, 347–351. https://doi.org/10.1016/j.scienta.2006.01.039 barbará, t., palma-silva, c., paggi, g.m., bered, f., fay, m.f., & lexer, c. (2007). cross-species transfer of nuclear microsatellite markers: potential and limitations. molecular ecology 16, 3759-3767. https://doi. org/10.1111/j.1365-294x.2007.03439.x bonin, a., bellemain, e., bronken, e.p., pompanon, f., brochmann, c., & taberlet, p. (2004). how to track and assess genotyping errors in population genetics studies. molecular ecology 13, 3261-3273. https:// doi.org/10.1111/j.1365-294x.2004.02346.x borges, d.b., amorim, m.b., waldschmidt, a.m., marianoneto, e., vivas, c.v., & pereira, d.g. (2012). optimization of dna extraction from fresh leaf tissues of melanoxylon brauna (fabaceae). genetics and molecular research 11, 1586-1591. https:// doi.org/10.4238/2012.may.22.8 butcher, p.a., decroocq, s., gray, y., & moran, g.f. (2000). development, inheritance and cross-species amplification of microsatellite markers from acacia bairu et al. new zealand journal of forestry science (2020) 50:6 page 8 https://doi.org/10.1111/j.1755-0998.2010.02944.x https://doi.org/10.1111/j.1755-0998.2010.02944.x https://doi.org/10.1111/j.1755-0998.2010.02944.x https://doi.org/10.1111/j.1755-0998.2010.02944.x https://doi.org/10.1016/j.mex.2019.01.010 https://doi.org/10.1016/j.mex.2019.01.010 https://doi.org/10.1038/nprot.2006.384 https://doi.org/10.1038/nprot.2006.384 https://doi.org/10.4238/gmr16039063 https://doi.org/10.4238/gmr16039063 https://doi.org/10.1016/j.scienta.2006.01.039 https://doi.org/10.1111/j.1365-294x.2007.03439.x https://doi.org/10.1111/j.1365-294x.2007.03439.x https://doi.org/10.1111/j.1365-294x.2004.02346.x https://doi.org/10.1111/j.1365-294x.2004.02346.x https://doi.org/10.4238/2012.may.22.8 https://doi.org/10.4238/2012.may.22.8 mangium. theoretical and applied genetics 101, 1282-1290.  https://doi.org/10.1007/ s001220051608 carlsson, j. (2008). effects of microsatellite null alleles on assignment testing. journal of heredity 99, 616623. https://doi.org/10.1093/jhered/esn048 chapuis, m.-p., & estoup, a. (2007). microsatellite null alleles and estimation of population differentiation. molecular biology and evolution 24, 621-631. https://doi.org/10.1093/molbev/msl191 cimmyt (centro internacional de mejoramiento de maíz y trigo). (2005). laboratory protocols: cimmyt applied molecular genetics laboratory. 3rd ed. el batán, mexico: cimmyt https://repository. cimmyt.org/xmlui/handle/10883/1333 dakin, e.e., & avise, j.c. (2004). microsatellite null alleles in parentage analysis. heredity 93, 504-509. https://doi.org/10.1038/sj.hdy.6800545 de beer, h. (1986). a. mearnsii. [weeds a. 24/1986]. pretoria, south africa: department of agriculture and water supply. dempster, a.p., laird, n.m., & rubin, d.b. (1977). maximum likelihood from incomplete data via the em algorithm. journal of the royal statistical society. series b (methodological) 39, 1-38. https:// doi.org/10.1111/j.2517-6161.1977.tb01600.x department of water affairs and forestry (2003). south african forestry facts for the year 2001/2002. [newsletter]. djibo, e.s., aichatou, a., manssour, a.m., bilassanou, h.i., djibo, m., moussa, t., & zouberou, a.m. (2017). exploring genetic diversity and structure of acacia senegal (l.) willd. to improve its conservation in niger. african journal of biotechnology 16,16501659. https://doi.org/10.5897/ajb2016.15846 doyle j.j. and doyle j.l. (1990). isolation of plant dna from fresh tissue. focus 12, 13-15. duminil, j., brown, r.p., ewédjè, e-e.b.k., mardulyn, p., doucet, j-l., & hardy o.j. (2013). large-scale pattern of genetic differentiation within african rainforest trees: insights on the roles of ecological gradients and past climate changes on the evolution of erythrophleum spp. (fabaceae). bmc evolutionary biology 13: 195. dunlop, r.w., barnes, r.d. nixon, k.m. & hagedorn, s.f. (2003). a breeding strategy for acacia measrnsii (black wattle) in south africa. southern african forestry journal 199, 55-63. https://doi.org/10.10 80/20702620.2003.10431749 elgailani, i.e.h., & ishak, c.y. (2014). determination of tannins of three common acacia species of sudan. advances in chemistry 2014, 1-5. https://doi. org/10.1155/2014/192708 fang, g., hammar, s., & grumet, r. (1992). a quick and inexpensive method for removing polysaccharides from plant genomic dna. biotechniques 13, 52-56. ferrie, g.m., cohen, o.r., schutz, p., leighty, k.a., plasse, c., bettinger, t.l., & hoffman, e.a. (2013). identifying parentage using molecular markers: improving accuracy of studbook records for a captive flock of marabou storks (leptoptilus crumeniferus). zoo biology 32, 556-564. https://doi.org/10.1002/ zoo.21090 gagneux, p., boesch, c., & woodruff, d.s. (1997). microsatellite scoring errors associated with non-invasive genotyping based on nuclear dna amplified from shed hair. molecular ecology 6, 861– 868. https://doi.org/10.1111/j.1365-294x.1997. tb00140.x goudet, j. (2001). fstat: a program to estimate and test gene diversities and fixation indices (version 2.9.3). available from http://www2.unil.ch/popgen/ softwares/fstat.htm [accessed 22 september 2018]. healey, a., furtado, a., cooper, t., & henry, r.j. (2014). protocol: a simple method for extracting nextgeneration sequencing quality genomic dna from recalcitrant plant species. plant methods 10: 21. https://doi.org/10.1186/1746-4811-10-21 huang, k., ritland, k., dunn, d.w., qi, x., guo, s., & li, b. 2016. estimating relatedness in the presence of null alleles. genetics 202, 247-260. https://doi. org/10.1534/genetics.114.163956 ildis. (2015). international legume database and information service. reading, uk:, university of reading school of plant sciences. http://www. ildis.org jordana, j. (2003). genetic structure of eighteen local south european beef cattle breeds by comparative f-statistics analysis. journal animal breeding and genetics 120, 73-87. https://doi.org/10.1046/ j.1439-0388.2003.00384.x kalinowski, s.t., taper, m.l., & marshall, t.c. (2007). revising how the computer program cervus accommodates genotyping error increases success in paternity assignment. molecular ecology 16, 1099-1106. https://doi.org/10.1111/j.1365294x.2007.03089.x kolosova n., miller b., ralph s., ellis b.e., douglas c., ritland k. and bohlmann j. (2004). isolation of high-quality rna from gymnosperm and angiosperm trees. biotechniques 35, 821-824. https://doi.org/10.2144/04365st06 koppolu, r., upadhyaya, h., dwivedi, s., hoisington, d., & varshney, r. (2010). genetic relationships among seven sections of genus arachis studied by using ssr markers. bmc plant biology 10:15. https://doi. org/10.1186/1471-2229-10-15 bairu et al. new zealand journal of forestry science (2020) 50:6 page 9 https://doi.org/10.1007/s001220051608 https://doi.org/10.1007/s001220051608 https://doi.org/10.1093/jhered/esn048 https://doi.org/10.1093/molbev/msl191 https://repository.cimmyt.org/xmlui/handle/10883/1333 https://repository.cimmyt.org/xmlui/handle/10883/1333 https://doi.org/10.1038/sj.hdy.6800545 https://doi.org/10.1111/j.2517-6161.1977.tb01600.x https://doi.org/10.1111/j.2517-6161.1977.tb01600.x https://doi.org/10.5897/ajb2016.15846 https://doi.org/10.1080/20702620.2003.10431749 https://doi.org/10.1080/20702620.2003.10431749 https://doi.org/10.1155/2014/192708 https://doi.org/10.1155/2014/192708 https://doi.org/10.1002/zoo.21090 https://doi.org/10.1002/zoo.21090 https://doi.org/10.1111/j.1365-294x.1997.tb00140.x https://doi.org/10.1111/j.1365-294x.1997.tb00140.x https://doi.org/10.1186/1746-4811-10-21 https://doi.org/10.1534/genetics.114.163956 https://doi.org/10.1534/genetics.114.163956 https://doi.org/10.1046/j.1439-0388.2003.00384.x https://doi.org/10.1046/j.1439-0388.2003.00384.x https://doi.org/10.1111/j.1365-294x.2007.03089.x https://doi.org/10.1111/j.1365-294x.2007.03089.x https://doi.org/10.2144/04365st06 https://doi.org/10.1186/1471-2229-10-15 https://doi.org/10.1186/1471-2229-10-15 lade, b.d., patil, a.s., & paikrao, h.m. (2014). efficient genomic dna extraction protocol from medicinal rich passiflora foetida containing high level of polysaccharide and polyphenol. springerplus 3: 457. https://doi.org/10.1186/2193-1801-3-457 le roux, j.j., richardson, d.m., wilson, j.r.u., & ndlovu, j. (2013). human usage in the native range may determine future genetic structure of an invasion: insights from acacia pycnantha. bmc ecology 13: 37. https://doi.org/10.1186/1472-6785-13-37 lee, g-a., kwon, s-j., park, y-j., lee, m.c., kim, h.h., lee, j.s., lee, s-y., gwag, j-g., kim, c-k., & ma, k-h. (2011). cross-amplification of ssr markers developed from allium sativum to other allium species. scientia horticulturae 128, 401-407. https://doi. org/10.1016/j.scienta.2011.02.014 leberg, p.l. (2002). estimating allelic richness: effects of sample size and bottlenecks. molecular ecology 11, 2445–2449. https://doi.org/10.1046/j.1365294x.2002.01612.x lodhi, m.a., ye, g.n., weeden, n.f., & reisch, b.i. (1994). a simple and efficient method for dna extraction from grapevine cultivars and vitis species. plant molecular biology reporter 12, 6-13. https://doi. org/10.1007/bf02668658 lucas, m.s., carvalho, c.s. hypolito, g.b., & côrtes, m.c. (2019). optimized protocol to isolate high quality genomic dna from different tissues of a palm species. hoehnea 46: e942018. https://doi. org/10.1590/2236-8906-94/2018 luro, f.l., costantino, g., terol, j., argout, x., allario, t., wincker, p., talon, m., ollitrault, p., & morillon, r. (2008). transferability of the est-ssrs developed on nules clementine (citrus clementina hort ex tan) to other citrus species and their effectiveness for genetic mapping. bmc genomics 9: 287. https:// doi.org/10.1186/1471-2164-9-287 millar, m.a., byrne, m., nuberg, i., & sedgley, m. (2008). high outcrossing and random pollen dispersal in a planted stand of acacia saligna subsp. saligna revealed by paternity analysis using microsatellites. tree genetics & genomes 4, 367-377. https://doi. org/10.1007/s11295-007-0115-z millar, m.a. 2009. characterisation of microsatellite dna markers for the rare acacia woodmaniorum (leguminosae: mimosaceae). conservation genetics resources 1, 441-445. https://doi.org/10.1007/ s12686-009-9102-z moncada x., payacán c., arriaza f., lobos s., seelenfreund d. and seelenfreund a. (2013). dna extraction and amplification from contemporary polynesian bark-cloth. plos one 8: e56549. https://doi. org/10.1371/journal.pone.0056549 moreira, p.a., & oliveira, d.a. (2011). leaf age affects the quality of dna extracted from dimorphandra mollis (fabaceae), a tropical tree species from the cerrado region of brazil. genetics and molecular research 10, 353-358. https://doi.org/10.4238/ vol10-1gmr1030 moyo, m., amoo, s.o., bairu, m.w., finnie, j.f., & van staden, j. (2008). optimising dna isolation for medicinal plants. south african journal of botany 74, 771–775. https://doi.org/10.1016/j. sajb.2008.07.001 murray, m.g., & thompson, w.f. (1980). rapid isolation of high molecular weight plant dna. nucleic acids research 8, 4321-4326. https://doi.org/10.1093/ nar/8.19.4321 ng, c.h., koh, s., lee, s., ng, k., mark, a., norwati, m., & wickneswari, r. (2005). isolation of 15 polymorphic microsatellite loci in acacia hybrid (acacia mangium × acacia auriculiformis). molecular ecology notes 5, 572-575. https://doi. org/10.1111/j.1471-8286.2005.00994.x omondi, s.f., kireger, e., dangasuk, o.g., chikamai, b., odee, d.w., cavers, s., & khasa, d.p. (2010). genetic diversity and population structure of acacia senegal (l) willd. in kenya. tropical plant biology 3, 59-70. https://doi.org/10.1007/s12042-009-9037-2 owen, d.l., & van der zel, d.w. (2000). chapter 1: trees, forests and plantations in southern africa. in d.l. owen (ed.) south african forestry handbook. pretoria, south africa: south african institute of forestry. pp. 3-8. peakall, r., & smouse, p.e. (2012). genalex 6.5: genetic analysis in excel. population genetic software for teaching and research-an update. bioinformatics 28, 2537-2539. https://doi.org/10.1093/ bioinformatics/bts460 peakall, r., & sydes, m.a. (1996). defining priorities for achieving practical outcomes from the genetic studies of rare plants. in s. stephens, & s. maxwell (eds.), back from the brink: refining the threatened species recovery process. (pp. 119-129). sydney: surrey beatty and sons. peakall, r., gilmore, s., keys, w., morgante, m., & rafalski, a. (1998). cross-species amplification of soybean (glycine max) simple sequence repeats (ssrs) within the genus and other legume genera: implications for the transferability of ssrs in plants. molecular biology and evolution 15, 12751287. https://doi.org/10.1093/oxfordjournals. molbev.a025856 pemberton, j.m., slate, j., bancroft, d.r., & barrett, j.a. (1995). non-amplifying alleles at microsatellite loci: a caution for parentage and population studies. molecular ecology 4, 249–252. https://doi. org/10.1111/j.1365-294x.1995.tb00214.x putman, a.i., & carbone, i. (2014). challenges in analysis and interpretation of microsatellite data for population genetic studies. ecology and evolution 4, 4399-4428. https://doi.org/10.1002/ece3.1305 bairu et al. new zealand journal of forestry science (2020) 50:6 page 10 https://doi.org/10.1186/2193-1801-3-457 https://doi.org/10.1186/1472-6785-13-37 https://doi.org/10.1016/j.scienta.2011.02.014 https://doi.org/10.1016/j.scienta.2011.02.014 https://doi.org/10.1046/j.1365-294x.2002.01612.x https://doi.org/10.1046/j.1365-294x.2002.01612.x https://doi.org/10.1007/bf02668658 https://doi.org/10.1007/bf02668658 https://doi.org/10.1590/2236-8906-94/2018 https://doi.org/10.1590/2236-8906-94/2018 https://doi.org/10.1186/1471-2164-9-287 https://doi.org/10.1186/1471-2164-9-287 https://doi.org/10.1007/s11295-007-0115-z https://doi.org/10.1007/s11295-007-0115-z https://doi.org/10.1007/s12686-009-9102-z https://doi.org/10.1007/s12686-009-9102-z https://doi.org/10.1371/journal.pone.0056549 https://doi.org/10.1371/journal.pone.0056549 https://doi.org/10.4238/vol10-1gmr1030 https://doi.org/10.4238/vol10-1gmr1030 https://doi.org/10.1016/j.sajb.2008.07.001 https://doi.org/10.1016/j.sajb.2008.07.001 https://doi.org/10.1093/nar/8.19.4321 https://doi.org/10.1093/nar/8.19.4321 https://doi.org/10.1111/j.1471-8286.2005.00994.x https://doi.org/10.1111/j.1471-8286.2005.00994.x https://doi.org/10.1007/s12042-009-9037-2 https://doi.org/10.1093/bioinformatics/bts460 https://doi.org/10.1093/bioinformatics/bts460 https://doi.org/10.1093/oxfordjournals.molbev.a025856 https://doi.org/10.1093/oxfordjournals.molbev.a025856 https://doi.org/10.1111/j.1365-294x.1995.tb00214.x https://doi.org/10.1111/j.1365-294x.1995.tb00214.x https://doi.org/10.1002/ece3.1305 r core team (2018). r: a language and environment for statistical computing. vienna, austria: r foundation for statistical computing. available from: http:// www.r-project.org. ravishankar, k., dinesh, m., nischita, p., & sandya, b. (2015). development and characterization of microsatellite markers in mango (mangifera indica) using next-generation sequencing technology and their transferability across species. molecular breeding 35, 1–13. https://doi.org/10.1007/ s11032-015-0289-2 reed, j.z., tollit, d.j., thompson, p.m., & amos, w. (1997). molecular scatology: the use of molecular genetic analysis to assign species, sex and individual identity to seal faeces. molecular ecology 6, 225-234. https://doi.org/10.1046/j.1365294x.1997.00175.x roberts, d.g., forrest, c.n., denham, a.j., & ayre, d.j. (2013). microsatellite markers for vulnerable australian arid zone acacias. conservation genetics resources 5, 199-201. https://doi.org/10.1007/ s12686-012-9767-6 rousset, f, (2008). genepop’007: a complete reimplementation of the genepop software for windows and linux. molecular and ecology resources 8, 103-106. https://doi.org/10.1111/ j.1471-8286.2007.01931.x sahu, s.k., thangaraj, m., & kathiresan, k. (2012). dna extraction protocol for plants with high levels of secondary metabolites and polysaccharides without using liquid nitrogen and phenol. international scholarly research network molecular biology 2012: 205049. https://doi. org/10.5402/2012/205049 sablok, g., gahlot, p., gupta, a.k., pareek, k., & shekhawat, n.s. (2009). extraction of pcr-usable dna from trees adapted to arid environment. plant omics 2, 103-109. schwartz, m.k., & monfort, s.l. (2008). genetic and endocrine tools for carnivore surveys. in: r.a. long, p. mackay, w.j. zielinksi, & j.c. ray (eds.), non-invasive survey methods for carnivores. (pp. 238–262). washington dc, usa: island press. son l., wickneswari r., harwood c.e., larcombe m.j., griffin r.a., koutoulis a., harbard j.l., cyer k.s., yee l.w., thinh h.h. and vaillancourt r.e. (2016). a multiplexed set of microsatellite markers for discriminating acacia mangium, a. auriculiformis, and their hybrid. tree genetics & genomes 12, 1-10. https://doi.org/10.1007/s11295-016-0990-2 tan, s.c., & yiap, b.c. (2009). dna, rna, and protein extraction: the past and the present. journal of biomedicine and biotechnology 2009: 574398. https://doi.org/10.1155/2009/574398 thermo scientific (2013). assessment of nucleic acid purity. [t042 technical bulletin, nanodrop spectrophotometers]. tiwari, k.l., jadhav, s.k., & gupta, s. (2012). modified ctab technique for isolation of dna from some medicinal plants. research journal of medicinal plants 6, 5-73. https://doi.org/10.3923/ rjmp.2012.65.73 varma, a., padh, h., & shrivastava, n. (2007). plant genomic dna isolation: an art or a science. biotechnology journal: healthcare nutrition technology 2, 386-392. https://doi.org/10.1002/ biot.200600195 varshney, r.k., graner, a., & sorrells, m.e. (2005). genetic microsatellite markers in plants: their features and applications. trends in biotechnology 23, 4855. https://doi.org/10.1016/j.tibtech.2004.11.005 vicente, s., máguas, c., & trindade, h. (2018). genetic diversity and differentiation of invasive acacia longifolia in portugal. web ecology 18, 91-103. https://doi.org/10.5194/we-18-91-2018 vidya, t.n.c., & sukumar, r. (2005). amplification success and feasibility of using microsatellite loci amplified from dung to population genetic studies of the asian elephant (elephas maximus). current science 88, 489-492. vieira, m.l.c., santini, l., diniz, a.l., & munhoz, c.f. (2016). microsatellite markers: what they mean and why they are so useful. genetics and molecular biology 39, 312-328. https://doi.org/10.1590/16784685-gmb-2016-0027 waits, l.p., luikart, g., & taberlet, p. (2001). estimating the probability of identity among genotypes in natural populations: cautions and guidelines. molecular ecology 10, 249-256. https://doi. org/10.1046/j.1365-294x.2001.01185.x wang, y. (2008). microsatellite loci transferability in chestnut. journal of american society of horticultural science 133, 692-700. https://doi. org/10.21273/jashs.133.5.692 wattier, r., engel, c.r., saumitou-laprade, p., & valero, m. (1998). short allele dominance as a source of heterozygote deficiency at microsatellite loci: experimental evidence at the dinucleotide locus gv1ct in gracilaria gracilis (rhodophyta). molecular ecology 7, 1569–1573. https://doi. org/10.1046/j.1365-294x.1998.00477.x weising, k., nybom, h., pfenninger, m., wolff, k., & kahl, g. (2005). dna fingerprinting in plants: principles, methods, and applications: chapter 4 – methodology. (pp. 81-106). new york: crc press. https://doi.org/10.1201/9781420040043 yasodha, r., ghosh, m., sumathi, r., & gurumurthi, k. (2005). cross-species amplification of eucalyptus ssr markers in casuarinaceae. acta botanica croatica 64, 115-120. bairu et al. new zealand journal of forestry science (2020) 50:6 page 11 http://www.r-project.org http://www.r-project.org https://doi.org/10.1007/s11032-015-0289-2 https://doi.org/10.1007/s11032-015-0289-2 https://doi.org/10.1046/j.1365-294x.1997.00175.x https://doi.org/10.1046/j.1365-294x.1997.00175.x https://doi.org/10.1007/s12686-012-9767-6 https://doi.org/10.1007/s12686-012-9767-6 https://doi.org/10.1111/j.1471-8286.2007.01931.x https://doi.org/10.1111/j.1471-8286.2007.01931.x https://doi.org/10.5402/2012/205049 https://doi.org/10.5402/2012/205049 https://doi.org/10.1007/s11295-016-0990-2 https://doi.org/10.1155/2009/574398 https://doi.org/10.3923/rjmp.2012.65.73 https://doi.org/10.3923/rjmp.2012.65.73 https://doi.org/10.1002/biot.200600195 https://doi.org/10.1002/biot.200600195 https://doi.org/10.1016/j.tibtech.2004.11.005 https://doi.org/10.5194/we-18-91-2018 https://doi.org/10.1590/1678-4685-gmb-2016-0027 https://doi.org/10.1590/1678-4685-gmb-2016-0027 https://doi.org/10.1046/j.1365-294x.2001.01185.x https://doi.org/10.1046/j.1365-294x.2001.01185.x https://doi.org/10.21273/jashs.133.5.692 https://doi.org/10.21273/jashs.133.5.692 https://doi.org/10.1046/j.1365-294x.1998.00477.x https://doi.org/10.1046/j.1365-294x.1998.00477.x https://doi.org/10.1201/9781420040043 n ew z ea la nd jo ur na l o f f or es tr y sc ie nc e a dd it io na l f il e 1: l is t o f 4 9 cr os ssp ec ie s ss r m ar ke rs u se d in th is s tu dy d ev el op m en t o f a n ss r -b as ed d n a fi ng er pr in ti ng m et ho d fo r bl ac k w at tl e (a ca ci a m ea rn si i d e w il d) n o. n am e d ye p ri m er s eq ue nc e (5 '-3 ') se le ct ed (y /n ) r ep ea t m ot if s n uc le ot id e re pe at a nn ea li ng t em p. si ze r an ge bp n o. o f al le le s r ef er en ce n ew z ea la nd jo ur na l o f f or es tr y sc ie nc e n o. n am e d ye p ri m er s eq ue nc e (5 '-3 ') se le ct ed (y /n ) r ep ea t m ot if s n uc le ot id e re pe at a nn ea li ng t em p. si ze r an ge bp n o. o f al le le s r ef er en ce n ew z ea la nd jo ur na l o f f or es tr y sc ie nc e n o. n am e d ye p ri m er s eq ue nc e (5 '-3 ') se le ct ed (y /n ) r ep ea t m ot if s n uc le ot id e re pe at a nn ea li ng t em p. si ze r an ge bp n o. o f al le le s r ef er en ce n ew z ea la nd jo ur na l o f f or es tr y sc ie nc e n o. n am e d ye p ri m er s eq ue nc e (5 '-3 ') se le ct ed (y /n ) r ep ea t m ot if s n uc le ot id e re pe at a nn ea li ng t em p. si ze r an ge bp n o. o f al le le s r ef er en ce an updated survey on the use of geospatial technologies in new zealand’s plantation forestry sector sarah de gouw, justin morgenroth*, cong xu new zealand school of forestry, university of canterbury, christchurch 8140, new zealand *corresponding author: justin.morgenroth@canterbury.ac.nz (received for publication 17 june 2020; accepted in revised form 10 october 2020) abstract background: geospatial technologies have developed rapidly in recent decades and can provide detailed, accurate data to support forest management. knowledge of the uptake of geospatial technologies, as well as barriers to adoption, in new zealand’s plantation forest management sector is limited and would be beneficial to the industry. this study provides an update to the 2013 benchmark study by morgenroth and visser. methods: an online survey was sent to 29 companies that own or manage plantation forests in new zealand. the survey was split into seven sections, composed of multiple-choice and open-ended questions, on the topics of: demographic information, data portals and datasets, global navigation satellite system (gnss) receivers, and four remote-sensing technologies. these included aerial imagery, multispectral imagery, hyperspectral imagery, and light detection and ranging (lidar). each section included questions relating to the acquisition, application and products created from each remotesensing technology. questions were also included that related to the barriers preventing the uptake of technologies. to determine the progression in the uptake of these technologies the results were compared to morgenroth and visser's study conducted five years earlier. results: twenty-three companies responded to the survey and together, those companies managed approximately 1,172,000 ha (or 69% of new zealand’s 1.706 million ha plantation forest estate (nzfoa, 2018)). the size of the estates managed by individual companies ranged from 1,000 ha to 177,000 ha (quartile 1 = 19,000 ha, median = 33,000 ha, quartile 3 = 63,150 ha). all companies used gnss receivers and acquired three-band, red-green-blue, aerial imagery. multispectral imagery, hyperspectral imagery and lidar data were acquired by 48%, 9% and 70% of companies, respectively. common applications for the products derived from these technologies were forest mapping and description, harvest planning, and cutover mapping. the main barrier preventing companies from acquiring most remotely-sensed data was the lack of staff knowledge and training, though cost was the main barrier to lidar acquisition. the uptake of all remote-sensing technologies has increased since 2013. lidar had the largest progression in uptake, increasing from 17% to 70%. there has also been a change in the way companies acquired the data. many of the companies used unpiloted aerial vehicles (uav) to acquire aerial and multispectral imagery in 2018, while in 2013 no companies were using uavs. esri arcgis continues to be the dominant geographic information system used by new zealand’s forest management companies (91%), though 22% of companies now use free gis software, like qgis or grass. the use of specialised software (e.g. fusion, lastools) for lidar or photogrammetric point cloud analysis increased since 2013, but most forestry companies who are processing .las files into various products (e.g. digital terrain model) are using arcgis. conclusions: this study showed that there had been a progression in the uptake of geospatial technologies in the new zealand plantation forest management sector. however, there are still barriers preventing the full utilisation of these technologies. the results suggest that the industry could benefit from investing in more training relating to geospatial technologies. new zealand journal of forestry science de gouw et al. new zealand journal of forestry science (2020) 50:8 https://doi.org/10.33494/nzjfs502020x118x e-issn: 1179-5395 published on-line: 05/11/2020 © the author(s). 2020 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access keywords: geospatial technologies; gnss; gps; remote sensing; gis; forestry; education; uav. http://creativecommons.org/licenses/by/4.0/), de gouw et al. new zealand journal of forestry science (2020) 50:8 page 2 introduction geospatial technologies and techniques are used to acquire, manipulate, and analyse geographic data (wang 2017). those that are commonly applied to forest description and management include the global navigation satellite system (gnss), geographic information systems (gis) and remote sensing (wing & sessions 2007). gnss allows for accurate geographic locations to be ascertained and navigation to be undertaken. remote sensing refers to acquiring information about features or processes without direct measurement or contact; it is reliant upon sensors designed to receive electromagnetic radiation after it has reflected off of a feature of interest (wulder & franklin 2003). aerial imagery, satellite imagery, and light detection and ranging (lidar) are examples of remote sensing. gis are software designed to manage, analyse, and communicate geographic data. the development of geospatial technologies over the past 50 years has occurred rapidly, producing data that are cheaper and faster to acquire and use (dash et al. 2016). the use of geospatial technologies, products, and analyses has been applied to a diverse range of forestry applications including forest health monitoring (coops et al. 2003), mapping forest disturbances (savage et al. 2017), harvest and road planning (abdi et al. 2009; akay et al. 2009; holopainen et al. 2014; olivera et al. 2016), forest inventory and resource mapping (dassot et al. 2011; pont et al. 2015; xu et al. 2019), as well as carbon inventory (stephens et al. 2012). the range of applications to which geospatial technologies are applied potentially reflects the widespread availability of the data, software, and technologies. while many forestry companies acquire their own geospatial data at considerable cost (e.g. aerial imagery, lidar data; (morgenroth & visser 2013)), data can often be freely downloaded from publicly available data repositories (dash et al. 2016). for instance, land information new zealand (linz) provides free public access to all orthorectified aerial imagery and lidar data for new zealand. likewise, the entire landsat satellite image archive, spanning from 1972 until present day, is freely available online (phiri & morgenroth 2017). in addition to data being freely available, the hardware required to collect data is becoming more widely available due to low purchase costs. the costs of unpiloted aerial vehicles (uavs) and accompanying sensors have decreased (marris 2013), essentially democratising the acquisition of some forms of remotely sensed data. the advances in geospatial technologies, their widespread availability, and their use in various applications explain why geospatial skills and knowledge have become requirements for many entrylevel jobs within forest companies (sample et al. 2015). merry et al. (2016) found that 71% of recent graduates from forestry education programmes used gis at least every second day in their jobs in the united states, which was a 28% increase from 2007 (merry et al. 2007). a study in the united states for entry-level forestry jobs found that 70% of job advertisements required that the applicant had knowledge and skills relating to mapping technologies (bettinger & merry 2018). in another study, 68.7% of forestry employers expected early-career foresters to have geospatial skills, including remote sensing, gis, and gnss (sample et al. 2015). it is clear that geospatial knowledge and skills are considered critical by forestry employers. the number of forestry education departments requiring a gis component as a part of the degree has increased. in 1989, 5% of forestry departments in canada required that undergraduates completed a geospatial or gis component to obtain their degree (sader et al. 1989). in a follow-up survey conducted in 1999, this rose to 10% (sader & vermillion 2000). a different survey conducted in the united states in 2012, reported that 94% of undergraduate forestry degrees required that a geospatial course was taken to complete the degree (merry et al. 2016). a study conducted in new zealand surveyed companies across a variety of industries and found that 44% of those companies believed that there was a shortage of trained gis specialists across the nation (de róiste 2014). this could be a barrier affecting the uptake of geospatial technologies as companies may lack staff with the knowledge or skills to process, analyse or apply the information and products produced using technologies such as lidar or multispectral imaging. in addition, the cost of acquiring data and using the hardware and software required for processing and analysing data can be another barrier for companies (bernard & prisley 2005; morgenroth & visser 2013; white et al. 2016). clearly there are manifold barriers to uptake of geospatial technologies, so organisational commitment is crucial. morgenroth and visser (2013) completed a study looking at the uptake of geospatial technologies within new zealand’s forest management sector five years ago. since then, the literature shows rapid developments and advances of geospatial technologies in new zealand forest research (dash et al. 2019; pearse et al. 2018; pearse et al. 2017; watt et al. 2016; watt et al. 2019; xu et al. 2019; xu et al. 2017). whether geospatial technology usage by companies in new zealand’s plantation forestry sector mirrors these research advancements is unknown. as such, there is a need for an update on the 2013 survey results. the objective of this study is to quantify the uptake of geospatial technologies by forestry companies and describe how the acquired data are being applied. additionally, it will identify the barriers that are limiting the uptake of geospatial technologies in the new zealand forestry sector. finally, this study compares these data to previous results to reflect on how geospatial technology adoption has changed over the last five years in new zealand’s plantation forestry sector. methods data a web-based questionnaire survey was developed in google forms and distributed to prospective respondents (see additional file). prior to distribution, a draft survey was sent to two representative respondents from the new zealand plantation forest industry and their feedback was used to revise the final survey, which was sent to all prospective respondents. the intended recipient of the survey was the company’s geospatial manager. on 5 may 2018, the final survey was distributed to 29 new zealand forest management companies. of these 29 companies, 19 were identified using the list of forest management companies in the 2016/17 new zealand plantation forest industry facts and figures publication (nz foa 2018). an additional ten companies were added to the list of survey recipients based on suggestions from forest industry professionals with knowledge of forest management and ownership structures. this approach excluded individual small-scale forest (<1000 ha) owners/managers, but included companies that manage small-scale forests or woodlots on behalf of their owners. when combined, the 29 companies invited to participate manage approximately 80% of new zealand’s plantation forest estate area (1,706,000 ha) (nz foa 2018). on 6 june 2018, a personalised follow-up email was sent out to those companies who had not completed the survey. this increased the response rate from 20 to 23 companies. on 22 june 2018, responses were no longer accepted. the questions developed by morgenroth and visser (2013) were used as the basis for the questions in the present survey. questions were updated to reflect changes in the available geospatial technologies. standardising the current survey to the 2013 survey allowed for a comparison of results to determine how uptake and barriers had changed over the past five years. the survey comprised seven sections, and asked recipients to answer questions about their company (demographic data), their use of freely available spatial data, their use of positioning technologies, and their acquisition and use of aerial imagery, multispectral imagery, hyperspectral imagery, and lidar data. to minimise confusion, we provided respondents with definitions of the various technologies to which the survey referred. of particular note, we defined three grades of gnss as consumer, mapping, and survey grade. these are generic terms to describe gnss receivers, whereby we described consumer grade receivers as being capable of low positioning accuracy (<10 m) and costing less than $1000. mapping grade receivers yielded <5 m accuracy and cost $1,000 – $20,000, and survey grade receivers yielded <0.5 m accuracy and cost more than $20,000. in terms of remotely-sensed data, we defined ‘aerial imagery’ as typically consisting of three bands in the visible wavelengths (red, green, blue) and being acquired from an aerial platform (e.g. aeroplane, uav). we defined ‘multispectral imagery’ as typically consisting of four or more bands in the visible and invisible wavelengths (red, green, blue, infrared, etc.) and being commonly acquired from an aeroplane, uav or satellite platform. ‘hyperspectral imagery’ was defined as typically consisting of hundreds of contiguous bands spanning the visible and infrared wavelengths and being acquired from an aeroplane, uav or satellite platform. lidar (also referred to as laser scanning), was described as being acquired aerially or terrestrially and yielding a three-dimensional point cloud that could be used to produce digital terrain models, canopy height models and structural descriptions of forests via lidar metrics. the survey questions were written in a manner that was directed at the company as opposed to the individual respondent. this reinforced to the respondent that they were answering on behalf of the company. multiplechoice questions were often followed by open-ended questions to allow respondents to provide additional details about their answer(s) in the preceding question. respondents were also given the opportunity to add an answer that was not provided as one of the default choices in the multiple-choice question by having an ‘other’ choice. most of the questions within the survey were compulsory and required an answer before the respondent could continue to the next section of the survey. this ensured that no questions were left unanswered. analysis descriptive statistics were used to summarise the survey results. the answers to open-ended questions were compiled and categorised to make it easier to analyse the data and identify trends. to analyse the progression of the uptake of geospatial technologies, the responses from morgenroth and visser (2013) study were compared to the results from the present survey. results demographic information of the 29 companies contacted, 23 responded to the survey (79% response rate). the total forest area managed by those companies was approximately 1,172,000 ha (69% of new zealand’s 1.706 million ha plantation forest estate (nz foa 2018)). the size of the estates managed by individual companies ranged from 1,000 ha to 177,000 ha (quartile 1 = 19,000 ha, median = 33,000 ha, quartile 3 = 63,150 ha). fifty-two percent of the companies that responded to the survey identified themselves as forest owners and managers, 44 percent were forest management companies. while the intended recipient of the survey was each company’s geospatial manager this was not always possible. one management company did not have a geospatial manager and outsourced all mapping, surveying and terrain planning, so the photography and mapping services contractor completed the survey on their behalf. other smaller management companies did not have an employee appointed as a geospatial manager, so the most appropriate staff member responded to the survey. data acquisition, processing, and barriers to uptake the acquisition of freely available data was common and supported plantation forest management for all 23 companies who responded to the survey. orthorectified aerial imagery was the most commonly used product (83% of respondents), while satellite imagery was acquired by 65% of respondents. with respect to derived geographic datasets, the land cover database (lcdb, manaaki whenua – landcare research) was de gouw et al. new zealand journal of forestry science (2020) 50:8 page 3 acquired by 70% of companies, followed by datasets in the fundamental soil layers (fsl, manaaki whenua – landcare research) (61%). other derived datasets, including the land use carbon analysis system (lucas) layers (ministry for the environment) (39%), s-map (manaaki whenua – landcare research) (30%), and virtual climate station network data (niwa) (17%) were acquired by fewer than 40% of respondents. in terms of online spatial data repositories, the land information new zealand (linz) data service portal was used by 91% of companies, while koordinates (78%), the land resource information system (lris) portal (57%), and the ministry for the environment data service (52%) were also used by more than half of the respondents. other datasets and online data portals were used by forestry companies, but less commonly. positioning technology all of the forest management companies used global navigation satellite system technology. sixty-one percent used two or more grades of receivers. consumer grade handheld receivers (e.g. garmin 60csx) were the most commonly used (83%). consumer grade receivers built in to devices such as a mobile phone or tablet were also used by 65% of the respondents. survey and mapping grade receivers that can provide more accurate and precise positioning, each had the same level of uptake (22%). recording the location of infrastructure and utilities such as landings, roads, fire ponds and trials were the most common uses of gnss receivers reported by respondents. boundary mapping and mapping for legal purposes, plot location, hazard and historic site location, as well as cutover mark-ups were also applications for gnss data. less common applications included gnssreferenced photos for resource consent compliance and ground control points for uav mapping. aerial imagery aerial imagery was the most commonly acquired form of remotely sensed data, with all responding companies acquiring aerial imagery. unpiloted aerial vehicles and aeroplanes were the most commonly used platforms to acquire aerial imagery, with 83% of respondents indicating that one or both platforms were used. one company used a helicopter to acquire their aerial imagery. when asked if the company acquired their aerial imagery on a regular basis, thirteen companies (57%) acquired aerial imagery on an as-needed basis. this was facilitated by the use of uavs to acquire imagery when collecting data on an irregular basis for areas of interest such as stands during harvest planning, mapping cutover areas after harvest completion, or assessing the effects of a windstorm. four companies acquired aerial imagery on an annual basis for their entire estate in addition to irregularly collecting imagery for areas of interest. eight companies (35%) only acquired aerial images on a regular cycle. these regular acquisition cycles ranged from quarterly up to three years. the spatial resolution of the aerial imagery acquired via uav was frequently finer than that acquired via an aeroplane. thirty-five percent of the companies de gouw et al. new zealand journal of forestry science (2020) 50:8 page 4 acquired aerial imagery at two or more differing spatial resolutions. the reported resolution of aerial imagery ranged from 0.1 m to 5 m, with the latter suggesting that despite respondents being provided with a definition of aerial imagery, there may have been some confusion or there was a typo in the response. this is because it is not likely that 3-band rgb aerial imagery, acquired by uav or aeroplane, would have a spatial resolution as poor as 5 m. nevertheless, respondents stated that the acquired imagery were used to produce true-colour orthophotos (91%) and photogrammetric point clouds (32%). multispectral imagery multispectral imagery was acquired by 48% (n = 11) of the forestry companies, with another two companies stating they were planning on acquiring the imagery in the near future. the lack of staff knowledge or training was the most common barrier preventing companies from acquiring multispectral imagery (50%), followed by the cost of acquiring multispectral imagery (42%). finally, one third of companies did not perceive any benefit from acquiring multispectral imagery. multispectral imagery was most frequently obtained from satellite platforms (82%), with sentinel imagery being most commonly acquired (73%). rapideye and landsat imagery were also used by 36% and 27% of companies, respectively. uavs and/or aeroplanes were also used to acquire multispectral imagery by two of the eleven (18%) companies that acquired multispectral imagery. ten of the eleven companies acquired multispectral imagery only when it was required. the one company that did acquire multispectral imagery on a regular basis did so annually. the spatial resolutions of the multispectral imagery differed depending on the platform from which data were acquired. the spatial resolution of multispectral imagery acquired using a uav was 10 cm. in contrast, acquisition of the imagery via satellite resulted in spatial resolutions ranging from 3 m to 30 m. three companies acquired multispectral imagery that had a spatial resolution of 5 m or less, another three companies acquired multispectral imagery that had a spatial resolution of 10 m. there were also several companies that acquired their imagery at 15 m or 30 m resolutions. companies derived true colour composites (91%) and false colour composites (82%), as well as vegetation indices from the multispectral imagery. seventy-three percent (n=8) of the companies who use multispectral imagery derived the normalised difference vegetation index (ndvi), while one company also derived the enhanced vegetation index (evi). hyperspectral imagery only 9% (n=2) of the companies that responded to the survey acquired hyperspectral imagery. the main barriers for companies not using hyperspectral imagery were the lack of staff knowledge and training (57%) as well as the cost of acquiring the imagery (48%). some companies did not believe there was any benefit of acquiring hyperspectral imagery (29%) or were unaware of it or its potential benefits (15%). despite respondents being provided with a definition of hyperspectral imagery, the detail in the responses suggest there may have been some confusion. both companies acquiring hyperspectral imagery responded that they did so via satellite, though neither specified which imagery they acquired. however, the companies did report the spatial resolution of their hyperspectral imagery, with one company reporting 3 m to 5 m resolution and the other 10 m to 20 m resolution. given that we are not aware of any hyperspectral sensors on satellite platforms capable of acquiring data at those resolutions, we suggest that the companies misconstrued multispectral satellite imagery as hyperspectral imagery. lidar data lidar data were used by 70% (n=16) of the companies, with two additional companies planning on acquiring lidar data in the future. the main barrier for companies not using lidar was the cost of acquiring it (57%). lack of estate scale and the lack of staff knowledge or training was a barrier for 29% and 14% of the companies, respectively. the smaller companies managing 16,000 ha or less did not acquire lidar data. aeroplanes were the most common platforms for acquiring lidar data (94%), but uavs were used by 13% of the companies. the density of the lidar point clouds ranged from 2 to 20 points m-². ten companies (63%) acquired lidar data with a resulting point cloud density of 4 points m-² or less. lidar data were only acquired as required by 81% of the respondents, with two other companies acquiring their lidar data on a regular three or five-year cycle. one company collected lidar data with no intention of acquiring it again in the future. all companies that acquired lidar data used it to derive digital terrain models (dtm). canopy height models (chm) (69%) were also commonly derived, while volume estimates and stem counts were estimated from lidar data by seven companies (44%). there were products that companies were not acquiring or deriving but would want to obtain in the future. these products, including estimates of stocking, biomass, individual tree height and volume, grade mix, and phenotyping could provide more detailed forest description for managers. processing of the tiled lidar files into various products (e.g. dtm, chm, lidar metrics) was outsourced to an aerial surveying company for 63% (n=10) of the companies, while 56% (n=9) of companies also outsourced parts of lidar processing to a third-party organisation. some lidar products were derived inhouse by 31% (n=5) of the companies, but no company produced all their lidar products in-house. application of remotely sensed imagery the most common application of aerial imagery was for general forest overviews and mapping. lidar and aerial imagery were commonly used for harvest planning (table 1). aerial imagery and lidar had other mutual applications which included site preparation, silvicultural planning and road mapping. lidar had the widest variety of applications, followed by aerial and multispectral imagery. all the technologies, except lidar, were used for cutover mapping. multispectral imagery and hyperspectral imagery were applied to tasks such as forest health evaluation and species identification. hyperspectral imagery, unlike multispectral imagery, was not used for mapping. multispectral imagery and aerial imagery were used for natural event assessment, examples of which include windthrow mapping, assessing snowfall damage and fire damage. software arcgis (environmental systems research institute, redlands, california, usa) was the most commonly used software for working with data collected from all four remote-sensing technologies (table 2). free gis software such as qgis (qgis development team) or geographic resources analysis support system (grass) (grass development team) were also commonly used geographic information systems. agisoft photoscan (agisoft llc, russia) (now called agisoft metashape) was used by two companies for photogrammetric point cloud data processing. fusion (mcgaughey 2018), lastools (rapidlasso gmbh, germany), and quick terrain (qt) modeller (applied imagery, united states of america) were used when working with lidar data, though each was only used by two companies; in contrast, five companies used arcgis to process .las tiles into products (i.e. dtm, chm). the majority of companies used two or more different types of software when working with remotely sensed data. atlas geomaster, a spatial stand record system, was used by ten companies. changes to uptake between 2013 and 2018 the uptake of use of different technologies and data has changed over the last five years. there was a progression in the uptake of gnss receivers, with the proportions of companies using each grade of gnss receiver having changed. five years ago, none of the companies surveyed reported using consumer grade receivers built into devices (such as a mobile phone or tablet). the results from the most recent survey showed that 65% of companies were using this grade of receiver (table 3). in 2018, there were fewer companies using consumer and mapping grade receivers compared to five years ago. in contrast to these decreases, the proportion of companies using survey grade receivers nearly doubled, going from 12% in 2013 to 22% in 2018. the uptake of the remote-sensing technologies included in the survey increased over the past five years. hyperspectral imagery was not included in morgenroth and visser’s 2013 study and consequently could not be compared to the uptake in 2018. lidar showed the greatest progression over the last five years with its uptake increasing from 17% in 2013 to 70% of companies in 2018 (table 4). comparably, the progression in the uptake of aerial imagery (+12%) and multispectral imagery (+13%) were modest, though both had much higher rates of use in 2013. there was an increase in the uptake of most software stated in the most recent survey compared to five years ago (table 5). the largest increase was in the uptake de gouw et al. new zealand journal of forestry science (2020) 50:8 page 5 de gouw et al. new zealand journal of forestry science (2020) 50:8 page 6 table 1: application of remote sensing imagery to forest management. n = number of companies adopting remote sensing for a particular application. % = percentage of all companies adopting a particular remote sensing type that applied it to a particular application. aerial imagery multispectral imagery hyperspectral imagery lidar data application n % n % n % n % general forest overview and mapping 15 68 5 45 harvest planning 13 59 12 75 cutover mapping 13 59 3 27 1 50 site preparation 5 23 4 25 silvicultural planning 3 14 1 6 road mapping 6 27 natural event assessment 2 14 3 27 historic site identification 3 19 hazard identification 2 9 2 6 species identification 5 45 1 50 forest health assessment 5 45 1 50 where aerial imagery is not available 2 18 wilding identification 1 9 inventory 5 31 slope management 3 19 forest valuation 3 19 3d models 1 6 aerial imagery multispectral imagery hyperspectral imagery lidar data software class software n % n % n % n % geographic information system esri arcgis 21 91 11 100 2 100 5 56 free gis 4 18 1 9 1 50 global mapper 2 9 image analysis erdas imagine image analysis software 1 5 1 9 trimble ecognition image analysis software 1 5 1 9 lidar or photogrammetric point cloud analysis and processing fusion 2 22 lastools 2 22 qt modeller 2 22 agisoft photoscan 2 9 table 2: software used when working with acquired imagery. n = number of companies using the software. of free gis software, from 6% in 2013 to 22% in 2018. arcgis saw the next largest increase in use, up 9% on 2013 use, while mapinfo use dropped from 18% in 2013 to 0% in 2018. in terms of image analysis software, erdas and trimble e-cognition software both showed small increases of 1% and 4%, respectively. only two companies (9% of respondents) reported using point cloud analysis and processing software in 2018, though none reported using it in 2013. discussion the results from the survey identify the geospatial technologies used within the new zealand plantation forest management sector, how they are used, and the barriers to their use. all the respondent companies used gnss receivers, however, there has been a change in the most commonly used grade of receiver compared to five years ago. the use de gouw et al. new zealand journal of forestry science (2020) 50:8 page 7 of dedicated consumer grade gnss receivers decreased from 100% in 2013 to 83% in 2018, which is possibly a consequence of companies using positional data from devices (e.g. tablets, mobile phones) with built-in gps receivers. these latter devices were not reported as being used by any companies in 2013, but were used by 65% of respondents in 2018. the increase in the uptake of gnss receivers within devices, such as tablets and smart phones, is likely aided by the improvement in technology and the versatility of these devices. the decrease in the uptake of mapping grade receivers, from 41% (2013) to 22% (2018) may be a consequence of the improved accuracy and precision of consumer grade receivers (tomaštík et al. 2016). companies may not be willing to pay for mapping grade receivers when consumer grade receivers can achieve similar accuracies for applications such as locating fire ponds, culverts and skid sites. in contrast, the increase in use of survey grade receivers may result from the need to ensure groundconsumer – handheld consumer – in device mapping survey percentage of respondents using the technology 2013 100 41 12 2018 83 65 22 22 table 3: progression of uptake of gnss receivers by grade. table 4: progression of uptake of remote sensing technologies. companies using software (%) software class software 2013 2018 change geographic information system esri arcgis 82 91 +9 mapinfo 18 0 -18 global mapper 0 9 n/a free gis 6 22 +16 image analysis erdas imagine image analysis software 12 13 +1 trimble ecognition image analysis software 0 4 +4 lidar or photogrammetric point cloud analysis and processing fusion 0 9 +9 lastools 0 9 +9 qt modeller 0 9 +9 agisoft photoscan 0 9 +9 specialist forestry software atlas geomaster 35 43 +8 table 5: progression of uptake of software used when processing and using products from the geospatial technologies included in the survey. aerial imagery multispectral imagery hyperspectral imagery lidar data percentage of respondents using remotely-sensed imagery 2013 88 35 17 2018 100 48 9 70 based data can be co-located with high-resolution remotely sensed data, in particular lidar data. this is because of the common practice of correlating groundbased inventory measurements with lidar plots, such that lidar metrics can subsequently be used for forest description across entire estates. moreover, the shift to survey grade receivers may also indicate a desire to minimise positional error associated with multi-pathing beneath forest canopy. the number of companies acquiring aerial imagery has increased by 14% since 2013, with all forest management companies stating that they acquired aerial imagery. the applications for aerial imagery have remained similar over the past five years but how the aerial imagery is acquired has changed. while in the 2013 survey, the cost of acquisition was the main barrier to uptake, that no longer appears to be an issue; perhaps this is because many companies (83% of them) are now using uavs to acquire aerial imagery. the use of uavs to capture aerial imagery was not reported by forest management companies in morgenroth and visser’s 2013 study, but presumably, the development of cheaper, smaller sensors and uavs has improved access to this technology. respondents reported that uavs allowed them to collect imagery when required for a target area and also when cloud cover would hinder the acquisition of imagery from a satellite or aeroplane. whether uavs completely replace aeroplanes to acquire aerial images in the foreseeable future will depend on whether the shortcomings of uavs (e.g. battery life, payload limitations, the regulatory framework, easy to use software; coops et al. (2003); heaphy et al. (2017)) can be solved. the uptake of multispectral imagery increased from 35% in 2013 to 48% in 2018. the most common barrier preventing companies from using multispectral imagery was the lack of staff education; this differs to the cost of the imagery being the most common barrier five years ago as multispectral imagery becomes cheaper and even free. the availability of free satellite imagery with spatial resolution ≤ 30 m (e.g. landsat 8, sentinel-2, planetscope), may have resulted in forestry companies experimenting with the utility of this data. the lack of best practice guidelines and the developing technical capacity of the industry needs to improve to fully utilise technologies such as multispectral imagery. this lack of knowledge is not specific to the forestry sector and can be seen across a range of industries in new zealand (de róiste 2014). more education and training for geospatial professionals will be required to process and analyse remotely sensed data such that it can be better utilised in the future. tertiary education and other training providers have a role to play here. while the survey was not designed to question educational or training preferences, previous research on gnss training found a strong preference for formal hands-on training, as opposed to online or in-person lectures (bettinger et al. 2019). in the future, the uptake of multispectral imagery may increase as more companies become aware of the technology and its benefits. as seen from the survey results there are already two other companies who are table 2: confusion matrix working towards acquiring multispectral imagery in the future. as with the increase in aerial imagery acquisition between 2013 and 2018, multispectral imagery acquisition may increase in future as suitable sensors become available for uavs at sufficiently low costs. all but one of the companies that acquired multispectral imagery also acquired lidar data. the combined use of multispectral imagery and lidar is seen in several published studies and was reviewed by xu et al. (2015). for example, watt et al. (2015) used a combination of satellite imagery and lidar data to estimate site index, other studies have used a combination of two technologies to determine biomass (estornell et al. 2012), volume (tonolli et al. 2011), stand age (xu et al. 2018) and to classify forest cover (dupuy et al. 2013). though no survey question specifically asked about data fusion, seventy-four percent of the companies acquired data for at least two of the remote-sensing technologies included in the survey. this suggests that it is possible for companies to combine the data from two technologies to optimise the information that can be extracted from analyses. though two companies did report acquiring hyperspectral imagery, we suspect their responses were incorrect; based on the reported spatial resolutions. it is likely that respondents had actually acquired multispectral satellite imagery. nevertheless, the low uptake of hyperspectral imagery was not unexpected. while its benefits can be considerable, the imagery and its acquisition suffer from a number of drawbacks (adão et al. 2017). the imagery contains hundreds of bands, spreading across the electromagnetic spectrum, and processing can be complex. moreover, the limited conditions under which imagery can be acquired and the cost of acquisition can have an influence on the uptake of hyperspectral imagery. anecdotally, there are few hyperspectral data providers in nz, thus the cost remains high in the absence of competition. finally, there can be a trade-off between spectral resolution and spatial resolution. it may be that companies value spatial resolution more than spectral resolution for many forestry applications. many of the survey respondents listed the factors above as reasons for not acquiring hyperspectral imagery. the uptake of lidar data and its products has seen the most significant increase of all the remote-sensing technologies included in the survey, increasing from 17% to 70% of companies since 2013 (morgenroth & visser 2013). since the 2013 survey, methods to operationalise aerial lidar data have been developed (dash et al. 2015), so the increased use shown in the present study is not surprising. the uptake of lidar in the new zealand forest sector is similar to the uptake in the us in which a recent study found that 42% of recent forestry programme graduates used lidar in their job (merry et al. 2016). this was a significant increase from only 10% of recent us forest graduates using lidar in a previous study conducted in 2007 (merry et al. 2007; merry et al. 2016). the main barrier preventing uptake of lidar in both 2013 and 2018 surveys was the cost. smaller companies de gouw et al. new zealand journal of forestry science (2020) 50:8 page 8 are not acquiring lidar data due to cost and estate scale. economies of scale apply as the cost per hectare of acquiring lidar typically decreases as the forest area increases. the connectivity of these forests will also affect the cost of acquiring lidar (adams et al. 2011). lidar is more expensive to acquire than aerial imagery or multispectral imagery (kelly & di tommaso 2015). however, the results from the survey indicate that forest management companies were willing to acquire lidar, despite the cost, due to perceived benefits. it should also be noted that the cost of acquiring lidar, when acquired and processed efficiently, is more cost-effective in comparison with intensive fieldwork (hummel et al. 2011). given the cost barrier for lidar data acquisition and the growing acquisition of aerial imagery via uav, perhaps future requirements for three-dimensional forest description will look towards photogrammetric point clouds, rather than lidar-derived point clouds. photogrammetric point clouds have been shown to be useful for a number of forest inventory purposes (iglhaut et al. 2019; pearse et al. 2018; white et al. 2013) and were already produced by 32% of surveyed companies. terrestrial lidar was not acquired by any of the forest management companies. terrestrial lidar is not suitable to collect data for large areas, but it can provide detailed tree information at a plot scale. terrestrial lidar is suited to measuring the below canopy structure, such as stem form, branching and stand density (dassot et al. 2011; white et al. 2016). the development and improvement of mobile handheld laser scanners, which are more portable than previous tripod-based scanners, may see a future increase in the uptake of terrestrial lidar. however, the limits imposed by steep terrain forests and the inaccuracy of these handheld scanners need to be improved first (dash et al. 2016). the products that companies wish to obtain from lidar in the future can be produced from data acquired via terrestrial lidar and may result in an increase in the uptake of terrestrial lidar. the survey response rate was imperfect, with 23 of 29 companies responding. nevertheless, the results presented are from companies with net stocked areas comprising 69% of new zealand’s 1.706 million ha plantation forest estate (nz foa 2018). we believe the results of this survey to be generalisable to mediumand large-scale plantation forest managers or owners in new zealand, though acknowledge the potential bias against small-scale forest owners or managers. moreover, we did not control for non-response bias, so it is possible that presented results are non-representative of new zealand’s entire plantation forestry sector. finally, it’s worth noting the importance of respondents having a clear understanding of the terms used in survey questions. a small number of responses to questions about aerial imagery and hyperspectral imagery suggested respondent confusion or misunderstanding. while we provided respondents with clear definitions of those terms, it may not have been sufficient to prevent incorrect responses to questions about spatial resolution. conclusions the results from this study have shown that all forestry companies that were surveyed were making use of online data portals and acquiring freely available datasets (e.g. aerial photography, soil and climate data). gnss and aerial imagery were the most commonly used geospatial technologies in new zealand’s forestry sector. companies were also making use of multispectral imagery and lidar data. the most common barriers preventing the uptake of geospatial technologies were the lack of staff education and the cost of acquiring the data. these barriers are comparable to barriers identified during morgenroth and visser’s 2013 survey and suggest that in order to get the most out of available technologies, forest industry may need to invest in more training. despite these barriers, over the last five years, there has been a progression in the uptake of all the technologies included in the survey, with lidar having the largest increase in uptake (from 17% to 70%). the results suggest that the application of geospatial technologies and remotely sensed data to plantation forest management is a rapidly growing field in new zealand. this supports similar rapid growth in these fields in other parts of the world (white et al. 2016). the collection and application of accurate and detailed data acquired from these technologies will support better forest management decisions. the improvement of forest management operations and decisions should lead to greater commercial gains (melville et al. 2015). the results of this survey will be informative for forest managers who have an interest in remaining on the cutting edge, educators who want to ensure teaching material is relevant, and the wider geospatial industry who are likely to be interested in the barriers, actual or perceived, to adoption of the technologies reported on in this study. list of abbreviations chm canopy height model dtm digital terrain model evi enhanced vegetation index gis geographic information system grass geographic resources analysis support system gnss global navigation satellite system lris land resource information system lucas land use carbon analysis system mth mean top height ndvi normalised difference vegetation index qgis quantum gis qt modeller quick terrain modeller uav unpiloted aerial vehicle de gouw et al. new zealand journal of forestry science (2020) 50:8 page 9 competing interests the authors declare that they have no competing interests. authors’ contributions jm conceived the study. sdg, jm, and cx contributed to the design of the study and distribution of the survey to the wider industry. sdg collated survey responses, summarised data, and wrote the manuscript. jm and cx advised on data interpretation and revised the manuscript. all authors read and approved the final manuscript. additional files additional file 1: survey sent to respondents. references abdi, e., majnounian, b., darvishsefat, a., mashayekhi, z., & sessions, j. (2009). a gis-mce based model for forest road planning. journal of forest science, 55(4), 171-176. https://doi.org/10.17221/52/2008-jfs adams, t., brack, c., farrier, t., pont, d., & brownlie, r. (2011). so you want to use lidar? a guide on how to use lidar in forestry. new zealand journal of forestry, 55(4), 19 23. adão, t., hruška, j., pádua, l., bessa, j., peres, e., morais, r., & sousa, j.j. (2017). hyperspectral imaging: a review on uav-based sensors, data processing and applications for agriculture and forestry. remote sensing, 9(11), 1110. https://doi.org/10.3390/ rs9111110 akay, a.e., oğuz, h., karas, i.r., & aruga, k. (2009). using lidar technology in forestry activities. environmental monitoring and assessment, 151(14), 117-125. https://doi.org/10.1007/s10661008-0254-1 bernard, a.m., & prisley, s.p. (2005). digital mapping alternatives: gis for the busy forester. journal of forestry, 103(4), 163-168. https://doi. org/10.1093/jof/103.4.163 bettinger, p., & merry, k. (2018). follow-up study of the importance of mapping technology knowledge and skills for entry-level forestry job positions, as deduced from recent job advertisements. mathematical and computational forestry & natural resource sciences, 10(1), 15. bettinger, p., merry, k., bayat, m., & tomaštík, j. (2019). gnss use in forestry-a multi-national survey from iran, slovakia and southern usa. computers and electronics in agriculture, 158, 369-383. https:// doi.org/10.1016/j.compag.2019.02.015 coops, n., stanford, m., old, k., dudzinski, m., culvenor, d., & stone, c. (2003). assessment of dothistroma needle blight of pinus radiata using airborne hyperspectral imagery. phytopathology, 93(12), 1524-1532. https://doi.org/10.1094/ phyto.2003.93.12.1524 dash, j.p., marshall, h.m., & rawley, b. (2015). methods for estimating multivariate stand yields and errors using k-nn and aerial laser scanning. forestry: an international journal of forest research, 88(2), 237247. https://doi.org/10.1093/forestry/cpu054 dash, j.p., pont, d., brownlie, r., dunningham, a., watt, m., & pearse, g. (2016). remote sensing for precision forestry. new zealand journal of forestry, 60(4), 1524. dash, j.p., watt, m.s., paul, t.s.h., morgenroth, j., & pearse, g.d. (2019). early detection of invasive exotic trees using uav and manned aircraft multispectral and lidar data. remote sensing, 11(15). https://doi. org/10.3390/rs11151812 dassot, m., constant, t., & fournier, m. (2011). the use of terrestrial lidar technology in forest science: application fields, benefits and challenges. annals of forest science, 68(5), 959-974. https://doi. org/10.1007/s13595-011-0102-2 de róiste, m. (2014). filling the gap: the geospatial skills shortage in new zealand. new zealand geographer, 70(3), 179-189. https://doi.org/10.1111/ nzg.12054 dupuy, s., lainé, g., tassin, j., & sarrailh, j. (2013). characterization of the horizontal structure of the tropical forest canopy using objectbased lidar and multispectral image analysis. international journal of applied earth observation and geoinformation, 25, 76-86. https://doi. org/10.1016/j.jag.2013.04.001 estornell, j., ruiz, l.a., velázquez-martí, b., & hermosilla, t. (2012). estimation of biomass and volume of shrub vegetation using lidar and spectral data in a mediterranean environment. biomass and bioenergy, 46, 710-721. https://doi.org/10.1016/j. biombioe.2012.06.023 heaphy, m., watt, m.s., dash, j.p., & pearse, g.d. (2017). uavs for data collection-plugging the gap. new zealand journal of forestry, 62(1), 23-30. holopainen, m., vastaranta, m., & hyyppä, j. (2014). outlook for the next generation’s precision forestry in finland. forests, 5(7), 1682-1694. https://doi. org/10.3390/f5071682 hummel, s., hudak, a.t., uebler, e.h., falkowski, m.j., & megown, k.a. (2011). a comparison of accuracy and cost of lidar versus stand exam data for landscape management on the malheur national forest. journal of forestry, 109(5), 267-273. iglhaut, j., cabo, c., puliti, s., piermattei, l., o’connor, j., & rosette, j. (2019). structure from motion photogrammetry in forestry: a review. current forestry reports, 5(3), 155-168. https://doi. org/10.1007/s40725-019-00094-3 kelly, m., & di tommaso, s. (2015). mapping forests with lidar provides flexible, accurate data with many uses. california agriculture, 69(1), 14-20. https:// doi.org/10.3733/ca.v069n01p14 marris, e. (2013). fly, and bring me data. nature, 498(7453), 156-158. https://doi. org/10.1038/498156a mcgaughey, r.j. (2018). fusion/ldv: software for lidar data analysis and visualization. fusion version 3.80. united states department of agriculture, de gouw et al. new zealand journal of forestry science (2020) 50:8 page 10 https://doi.org/10.17221/52/2008-jfs https://doi.org/10.3390/rs9111110 https://doi.org/10.3390/rs9111110 https://doi.org/10.1007/s10661-008-0254-1 https://doi.org/10.1007/s10661-008-0254-1 https://doi.org/10.1093/jof/103.4.163 https://doi.org/10.1093/jof/103.4.163 https://doi.org/10.1016/j.compag.2019.02.015 https://doi.org/10.1016/j.compag.2019.02.015 https://doi.org/10.1094/phyto.2003.93.12.1524 https://doi.org/10.1094/phyto.2003.93.12.1524 https://doi.org/10.1093/forestry/cpu054 https://doi.org/10.3390/rs11151812 https://doi.org/10.3390/rs11151812 https://doi.org/10.1007/s13595-011-0102-2 https://doi.org/10.1007/s13595-011-0102-2 https://doi.org/10.1111/nzg.12054 https://doi.org/10.1111/nzg.12054 https://doi.org/10.1016/j.jag.2013.04.001 https://doi.org/10.1016/j.jag.2013.04.001 https://doi.org/10.1016/j.biombioe.2012.06.023 https://doi.org/10.1016/j.biombioe.2012.06.023 https://doi.org/10.3390/f5071682 https://doi.org/10.3390/f5071682 https://doi.org/10.1007/s40725-019-00094-3 https://doi.org/10.1007/s40725-019-00094-3 https://doi.org/10.3733/ca.v069n01p14 https://doi.org/10.3733/ca.v069n01p14 https://doi.org/10.1038/498156a https://doi.org/10.1038/498156a forest service. pacific northwest research station. melville, g., stone, c., & turner, r. (2015). application of lidar data to maximise the efficiency of inventory plots in softwood plantations. new zealand journal of forestry science, 45: 9. https://doi.org/10.1186/ s40490-015-0038-7 merry, k.l., bettinger, p., clutter, m., hepinstall, j., & nibbelink, n.p. (2007). an assessment of geographic information system skills used by field-level natural resource managers. journal of forestry, 105(7), 364-370. merry, k.l., bettinger, p., grebner, d.l., boston, k., & siry, j. (2016). assessment of geographic information system (gis) skills employed by graduates from three forestry programs in the united states. forests, 7(12), 304. https://doi.org/10.3390/ f 7120304 morgenroth, j., & visser, r. (2013). uptake and barriers to the use of geospatial technologies in forest management. new zealand journal of forestry science, 43(1), 16. https://doi.org/10.1186/11795395-43-16 nz foa. (2018). new zealand plantation forest industry facts and figures. https://nzfoa.org.nz/resources/ publications/facts-and-figures olivera, a., visser, r., acuna, m., & morgenroth, j. (2016). automatic gnss-enabled harvester data collection as a tool to evaluate factors affecting harvester productivity in a eucalyptus spp. harvesting operation in uruguay. international journal of forest engineering, 27(1), 15-28. https://doi.org/ 10.1080/14942119.2015.1099775 pearse, g.d., dash, j.p., persson, h.j., & watt, m.s. (2018). comparison of high-density lidar and satellite photogrammetry for forest inventory. isprs journal of photogrammetry and remote sensing, 142, 257-267. https://doi.org/10.1016/j. isprsjprs.2018.06.006 pearse, g.d., morgenroth, j., watt, m.s., & dash, j.p. (2017). optimising prediction of forest leaf area index from discrete airborne lidar. remote sensing of environment, 200, 220-239. https://doi. org/10.1016/j.rse.2017.08.002 phiri, d., & morgenroth, j. (2017). developments in landsat land cover classification methods: a review. remote sensing, 9(9), 967. https://doi. org/10.3390/rs9090967 pont, d., kimberley, m., brownlie, r., morgenroth, j., & watt, m.s. (2015). tree counts from airborne lidar. new zealand journal of forestry, 60(1), 39. sader, s.a., hoffer, r.m., & johnson, e.w. (1989). the status of remote-sensing education. journal of forestry, 87(10), 25-30. sader, s.a., & vermillion, s. (2000). remote sensing education: an updated survey. journal of forestry, 98(4), 31-37. sample, v.a., bixler, r.p., mcdonough, m.h., bullard, s.h., & snieckus, m.m. (2015). the promise and performance of forestry education in the united states: results of a survey of forestry employers, graduates, and educators. journal of forestry, 113(6), 528-537. https://doi.org/10.5849/jof.14122 savage, s.l., lawrence, r.l., & squires, j.r. (2017). mapping post-disturbance forest landscape composition with landsat satellite imagery. forest ecology and management, 399, 9-23. https://doi. org/10.1016/j.foreco.2017.05.017 stephens, p.r., kimberley, m.o., beets, p.n., paul, t.s.h., searles, n., bell, a., brack, c., & broadley, j. (2012). airborne scanning lidar in a double sampling forest carbon inventory. remote sensing of environment, 117, 348-357. https://doi. org/10.1016/j.rse.2011.10.009 tomaštík, j., saloň, š., & piroh, r. (2016). horizontal accuracy and applicability of smartphone gnss positioning in forests. forestry: an international journal of forest research, 90(2), 187-198. https:// doi.org/10.1093/forestry/cpw031 tonolli, s., dalponte, m., neteler, m., rodeghiero, m., vescovo, l., & gianelle, d. (2011). fusion of airborne lidar and satellite multispectral data for the estimation of timber volume in the southern alps. remote sensing of environment, 115(10), 24862498. https://doi.org/10.1016/j.rse.2011.05.009 wang, j. (2017). introduction to geospatial technology applications in forest management. in: j. wang (ed.), introduction to computing applications in forestry and natural resource management. boca raton, fl, usa: crc press. watt, m.s., dash, j.p., bhandari, s., & watt, p. (2015). comparing parametric and non-parametric methods of predicting site index for radiata pine using combinations of data derived from environmental surfaces, satellite imagery and airborne laser scanning. forest ecology and management, 357, 1-9. https://doi.org/10.1016/j. foreco.2015.08.001 watt, m.s., dash, j.p., watt, p., & bhandari, s. (2016). multisensor modelling of a forest productivity index for radiata pine plantations. new zealand journal of forestry science, 46: 9 https://doi.org/10.1186/ s40490-016-0065-z watt, m.s., pearse, g.d., dash, j.p., melia, n., & leonardo, e.m.c. (2019). application of remote sensing technologies to identify impacts of nutritional deficiencies on forests. isprs journal of photogrammetry and remote sensing, 149, 226-241. https://doi.org/10.1016/j.isprsjprs.2019.01.009 white, j.c., coops, n.c., wulder, m.a., vastaranta, m., hilker, t., & tompalski, p. (2016). remote sensing technologies for enhancing forest inventories: a review. canadian journal of remote sensing, 42(5), 619-641. https://doi.org/10.1080/07038992.201 6.1207484 white, j.c., wulder, m.a., vastaranta, m., coops, n.c., pitt, d., & woods, m. (2013). the utility of image-based point clouds for forest inventory: a comparison with airborne laser scanning. forests, 4, 518-536. https://doi.org/10.3390/f4030518 wing, m.g., & sessions, j. (2007). geospatial technology education. journal of forestry, 105(4), 173-178. de gouw et al. new zealand journal of forestry science (2020) 50:8 page 11 https://doi.org/10.1186/s40490-015-0038-7 https://doi.org/10.1186/s40490-015-0038-7 https://doi.org/10.3390/f7120304 https://doi.org/10.3390/f7120304 https://doi.org/10.1186/1179-5395-43-16 https://doi.org/10.1186/1179-5395-43-16 https://doi.org/10.1080/14942119.2015.1099775 https://doi.org/10.1080/14942119.2015.1099775 https://doi.org/10.1016/j.isprsjprs.2018.06.006 https://doi.org/10.1016/j.isprsjprs.2018.06.006 https://doi.org/10.1016/j.rse.2017.08.002 https://doi.org/10.1016/j.rse.2017.08.002 https://doi.org/10.3390/rs9090967 https://doi.org/10.3390/rs9090967 https://doi.org/10.5849/jof.14-122 https://doi.org/10.5849/jof.14-122 https://doi.org/10.1016/j.foreco.2017.05.017 https://doi.org/10.1016/j.foreco.2017.05.017 https://doi.org/10.1016/j.rse.2011.10.009 https://doi.org/10.1016/j.rse.2011.10.009 https://doi.org/10.1093/forestry/cpw031 https://doi.org/10.1093/forestry/cpw031 https://doi.org/10.1016/j.rse.2011.05.009 https://doi.org/10.1016/j.foreco.2015.08.001 https://doi.org/10.1016/j.foreco.2015.08.001 https://doi.org/10.1186/s40490-016-0065-z https://doi.org/10.1186/s40490-016-0065-z https://doi.org/10.1016/j.isprsjprs.2019.01.009 https://doi.org/10.1080/07038992.2016.1207484 https://doi.org/10.1080/07038992.2016.1207484 https://doi.org/10.3390/f4030518 wulder, m.a., & franklin, s.e. (2003). remote sensing of forest environments, introduction. in m.a. wulder & s.e. franklin (eds.), remote sensing of forest environments: concepts and case studies. boston, ma, usa: springer. https://doi.org/10.1007/9781-4615-0306-4_1 xu, c., manley, b., & morgenroth, j. (2018). evaluation of modelling approaches in predicting forest volume and stand age for small-scale plantation forests in new zealand with rapideye and lidar. international journal of applied earth observation and geoinformation, 73, 386-396. https://doi. org/10.1016/j.jag.2018.06.021 xu, c., manley, b., & morgenroth, j. (2019). describing area and yield for small-scale plantation forests in wairarapa region of new zealand using rapideye and lidar. new zealand journal of forestry science, 49: 10. https://doi.org/10.33494/ nzjfs492019x16x xu, c., morgenroth, j., & manley, b. (2015). integrating data from discrete return airborne lidar and optical sensors to enhance the accuracy of forest description: a review. current forestry reports, 1(3), 206-219. https://doi.org/10.1007/s40725015-0019-3 xu, c., morgenroth, j., & manley, b. (2017). mapping net stocked plantation area for small-scale forests in new zealand using integrated rapideye and lidar sensors. forests, 8(12): 487. https://doi. org/10.3390/f8120487 de gouw et al. new zealand journal of forestry science (2020) 50:8 page 12 https://doi.org/10.1007/978-1-4615-0306-4_1 https://doi.org/10.1007/978-1-4615-0306-4_1 https://doi.org/10.1016/j.jag.2018.06.021 https://doi.org/10.1016/j.jag.2018.06.021 https://doi.org/10.33494/nzjfs492019x16x https://doi.org/10.33494/nzjfs492019x16x https://doi.org/10.1007/s40725-015-0019-3 https://doi.org/10.1007/s40725-015-0019-3 https://doi.org/10.3390/f8120487 https://doi.org/10.3390/f8120487 new zealand journal of forestry science an updated survey on the use of geospatial technologies in new zealand’s plantation forestry sector additional file 1: survey sent to respondents. survey title: uptake of geospatial technologies in the new zealand forest industry company profile 1. what is your name? 2. what is your position title? 3. what is the name of your company? 4. type of company?  forest owner and manager  forest manager  forest consultant  other: 5. what is the net stocked area (hectares) of forests that your company manages? data acquisition 6. which of the following geographic data portals does your company use?  stats nz data service  koordinates  ministry for the environment (mfe) data service  land resource information systems (lris) portal  land information new zealand (linz) data service  none  other: 7. which of the following datasets does your company use?  fundamental soils layer from landcare research  landcover database from landcare research  s-map from landcare research  aerial photography from land information new zealand (linz)  satellite imagery from land information new zealand (linz)  virtual climate station network from niwa  lucas land use map from ministry for the environment  none  other:  new zealand journal of forestry science positioning technology 8. what grade of global positioning system does your company use?  consumer grade receiver built into device (e.g. iphone)capable of <10 m accuracy  consumer grade receiver (e.g. garmin 60 csx)capable of <10 m accuracy, cost <$1,000  mapping grade receiver (e.g. trimble nomad)capable of <5 m accuracy, cost $1,000$20,000  survey grade receiver (e.g. trimble geoexplorer 6000)capable of <0.5 m accuracy, cost >$20,000  none 9. how does your company use its gps receiver(s)? e.g. boundary mapping, plot centre location. aerial photography 10. does your company use aerial photography? aerial photography typically consists of three bands (red, green, blue) and is acquired from an aerial platform (e.g. plane, uav)  yes – go to question 12  no – go to question 11 11. what are the reasons for not using aerial photography?  cost  no perceived benefit  current staff lack knowledge or training to use aerial photography  was not aware of aerial photography  other: 12. how is your aerial photography acquired?  unmanned aerial vehicle (aka drone)  airplane  helicopter  other: 13. what products does your company derive from aerial photography?  true colour composites (this imagery includes only red, green and blue bands (rgb))  photogrammetric point clouds  none  other: 14. for what applications do you use your aerial photography? e.g. harvest planning. 15. does your company acquire aerial photographs on a regular cycle? e.g. every two years or only as required. 16. what software do you use when working with your aerial photography?  esri arcgis  mapinfo  atlas geomaster  open/free gis (e.g. qgis, grass)  envi image analysis software  trimble e-cognition image analysis software  erdas imagine image analysis software  other: 17. what is the spatial resolution of your aerial photography? e.g. 2 metres. new zealand journal of forestry science multispectral imagery 18. does your company use multispectral imagery? multispectral imagery typically consists of four or more bands (red, green, blue, infrared, etc) and is acquired from an airplane, uav or satellite.  yes – go to question 20  no – go to question 19 19. what are the reasons for not using multispectral imagery?  cost  no perceived benefit  current staff lack knowledge or training to use multispectral imagery  was not aware of multispectral imagery  other: 20. how is your multispectral imagery acquired?  airplane  satellite  unmanned aerial vehicle (aka drone)  helicopter  other: 21. if you acquire satellite imagery which sensor do you use?  landsat  sentinel  rapid eye  spot  ikonos  geoeye  pleiades  worldview  other: 22. what products does your company derive from the multispectral imagery?  true-colour composites (includes only red, green and blue bands (rgb))  false-colour composites (including rgb and other bands)  ndvi (normalized difference vegetation index)  other vegetation indices (e.g. savi, evi, sr)  none  other: 23. if you use an alternative vegetation index to ndvi what is it? e.g. savi. 24. for what applications do you use your multispectral imagery? 25. does your company acquire multispectral imagery on a regular cycle? e.g. every two years or only as required. 26. what software do you use when working with your multispectral imagery?  esri arcgis  mapinfo  atlas geomaster  open/free gis (e.g. qgis, grass)  envi image analysis software  trimble e-cognition image analysis software new zealand journal of forestry science  erdas imagine image analysis software  other: 27. what is the spatial resolution of your multispectral imagery? e.g. 10 metres. hyperspectral imagery 28. does your company use hyperspectral imagery? hyperspectral imagery typically consists of hundreds of bands spanning the visible and infrared wavelengths and being acquired from an airplane, uav or satellite platform  yes – go to question 30  no – go to question 29 29. what are the reasons for not using hyperspectral imagery?  cost  no perceived benefits  current staff lack knowledge or training to use hyperspectral imagery  was not aware of hyperspectral imagery  other: 30. how is your hyperspectral imagery acquired?  unmanned aerial vehicle (aka drone)  airplane  helicopter  satellite  other: 31. for what applications do you use your hyperspectral imagery? 32. does your company acquire hyperspectral imagery on a regular cycle? e.g. every two years or only as required. 33. what software do you use when working with your hyperspectral imagery?  esri arcgis  mapinfo  atlas geomaster  open/free gis (e.g. qgis, grass)  envi image analysis software  trimble e-cognition image analysis software  erdas imagine image analysis software  other: 34. what is the spatial resolution of your hyperspectral imagery? e.g. 20 metres. lidar 35. does your company use lidar data? lidar stands for light detection and ranging, it is also known as laser scanning. lidar data is acquired aerially or terrestrially and yields a threedimensional pointcloud that can be used to produce digital terrain models, canopy height models and structural descriptions of forests via lidar metrics.  yes – go to question 37  no – go to question 36 36. what are the reasons for not using lidar imagery?  cost  no perceived benefits new zealand journal of forestry science  current staff lack knowledge or training to use lidar  was not aware of lidar  other: 37. how is your lidar data acquired?  unmanned aerial vehicle (aka drone)  airplane  helicopter  terrestrial platform (e.g. lidar sensor mounted on tripod)  vehicular platform (e.g. lidar sensor mounted on ute)  other: 38. what is the point cloud density (points/m²) of the lidar data you acquire? 39. does your company acquire lidar data on a regular cycle? e.g. every two years or only as required. 40. do you process the raw. las files in-house or do you use lidar products (e.g. digital elevation model) produced by an external provider?  products are derived in-house from raw lidar data (i.e. las files)  products are provided by an aerial surveying company  products are derived by a third-party organisation (e.g. consultants) from raw data provided by surveying company 41. if you process raw .las files what software do you use?  fusion  lastools  esri  r  other: 42. what product(s) does your company derive from lidar data collection and processing?  digital elevation model  canopy height model  mean top height estimates  volume or biomass estimates  stem count  other: 43. for what applications do you use your lidar products? 44. what products would your company want to obtain from lidar data collection and processing in the future? prediction of noise emission in the machining of wood materials by means of an artificial neural network şükrü özşahin1 and hilal singer2,* 1 department of industrial engineering, karadeniz technical university, trabzon, turkey 2 department of industrial engineering, bolu abant izzet baysal university, bolu, turkey *corresponding author: hilal.singer@hotmail.com (received for publication 20 january 2020; accepted in revised form 18 march 2022) abstract background: noise produced during machining of wood materials can be a source of harm to workers and an environmental hazard. understanding the factors that contribute to this noise will aid the development of mitigation strategies. in this study, an artificial neural network (ann) model was developed to model the effects of wood species, cutting width, number of blades, and cutting depth on noise emission in the machining process. methods: a custom application created with matlab codes was used for the development of the multilayer feed-forward ann model. model performance was evaluated by numerical indicators such as mape, rmse, and r2. results: the ann model performed well with acceptable deviations. the mape, rmse, and r2 values were 0.553%, 0.600, and 0.9824, respectively, in the testing phase. furthermore, this study predicted the intermediate values not provided from the experimental study. the model predicted that lower noise emissions would occur with decreased cutting width and cutting depth. conclusions: anns are quite effective in predicting the noise emission. practitioners relying on the ann approach for investigating the effects of various factors on noise emission can save time and costs by reducing the number of experimental combinations studied to generate predictive models. new zealand journal of forestry science özşahin and singer new zealand journal of forestry science (2022) 52:11 https://doi.org/10.33494/nzjfs522022x92x e-issn: 1179-5395 published on-line: 12/04/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access designed and operated, otherwise, noise problems may arise. noise is generally defined as an unwanted sound (engin et al. 2019) and is a major occupational and environmental hazard. the continuous exposure of workers to high noise levels can cause detrimental health effects such as hearing loss, sleep disturbance, fatigue, and hypertension (hong et al. 2013). according to the national institute of occupational safety and health, an estimated 14% of workers are exposed to noise higher than the permissible limit (85 db(a)) (lee et al. 2009; ismaila & odusote 2014). occupational exposure to noise is unavoidable in the wood processing industry; however, this exposure could be minimised by better understanding the factors affecting noise. the most important main factors influencing the noise level are wood properties and machining parameters. therefore, it is important to introduction wood is a naturally occurring material consisting of cellulose, hemicelluloses, lignin, extractives, and inorganic components (uysal & yorur 2013). it can be used in both a solid form or further processed into woodbased composites (sedleckỳ & gašparík 2017). one of the most important wood-based composites is mediumdensity fiberboard (mdf). mdf is made from wood fibers that are glued together with heat and pressure. the physical and mechanical properties and surface qualities of mdf panels are relatively standardised and uniform. these characteristics make the panels a suitable alternative to solid wood for industrial manufacturing of furniture (fathollahzadeh et al. 2013). the production of furniture and decoration elements requires a series of transformation processes. the machines used in these processes must be properly keywords: artificial neural network, noise emission, machining, wood, prediction http://creativecommons.org/licenses/by/4.0/), özşahin and singer new zealand journal of forestry science (2022) 52:11 page 2 evaluate subfactors related to both wood properties and machining parameters for the reduction of noise emission in the machining process (owoyemi et al. 2017; çota et al. 2019). in recent years, several attempts have been made to examine the influences of various factors on noise emission in wood machining. ratnasingam and scholz (2008) stated that the use of smaller engines and breaking of fewer chips led to lower noise emission. svoreň et al. (2010) reported that the circular saw blade with sigmoid compensating slots had the lowest noise levels in the range of (2-5) db(a). pinheiro et al. (2015) determined that an increase in the moisture content of wood led to a decrease in noise emission. krilek et al. (2016) observed that the number of saw blade teeth had a significant effect on noise emission. this observation was also confirmed by kvietková et al. (2015). durcan and burdurlu (2018) noted that decreasing the blade number led to higher noise emission, while çota et al. (2019) reported that noise emission increased with increasing feed speed. it is clear that plenty of values for factors have to be investigated to detect a change in noise emission. however, the measuring of the effect of each factor on noise emission is expensive, and conducting tests is also time-consuming. therefore, it is important to find reliable and economic methods providing the desired results (mckenzie et al. 2003). owing to the heterogeneous nature of wood, wood-related factors possess nonlinear changes. hence, traditional linear models are inadequate in describing the characteristics of these factors. ignoring the presence of nonlinearities leads to misleading results. machine learning techniques are more appropriate for modeling and optimisation purposes. artificial neural networks (anns), one of the most attractive branches in artificial intelligence, are able to deal with linear and nonlinear problems and learn complex cause-and-effect relationships among inputs and outputs. anns are good for tasks involving fuzzy or incomplete information. they can be faster, cheaper, and more adaptable than conventional methods (ozsahin & murat 2018). the ann approach has brought a new insight into the solution of many problems in wood science. this approach has been employed for analyzing moisture in wood (avramidis & wu 2007), prediction of fracture toughness (samarasinghe et al. 2007), classification of veneer defects (castellani & rowlands 2008), wood recognition (khalid et al. 2008), modeling of some properties of oriented strand board (özşahin 2012; ozsahin 2013), determination of optimum power consumption in wood machining (tiryaki et al. 2016), prediction of formaldehyde emission (akyüz et al. 2017), and modeling of physical properties of heattreated wood (ozsahin & murat 2018). these studies have shown that the ann approach produces highly successful results. consequently, the existing literature has a gap in the prediction of noise emission by the ann approach. therefore, the objectives of this study are to: 1) develop an ann model for modeling the effects of wood species, cutting width, number of blades, and cutting depth on noise emission in the machining process; 2) to present a road map for the wood processing industry seeking to enhance worker health and safety; and 3) to fill the gap in the literature. methods dataset the data used in this study were taken from durcan and burdurlu (2018). the experimental process conducted by the authors can be briefly explained as follows. lombardy poplar (populus nigra l.), oriental beech (fagus orientalis l.), and mdf were selected as materials for the experiments. in the planing of the samples, five different levels of cutting width (6, 12, 18, 25, and 30 mm), three different levels of cutting depth (1, 2, and 3 mm), and two different levels of number of blades (1 blade and 4 blades) were tested. the cutting speed was chosen as 26.7 m/s, and the feed rate was 5 m/min. the extech hd 600 device (extech instruments, nh, usa) was used for the measurement of noise emission. a total of 1800 measurements were recorded with 20 measurements (replications) for each combination of factors. more information about the experimental procedure can be found in durcan and burdurlu (2018). artificial neural network approach the ann is a computational model that is inspired by the human brain (mia & dhar 2016). the ann approach offers many advantages over traditional statistical methods because it is capable of describing the relationship between input and output variables without any prior knowledge (venkata ramana et al. 2013; shebani & iwnicki 2018). anns can be used for data sorting, pattern recognition, optimisation, clustering, and simulation (yadav & chandel 2014). the most widely used network is the multilayer perceptron (mlp). it consists of one input layer, t hidden layer(s), and one output layer (drouillet et al. 2016). the input layer receives the data and transmits them to the hidden layer(s). the hidden layer(s) processes the information and sends the result to the output layer. the output layer provides the outputs of the network (kara et al. 2016). the mlp network comprises a number of neurons (nodes) organised in layers (ghorbani et al. 2016). each node is connected to other nodes by communication links (connections). each connection has a weight (özşahin 2012). in order to obtain the net input, inputs are multiplied by weights and combined with the relevant bias. outputs are calculated by applying a mathematical function to the net input. this process is summarised in equations (1) and (2) (ozsahin 2013). (1) (2) where: xi is the input signal, wij is the weight between the ith node and the jth node, θj is the bias, netj is the net input of the jth node, f(.) is one of the activation functions, and yj is the output of the jth node. input nodes and output nodes represent inputs and outputs, respectively. however, hidden nodes vary depending on the complexity level of the handled problem (beltramo et al. 2016). if too few hidden nodes are used, the network does not have enough ability to model complex relationships between inputs and outputs. on the other hand, if too many hidden nodes are used, overfitting problems may arise (quintana et al. 2011). neural networks must be trained with known inputoutput data. during the training process, the values of weights and biases are changed to obtain the best prediction results (haghdadi et al. 2013). when the error reaches a determined value or the specified number of iterations is reached, the training of anns is finished (ertunc et al. 2013). if the model responds correctly to input values that are not employed in training, the weights and biases of the trained network are saved. these weights and biases can be used to predict outputs for new input vectors (yildirim et al. 2011). artificial neural network analysis in this study, the noise emission values were predicted with the ann approach. the wood species, cutting width, number of blades, and cutting depth were considered as inputs, while the noise emission was the output of the ann model. we ran the ann model with a range of values for the given parameters. the other process parameters, environmental conditions, and wood-related parameters were held constant. the ann modeling steps were performed using matlab (mathworks, ma, usa). figure 1 shows the steps of this study. the data were grouped randomly and homogeneously in the form of training and testing data. 60 data points (66.67% of total data) were used for the training process and 30 data points (33.33% of total data) were used to test the validity of the ann model. different data groups were constituted from the data. each data group was tested to detect suitable data sets. the subsets used in the ann analysis are shown in table 1. in modeling, a feed-forward backpropagation neural network was used. the activation functions were the hyperbolic tangent sigmoid function (tansig) and the linear transfer function (purelin). the levenbergmarquardt algorithm (trainlm) was employed for training, and the gradient descent with a momentum backpropagation algorithm (traingdm) was considered as the learning rule. the training progress was monitored through the mean square error (mse) [equation (3)]: (3) where, ti refers to the actual value, tdi refers to the model output, and n refers to the number of measurements. normalising the data before the training and testing of anns is recommended to equalise the significance of variables (canakci et al. 2015). as the tansig function was used as the activation function, the experimental data were normalised between −1 and 1. the mapping of each variable to a value between −1 and 1 was carried out using equation (4). the outputs of the ann model were converted into the real values by using a reverse normalising process. (4) özşahin and singer new zealand journal of forestry science (2022) 52:11 page 3 table 1: description of the study sites figure 1: the steps of this study based on the ann approach. özşahin and singer new zealand journal of forestry science (2022) 52:11 page 4 o pe ra ti on n oi se e m is si on le ve ls ( db a ) cu tt in g w id th (m m ) b la de n o cu tt in g de pt h (m m ) po pl ar b ee ch m d f sa m pl e id m ea su re d p re di ct ed er ro r (% ) sa m pl e id m ea su re d p re di ct ed er ro r (% ) sa m pl e id m ea su re d p re di ct ed er ro r (% ) 6 1 1 b 1 83 .0 6 83 .1 0 -0 .0 5 a 21 80 .0 7 80 .3 5 -0 .3 5 a 41 78 .6 2 78 .4 8 0. 18 6 1 2 a 1 85 .2 6 84 .9 7 0. 34 b 11 84 .2 3 84 .4 2 -0 .2 3 b 21 83 .7 5 83 .5 4 0. 25 6 1 3 a 2 86 .8 7 86 .7 0 0. 20 a 22 86 .7 0 87 .1 6 -0 .5 3 a 42 85 .8 5 86 .4 8 -0 .7 4 6 4 1 a 3 82 .7 5 82 .0 0 0. 91 b 12 79 .7 4 79 .1 7 0. 71 b 22 78 .1 4 76 .9 3 1. 54 6 4 2 b 2 84 .0 1 83 .9 3 0. 10 a 23 82 .9 0 82 .7 9 0. 13 a 43 82 .2 4 82 .2 9 -0 .0 6 6 4 3 a 4 84 .9 4 85 .0 8 -0 .1 6 a 24 84 .5 9 84 .7 1 -0 .1 4 a 44 84 .3 9 84 .2 0 0. 22 12 1 1 a 5 84 .5 6 84 .6 3 -0 .0 8 b 13 81 .4 3 82 .0 1 -0 .7 1 a 45 83 .4 9 82 .8 9 0. 71 12 1 2 a 6 87 .8 1 86 .4 3 1. 57 a 25 87 .9 0 88 .1 0 -0 .2 2 a 46 84 .8 9 85 .2 9 -0 .4 7 12 1 3 b 3 90 .8 6 89 .7 8 1. 19 a 26 91 .1 2 90 .4 6 0. 73 b 23 88 .0 5 88 .3 1 -0 .3 0 12 4 1 b 4 82 .9 8 83 .5 8 -0 .7 2 a 27 80 .7 6 80 .8 7 -0 .1 4 a 47 81 .3 2 81 .3 1 0. 02 12 4 2 a 7 85 .8 3 85 .4 3 0. 46 a 28 86 .6 1 86 .9 2 -0 .3 6 b 24 83 .1 5 84 .0 9 -1 .1 3 12 4 3 a 8 89 .4 6 88 .9 6 0. 55 b 14 87 .6 0 87 .6 0 0. 00 a 48 86 .5 1 86 .2 3 0. 33 18 1 1 a 9 86 .2 0 86 .0 9 0. 13 a 29 82 .8 6 83 .6 5 -0 .9 6 b 25 84 .2 0 84 .9 1 -0 .8 4 18 1 2 b 5 88 .2 9 87 .9 4 0. 40 a 30 89 .1 6 89 .6 7 -0 .5 7 a 49 86 .9 4 86 .9 7 -0 .0 4 18 1 3 a 10 91 .1 9 91 .2 1 -0 .0 2 b 15 92 .7 1 92 .3 7 0. 37 a 50 89 .9 9 90 .0 6 -0 .0 8 18 4 1 a 11 84 .6 6 85 .0 9 -0 .5 0 a 31 81 .9 2 82 .5 5 -0 .7 6 a 51 83 .5 0 83 .7 0 -0 .2 3 18 4 2 a 12 86 .9 3 86 .9 5 -0 .0 3 b 16 89 .0 5 88 .6 0 0. 51 b 26 85 .7 5 85 .8 2 -0 .0 8 18 4 3 b 6 89 .9 4 90 .4 4 -0 .5 5 a 32 89 .5 2 89 .7 4 -0 .2 5 a 52 88 .4 8 88 .1 7 0. 35 25 1 1 a 13 87 .1 3 87 .7 0 -0 .6 6 b 17 86 .6 3 86 .0 4 0. 68 a 53 86 .2 3 86 .8 8 -0 .7 5 25 1 2 b 7 91 .8 7 90 .7 3 1. 24 a 33 91 .2 2 91 .4 0 -0 .2 0 a 54 88 .2 2 88 .8 5 -0 .7 1 25 1 3 a 14 93 .9 1 93 .4 9 0. 45 a 34 94 .5 6 94 .3 1 0. 27 b 27 92 .8 6 92 .0 1 0. 92 25 4 1 b 8 85 .9 9 86 .7 6 -0 .8 9 a 35 85 .4 6 84 .8 2 0. 75 b 28 85 .5 4 85 .7 2 -0 .2 1 25 4 2 a 15 89 .1 6 89 .4 6 -0 .3 3 b 18 90 .5 2 90 .3 8 0. 15 a 55 87 .4 6 87 .7 5 -0 .3 3 25 4 3 a 16 91 .3 5 92 .1 4 -0 .8 6 a 36 91 .2 2 91 .8 8 -0 .7 2 a 56 90 .2 7 90 .3 3 -0 .0 6 30 1 1 a 17 89 .5 0 88 .8 0 0. 79 a 37 88 .8 2 89 .3 9 -0 .6 5 a 57 89 .8 6 88 .2 3 1. 81 30 1 2 a 18 94 .5 9 94 .4 6 0. 14 b 19 92 .1 9 92 .5 8 -0 .4 2 b 29 91 .3 1 90 .1 3 1. 29 30 1 3 b 9 96 .3 3 96 .5 2 -0 .2 0 a 38 95 .1 1 95 .6 0 -0 .5 2 a 58 93 .7 1 93 .3 3 0. 41 30 4 1 a 19 86 .7 5 87 .8 9 -1 .3 2 a 39 87 .9 4 87 .7 8 0. 18 a 59 87 .5 7 87 .1 1 0. 52 30 4 2 b 10 92 .6 0 93 .1 3 -0 .5 7 a 40 91 .5 6 91 .6 0 -0 .0 5 a 60 89 .9 7 89 .0 7 1. 00 30 4 3 a 20 94 .5 4 93 .9 7 0. 61 b 20 93 .5 4 93 .3 2 0. 23 b 30 91 .8 8 91 .7 9 0. 10 ta b le 1 : t he m ea su re d an d pr ed ic te d va lu es o f n oi se e m is si on a nd th ei r pe rc en ta ge e rr or s. b ol d va lu es : t es ti ng d at a, th e ot he r va lu es : t ra in in g da ta a a nd b d en ot e sa m pl e id s in tr ai ni ng a nd te st in g, r es pe ct iv el y where, xnorm is the normalised value, x is the real value, and xmin and xmax are the minimum and maximum values of x, respectively. the performance of ann-based models is affected by many factors such as activation functions, learning rule, momentum, and the number of nodes in the hidden layer(s) (mohanraj et al. 2012). therefore, different network parameters and configurations were tried until the difference between the measured and predicted values was minimised. the established models were checked by employing the testing data. as a result, the ann model yielding the nearest values to the experimental results was run for predictions. the optimum values of weights and biases of the ann model are shown in table 2. figure 2 shows the developed model. the input layer of the ann model consists of four nodes representing wood species, cutting width, number of blades, and cutting depth. the output node represents the output parameter called noise emission. anns should be not too large to prevent the loss of generalisation. the attention should be paid to the number of nodes in each hidden layer (muralitharan et al. 2018). in this study, the ann model was designed on the trial-and-error basis. the best performance was obtained with 3-3 hidden nodes. the proposed model is mathematically logical and defined because the number of the connections is lower than the number of data points available for training. the performance of prediction models can be evaluated by using various statistical measures. in this study, the mean absolute percentage error (mape), the root mean square error (rmse), and the coefficient of determination (r2) were used to compare the established models. the mape, rmse, and r2 values were calculated by using the following equations: (5) (6) (7) where is the average of predicted values. özşahin and singer new zealand journal of forestry science (2022) 52:11 page 5 hidden layer 1 hidden layer 2 output layer neuron 1 neuron 2 neuron 3 bias 1 neuron 1 neuron 2 neuron 3 bias 2 neuron 1 bias 3 0.01109 -3.27308 -0.04546 -0.51600 2.13712 3.20086 -8.24474 -6.65524 0.26429 -2.53834 -0.01309 -1.92293 -0.01902 0.03776 -12.69931 -11.60765 0.00014 12.65604 0.48143 0.00226 0.08328 0.03226 5.17537 -14.63688 0.50336 0.11658 -3.50894 5.29178 -0.00803 -3.63630 -4.84149 figure 2: the proposed network architecture. table 2: the optimum values of weights and biases. results and discussion in this study, a feed-forward backpropagation neural network was designed for the prediction of noise emission. the network was trained and tested using 90 data points. as a result of the modeling process, the 4:3:3:1 architecture was selected to make predictions. the actual and predicted values and their percentage errors are given in table 1. the mape, rmse, and r2 values were employed as the main criteria to evaluate the performance of the ann model. table 3 shows the mape, rmse, and r2 statistics calculated for the ann model. according to lewis (1982), typical mape values for performance evaluation are categorised as follows: mape ≤ 10% – high, 10% ≤ mape ≤ 20% – good, 20% ≤ mape ≤ 50% – reasonable, and mape ≥ 50% – inaccurate. in this study, the mape values were calculated as 0.46% for the training phase and 0.55% for the testing phase. as seen from the results, the ann model has an excellent performance in the prediction of noise emission. rmse measures the deviation between actual and predicted values. the lower value of rmse suggests better model performance (chen & chau 2016). in this study, the rmse values were calculated as 0.521 db(a) and 0.600 db(a) for the training and testing phases, respectively. it can be thus said that the prediction of noise emission is successful in terms of the rmse criterion. r2 is an indicator of the strength of the relationship between measured and predicted values. if the r2 value of a model is above 0.90, the model has a high performance (özşahin 2012). in this study, the regression analysis was carried out to calculate the r2 values of the proposed model. the r2 values were calculated as 0.98 and 0.98 for the training and testing phases, respectively. the values of the r2 criterion show that the established network has the ability to explain at least 98% of the observed variation in noise emission. the comparisons between the measured and predicted values are presented in figure 3. the predicted values showed a close match with the measured values. therefore, it is concluded that the ann model can be used as an appropriate tool to predict noise emission. the investigation of the influence of each factor on noise emission requires a large number of experimental studies. however, extra experiments are time-consuming and give rise to an increase in costs. the combinations obtained by anns may be used to improve experimental processes. in this respect, the use of the ann approach is important because it is capable of predicting the özşahin and singer new zealand journal of forestry science (2022) 52:11 page 6 phase performance criterion mape rmse r2 training 0.461 0.521 0.9811 (y = 0.9811x + 1.6467) testing 0.553 0.600 0.9824 (y = 0.9792x + 1.7365) table 3: performance evaluation criteria for the noise emission prediction. figure 3: the comparison of the measured and predicted values: (a) training; and (b) testing. (a) (b) untested experimental results (akyüz et al. 2017). in this study, wood species and number of blades were fixed, and cutting width and cutting depth were changed. the intermediate values not obtained from the experimental study were determined by the ann model for different cutting widths and cutting depths. the surface plots showing the changes in noise emission are given in figure 4. as seen in this figure, noise emission decreases with decreased cutting width and cutting depth. the optimisation can be performed via an analysis of responses of the model. each wood type possesses a different structure. this differentiates the changes in noise emissions. as can be seen in figure 4, the structural heterogeneities of the poplar and beech woods give rise to nonuniform changes in noise emissions. the mdf material has a more homogeneous structure than the others. hence, the changes in the noise levels emitted during the cutting of the mdf boards show homogeneous-like behaviour. the modeling results provided a better understanding of the effect of wood structure on machining noise. the improper setting of machining parameters leads to high noise levels. revealing the mutual relations of different factors is very important for obtaining the best results. because the developed model operates with an average error of 0.55%, the results are acceptable and guiding. by taking into account interval values, the ann model can allow earlier detection of noise levels and help to control the noise. it has been reported that approximately 16% of adult-onset hearing loss is caused by workplace noise (thepaksorn et al. 2019) and the wood processing industry is one of the noisiest industries. in order to reduce noise emissions, processing conditions and workplace-specific factors must be properly set via scientific approaches. it is clear that each change in noise emission will affect workers. loud noise can cause workplace accidents and injuries. hence, preventive measures must be applied to reduce the severity of high noise. some important control strategies are as follows: changing the loudest technological processes and machines, performing routine maintenance on machinery and equipment, preserving the sharpness of blades, ensuring the balance of rotating parts, installing isolation dampers, utilising helicoidal gears, clamping of parts or panels, using flexible connections, ensuring pressure tightness and homogeneity, and using acoustic silencers and sound insulating control cabins. furthermore, effective hearing loss prevention programs that comprise exposure assessments, noise controls, regular audiometric monitoring, usage of hearing protectors for exposure >85 db(a), worker training, and good record keeping are required to reduce adverse results. the modeling results show that there is a good agreement between the actual and predicted values. based on the results of this study, it can be said that the effects of various factors on noise emission can be predicted by anns without the need for experimental studies that require much time and high costs. in further research, different variables can be used to predict noise emission. conclusions the use of the ann approach for modeling the effects of wood species, cutting width, number of blades, and cutting depth on noise emission in the machining process has been studied. the main results obtained özşahin and singer new zealand journal of forestry science (2022) 52:11 page 7 figure 4: predictive changes in noise emission for different cutting widths and cutting depths. from this study are summarised below. 1. the values obtained with the ann model are very close to the measured values. 2. the ann model provides very satisfactory results with acceptable deviations. the mape, rmse, and r2 values are 0.55%, 0.60 db(a), and 0.98, respectively, in the testing phase. these values demonstrate that the developed model can provide accurate, fast, and acceptable results. 3. in the predictive examples, it is seen that noise emission increases with increased cutting width and cutting depth. the usage of the ann approach would be useful for the wood processing industry in obtaining the emission values of the noise which creates a potential threat for worker health. 4. anns are quite effective in predicting the noise emission. this capability to prediction and faster decision-making help the wood processing industry to get precautions and achieve better results. hence, the ann model can reduce the experimental time and costs. competing interests the authors declare that they have no competing interests. authors' contributions the first author planned the study and carried out the ann analysis. the second author wrote the manuscript. acknowledgements the authors would like to thank fatih mehmet durcan and dr. erol burdurlu for providing the data used in this paper. references akyüz, i̇., özşahin, ş., tiryaki, s., & aydın, a. (2017). an application of artificial neural networks for modeling formaldehyde emission based on process parameters in particleboard manufacturing process. clean technologies and environmental policy, 19(5), 1449–1458. https:// doi.org/10.1007/s10098-017-1342-0 avramidis, s., & wu, h. (2007). artificial neural network and mathematical modeling comparative analysis of nonisothermal diffusion of moisture in wood. holz als roh und werkstoff, 65(2), 89–93. https:// doi.org/10.1007/s00107-006-0113-0 beltramo, t., ranzan, c., hinrichs, j., & hitzmann, b. (2016) artificial neural network prediction of the biogas flow rate optimised with an ant colony algorithm. biosystems engineering, 143, 68–78. https://doi. org/10.1016/j.biosystemseng.2016.01.006 canakci, a., varol, t., & ozsahin, s. (2015). artificial neural network to predict the effect of heat treatment, reinforcement size, and volume fraction on alcumg alloy matrix composite properties fabricated by stir casting method. international journal of advanced manufacturing technology, 78(1-4), 305–317. https://doi.org/10.1007/ s00170-014-6646-1 castellani, m., & rowlands, h. (2008). evolutionary feature selection applied to artificial neural networks for wood-veneer classification. international journal of production research, 46(11), 3085–3105. https:// doi.org/10.1080/00207540601139955 chen, x.y., & chau, k.w. (2016). a hybrid double feedforward neural network for suspended sediment load estimation. water resources management, 30(7), 2179–2194. https://doi. org/10.1007/s11269-016-1281-2 çota, h., lato, e., & quku, d. (2019). analysis of noise level at mdf and particleboard processing with different feeding speed. pro ligno, 15(2), 10–17. drouillet, c., karandikar, j., nath, c., journeaux, a.c., el mansori, m., & kurfess, t. (2016). tool life predictions in milling using spindle power with the neural network technique. journal of manufacturing processes, 22, 161–168. https://doi. org/10.1016/j.jmapro.2016.03.010 durcan, f.m., & burdurlu, e. (2018). effects of some machining parameters on noise level in planing of some wood materials. bioresources, 13(2), 2702–2714. https://doi.org/10.15376/ biores.13.2.2702-2714 engin, i.c., ozkan, e., & cetin, s. (2019). determination of risky areas at the marble workshops in terms of noise. acoustics australia, 47(1), 79–90. https:// doi.org/10.1007/s40857-018-0146-7 ertunc, h.m., ocak, h., & aliustaoglu, c. (2013). ann and anfis-based multi-staged decision algorithm for the detection and diagnosis of bearing faults. neural computing and applications, 22, 435–446. https://doi.org/10.1007/s00521-012-0912-7 fathollahzadeh, a., enayati, a.a., & erdil, y.z. (2013). effect of laboratory-accelerated aging treatment on the ultimate strength of a 4-sided mdf kitchen cabinet. turkish journal of agriculture and forestry, 37(5), 649–656. https://doi.org/10.3906/tar1208-43 ghorbani, m.a., zadeh, h.a., isazadeh, m., & terzi, o. (2016). a comparative study of artificial neural network (mlp, rbf) and support vector machine models for river flow prediction. environmental earth sciences, 75(6), 476. https://doi. org/10.1007/s12665-015-5096-x haghdadi, n., zarei-hanzaki, a., khalesian, a.r., & abedi, h.r. (2013). artificial neural network modeling to özşahin and singer new zealand journal of forestry science (2022) 52:11 page 8 https://doi.org/10.1007/s10098-017-1342-0 https://doi.org/10.1007/s10098-017-1342-0 https://doi.org/10.1007/s00107-006-0113-0 https://doi.org/10.1007/s00107-006-0113-0 https://doi.org/10.1016/j.biosystemseng.2016.01.006 https://doi.org/10.1016/j.biosystemseng.2016.01.006 https://doi.org/10.1007/s00170-014-6646-1 https://doi.org/10.1007/s00170-014-6646-1 https://doi.org/10.1080/00207540601139955 https://doi.org/10.1080/00207540601139955 https://doi.org/10.1007/s11269-016-1281-2 https://doi.org/10.1007/s11269-016-1281-2 https://doi.org/10.1016/j.jmapro.2016.03.010 https://doi.org/10.1016/j.jmapro.2016.03.010 https://doi.org/10.15376/biores.13.2.2702-2714 https://doi.org/10.15376/biores.13.2.2702-2714 https://doi.org/10.1007/s40857-018-0146-7 https://doi.org/10.1007/s40857-018-0146-7 https://doi.org/10.1007/s00521-012-0912-7 https://doi.org/10.3906/tar-1208-43 https://doi.org/10.3906/tar-1208-43 https://doi.org/10.1007/s12665-015-5096-x https://doi.org/10.1007/s12665-015-5096-x predict the hot deformation behavior of an a356 aluminum alloy. materials and design, 49, 386–391. https://doi.org/10.1016/j.matdes.2012.12.082 hong, o.s., kerr, m.j., poling, g.l., & dhar, s. (2013). understanding and preventing noise-induced hearing loss. disease-a-month, 59(4), 110–118. https://doi.org/10.1016/j.disamonth.2013.01.002 ismaila, s.o., & odusote, a. (2014). noise exposure as a factor in the increase of blood pressure of workers in a sack manufacturing industry. beni-suef university journal of basic and applied sciences, 3(2), 116–121. https://doi.org/10.1016/j. bjbas.2014.05.004 kara, f., aslantaş, k., & çiçek, a. (2016). prediction of cutting temperature in orthogonal machining of aisi 316l using artificial neural network. applied soft computing, 38, 64–74. https://doi. org/10.1016/j.asoc.2015.09.034 khalid, m., lee, e.l.y., yusof, r., & nadaraj, m. (2008). design of an intelligent wood species recognition system. international journal of simulation: systems, science and technology, 9(3), 9–19. krilek, j., kováč, j., barcik, š., svoreň, j., štefánek, m., & kuvik, t. (2016). the influence of chosen factors of a circular saw blade on the noise level in the process of cross cutting wood. wood research, 61(3), 475–486. kvietková, m., gaff, m., gašparík, m., kminiak, r., & kriš a. (2015). effect of number of saw blade teeth on noise level and wear of blade edges during cutting of wood. bioresources, 10(1), 1657–1666. https:// doi.org/10.15376/biores.10.1.1657-1666 lee, j.h., kang, w., yaang, s.r., choy, n., & lee, c.r. (2009). cohort study for the effect of chronic noise exposure on blood pressure among male workers in busan, korea. american journal of industrial medicine, 52(6), 509–517. https://doi. org/10.1002/ajim.20692 lewis, c.d. (1982). international and business forecasting methods. butterworths: london. mckenzie, h.m., shelbourne, c.j.a., kimberley, m.o., mckinley, r.b., & britton, r.a.j. (2003). processing young plantation-grown eucalyptus nitens for solid-wood products. 2: predicting product quality from tree, increment core, disc, and 1-m billet properties. new zealand journal of forestry science, 33(1), 79–113. mia, m., & dhar, n.r. (2016). prediction of surface roughness in hard turning under high pressure coolant using artificial neural network. measurement, 92, 464–474. https://doi. org/10.1016/j.measurement.2016.06.048 mohanraj, m., jayaraj, s., & muraleedharan, c. (2012). applications of artificial neural networks for refrigeration, air-conditioning and heat pump systems a review. renewable and sustainable energy reviews, 16(2), 1340–1358. https://doi. org/10.1016/j.rser.2011.10.015 muralitharan, k., sakthivel, r., & vishnuvarthan, r. (2018). neural network based optimization approach for energy demand prediction in smart grid. neurocomputing, 273, 199–208. https://doi. org/10.1016/j.neucom.2017.08.017 owoyemi, m.j., falemara, b.c., & owoyemi, a.j. (2017). noise pollution and control in mechanical processing wood industries. biomedical statistics and informatics, 2(2), 54–60. https://doi. org/10.20944/preprints201608.0236.v1 ozsahin, s. (2013). optimization of process parameters in oriented strand board manufacturing with artificial neural network analysis. european journal of wood and wood products, 71(6), 769–777. https://doi.org/10.1007/s00107-013-0737-9 ozsahin, s., & murat, m. (2018). prediction of equilibrium moisture content and specific gravity of heat treated wood by artificial neural networks. european journal of wood and wood products, 76(2), 563–572. https://doi.org/10.1007/s00107017-1219-2 özşahin, s. (2012). the use of an artificial neural network for modeling the moisture absorption and thickness swelling of oriented strand board. bioresources, 7(1), 1053–1067. pinheiro, c., de sampaio alves, m.c., & simões amaral, s. (2015). moisture content and its influence on the roughness and noise emission during wood machining. advanced materials research, 1088, 680–685. https://doi.org/10.4028/www. scientific.net/amr.1088.680 ratnasingam, j., & scholz, f. (2008). noise level evaluation in the malaysian wood working industry (report no:10/08). international furniture research group: usa. quintana, g., garcia-romeu, m.l., & ciurana, j. (2011) surface roughness monitoring application based on artificial neural networks for ball-end milling operations. journal of intelligent manufacturing, 22(4), 607–617. https://doi.org/10.1007/s10845009-0323-5 samarasinghe, s., kulasiri, d., & jamieson, t. (2007). neural networks for predicting fracture toughness of individual wood samples. silva fennica, 41(1), 105–122. https://doi.org/10.14214/sf.309 sedleckỳ, m., & gašparík, m. (2017). power consumption during edge milling of medium-density fiberboard and edge-glued panel. bioresources, 12(4), 7413– 7426. shebani, a., & iwnicki, s. (2018). prediction of wheel and rail wear under different contact conditions using artificial neural networks. wear, 406-407, 173– özşahin and singer new zealand journal of forestry science (2022) 52:11 page 9 https://doi.org/10.1016/j.matdes.2012.12.082 https://doi.org/10.1016/j.disamonth.2013.01.002 https://doi.org/10.1016/j.bjbas.2014.05.004 https://doi.org/10.1016/j.bjbas.2014.05.004 https://doi.org/10.1016/j.asoc.2015.09.034 https://doi.org/10.1016/j.asoc.2015.09.034 https://doi.org/10.15376/biores.10.1.1657-1666 https://doi.org/10.15376/biores.10.1.1657-1666 https://doi.org/10.1002/ajim.20692 https://doi.org/10.1002/ajim.20692 https://doi.org/10.1016/j.measurement.2016.06.048 https://doi.org/10.1016/j.measurement.2016.06.048 https://doi.org/10.1016/j.rser.2011.10.015 https://doi.org/10.1016/j.rser.2011.10.015 https://doi.org/10.1016/j.neucom.2017.08.017 https://doi.org/10.1016/j.neucom.2017.08.017 https://doi.org/10.20944/preprints201608.0236.v1 https://doi.org/10.20944/preprints201608.0236.v1 https://doi.org/10.1007/s00107-013-0737-9 https://doi.org/10.1007/s00107-017-1219-2 https://doi.org/10.1007/s00107-017-1219-2 https://doi.org/10.4028/www.scientific.net/amr.1088.680 https://doi.org/10.4028/www.scientific.net/amr.1088.680 https://doi.org/10.1007/s10845-009-0323-5 https://doi.org/10.1007/s10845-009-0323-5 https://doi.org/10.14214/sf.309 özşahin and singer new zealand journal of forestry science (2022) 52:11 page 10 184. https://doi.org/10.1016/j.wear.2018.01.007 svoreň, j., javorek, l., & murin, l. (2010). effect of the shape of compensating slots in the body of a circular saw blade on noise level in the cutting process. pro ligno, 6(4), 5–12. thepaksorn, p., koizumi, a., harada, k., siriwong, w., & neitzel, r.l. (2019). occupational noise exposure and hearing defects among sawmill workers in the south of thailand. international journal of occupational safety and ergonomics, 25(3), 458– 466. https://doi.org/10.1080/10803548.2017.13 94710 tiryaki, s., malkoçoğlu, a., & özşahin, ş. (2016). artificial neural network modeling to predict optimum power consumption in wood machining. drewno, 59(196), 109-125. https://doi.org/10.12841/ wood.1644-3985.140.08 uysal, b., & yorur, h. (2013). the effect of steam treatment on bonding strength of impregnated wood materials. journal of adhesion science and technology, 27(8), 896–904. https://doi.org/10.1 080/01694243.2012.727161 venkata ramana, r., krishna, b., kumar, s.r., & pandey, n.g. (2013). monthly rainfall prediction using wavelet neural network analysis. water resources management, 27(10), 3697–3711. https://doi. org/10.1007/s11269-013-0374-4 yadav, a.k., & chandel, s.s. (2014). solar radiation prediction using artificial neural network techniques: a review. renewable and sustainable energy reviews, 33, 772–781. https://doi. org/10.1016/j.rser.2013.08.055 yildirim, i., ozsahin, s., & akyuz, k.c. (2011). prediction of the financial return of the paper sector with artificial neural networks. bioresources, 6(4), 4076–4091. https://doi.org/10.1016/j.wear.2018.01.007 https://doi.org/10.1080/10803548.2017.1394710 https://doi.org/10.1080/10803548.2017.1394710 https://doi.org/10.12841/wood.1644-3985.140.08 https://doi.org/10.12841/wood.1644-3985.140.08 https://doi.org/10.1080/01694243.2012.727161 https://doi.org/10.1080/01694243.2012.727161 https://doi.org/10.1007/s11269-013-0374-4 https://doi.org/10.1007/s11269-013-0374-4 https://doi.org/10.1016/j.rser.2013.08.055 https://doi.org/10.1016/j.rser.2013.08.055 regression kriging to improve basal area and growing stock volume estimation based on remotely sensed data, terrain indices and forest inventory of black pine forests ferhat bolat*,#, sinan bulut#, alkan günlü, i̇lker ercanlı, muammer şenyurt deparment of forest engineering, faculty of forestry, çankırı karatekin university, turkey *corresponding author: fbolat@karatekin.edu.tr # these authors contributed equally to this work (received for publication 5 april 2019; accepted in revised form 20 july 2020) abstract background: the use of satellite imagery to quantify forest metrics has become popular because of the high costs associated with the collection of data in the field. methods: multiple linear regression (mlr) and regression kriging (rk) techniques were used for the spatial interpolation of basal area (g) and growing stock volume (gsv) based on landsat 8 and sentinel-2. the performance of the models was tested using the repeated k-fold cross-validation method. results: the prediction accuracy of g and gsv was strongly related to forest vegetation structure and spatial dependency. the nugget value of semivariograms suggested a moderately spatial dependence for both variables (nugget/sill ratio~70%). landsat 8 and sentinel-2 based rk explained approximately 52% of the total variance in g and gsv. root-mean-square errors were 7.84 m2 ha-1 and 49.68 m3 ha-1 for g and gsv, respectively. conclusions: the diversity of stand structure particularly at the poorer sites was considered the principal factor decreasing the prediction quality of g and gsv by rk. new zealand journal of forestry science bolat et al. new zealand journal of forestry science (2020) 50:4 https://doi.org/10.33494/nzjfs502020x49x e-issn: 1179-5395 published on-line: 01/08/2020 © the author(s). 2020 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access nevertheless, there remain a number of issues like unequal or fragmented forest distribution, differing tree species and age classes, which can lead to difficulty when trying to maximise the spatial variance explained when modeling forest metrics (chirici et al. 2008; gebreslasie et al. 2008; ingram et al. 2005; lu et al. 2004). because of these drawbacks, and in order to estimate forest metrics at an acceptable level of confidence and at a fine level of detail, scientists have combined remotely sensed data and ground measurements using methods such as ordinary least square, machine learning, and geo-statistic methods in the last ten years (mallinis et al. 2004; franco-lopez et al. 2001; ingram et al. 2005; meng et al. 2009; dos reis et al. 2018). for example, meng et al. (2009) evaluated ordinary kriging, universal kriging, cokriging and regression kriging methods combined with landsat 7 etm+ for spatially predicting g of pine forests. introduction forest inventory studies are conducted to quantify forest attributes such as basal area (g), growing stock volume (gsv), biomass and carbon sequestration that are providing essential information for forest managers. g has been often used as an important auxiliary variable to determine competition indices (contreras et al. 2011), stand density (curtis 1982), and diameterheight increment and mortality (kuehne et al. 2016). determination of the gsv (i.e. standing volume) is needed to assess silvicultural treatments for ensuring sustainable wood production in a managed forest (o’hehir & nambiar 2010). measuring g and gsv can be time consuming particularly for precision forestry. as a consequence, foresters are using remote sensing for the estimation of forest metrics in remote and difficult locations. keywords: forestry, k-fold cross-validation, landsat 8, sentinel-2, semi-arid region bolat et al. new zealand journal of forestry science (2020) 50:4 page 2 they found that using regression kriging resulted in the greatest precision and almost unbiased estimates of g for loblolly pine and slash pine. the authors concluded that kriging is robust for the interpolation of forest variables such as g, gsv, and other carbon and forestry metrics such as stand density, stand age and dominant height. maselli and chiesi (2006) evaluated three methods relying on remotely sensed data including k-nearest neighbour method, a locally calibrated regression analysis and a kriging method for forest volume estimation in a mediterranean region. results showed that all three methods were similar in terms of their correlation coefficient and root-mean-square error, but the kriging method was slightly better with regard to lower residual variance. dos reis et al. (2018) assessed a variety of methods including a random forest algorithm and a kriging method integrated with landsat thematic mapper (tm) data for the spatial prediction of g and gsv. multiple linear regression and artificial neural network models had the poorest performance for the estimation of timber volume of eucalyptus clonal stands. in forest inventories, a wide range of data types are gathered in terms of forest variables as well as geographical locations. the relationships between a response variable and explanatory variable(s) can differ from one sample unit to another because of sampling density, site index and stand age. in order to obtain unbiased residual estimates, local (random) effects should be introduced in a model (fortin et al. 2007). if a model accounts for local effects adequately, it may predict the target variable at an acceptable level of accuracy (magnussen et al. 2017). geostatistical approaches have the ability to account for unknown random effects in a local forest area, providing approximately unbiased estimates of residuals (isaaks & srivastava 1989). also, the forest metrics are spatially correlated within the same locations. although multiple regression analysis is user-friendly and easy to use for forest managers, it does not account for spatial autocorrelation in data. therefore, especially in case of a trend in the error variance across predictions, this method can be inappropriate in the prediction of forest metrics due to its biased parameter estimates. rk uses residuals from the least-square methods (i.e. multivariate regression). the improvements on the predictions can be important when taking spatial autocorrelation in the data into account. although there are a variety of approaches to consider spatial autocorrelation in data, the development and implementation of these methods require a strong statistical background. in this context, rk is both userfriendly and relatively simple, which may suit the needs of forest managers. therefore, the objective of this study was to evaluate the performance of landsat 8 and sentinel-2 based rk and mlr for improving the spatial predictions of g and gsv. methods study area the study area is within the inner anatolia semi-arid climate in the black sea steppe transitional zone in turkey (figure 1). it is bounded by latitudes 40° 29′ 09″–40° 30′ 44″ n and longitudes 33° 25′ 47″–33° 27′ 19″ e (wgs 1984 utm zone 36n). in the study area, average annual temperature and total precipitation are approximately 11.3 °c and 412 mm, respectively, while elevations range from 1280 to 1642 m (turkish state meteorological service n.d.). terrain variables of the study area were obtained from a digital elevation map with a twelve-metre resolution (table 1). the figure 1. location of the study area and its location in turkey study area consists of 380 hectares, where anatolian black pine (pinus nigra arnold. subsp. pallasiana lamb holmboe) is widely distributed. in this region, harsh ecological conditions such as low precipitation, high temperatures, and poor soils contribute towards poor forest growth (barbati et al. 2014; erşahin et al. 2016; göl & abay 2003). such conditions adversely influence the distribution and abundance of conifer forests in this region (çolak & rotherham 2006). anatolian black pine has been frequently planted because it adapts well to the semi-arid conditions of this region (konukcu 2001). field research the study area was afforested in the 1961-1965 period and have been thinned every ten years since establishment. a systematic sampling with 51 plots distributed within a 200 x 200 m grid were measured in 2016. circular plot sizes ranged from 400 to 800 m2 depending on the crown closure as suggested by the forest management guidelines for turkey (anonymous 2008). we calculated descriptive statistics for basal area (g, m2 ha-1), and growing stock volume (gsv, m3 ha-1) using the sas software®. (table 2). g and gsv of each sample plot were calculated by eq. 1 and eq. 2 (şenyurt 2017). in each sample plot, tree diameters at breast height (dbh) were measured with 0.1 cm-precision for each living tree (dbh ≥ 8 cm) using calipers. trees with unsuppressed growth (due to lack of competition) representing potential site productivity were selected to assess the site index of each sample plot. tree heights were measured with 1% precision using the vertex iv ultrasound instrument. top height was calculated as the mean height of the 100 tallest trees within a hectare. the mean age of each sample plot was measured using increment cores extracted from five trees that were selected based on the mean square diameter of the sample plot. subsequently, the site index of each sample plot was assessed using the site index equation developed by kalıpsız (1963) for black pine stands for a base age of 100 years. relative density (rd) of each sample plot was calculated by the stand density index of curtis (1982) (eq.3). satellite images, processing, and data landsat 8 and sentinel-2 satellite images were obtained from the united states geological survey earth explorer data portal (usgs 2000) and acquired on 02 and 20 august 2016, respectively. wgs 1984 (utm zone 36) projection system was used for orthorectification and georeferencing of the satellite images with first order nearest neighbourhood rules. the atmospheric correction was applied to landsat 8 and sentinel-2 images using the atcor module of qgis® software. then, the geometric correction was applied using twenty ground control points such as crossings, bridges, and hill tops through a global positioning system (gps). inventory was carried out using circular plots of 400, 600 and 800 m2. a single pixel obtained from landsat 8 covers an area of 900 m2 (30x30 m). sentinel-2 has bands with a spatial resolution of 10, 20 and 60 m, which cover an area of 100, 200 and 3600 m2, respectively. since landsat 8 and sentinel-2 spatial resolution (i.e. b9 and b10 bands of sentinel-2) are higher than sample plot size, the position of some sample plots may not be the center of a pixel. in such a circumstance, the buffer zone was applied for obtaining more representative data. for those plots that did not coincide with the center of the corresponding pixel, the average of all those pixels comprised by a plot was calculated. detail descriptions and formulae for vegetation indices calculated from landsat 8 and sentinel-2 bands can be found in vescovo et al. (2012), chrysafis et al. (2017), and korhonen et al. (2017). table 3 lists all bands and vegetation indices from both satellite sensors used in this study. the six bands with 30 m spatial resolution of landsat 8 were used, i.e. b2 (minimum value= 450; maximum value=515 nm), b3 (525-600 nm), bolat et al. new zealand journal of forestry science (2020) 50:4 page 3 variable min. max. mean std. dev. aspect (°) 0.00 354.09 221.98 120.30 slope (°) 4.00 49.74 26.16 12.07 elevation (m) 1264.00 1547.00 1434.53 71.66 twi 3.00 16.17 5.70 2.19 sti 0.00 0.98 0.03 0.15 ci -50.00 32.38 1.71 13.05 twi: topographic wetness index, sti: sediment transport index, ci: convergence index table 1: descriptive statistics of terrain attributes of the study area variable #n min. max. mean std. dev. cv (%) g (m2ha-1) 51 6.64 55.21 27.89 11.47 41.13 gsv (m3ha-1) 51 38.95 341.90 170.66 72.42 42.44 cv: coefficient of variation, #n: number of plots table 2. descriptive statistics of the forest inventory data b4 (630-680 nm), b5 (845-885 nm), b6 (1560-1660 nm) and b7 (2100-2300 nm). the twelve bands of sentinel-2 were used, i.e. b2 (458-523 nm), b3 (543-578 nm), b4 (650-680 nm), b8 (785-900 nm) with 10 m resolution, b5 (698-713 nm), b6 (733-748 nm), b7 (773-793 nm), b8a (855-875 nm), b11 (1565-1655 nm), b12 (21002280 nm) with 20 m resolution, and b9 (930-950 nm), and b10 (1365-1385 nm) with 60 m spatial resolution. statistical and geostatistical analysis multiple linear regression (mlr) mlr has been used to assess the relationships between forest structural attributes and remotely sensed data (næsset 2002). we focused on the predictability of g and gsv using mlr and regression kriging (rk) based on landsat 8, sentinel-2 data, and terrain indices. for this purpose, the statistical significance of the terrain variables and of the variables obtained from landsat 8 and sentinel-2 were tested at 0.05 significance level using the forward variable selection technique in sas® software. (4) where; b0 is a constant coefficient, bi is the vector of independent variables, xi is an independent variable that accounts for variation of a dependent variable and e is the model residual. the model residuals are expected to have a normal distribution. regression kriging (rk) the sample plots in the same location are inherently interdependent (spatially correlated). therefore, in fitting a model, it is highly important to consider the spatial autocorrelation in data in order to improve the model performance. rk uses the spatial autocorrelation in the residuals from mlr, and therefore may improve the predictions, as mlr does not consider spatial autocorrelation in data. kriging methods have been increasingly used for interpolating spatially dependent data in recent decades. rk consists of three steps. initially, mlr is carried out to estimate the regression parameters. then, the residuals from mlr are incorporated into ordinary kriging to account for prediction uncertainty using arcgis® software. finally, the values of the target variable were calculated by adding the predictions of mlr and the kriged values of the residuals (odeh et al. bolat et al. new zealand journal of forestry science (2020) 50:4 page 4 1995) (figure 2). (5) where: φ0 and φi are the estimated values by mlr, xi is an independent variable that accounts for dependent variable variation, n is the number of observations, λi is ordinary kriging weight and, ε is the kriged model residual at measurement locations. predictive performance and validation of mlr and rk the repeated k-fold cross-validation with n repetition procedure was used to adequately exploit our small sample data (n=51) using sas® software. five-fold with 10 repetitions was applied in the cross-validation procedure, which allowed to reduce prediction biases. the data were randomly split into two groups (training and testing subsamples) using the unrestricted random sampling method at each repetition, and the different random subsamples were selected for the training and testing purposes. this process was repeated ten times for determining the best model parameters at the 0.05 significance level. finally, the best predictive mlr and rk models were compared using goodness-of-fit statistics including the root-mean-square error (rmse), the mean absolute error (mae) and the adjusted coefficient of determination (r2adj) (kozak & kozak 2003) (eq. 6 to 8). the similarities between the observed and the predicted values were assessed by the correlation coefficient (r). rk and mlr were further evaluated by plotting the residuals for each model. also, the distribution of the relative error percent, re (%), was used for model comparison. satellite indices satellite indices satellite indices satellite indices la nd sa t 8 b1-b7, b9 se nt in el -2 s1-s12, s8a se nt in el -2 msrre1 se nt in el -2 nlinirn2 ndvi ndvi msrre2 rsrre1 evi dvi msrnirn1 rsrre2 savi msrnir msrnirn2 rsrnirn1 msavi nlinir nlire1 rsrnirn2 ndmi rsrnir nlire2 evire1 nbr evinir nlinirn1 evire2 nbr2 ctvi table 3. bands (b) and vegetation indices of landsat 8 and sentinel-2 considered as explanatory variables for g and gsv estimation. for calculations of spectral indices see chrysafis et al. (2017) where; di and di are the observed and the predicted values, respectively, di is the mean of the observed values, and nt and k are the total number of the observations and independent variables, respectively. results we used multiple linear regression (mlr) to predict basal area (g) and growing stock volume (gsv) based on landsat 8 and sentinel-2 data and terrain indices choosing explanatory variable(s) at the 0.05 level of significance. the band values and vegetation indices of landsat 8 and sentinel-2 (table 3) and terrain variables (table 1) were analyzed to predict g and gsv. when landsat 8 and the terrain variables were considered, evi, elevation and sti independent variables were significant at the 0.05 level of significance. when sentinel-2 and the terrain variables were considered, nlinirn2, elevation and sti were found to be significant at p<0.05 (table 5). goodness-of-fit statistics for mlr and rk are given in table 4. rk performed better than mlr for both prediction and validation datasets, particularly when using rk based on landsat 8 data. in both data sets, the models based on sentinel-2 data performed worse than those based on landsat 8. in summary, the prediction and validation statistics suggested using rk based on landsat 8 data to predict g and gsv (table 4). residual plots for g and gsv are displayed in figure 3. the residual patterns of rk and mlr had no trend, suggesting that the predictions were unbiased for all cases. figure 4a shows the relative errors (res) for the model predictions against different values of g and gsv. the number of observed data used in the predictions is an important factor determining modeling performance or prediction quality. interestingly, in general, figure 4a indicated that the models performed better at g ≤20 and g ≥40 m2ha-1, while the number of observed data were lower at these gs. however, we believe that those lower re values for g predictions at ≤20 and ≥40 m2ha-1 could be biased. figure 4b shows a similar case for the gsv models (the number of observed data are not shown in the figure). a graph of the observed versus modelled (predicted) gs, are shown in figure 5. these graphics suggest that g estimates obtained from landsat 8 showed greater similarity to the observations than those obtained from sentinel-2 (r=0.72). in contrast, gsv estimates obtained from sentinel-2 were more precise than those obtained from landsat 8 (r=0.75). the experimental semivariograms for the residuals of g and gsv are shown in figure 6. nugget values for landsat 8 are lower than those for sentinel-2. the nugget values are 70% for both g and gsv when using landsat 8 and 70% for g and 65% for gsv when using sentinel-2. cambardella et al. (1994) proposed that a variable with a nugget ratio <25% is assumed to be strongly spatially dependent, between 25% and 75% moderately spatially bolat et al. new zealand journal of forestry science (2020) 50:4 page 5 ̭ ̵ figure 2: flowchart of regression kriging application used to spatially interpolate g and gsv dependent, and >75% weakly spatially dependent. the nugget value was quite high in our research field, suggesting small-scale variabilities in the study area, in other words, a weak autocorrelation in our data. the surface maps of g and gsv obtained by rk are shown in figure 7. discussion the goodness-of-fit-statistics suggested that rk models were adequate for both g and gsv predictions. although the residual distributions showed that rk models were unbiased for g and gsv estimates, the relative error distributions suggested that rk models were biased for lower and higher values of g and gsv. since plots within the same sampling unit are inherently correlated, the assumption of independence of observations is generally violated in the ordinary least square methods (e.g. multivariate regression) (gregorie 1987). this feature results frequently in large error variance in residuals achieved by mlr (fortin et al. 2007). in this study, the small error variance occurred due to low spatial dependence in the residuals bolat et al. new zealand journal of forestry science (2020) 50:4 page 6 prediction set (n=30) validation set (n=21) rmse mae rmse mae g landsat 8 mlr 9.23 7.22 0.34 12.44 8.31 0.28 rk 7.84 5.89 0.52 10.30 8.98 0.28 sentinel-2 mlr 9.08 7.36 0.36 17.11 9.44 0.15 rk 8.40 6.62 0.45 9.79 7.97 0.12 gsv landsat 8 mlr 57.39 45.24 0.36 73.93 51.49 0.14 rk 49.68 37.54 0.52 62.86 53.32 0.31 sentinel-2 mlr 59.55 46.85 0.31 103.37 58.77 0.15 rk 47.47 37.20 0.56 68.44 57.35 0.17 table 4. the goodness-of-fit statistics of the fitted mlr and rk models for g (m2 ha-1) and gsv (m3 ha-1) in the anatolian black pine forest. note: bold numbers indicate best g and gsv models response variable parameter estimate s.e. t-value p-value ga intercept 122.548 35.196 3.482 0.002 evi 104.102 29.968 3.474 0.002 sti -22.835 8.472 -2.695 0.013 elevation -0.091 0.024 -3.835 0.001 gsva intercept 766.719 221.901 3.455 0.002 evi 690.130 188.942 3.653 0.001 sti -137.790 53.416 -2.580 0.016 elevation -0.581 0.149 -3.891 0.001 gb intercept -49155.840 16092.504 -3.055 0.005 nlinirn2 49282.437 16090.139 3.063 0.005 sti -18.741 10.135 -1.849 0.076 elevation -0.055 0.026 -2.130 0.043 gsvb intercept -212171.607 86662.410 -2.448 0.022 nlinirn2 213198.437 86702.293 2.459 0.022 sti -136.718 59.549 -2.296 0.031 elevation -0.532 0.166 -3.214 0.004 table 5. parameter estimates, standard errors (s.e.), t-values, and p-values of rk models predicting basal area (g, m2 ha-1) and growing stock volume (gsv, m3 ha-1) as a function of landsat 8, sentinel-2 data and terrain variables for the anatolian black pine a landsat 8 data; b sentinel-2 data r2adj r 2 adj (figure 6). while mlr may be more suitable in the case of low spatial dependency in the data, it does not take into account the spatial structure in data (palmer et al. 2009). rk has the capability to account for these structures in continuous variables and has the advantages of the variable selection (odeh et al. 1995). therefore, we recommend the use of rk thus providing the surface map of the predicted values and accounting for measurement uncertainties. bolat et al. new zealand journal of forestry science (2020) 50:4 page 7 the study area is located in the anatolian steppe transitional zone, covered by a semi-arid ecosystem (göl & abay 2003). in this region, tree (or stand) growth is limited due to the ecological conditions such as low precipitation, extreme temperature, and also poor soils. in this type of ecosystems, the general directorate of forestry of turkey frequently recommends the anatolian black pine for afforestation (konukcu 2001). however, the harsh environmental conditions may lead figure 3. residuals for g and gsv against predicted values from the rk and mlr models based on landsat 8 and sentinel-2 data figure 4. line plots of relative errors, re (%), of rk and mlr based on landsat 8 and sentinel-2 data for: (a) basal area (g); and (b) growing stock volume (gsv). table 2: confusion matrix bolat et al. new zealand journal of forestry science (2020) 50:4 page 8 figure 5. scatterplots of observed and predicted values of g and gsv using rk based on landsat 8 and sentinel-2 data figure 6. experimental (circles) and theoretical (lines) semivariograms for basal area (g) and growing stock volume (gsv) based on landsat 8 and sentinel-2 data to highly variable stands. therefore, the explanatory variables (e.g. ndvi) are likely to be less correlated with the response variable, as suggested by our study. in our view, the forest vegetation structure limited the performance of rk, as the study site was composed of poor growth and thinned patches. in other words, sample plots with similar diameters at breast height may differ in tree heights ranging from 20 to 24 m (8.2%), 15 to 19 m (57.4%) and 10 to 14 m (34.4%). these conditions led to poor correlations between the spectral data and the ground measurements. another reason for suboptimal predictions was the thinning treatments in the study area. since the stands were partially thinned, the sample plots with similar g showed the differences in terms of crown closure and stem density, which results in a discrepancy between the spectral data and the stand characteristics, leading to imprecise and biased models. thinning treatments and harsh environmental conditions led to the spatial discontinuities, suggesting a weak spatial dependency of the g and gsv in the study area. in other words, the high short-range variability as evidenced by high nugget effect (~70%) occurred in the data (dai et al. 2014; gilbert & lowell 1997; gunnarsson et al. 1998; maselli & chiesi 2006; viana et al. 2012). the nugget effect is one of the key factors decreasing the interpolation quality by kriging (isaaks & srivastava 1989). our results suggest that a systematic sampling (200 x 200 m grid) resulted in a high nugget effect in a semi-arid region (nugget effect ~70% for our data). in this context, the results of this study are important, evidencing that the sampling distance and scheme used in national forest inventory is not proper for geostatistical studies and that additional random subsamples are needed in finer resolution for a successful geostatistical analysis and interpolation. in these types of forests, destan et al. (2013) and corona et al. (2014) advised that inventory should be carried out with smaller sampling distances and that including more auxiliary variables representing spatial variability (e.g. soil and topography) may improve the performance of the kriging methods. in this study, aspect, slope, elevation, twi, sti, and ci were used as explanatory variables to improve the predictions. along with evi indices of landsat 8, nlinirn2 indices of sentinel-2, elevation and sti contributed to explain the total variance of g and gsv at the 0.05 significance level. consequently, our results are promising as rk explained approximately fifty percent of total variance observed in both g and gsv obtained from landsat 8 and sentinel-2. conclusions evi obtained from landsat 8, nlinirn2 obtained from sentinel-2, elevation and sti were the best independent variables explaining g and gsv. rk performed adequately to predict g from landsat 8 and gsv from sentinel-2. rk and mlr had similar residual scatterplots. however, the relative error distributions showed that especially in gsv estimates rk based on landsat 8 performed better than mlr based on both landsat 8 and sentinel-2. a highly random distribution of g and gsv occurred in our study site as shown by high nugget variance, suggesting a weak autocorrelation from a geostatistical perspective, which might be attributed to the coexistence in short distances of high-low productive patches, the coarse sampling scheme and undefined thinning practices. this suggests the sampling scheme may have been inadequate to detect the autocorrelation in g and gsv. across such sites, a geostatistical sampling scheme should be performed with shorter sampling distances to improve modeling of semivariograms around the origin, which may help to decrease the nugget effect. sampling sites having equal distributions of site index values can increase the performance of mlr and rk. the sampling sites with thinning practices should be excluded to improve the variance explained. in conclusion, we found that reasonably accurate predictions for forest planning can be achieved using landsat 8 and sentinel-2 through the rk method. list of abbreviations dbh: diameter breast height g: basal area gsv: growing stock volume mae: mean absolute error mlr: multiple linear regression r2adj: adjusted coefficient of determination rd: relative density re: relative error rk: regression kriging rmse: root-mean-square error competing interests the authors have no competing interests. authors’ contributions all authors participated in the forest inventory. fb and sb analysed the data, interpreted the results, and wrote the entire paper. bolat et al. new zealand journal of forestry science (2020) 50:4 page 9 figure 7. the surface maps of g and gsv predictions from the rk model based on landsat 8 and sentinel-2 data acknowledgements the authors thank dr. sabit erşahin and two anonymous reviewers for his valuable comments which helped to improve the quality of the manuscript. also, the authors thank nesrin bolat and rabia kalkan for helping to collect data in the field. this study was supported by çankırı karatekin university (project no.: of090316b05). references anonymous. (2008). forest management guidelines, republic of turkey, general directorate of turkey, forest management and planning department. ankara. barbati, a., marchetti, m., chirici, g., & corona, p. (2014). european forest types and forest europe sfm indicators: tools for monitoring progress on forest biodiversity conservation. forest ecology and management, 321, 145-157. https://doi. org/10.1016/j.foreco.2013.07.004 cambardella, c., moorman, t., parkin, t., karlen, d., novak, j., turco, r., & konopka, a. (1994). fieldscale variability of soil properties in central iowa soils. soil science society of america journal, 58(5), 1501-1511. https://doi.org/10.2136/ sssaj1994.03615995005800050033x chirici, g., barbati, a., corona, p., marchetti, m., travaglini, d., maselli, f., & bertini, r. (2008). non-parametric and parametric methods using satellite images for estimating growing stock volume in alpine and mediterranean forest ecosystems. remote sensing of environment, 112(5), 2686-2700. https://doi. org/10.1016/j.rse.2008.01.002 chrysafis, i., mallinis, g., siachalou, s., & patias, p. (2017). assessing the relationships between growing stock volume and sentinel-2 imagery in a mediterranean forest ecosystem. remote sensing letters, 8(6), 508-517. https://doi.org/10.1080/215070 4x.2017.1295479 contreras, m.a., affleck, d., & chung, w. (2011). evaluating tree competition indices as predictors of basal area increment in western montana forests. forest ecology and management, 262(11), 1939-1949. https://doi.org/10.1016/j.foreco.2011.08.031 corona, p., fattorini, l., franceschi, s., chirici, g., maselli, f., & secondi, l. (2014). mapping by spatial predictors exploiting remotely sensed and ground data: a comparative design-based perspective. remote sensing of environment, 152, 29-37. https://doi.org/10.1016/j.rse.2014.05.011 curtis, r.o. (1982). a simple index of stand density for douglas-fir. forest science, 28(1), 92-94. çolak, a.h., & rotherham, i.d. (2006). a review of the forest vegetation of turkey: its status past and present and its future conservation. paper presented at the biology and environment: proceedings of the royal irish academy. https://doi.org/10.3318/ bioe.2006.106.3.343 dai, f., zhou, q., lv, z., wang, x., & liu, g. (2014). spatial prediction of soil organic matter content integrating artificial neural network and ordinary kriging in tibetan plateau. ecological indicators, 45, 184-194. https://doi.org/10.1016/j.ecolind.2014.04.003 destan, s., yilmaz, o., & sahin, a. (2013). making objective forest stand maps of mixed managed forest with spatial interpolation and multi-criteria decision analysis. iforest-biogeosciences and forestry, 6(5), 268. https://doi.org/10.3832/ifor0099-006 dos reis, a.a., carvalho, m.c., de mello, j.m., gomide, l.r., ferraz filho, a.c., & junior, f.w.a. (2018). spatial prediction of basal area and volume in eucalyptus stands using landsat tm data: an assessment of prediction methods. new zealand journal of forestry science, 48(1), 1. https://doi. org/10.1186/s40490-017-0108-0 erşahin, s., bilgili, b.c., dikmen, ü., & ercanli, i. (2016). net primary productivity of anatolian forests in relation to climate, 2000–2010. forest science, 62(6), 698-709. https://doi.org/10.5849/ forsci.15-171 fortin, m., daigle, g., ung, c.-h., bégin, j., & archambault, l. (2007). a variance-covariance structure to take into account repeated measurements and heteroscedasticity in growth modeling. european journal of forest research, 126(4), 573-585. https://doi.org/10.1007/s10342-007-0179-1 franco-lopez, h., ek, a.r., & bauer, m.e. (2001). estimation and mapping of forest stand density, volume, and cover type using the k-nearest neighbors method. remote sensing of environment, 77(3), 251-274. https://doi.org/10.1016/s00344257(01)00209-7 gebreslasie, m., ahmed, f., & van aardt, j. (2008). estimating plot-level forest structural attributes using high spectral resolution aster satellite data in even-aged eucalyptus plantations in southern kwazulu-natal, south africa. southern forests: a journal of forest science, 70(3), 227-236. https:// doi.org/10.2989/sf.2008.70.3.6.667 gilbert, b., & lowell, k. (1997). forest attributes and spatial autocorrelation and interpolation: effects of alternative sampling schemata in the boreal forest. landscape and urban planning, 37(3-4), 235-244. https://doi.org/10.1016/s0169-2046(97)800072 göl, c., & abay, g. (2003). researches on general soil properties and nature plant species of plantatian area of cankın forestry faculty of ankara university in turkish. journal of south-west anatolia forest research institute, 5(211). gregorie, t.g. (1987). generalized error structure for forestry yield models. forest science, 33(2), 423444. gunnarsson, f., holm, s., holmgren, p., & thuresson, t. (1998). on the potential of kriging for forest management planning. scandinavian journal of bolat et al. new zealand journal of forestry science (2020) 50:4 page 10 forest research, 13(1-4), 237-245. https://doi. org/10.1080/02827589809382981 ingram, j.c., dawson, t.p., & whittaker, r.j. (2005). mapping tropical forest structure in southeastern madagascar using remote sensing and artificial neural networks. remote sensing of environment, 94(4), 491-507. https://doi.org/10.1016/j. rse.2004.12.001 isaaks, e.h., & srivastava, r.m. (1989). an introduction to applied geostatistics. oxford university press, 575 p. kalıpsız, a. (1963). türkiye’de karaçam meşcerelerinin tabii bünyesi ve verim kudreti üzerine araştırmalar. ogm yayınları, i̇stanbul. konukcu, m. (2001). ormanlar ve ormancılığımız, devlet planlama teşkilatı yayını, no; 2630, ankara. korhonen, l., packalen, p., & rautiainen, m. (2017). comparison of sentinel-2 and landsat 8 in the estimation of boreal forest canopy cover and leaf area index. remote sensing of environment, 195, 259-274. https://doi.org/10.1016/j. rse.2017.03.021 kozak, a., & kozak, r. (2003). does cross validation provide additional information in the evaluation of regression models? canadian journal of forest research, 33(6), 976-987. https://doi. org/10.1139/x03-022 kuehne, c., weiskittel, a.r., wagner, r.g., & roth, b.e. (2016). development and evaluation of individual tree-and stand-level approaches for predicting spruce-fir response to commercial thinning in maine, usa. forest ecology and management, 376, 84-95. https://doi.org/10.1016/j. foreco.2016.06.013 lu, d., mausel, p., brondızio, e., & moran, e. (2004). relationships between forest stand parameters and landsat tm spectral responses in the brazilian amazon basin. forest ecology and management, 198(1-3), 149-167. https://doi.org/10.1016/j. foreco.2004.03.048 magnussen, s., mauro, f., breidenbach, j., lanz, a., & kändler, g. (2017). area-level analysis of forest inventory variables. european journal of forest research, 136(5-6), 839-855. https://doi. org/10.1007/s10342-017-1074-z mallinis, g., koutsias, n., makras, a., & karteris, m. (2004). forest parameters estimation in a european mediterranean landscape using remotely sensed data. forest science, 50(4), 450-460. maselli, f., & chiesi, m. (2006). evaluation of statistical methods to estimate forest volume in a mediterranean region. ieee transactions on geoscience and remote sensing, 44(8), 2239-2250. https://doi.org/10.1109/tgrs.2006.872074 meng, q., cieszewski, c., & madden, m. (2009). large area forest inventory using landsat etm+: a geostatistical approach. isprs journal of photogrammetry and remote sensing, 64(1), 27-36. https://doi.org/10.1016/j.isprsjprs.2008.06.006 næsset, e. (2002). predicting forest stand characteristics with airborne scanning laser using a practical two-stage procedure and field data. remote sensing of environment, 80(1), 88-99. https://doi. org/10.1016/s0034-4257(01)00290-5 o’hehir, j., & nambiar, e. (2010). productivity of three successive rotations of p. radiata plantations in south australia over a century. forest ecology and management, 259(10), 1857-1869. https://doi. org/10.1016/j.foreco.2009.12.004 odeh, i.o., mcbratney, a., & chittleborough, d. (1995). further results on prediction of soil properties from terrain attributes: heterotopic cokriging and regression-kriging. geoderma, 67(3-4), 215-226. https://doi.org/10.1016/0016-7061(95)00007-b palmer, d.j., höck, b., kimberley, m.o., watt, m.s., lowe, d.j., & payn, t.w. (2009). comparison of spatial prediction techniques for developing pinus radiata productivity surfaces across new zealand. forest ecology and management, 258(9), 2046-2055. https://doi.org/10.1016/j.foreco.2009.07.057 şenyurt, m. (2017). çnü araştırma ormanı karaçam meşcereleri için artım ve büyüme tahminlerinin yapay sinir ağları tekniği ile elde edilmesi. project no: of090316b05. turkish state meteorological service. (n.d.) retrieved from https://mgm.gov.tr/eng/forecast-cities.aspx vescovo, l., wohlfahrt, g., balzarolo, m., pilloni, s., sottocornola, m., rodeghiero, m., & gianelle, d. (2012). new spectral vegetation indices based on the near-infrared shoulder wavelengths for remote detection of grassland phytomass. international journal of remote sensing, 33(7), 2178-2195. https://doi.org/10.1080/01431161.2011.607195 viana, h., aranha, j., lopes, d., & cohen, w.b. (2012). estimation of crown biomass of pinus pinaster stands and shrubland above-ground biomass using forest inventory data, remotely sensed imagery and spatial prediction models. ecological modelling, 226, 22-35. https://doi.org/10.1016/j. ecolmodel.2011.11.027 bolat et al. new zealand journal of forestry science (2020) 50:4 page 11 effect of pre-germinative treatments on nothofagus glauca seed germination and seedling growth ángel cabello1, nicolás espinoza2, sergio espinoza3, antonio cabrera3,4, rómulo santelices3,* 1 jardín botánico chagual. santiago, chile 2 aela energía. santiago, chile 3 centro de desarrollo para el secano interior, facultad de ciencias agrarias y forestales, universidad católica del maule. talca, chile 4 vicerrectoría de investigación y postgrado, universidad católica del maule, talca, chile *corresponding author: rsanteli@ucm.cl (received for publication 22 august 2018; accepted in revised form 12 april 2019) abstract background: nothofagus glauca (phil.) krasser (nothofagaceae, “hualo”) is an endemic tree of the mediterranean zone of chile. the natural forests in this area have been severely fragmented as a result of human causes such as replacement by agricultural crops and fast-growing tree species. from 1975, these forests have declined from 900,000 ha to 145,000 ha, so it is categorised on the iucn red list as ‘vulnerable’. in restoring this ecosystem, efforts should focus, in part, on the propagation of quality stock. however, information on propagation systems is still insufficient. methods: we aimed to analyse the effect of different pre-germinative treatments and sowing times on seed germination, and seedling growth and quality. the pre-germinative treatments were: (i) cold stratification; (ii) soaking in gibberellic acid (ga3) and thiourea solution; and (iii) nursery cultivation, while the sowing times were july, august and september. results: a high germination capacity was achieved by: soaking the seeds in ga3 solution irrespective of concentration; stratifying, irrespective of period; or soaking in 7.5 mg l-1 thiourea solution, values significantly varied from that of the control treatments. the sowing time was not relevant in terms of the percentage of germination or seedling development. stratification at 5°c for 60 days produced the best quality indices for n. glauca seedlings but no significant differences were found in any of the morphological attributes tested as a result of the pre-germinative treatments. conclusions: the pre-germinative treatments significantly improved the germination and seedlings growth of n. glauca. cold stratification at 5°c for 60 days is recommended as it produced suitable seedlings for field establishment. gibberellic acid and thiourea did not produce important effects on seedling growth. our results suggest the presence of endogenous physiological dormancy of the n. glauca seeds. the results of this study provide important information on propagation and nursery techniques of n. glauca, which can be used in restoration programmes. keywords: stratification; gibberellic acid (ga3); thiourea; sowing time; seeds; hualo new zealand journal of forestry science cabello et al. new zealand journal of forestry science (2019) 49:3 https://doi.org/10.33494/nzjfs492019x34x e-issn: 1179-5395 published on-line: 05/05/2019 © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access niches and habitats for flora, fauna, and associated microbiota (arroyo et al. 1996). however, the highest human population density in chile is also concentrated in this zone and there is strong anthropogenic pressure on natural resources. this situation led to a reduction of forest cover in this region (donoso and lara 1996) and the species that the forests contain. between 1975 and 2013, it is estimated that the extent of these forests introduction the mediterranean zone of chile contains the country’s greatest diversity of native flora and fauna (myers et al. 2000). this area also supports deciduous forests that are adapted to prolonged periods of dry summers. these forests play very important roles in the conservation of water and organic soil and in the biochemical cycle of carbon, and they provide a great variety of ecological cabello et al. new zealand journal of forestry science (2019) 49:3 page 2 declined from 900,000 ha to 145,500 ha (santelices et al. 2013a; urzúa 1975). one example is nothofagus glauca (phil.) krasser (nothofagaceae) (hualo), which is an endemic species of this zone and is classified on the iucn red list as vulnerable (barstow et al. 2017), but is not currently listed as an endangered species by the chilean government (d.s. 151/2007 ministerio secretaría general de la presidencia). its natural range includes mid-elevations, both in the andes mountains and the coastal range, in the central zone of chile (santelices et al. 2013a). unfortunately, the remaining forests are frequently damaged or destroyed by wildfires, which leads to temporary changes in the soil properties (litton and santelices 2003). such changes favour the invasion of exotic species such as pinus radiata d.don, which is particularly problematic (litton and santelices 2002). nothofagus glauca forests are likely to be affected by climate change (i.e. an increase in temperatures and prolonged periods of drought) so it is important to study and manage the remaining native vegetation in the area where the species is naturally distributed, which belongs to one of the 25 biodiversity hotspot conservation areas that have been declared (myers et al. 2000). restoration of this degraded ecosystem is a priority task and, for this, it is essential to understand how to successfully propagate and cultivate nursery plants. most species in the genus nothofagus exhibit seed dormancy (cabello 1987, 2004; leon-lobos and ellis 2005; wilcox and ledgard 1983), although not all to the same extent (wardle and campbell 1976). cold stratification and treatments with growth regulators are most commonly used to overcome this type of dormancy (baskin and baskin 2014; hartmann and kester 1999). in nothofagus spp., stratification shortens the time to germination but this is highly variable between seedlots and species (wardle 1984). it has been observed that n. glauca seeds submitted to cold and humidity treatments increased their germinative capacity (donoso and cabello 1978). gibberellic acid is a plant hormone that is known to improve plant development and has been shown to break dormancy and increase germination in seeds of several genera (bewley and black 1982; bewley and black 1985; de mello et al. 2009). there is also evidence that treatment with a solution of gibberellic acid can break the state of internal dormancy of the seeds in some nothofagus species in mediterranean environments (cabello 2004; cabello et al. 2016; rocuant 1984; santelices et al. 2011). previous work has shown that either cold stratification or gibberellic acid treatments can improve seed germination of related species including nothofagus macrocarpa (a.dc.) vasq. & rodr. (cabello et al. 2016), n. alessandrii esp. (gordon and rowe 1982; rocuant 1984; santelices et al. 2011), n. antarctica (g.forst.) oerst., n. betuloides (mirb.) oerst. (gordon and rowe 1982), and n. obliqua (mirb.) oerst. (gordon and rowe 1982; rocuant 1984; rowe and gordon 1981; shafiq 1981; subiri 1997). treatment with a solution of thiourea, a chemical promotor of germination (baskin and baskin 2014), has also been recommended to break the dormancy of seeds but its effectiveness in overcoming the latent dormancy in some nothofagus (n. alessandrii, n. dombeyi, n. obliqua, n. pumilio (rocuant 1984)) is unclear. little information exists about the effects of these promotors on n. glauca seeds, and there has been no documented effect in the cultivation of plants. thus, in order to make the germination process be more efficient (i.e. having a better quality of germinated seeds in the shortest period), it is essential to carry out some type of pre-germinative treatment, to provoke physiological changes in the seed in order to accelerate germination. this can be achieved by simulating natural conditions or with the application of chemical growth regulators (bonner 2008). previous works on the cultivation of n. glauca indicated that plants must be protected from direct radiation (gonzález et al. 2009; santelices et al. 2013b). these authors suggest that plants with morphological attributes suitable for forestation can be produced using 50−65% of shade although plants of an acceptable size and of a moderate slenderness index can be obtained using shading levels ranging from 18−35%. in addition to shade, the application of a controlled-release fertiliser in doses of 7.5 to 10 g l-1 significantly improved the growth of n. glauca compared with 3−4 g l-1, which is currently being used (santelices et al. 2013b). the objective of this work was to analyse the germination process of n. glauca seeds submitted to different pre-germinative treatments and to evaluate the subsequent growth of seedlings in nursery subjected to these treatments. methods seed collection nothofagus glauca seed were collected in march 1998, near the town of empedrado (talca province, maule region, chile). the trees selected for seed collection were at least 50-m apart to increase the likelihood of possible genetic variation. the seeds were collected from 15 phenotypically similar trees, placing a plastic mesh on the ground, and they were transported to the universidad de chile nursery in santiago, where they were cleaned, characterized using a purity analysis, and the weight and moisture content were determined using standard methods (ista 2006). in addition, the viability was determined by a cut test, for which three samples from each set of 50 seeds were chosen randomly. the determined characteristics were: 98.2% purity, 7.9% moisture content, 2,087 number of seeds kg-1, and 66% viability. seeds were stored in polyethylene bags at 4°c for five months, to be germinated the following spring. pre-germination treatments and laboratory cultivation seeds were selected based on their phenological and sanitary appearance. all seeds that were damaged or showed some external anomaly were discarded. the remaining seeds were placed in distilled water for 24 hours. seeds that floated were considered non-viable so were separated and discarded. only viable seeds were used in subsequent experiments. all the treatments were carried out sequentially so that the germination process could be started on the same day, thus maintaining the same conditions for all of them. seeds soaked in distilled water for 24 hours were used as the control treatment. cold stratification − seeds were mixed with wet sand (which had been treated previously at 150°c for 60 minutes) and placed in polyethylene bags in a refrigerator at a temperature of 5°c (± 1°c) for periods of 30, 45, or 60 days. gibberellic acid − seeds were soaked for 24 hours in aqueous gibberellic acid solutions at concentrations of 25, 50, 100, 200, 400, or 800 mg l-1. thiourea − seeds were soaked for 24 hours in aqueous thiourea solutions at concentrations of 7.5, 15, or 30 g l-1. seed germination was carried out in the dark in a chamber at 20°c located at the laboratorio de semillas of the universidad de chile for 34 days. a completely randomised experimental design of fixed effects was used, with three replicates for each treatment, and 25 seeds per experimental unit. the germination process was controlled on a daily basis and, after 34 days, evaluated for germination capacity (percentage of germinated seeds with respect to the total number of seeds sown and the maximum czabator value (maximum ratio from cultivated germination percentage on day x, divided by x (czabator 1962)). in addition, the germinative energy was determined (accumulated percentage of germination on the day when the maximum value occurs); and the energy period (number of days in which the maximum value occurs). seeds were considered to have germinated when the emerging radicles were over 2-mm long. nursery cultivation using selected pre-germination treatments once the laboratory experiments were completed, the best treatments from each experiment were selected based on the level of germination (i.e., 60 days cold stratification, soaking in 800 mg l-1 gibberellic acid solution and soaking in 7.5 g l-1 thiourea solution, plus a control treatment). these treatments were applied to viable seeds, which were subsequently sown directly onto a seedbed in a nursery at the justo pastor león experimental center from the universidad de chile, located in the maule region. the characteristics of the soil used in the experiment are the following: loamy sandy texture, acidic ph (5.5); 1.58 ppm of organic matter; 5.18 ppm of nitrogen content; 13.49 ppm of phosphorus; and 17.84 ppm of potassium. sowing was carried out during the last week of september. in the nursery experiment, plants were cultivated with bare roots in 1.1-m wide seedbeds with 8 rows of plants and 33 plants per metre. during cultivation, the plants were protected with a plastic mesh of 50% shade. a randomised complete block design of fixed effects was used, with three repetitions for each treatment and 40 plants per experimental unit, obtained from the centre of the seedbed to avoid edge effects. after one vegetative growth season in the nursery (8 months), the morphological attributes of stem length (l), root-collar diameter (d), above-ground biomass (ab), below-ground biomass (rb) and total biomass (tb = ab + rb), were measured. with this information, the slenderness index (si) and the shoot to root ratio (sri) were calculated, according to the following formulae: si = l (cm) / d (mm) sri = ab (g) / rb (g) laboratory or nursery cultivation of control seeds during different seasons independent experiments were conducted to examine the effect of sowing season on germination. viable seeds obtained by soaking in distilled water for 24 hours were sown in the nursery during the last week of july, august, and september. the same experimental designs were used here as in the laboratory and nursery cultivation experiments described above. data analysis analysis of variance (anovas) and the comparisons of means were done using the glm (generalized linear model) procedure in spss for windows v.18. data were transformed when necessary to ensure the assumptions of normality and homogeneity of variance. the variables expressed as percentages were transformed into angular values prior to determining the effects of the anovas. data were transformed logarithmically in those cases where the variables showed dispersion. average values that were considered significantly different were compared using the tukey test at the 5% level. results pre-germination treatments and laboratory cultivation significant differences were observed in the germination capacity and the maximum value of the seeds as a result of the pre-germination treatments (table 1). a high germination capacity was achieved by: soaking the seeds in gibberellic acid solution irrespective of concentration; stratifying seeds at 5°c irrespective of period; or soaking the seeds in 7.5 mg l-1 thiourea solution. stratification of seeds at 5°c for 60 days also produced significant differences in the maximum value (seed speed germination). the germination capacity and maximum value decreased as the concentration of thiourea increased, with the highest concentration of thiourea decreasing germination capacity to levels comparable with untreated seeds. based on these results and in the limited amount of available seeds, three treatments (soaking the seeds in 800 mg l-1 gibberellic acid solution; stratifying seeds at 5°c for 60 days; or soaking the seeds in 7.5 mg l-1 thiourea solution) were used for the nursery experiment. cabello et al. new zealand journal of forestry science (2019) 49:3 page 3 pre-germination treatments and nursery cultivation stratification at 5°c for 60 days produced the best quality indices for n. glauca seedlings but no significant differences were found in any of the morphological attributes tested as a result of the pre-germinative treatments used (table 2). cultivation of control seeds during different seasons the highest germination of control seeds in the laboratory (73%) occurred in july (table 1) but the energy level was much higher than that of the pregermination treatments and even with untreated seeds. the percentage germination decreased significantly for later sowing times. the only significant effect of sowing time on nursery-grown plants was on root-collar diameter (table 2). discussion the results obtained in this study indicate that n. glauca exhibits internal dormancy. this is consistent with findings reported by others for the same species (donoso and cabello 1978; santelices et al. 2013a; santelices et al. 1996), and for other nothofagus species from south america (arana et al. 2015; cabello 2004; cabello et al. 2016; donoso 2013; donoso et al. 2013), and australasia (fountain and outred 1991; wardle and campbell 1976; wilcox and ledgard 1983). the reported germination percentages differ but this may be explained by differences in season of collection, the mast seeding habit of the species, and the geographical origin of seeds. santelices et al. (2017) pointed out that patterns of seed germination in n. glauca are strongly influenced by provenance variability and suggested a potential cabello et al. new zealand journal of forestry science (2019) 49:3 page 4 treatment germinative capacity (%) maximum value (% per day) germinative energy (%) energy period (days) pre-germinative treatment in the laboratory (sowing time september): control 18.7 ± 1.3 c 0.60 ± 0.03 de 17.3 ± 1.3 29 ± 2.9 stratification 30 days 80.0 ± 2.9 a 7.62 ± 0.70 bc 37.3 ± 5.8 5 ± 1.0 stratification 45 days 72.0 ± 14.4 ab 8.78 ± 3.11 b 33.3 ± 13.5 4 ± 0.3 stratification 60 days 94.7 ± 3.5 a 27.33 ± 1.33 a 54.7 ± 2.6 2 ± 0.0 gibberellic acid 25 mg l-1 86.7 ± 1.3 a 3.28 ± 0.28 cde 65.3 ± 10.4 20 ± 1.4 gibberellic acid 50 mg l-1 85.3 ± 2.6 a 3.69 ± 0.20 bcde 60.0 ± 2.3 16 ± 0.6 gibberellic acid 100 mg l-1 92.0 ± 2.3 a 3.75 ± 0.38 bcde 69.0 ± 5.8 19 ± 1.3 gibberellic acid 200 mg l-1 96.0 ± 4.0 a 4.39 ± 0.38 bcde 78.7 ± 7.4 18 ± 2.9 gibberellic acid 400 mg l-1 84.0 ± 6.1 a 3.85 ± 0.75 bcde 56.0 ± 10.5 16 ± 3.7 gibberellic acid 800 mg l-1 97.3 ± 2.6 a 5.43 ± 0.19 bcd 81.3 ± 3.5 15 ± 0.5 thiourea 7.5 g l-1 92.0 ± 4.0 a 4.81 ± 0.52 bcde 58.7 ± 3.5 13 ± 2.1 thiourea 15 g l-1 50.7 ± 2.6 b 2.55 ± 0.21 cde 42.7 ± 9.6 16 ± 2.6 thiourea 30 g l-1 5.3 ± 3.5 c 0.18 ± 0.12 e 5.3 ± 3.5 31 ± 1.6 sowing times for control seeds in the nursery: july 73.3 ±11.6 a 0.84 ± 0.13 a 72.5 ± 11.4 86 ± 1.2 august 59.2 ± 6.0 ab 0.70 ± 0.06 a 58.3 ± 5.4 84 ± 1.2 september 24.2 ± 5.0 b 0.47 ± 0.10 a 24.2 ± 5.0 52 ± 1.4 table 1. effect of pre-germinative treatments and sowing times on the germination of nothofagus glauca (means ± s.e. values with the same letter are not significantly different, p < 0.05). cabello et al. new zealand journal of forestry science (2019) 49:3 page 5 tr ea tm en t l (c m ) d ( m m ) a b ( g) r b ( g) tb ( g) q ua lit y in de x si sr i pr ege rm in at iv e tr ea tm en t (s ow in g ti m e se pt em be r) : co nt ro l 9 .9 ± 1 .2 b 2. 96 ± 0 .3 1 b 0. 18 ± 0 .0 4 b 0. 21 ± 0 .0 1 a 0. 40 ± 0 .0 5 b 3. 4 ± 0. 3 b 0. 9± 0. 2 b st ra ti fic at io n 60 d ay s 22 .1 ± 2 .4 a 4. 72 ± 0 .4 4 a 0. 87 ± 0 .1 4 a 0. 64 ± 0 .1 6 a 1. 52 ± 0 .3 0 a 4. 7 ± 0. 1 a 1. 7± 0. 1 a g ib be re lli c ac id 8 00 m g l1 16 .1 ± 2 .5 a b 4. 06 ± 0 .2 5 ab 0. 60 ± 0 .1 3 ab 0. 56 ± 0 .1 3 a 1. 16 ± 0 .2 6 ab 3. 9 ± 0. 3 ab 1. 0± 0. 1 b th io ur ea 7 .5 g l -1 17 .2 ± 2 .4 a b 3. 78 ± 0 .2 7 ab 0. 57 ± 0 .1 3 ab 0. 47 ± 0 .1 2 a 1. 04 ± 0 .2 5 ab 4. 5 ± 0. 3 ab 1. 2± 0. 0 ab so w in g ti m es fo r co nt ro l se ed s: ju ly 8 .0 ± 0 .2 a 2. 11 ± 0 .0 2 b 0. 08 ± 0 .0 0 a 0. 16 ± 0 .0 1 a 0. 25 ± 0 .0 2 a 3. 8 ± 0. 1 a 0. 6 ± 0. 05 a a ug us t 7 .9 ± 0 .9 a 2. 09 ± 0 .0 8 b 0. 08 ± 0 .1 6 a 0. 16 ± 0 .0 2 a 0. 25 ± 0 .0 4 a 3. 9 ± 0. 6 a 0. 5 ± 0. 04 a se pt em be r 9 .9 ± 1 .2 a 2. 96 ± 0 .3 1 a 0. 18 ± 0 .0 4 a 0. 21 ± 0 .0 1 a 0. 40 ± 0 .0 5 a 3. 4 ± 0. 3 a 0. 9 ± 0. 2 a ta b le 2 . e ffe ct o f p re -g er m in at iv e tr ea tm en ts a nd s ow in g ti m es o n th e m or ph ol og ic al a tt ri bu te s an d qu al it y in di ce s of n ot ho fa gu s gl au ca s ee dl in gs ( m ea ns ± s .e . v al ue s w it h th e sa m e le tt er a re n ot s ig ni fic an tly d iff er en t, p < 0. 05 ). l= s te m le ng th , d = r oo tco lla r di am et er , a b = a bo ve -g ro un d bi om as s, r b = b el ow -g ro un d bi om as s, t b = to ta l b io m as s, s i= s le nd er ne ss in de x, s r i= s ho ot to r oo t r at io . capacity of the species to adapt to climate. dormancy and germination are important constraining factors that should be considered in restoration and reforestation programmes (pérez-fernández and gómez-gutiérrez 2007). for those species with seeds that show dormancy, it is advisable to develop effective strategies to ensure a greater and more homogenous percentage of germination and uniform seedling development. previous work on related species (n. macrocarpa, n. alessandrii, n. obliqua) showed that cold stratification at 4°c (± 1°c) for 30, 45 or 60 days produced the greatest increase in germination percentage (cabello et al. 2016; santelices et al. 2011; subiri 1997) while cold stratification at 5°c (± 1°c) for periods between 30 and 60 days were equally effective for n. glauca. although cold stratification produced the best results, it requires more time as well as facilities that maintain cold conditions, so treatment with gibberellic acid or low levels of thiourea may be more cost-effective alternatives. cold stratification for 60 days was the most efficient treatment tested as almost 55% of germinative energy was attained in an energy period of two days. some doses of growth regulators generated lower maximum values than cold stratification but had higher germinative energy average values although with a longer energy period. in this context, the degree of difficulty of the treatments should be evaluated (ga3 at concentrations higher than 100 mg l-1 are effective for the promotion of germination). in no case should the seeds be treated with doses equal to or greater than 30 g l-1 of thiourea, since concentrations greater than 30 g l-1 could negatively affect seed germination and seedling growth (hartmann and kester 1999). although thiourea facilitates germination, its application in high elevated doses is not always accompanied by the necessary processes for normal radicle activity or for a normal cell division process, as it has been observed to inhibit the synthesis of dna (rodríguez et al. 1983). evaluating the effect of sowing season in terms of germinative capacity and maximum value of control seeds suggests that july and september may be the best months for sowing n. glauca seeds, although this result needs further research. however, the average germinative capacity observed was lower and the energy period was higher compared with the use of the pregerminative treatments of cold stratification or soaking in gibberellic acid or thiourea. it is known that growth regulators are involved in the mechanisms that control the induction and breaking of dormancy (leadem 1987) and with the exogenous application of these germination promoters, the dormancy is overcome quicker. because of the variable masting/fruiting nature of the species, and due to the low seed availability during the season when the experiment was performed, only control seeds were used to test the effect of sowing season; however, future research should include the effect of different pre-germination treatments at each one of the sowing seasons. cold stratification is one of the most used treatments to break the internal dormancy of seeds of forest tree species (baskin and baskin 2014; bonner 2008). a higher germinative percentage is obtained by stratifying the seeds for 60 days and the effect of this treatment is also reflected in seedling growth. all the morphological attributes we evaluated, with the exception of root biomass, were significantly different from the plants whose seeds did not receive treatment. seedling height after one growing season (22.1 cm) was not exceptional but the diameter growth was 4.7 mm, which generated a slenderness index within the range suggested for species (like n. glauca) that live in mediterranean environments (villar-salvador 2003). by observing the biomass partition and the shoot to root ratio, it is evident that the plants from this treatment tend to invest more in aboveground biomass than in below-ground biomass. there is not enough information about what sri ranges may be optimal for the different mediterranean forest tree species and there is no universal sri associated with an optimal development of the plantations, but each species will have a range of optimal values (villar-salvador 2003). thomson (1985) suggests a well-balanced sri system for good seedling field performance, however, this contention must be taken with care for n. glauca. therefore, it can be considered that the produced seedlings have desirable functional characteristics to be established in the field, although this hypothesis needs further research. gibberellic acid is a growth regulator that has been widely used as a promoter of seed germination in species with physiological dormancy (baskin and baskin 2014; bonner 2008). in nothofagus betuloides and n. alessandrii, it has been successfully used to break dormancy (martínez-pastur et al. 1994; santelices et al. 2011). the latter authors, similar to that reported in our study, suggested low doses and short immersion times. however, there is a risk that plants with unwanted characteristics can be produced (cabello 2004; rascio et al. 1998), and this was observed in the first developmental stages of cultivated seedlings (e.g. gigantism, expressed by very elongated and succulent internodes). nevertheless, there were no significant differences from the other pre-germinative treatments tested here, so pre-treatment with gibberellic acid could have a beneficial effect on the development of nursery seedlings. a similar situation was observed with the thiourea pre-treatment. in both cases, the morphological attributes of the seedlings obtained indicated that these plants also have suitable functional characteristics for forestation or restoration even though they were smaller than plants grown from cold-stratified seeds. however, according to the slenderness index, they could be somewhat robust, therefore they could have a lower growth capacity and survival due to the eventual deterioration of their carbon balance (villar-salvador 2003). thiourea has been less widely used than gibberellic acid for breaking seed dormancy and has been reported to inhibit root production of seedlings in some species (baskin and baskin 2014). there was less root biomass production with respect to the aerial biomass but a similar tendency as for the other pre-germinative treatments was also observed. it is noteworthy that, with the control treatment, seedlings invested more in below cabello et al. new zealand journal of forestry science (2019) 49:3 page 6 than in aboveground biomass. however, this result should be interpreted with caution since the difference in aboveand belowground biomass (i.e., 0.3 mg) of this treatment was very small and has little physiological significance. the effect of sowing season for control seeds in the nursery was only significant for diameter. seedlings were short and somewhat robust, thus would not be suitable for forestation or restoration. a similar, but not significant, trend was observed with seedlings whose seeds did not receive pre-germinative treatment, that is, to invest more in root biomass than in shoot biomass (in some cases close to 50%). conclusions the results obtained in this investigation show that the seeds of n. glauca have an endogenous physiological dormancy, which can be overcome by cold stratification treatments (during 30, 45, or 60 days), with immersions in gibberellic acid (from 25 to 800 mg l-1) and in thiourea (7.5 g l-1). with sowing at the beginning of winter (july), dormancy is also broken, although the germination capacity is lower than that achieved with the other treatments. by stratifying the seeds at 5°c for 60 days, seedlings with functional attributes in order to be established in the field can be produced. however, with gibberellic acid and thiourea, no differences in the growth of the plants are observed. the sowing time did not have, in general, a significant effect on the development of the seedlings (except for the diameter) and the size of these plants is lower than that achieved with the other treatments. the results of this study provide important information on propagation and nursery techniques of n. glauca, which can be used in restoration programmes. ethics approval not applicable. consent for publication not applicable. availability of data please contact the primary author for further information. competing interests the authors declare that they have no competing interests. funding this research was carried out with funding from the universidad de chile and the universidad católica del maule. author’s contributions angel cabello designed the experiment, supervised the fieldwork, and wrote and provided critical revisions of the manuscript. nicolás espinoza installed and assessed cabello et al. new zealand journal of forestry science (2019) 49:3 page 7 the experiments, and provided critical revisions of the manuscript. sergio espinoza evaluated the experiments, conducted the statistical analysis of the data, and provided critical revisions of the manuscript. antonio cabrera evaluated the experiments and provided critical revisions of the manuscript. rómulo santelices supervised the entire research, evaluated the experiments and wrote the manuscript. all authors read and approved the final manuscript. acknowledgements the authors are grateful to the departamento de silvicultura y conservación de la naturaleza of the universidad de chile and the departamento de ciencias forestales of the universidad católica del maule for the support to carry out this research. references arana, m., gonzalez polo, m., martinez-meier, a., gallo, l., benech-arnold, r., sanchez, r., & batlla, d. (2015). seed dormancy responses to temperature relate to nothofagus species distribution and determine temporal patterns of germination across altitudes in patagonia. new phytologist, 209(2), 507−520. arroyo, m. t. k., riveros, m., peñaloza, a., cavieres, l., & faggi, a. m. (1996). phytogeographic relationships and regional richness patterns of the cool temperate rainforest flora of southern south america. in rg lawford, pb alaback, & e fuentes (eds.), highlatitude rainforests and associated ecosystems of the west coasts of the americas: climate, hydrology, ecology and conservation (pp. 134−172). new york: springer. barstow, m., rivers, m. c., & baldwin, h. (2017). nothofagus glauca. the iucn red list of threatened species 2017: e.t32034a2809142. http://dx.doi.org/10.2305/iucn.uk.2017-3.rlts. t32034a2809142.en. accessed january 2018. baskin, c. c., & baskin, j. m. (2014). seeds. ecology, biogeography, and evolution of dormancy and germination. san diego, ca, usa: academic press. bewley, j. d., & black, m. (1982). physiology and biochemistry of seeds in relation to germination. new york: springer. bewley, j. d., & black, m. (1985). seeds: physiology of development and germination. new york, ny: plenum press. bonner, f. t. (2008). seed biology. in ft bonner, & rp karrfalt (eds.), the woody plant seed manual. agriculture handbook 727 (pp. 3−37). washington, dc: united states department of agriculture, forest service. cabello, a. (1987). proyecto de protección y recuperación de especies arbóreas y arbustivas amenazadas de extinción. especies amenazadas: colecta de semillas y producción de plantas en vivero. parte i. (chile forestal). santiago, chile: corporación nacional forestal (conaf). cabello, a. (2004). nothofagus macrocarpa “roble de santiago” y las dificultades de propagación de las especies del género nothofagus. revista chagual, 2, 37−41. cabello, a., gallegos, m., & suazo, d. (2016). efecto de la temperatura de cultivo y de tratamientos pregerminativos en la germinación de semillas de nothofagus macrocarpa (a. dc.) vasqz. et rodr. (nothofagaceae). revista chagual, 14, 44−50. czabator, f. j. (1962). germination value: an index combining speed and completeness of pine seed germination. forest science, 8, 386−396. de mello, a. m., streck, n. a., blankenship, e. e., & paparozzi, e. t. (2009). gibberellic acid promotes seed germination in penstemon digitalis cv. husker red. hortscience, 44(3), 870−873. donoso, c. (2013). nothofagus leonii espinosa. (nothofagus obliqua x glauca) huala (cordillera de los andes), hualo (cordillera de la costa). familia: nothofagaceae. in c donoso (ed.), las especies arbóreas de los bosques templados de chile y argentina. autoecología (pp. 443−447). valdivia, chile: marisa cuneo ediciones. donoso, c., & cabello, a. (1978). antecedentes fenológicos y de germinación de especies leñosas chilenas. ciencias forestales, 1, 31−40. donoso, c., & lara, a. (1996). utilización de los bosques nativos en chile: pasado, presente y futuro. in jj armesto, c villagrán, & mk arroyo (eds.), ecología de los bosques nativos de chile (pp. 363−388). santiago, chile: editorial universitaria. donoso, p., donoso, c., gallo, l., azpilicueta, m., baldini, a., & escobar, b. (2013). nothofagus obliqua (mirb.) oerst. roble, pellín, hualle. familia nothofagaceae. in c donoso (ed.), las especies arbóreas de los bosques templados de chile y argentina: autoecología (segunda edición ed., pp. 471−500). valdivia, chile: marisa cuneo ediciones. fountain, d. w., & outred, h. a. (1991). germination requirements of new zealand native plants: a review. new zealand journal of botany, 29(3), 311−316. gonzález, m., quiroz, i., garcía, e., & soto, h. (2009). plantas de nothofagus glauca (phil.) krasser, ensayos de germinación y producción. chile forestal, 344, 40−44. gordon, a. g., & rowe, d. c. f. (1982). seed manual for ornamental trees and shrubs. bulletin 59. london, england: forestry commission. her majesty’s stationery office. table 2: confusion matrix hartmann, h. t., & kester, d. e. (1999). propagación de plantas, principios y prácticas (séptima reimpresión ed.). méxico: editorial continental. ista (2006). international rules for seed testing. zurich, switzerland: ista (international seed testing association). leadem, c. l. (1987). the role of plant growth regulators in the germination of forest tree seeds. plant growth regulation, 6(1), 61−93. leon-lobos, p., & ellis, r. h. (2005). seed survival in chilean nothofagus in response to desiccation and storage. seed science research, 15(2), 113−123. litton, c. m., & santelices, r. (2002). early post-fire succession in a nothofagus glauca forest in the coastal cordillera of south-central chile. international journal of wildland fire, 11(2), 115−125. litton, c. m., & santelices, r. (2003). effect of wildfire on soil physical and chemical properties in a nothofagus glauca forest, chile. revista chilena de historia natural, 76(4), 529−542. martínez-pastur, c., arena, m., & fernández, c. (1994). nota sobre la influencia del ácido giberélico y del nitrato de potasio en la germinación de semillas de nothofagus betuloides (mirb.) oerst. investigación agraria: sistesmas y recursos forestales, 3(1), 83−89. myers, n., mittermeier, r. a., mittermeier, c. g., da fonseca, g. a. b., & kent, j. (2000). biodiversity hotspots for conservation priorities. nature, 403(6772), 853−858. pérez-fernández, m., & gómez-gutiérrez, j. (2007). importancia e interpretación de la latencia y germinación de semillas en ambientes naturales. in jm rey-benayas, t espigares-pinilla, & jm nicolau-ibarra (eds.), restauración de ecosistemas mediterráneos (pp. 87−112). alcalá de henares, españa: universidad de alcalá. rascio, n., mariani, p., dalla vecchia, f., la rocca, n., profumo, p., & gastaldo, p. (1998). effects of seed chilling or ga(3) supply on dormancy breaking and plantlet growth in cercis siliquastrum l. plant growth regulation, 25(1), 53−61. rocuant, l. (1984). efecto de giberelina y de tiourea en la germinación de semillas: especies del género nothofagus. bosque, 5, 53−58. rodríguez, d., matilla, a., aldasoro, j., hernándeznistal, j., & nicoläs, g. (1983). germination of cicer arietinum seeds and thiourea-induced phytotoxicity. physiologia plantarum, 57, 267−272. rowe, d. c. f., & gordon, a. g. (1981). studies on effects of prechilling periods or gibberellins used to stimulate cabello et al. new zealand journal of forestry science (2019) 49:3 page 8 the seed germination of nothofagus obliqua and n. procera. seed sciences and technology, 9, 823−838. santelices, r., donoso, c., & cabello, a. (2013a). nothofagus glauca (phil.) krasser, hualo, roble maulino, roble colorado (maule). familia: nothofagaceae. in c donoso (ed.), las especies arbóreas de los bosques templados de chile y argentina: autoecología (segunda edición ed., pp. 433−442). valdivia, chile: marisa cuneo ediciones. santelices, r., espinoza, s., cabrera-ariza, a., peña-rojas, k., & donoso, s. r. (2013b). effect of shading and fertilisation on the development of containergrown nothofagus glauca seedlings, a threatened species from central chile. southern forests, 75(3), 145-148. santelices, r., espinoza, s., magni, c., cabrera, a., donoso, s., & peña, k. (2017). variability in seed germination and seedling growth at the intra and inter-provenance levels of nothofagus glauca (lophozonia glauca), an endemic species of central chile. new zealand journal of forestry science, 47(10), 1−9. santelices, r., navarro-cerrillo, r. m., & drake, f. (2011). propagation and seedling cultivation of the endemic species nothofagus alessandrii espinosa in central chile. restoration ecology, 19, 177−185. santelices, r., riquelme, m., & rojas, f. (1996). aspectos sobre la semilla y germinación de nothofagus glauca (phil.) krasser de dos procedencias de la vii región. ciencia e investigación forestal, 10, 297−305. shafiq, y. (1981). effect of gibberellic acid (ga3) and prechilling on germination percent of nothofagus obliqua mirb. and nothofagus procera oerst. seeds. turrialba, 31(4), 365−368. subiri, m. (1997). técnicas de propagación por semillas y de producción de plantas en vivero de nothofagus obliqua de tres procedencias. memoria de título ingeniería forestal, universidad de chile, santiago, chile. thomson, s. (1985). seedling morphological evaluation what you can tell by looking. in ml durvea (ed.), proceedings: evaluating seedling quality: principles, procedures, and predictive abilities of major tests. workshop held october 16−18, 1984 (pp. 59−70). corvallis, or, usa: forest research laboratory, oregon state university. urzúa, a. (1975). cambio de estructura en el bosque de nothofagus glauca (phil.) krasser. tesis ingeniería forestal, universidad de chile, santiago, chile. villar-salvador, p. (2003). importancia de la calidad de planta en los proyectos de revegetación. in reybenayas j.m, t espigares-pinilla, & jm nicolauibarra (eds.), restauración de ecosistemas mediterráneos (pp. 65−86). alcalá de henares, españa: universidad de alcalá. wardle, j. (1984). the new zealand beeches: ecology, utilisation and management. wellington, new zealand: new zealand forest service. wardle, p., & campbell, a. d. (1976). winter dormancy in seedlings of mountain beech (nothofagus solandri var. cliffortioides) near timber line. new zealand journal of botany, 14(2), 183−186. wilcox, m. d., & ledgard, n. j. (1983). provenance variation in the new-zealand species of nothofagus. new zealand journal of ecology, 6, 19−31. list of abbreviations l= stem length d= root-collar diameter ab= aboveground biomass rb= belowground biomass tb= total biomass si= slenderness index sri= shoot to root ratio. cabello et al. new zealand journal of forestry science (2019) 49:3 page 9 allometric biomass equations for young trees of four broadleaved species in albania erion istrefi1*, elvin toromani2, nehat çollaku2 and bashkim thaçi3 1 ministry of tourism and environment, environmental services project, blv zhan d’ark, tirana, albania. 2 agricultural university of tirana, faculty of forestry sciences,1029 koder-kamez, tirana, albania. 3university of applied sciences in ferizaj. rr.universiteti, ferizaj, kosovo. *corresponding author: erion.istrefi@yahoo.com (received for publication 5 april 2019; accepted in revised form 6 may 2019) abstract background: biomass assessment of young forest stands is important because of their role in the carbon cycling. the aim of this study was to develop biomass equations for young broadleaved species growing in natural conditions in albania. methods: five forest stands were investigated using circular sample plots. diameter at breast height (dbh) and height (h) from 58 sampled trees ranging in age from 4 to 34 years old of turkey oak (quercus cerris l.), sweet chestnut (castanea sativa mill.), european hornbeam (carpinus betulus l.) and manna ash (fraxinus ornus l.) were measured in situ. logarithmic regression equations were used and tested for their performance to estimate aboveground and tree-components biomass for each species using dbh, h and their combination dbh2 x h as predictors. results: we found that dbh was a reliable predictor for estimation of aboveground and components biomass for young trees but the inclusion of height in biomass allometry did not improve the biomass estimation. we observed differences in scale (β0) and exponent (β1) coefficients of biomass models, not only between broadleaved species, but also among treecomponents within species. both coefficients were strongly species-specific and their values reflect differences in biomass stocking rate due to different growth strategies of each species in early development phases. conclusions: allometric equations to estimate aboveground and tree-component biomass appeared to be species-specific, meaning that such models are applicable for species growing at sites with similar ecological conditions. from the tree variables used, dbh was the most reliable predictor of aboveground and individual components biomass, whereas height proved to be a promising predictor for stand biomass. these allometric equations developed for young trees will improve the accuracy of current estimates of forest carbon stock in albania. new zealand journal of forestry science istrefi et al. new zealand journal of forestry science (2019) 49:8 https://doi.org/10.33494/nzjfs492019x51x e-issn: 1179-5395 published on-line: 22 july 2019 © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access equations. such allometric models are commonly used to estimate biomass in mature forests, but these are missing for young forests. they are adequate tools to express biomass in terms of dry mass on the basis of easily measured variables. most of these equations in the albanian and international literature use diameter at breast height (dbh at 1.3 m from the ground) and tree height (h) to estimate aboveground biomass (agb) or tree-components biomass. other variables such as wood density, crown length, height-to-diameter ratio, or tree introduction during the last decades, the area of young forests in albania has been expanding due to intensive management and damage incurred by forest fires. since young forest stands play a critical role in the forest carbon cycling due to their capacity in dioxide carbon uptake in early growth phases, we urgently need to develop and improve techniques for accurate prediction of their biomass (lehtonen 2005). one of the most efficient ways to achieve this goal is to use allometric keywords: biomass model performance, carpinus betulus, castanea sativa, fraxinus ornus, quercus cerris, tree biomass istrefi et al. new zealand journal of forestry science (2019) 49:8 page 2 age are also used to improve the accuracy of allometric equations in biomass prediction. the information on biomass of forest stands is required to assess the amount of carbon stocked in albanian forests and to estimate changes in carbon stocks to meet reporting requirements of albania to the united nations framework convention on climate change (unfccc). for that reason, the development of biomass functions in each country has become an important issue and it is encouraged to be done for most relevant tree species (ipcc 2003). despite recent developments in remote sensing techniques, biomass allometric models are fundamental to estimating biomass at the tree or stand level. few studies on biomass and biomass expansion factors (befs) have been reported for forest species in albania. the first results regarding biomass estimation at country level were provided in the framework of the national forest inventory (anfi 2004). biomass was calculated using inventory data on stand volume per hectare and biomass expansion factors (befs) from other countries with similar site conditions. omuri (2006) generated allometric equations and determined befs for austrian pine (pinus nigra arn.), beech (fagus sylvatica l.) and birch (betula pendula l.) ≤50 years old in the shishtavec area, northern-east albania. toromani et al. (2011) developed allometric equations for young poplar plantations growing in eastern albania, in the region of pogradeci, based on several tree variables. istrefi et al. (2018) conducted research in several areas in eastern and central albania to assess aboveground and tree components biomass of turkey oak (quercus cerris l.) ≤0 years by using linear and non-linear allometric models. several studies on allometry have been conducted worldwide, where 279 equations have been developed for different tree species and ecological regions (zianis et al. 2005; zianis and mencuccini 2004). these biomass equations were developed for young trees with a threshold of dbh ≥10 cm omitting trees with a smaller diameter. based on our review of the literature, we noted that most biomass equations were developed for mature stands (wirth et al. 2004; joosten et al. 2004; zianis et al. 2005) and only few for young stands (dutca et al. 2010). despite the limited number of published equations, for many other forest species growing in albania the availability of allometric equations is still limited. therefore, there is still a need for improving existing biomass models and developing new ones for other forest species such as turkey oak, sweet chestnut (castanea sativa mill.), european hornbeam (carpinus betulus l.), manna ash (fraxinus ornus l.) and others. from the review in the globallometree platform (http://globallometree.org, henry et al. 2013), the only available allometric equations for young trees were for cas. sativa, and car. betulus. due to the lack of biomass equations for young broadleaved species in albania, the development of allometric models for biomass estimates remains a challenge for researchers in albania. the main objective of this study was to develop species-specific allometric equations for predicting aboveground and tree-components biomass (i.e. stem, branches, and foliage) using several tree variables and to test their accuracy and precision for biomass prediction. methods study area forests in albania cover an area of 941957 ha (anfi 2004) distributed over the whole territory. within project “assisted natural regeneration of degraded lands in albania”, the project area was stratified in four ecological zones based on climate, soil, elevation and geology (unfccc 2009). the first ecological zone comprises forest areas of mediterranean shrubs and gariga, while the second ecological zone includes mixed oak forest stands (quercus spp.) with hornbeam (carpinus betulus l.) and other species. the third ecological zone includes european box (buxus sempervirens l.) with common juniper (juniperus communis l.), while the fourth ecological zone comprises alpine vegetation areas with dwarf juniper (juniperus nana miquel.). our study was focused on the second ecological zone dominated by pure and mix oak forest stands (fig. 1). a systematic 1 x 1 km geo-referenced grid using arcgis program was established, where 19 intersections of this grid were located in the second ecological zone. eight out of 19 intersections located in young natural forest stands were selected randomly from north-east to central part of albania. the main species at the research sites were: turkey oak (quercus cerris), sweet chestnut (cas. sativa), european hornbeam (car. betulus) and manna ash (fraxinus ornus). the study areas have a distinct mediterranean climate; the mean annual temperatures vary across sites ranging from 8.8°c to 14.4°c (table 1), with extreme minimum and maximum temperatures of -14°c to 6.2°c and 23.7°c to 30°c, respectively (harris et al. 2014; www.climexp.knmi.nl). the rainfall time series indicated the presence of spatial and temporal variability in annual and seasonal precipitation sums at all sampling sites. cinnamon and dark mountain forest soils, the most common in the sampling sites, are deep with moderate fertility as well as high clay content and strong alkalinity (fao 2015). sampling methodology and procedure starting from the main grid (1 x 1 km), biomass sampling plots were established at the intersections of a 200 x 200 m grid. five sample plots were randomly selected and trees inside circular sample plots (400 m2) were measured. before conducting the destructive sampling, the diameter at breast height [dbh (cm)] of trees was measured by calliper with ±1 mm accuracy in two perpendicular directions. tree heights [h (m)], were measured by ultrasonic hypsometer vertex iii (haglöf, sweden). the position of each tree in relation to the plot centre was recorded by measuring the distance from the centre and tree azimuth. in total, three to five trees per sample plot were selected and felled at ground level during the period from june 2016 to october 2017. fifty-eight sampled trees were partitioned into the three main components (stem, branches and foliage), whereas plant roots were not investigated. the fresh weight of the stem, branches and leaves was weighed in the field using a precision scale with 50 kg capacity and accuracy ±1%. the same procedure of destructive analysis was followed at all research sites. three to five subsamples from each tree-component were stored in sealed plastic bags and then sent to the laboratory of the faculty of forestry sciences in tirana. masses of stem discs (2–3 cm thick), branches and foliage samples were weighed. subsequently, all samples were dried at 70°c temperature, until a constant weight was achieved. treecomponents biomass was assessed on the basis of fresh to oven-dry weight ratio (eq. 1) of subsamples collected from sampled trees. c = dwsubsample/gwsubsample (1) in those cases where the stem could not be weighed in the field, the diameter over bark was measured at 1-metre intervals from the stem base to the top to determine the log volumes using newton–riecke’s equation (van laar & akça 1997). the total agb of each tree was estimated by the sum of stem-wood, branches and foliage weights. development of biomass equations so far, researchers have used a variety of regression models for estimating tree and tree-components biomass. biomass data exhibit heteroscedasticity because the variance differs across observations and for that reason the power function is considered an appropriate model in our study. we transformed the observed data using logarithmic transformation which is commonly used in dimension analysis studies to fit appropriate allometric equations. the following models with this form were fitted to the sampled tree data: model 1: ln dw=ln β0+β1× ln dbh (2) model 2: ln dw=ln β0+β1 × ln h (3) model 3: ln dw=ln β0+β1× ln dbh 2 × ln h (4) model 4: ln dw=ln β0+β1ln dbh+β2 ln h (5) where dw is the aboveground dry biomass, or any of three components (stem, branches, foliage); dbh the diameter at breast height (cm); h the total tree height (m); and β0, β1 and β2 are the regression coefficients. in the biomass models 2, 3 and 4, β0 represents the scaling coefficient, while β1 is the scaling exponent. the natural logarithm transformation linearises the allometric relationships, equalises the variance over the entire range of the dataset and provides comparability with results of previous studies (zianis et al. 2005; niklas 2006). however, the transformation introduces a systematic bias into the calculation, therefore to eliminate the bias, the final result is usually multiplied istrefi et al. new zealand journal of forestry science (2019) 49:8 page 3 figure 1: location of the research sites and sampled species by a correction factor (cf) calculated from the standard error of the estimate (see) of the regression (sprugel 1983): cf=exp×(see2/2) (6) models were fitted using the regression procedure within the spss software (version 24) for windows (ibm 2006) to relate aboveground tree mass to different measures of felled tree variables including diameter at breast height (dbh), tree height (h), and their combination (dbh2 × h). biomass model fitting and evaluation model fitting was carried out in two phases. first, we eliminated non-significant models (p>0.05) and those with low values of coefficients of determination (r2 <0.5). secondly, in order to select the best equations for each species, several goodness-of-fit statistics were examined: the coefficient of determination (r2), the root mean squared error (rmse), the root mean squared prediction error (rmspe) and the mean error (me). the coefficient of determination (r2) indicates the proportion of the total variance explained by the model, whereas the other fitting statistics analyse the accuracy of the biomass estimates. the smaller their values, the better is the biomass model prediction performance (zeng & tang 2011). we also applied the reduced major axis (rma) regression, because the observed data are subject to errors. rma is more appropriate than standard ordinary least squares (ols) regression when the independent variable is measured with error (sokal & rohlf 2012). the tree variables (dbh or h) used to predict aboveground biomass are subject to natural variation and measurement errors (especially h) and this fact cannot be neglected (niklas 2006; kaitaniemi 2004). the scaling exponent of the rma regression analysis was estimated using the equation 7: brma=β1/ryx (7) where brma is the scaling exponent based on the rma model, β1 is the scaling exponent estimated by the least istrefi et al. new zealand journal of forestry science (2019) 49:8 page 4 square model regression (ols) applied to eqs. (2–4), and ryx is the correlation coefficient determined from the least square regression (henry & aarssen 1999). uncertainty of biomass models uncertainty is defined as the lack of knowledge of the true value which can be described as a probability density function characterising the range and likelihood of possible values of aboveground or tree-components biomass. it depends on the quality and quantity of applicable data as well as on the predictive ability of allometric equation to estimate biomass. we evaluated the reliability of the biomass models in terms of differences between the estimates obtained from allometric equation and the true value. for that reason, the biomass models developed for each species were used to predict the aboveground and treecomponents biomass using sampled tree variables as predictors. since the biomass equations were directly applied to sampled tree data of the investigated forest species, the components of the errors accounted for are the sampling and biomass model errors (assumed to be small). the key approach is that uncertainty in biomass estimates can be propagated from uncertainties in the biomass data and allometric equation prediction accuracy and its estimation is based on the error propagation equation (bevington and robinson 1992). using this interpretation, a simple equation is used for the uncertainty of the sum, expressed in percentage terms: (8) where: u is the percentage uncertainty of aboveground biomass in the sum of the quantities (half the 95 percent confidence interval divided by the total (i.e., mean) and expressed as a percentage), dwst, dwbr, dwf are the uncertain quantities of stem, branches and foliage dry weight (kg) predicted by allometric equations, and ust, ubr, and uf are the percentage uncertainties associated with biomass estimates of tree-components (i.e., stem, branches, foliage). site long (°e) lat (°n) elevation (m) mat (°c) maps (mm) sllove 20°24’’16’’ 41°45’35’’ 750 9.6 959 bushtrice 20°25’02’’ 41°53’34’’ 780 9.7 1010 maqellare 20°28’03’’ 41°39’17’’ 850 9.3 1000 zerqan 20°22’’04’’ 41°30’48’’ 715 9.8 1008 paper 19°57’42’’ 41°04’43’’ 160 14.4 1133 tomin 20°25’’20’’ 41°41’18’’ 647 9.8 1030 rrape 19°57’02’’ 42°02’45’ 805 10.1 1000 qafe-mali 20°06’28’’ 42°08’03’ 1160 8.8 1157 *source: www.climexp.knmi.nl table 1. site description including location, longitude (long), latitude (lat), elevation above sea level, mean annual temperature (mat)* and mean annual precipitation sum (maps)*. 𝑈𝑈 = √(𝑈𝑈𝑠𝑠𝑠𝑠 ∙ 𝐷𝐷𝐷𝐷𝑠𝑠𝑠𝑠 )2 + (𝑈𝑈𝑏𝑏𝑏𝑏 ∙ 𝐷𝐷𝐷𝐷𝑏𝑏𝑏𝑏 )2 + (𝑈𝑈𝑓𝑓 ∙ 𝐷𝐷𝐷𝐷𝑓𝑓 )2 (𝐷𝐷𝐷𝐷𝑠𝑠𝑠𝑠 + 𝐷𝐷𝐷𝐷𝑏𝑏𝑏𝑏 + 𝐷𝐷𝐷𝐷𝑓𝑓 ) this error propagation equation is applicable for estimation of the overall uncertainty derived by the summed quantities of tree-components biomass estimates by allometric models. results sampled tree information the basic characteristics of sampled trees are shown in table 2. the sampled trees ranged in age from 6 to 34 years, in dbh from 2 to 16 cm and in height from 1.90 to 9.83 m. a total of 58 sampled trees were analysed in this study. the aboveground and tree-components biomass was computed for the four species (table 3). for q. cerris the total tree biomass ranged between 1.07 and 56.34 kg, stem biomass from 0.84 to 34.65 kg, branch biomass from 0.19-17.84 kg, and foliage biomass between 0.04 and 3.84 kg. the relative contribution of stem biomass to agb in q. cerris increased from 62% for large diameter classes to 78% for small diameter classes (fig. 2). the proportion of branch biomass for this species increased from 18% for the small diameter class to 32% for the large diameter class. the ratio of foliage biomass versus agb increased proportionally with dbh from 4% to7%, indicating that leaf mass gives a marginal contribution in aboveground biomass. for car. betulus, the agb ranged from 0.72 to 21.80 kg, stem biomass from 0.40 to 13.12 kg, branch biomass from 0.23 to 6.25 kg, and foliage biomass from 0.09 to 2.43 kg. for this species, the proportion of stem increased from 55% for the small diameter classes to 60% for large diameter classes, while the contribution of branch biomass declined from 32% for the small diameter classes to 29% for large diameter classes. regardless of the dbh values of car. betulus trees, the relative contribution of foliage biomass was increased from 11.0 to 12.5%. for f. ornus, the agb per tree ranged from 1.83 to 12.09 kg, stem biomass from 0.96 to 7.20 kg, branch biomass istrefi et al. new zealand journal of forestry science (2019) 49:8 page 5 tree species number of trees age (years) dbh (cm) h (m) quercus cerris 18 8–27 2.00–15.00 2.75–9.83 carpinus betulus 17 6–34 2.00–16.00 2.30–9.10 fraxinus ornus 16 6–26 2.20–7.50 1.90–8.00 castanea sativa 7 7-15 3.20–12.00 2.75–9.60 tree variable species number of trees values dbh h dwst dwbr dwf agb q. cerris 18 min 2.00 2.75 0.84 0.19 0.04 1.07 max 15.00 9.83 34.65 17.84 3.84 56.34 mean 6.51 4.43 7.97 3.46 0.74 12.16 sd 3.40 1.77 9.24 4.25 0.92 14.23 car. betulus 17 min 2.00 2.30 0.40 0.23 0.09 0.72 max 16.00 9.10 13.12 6.25 2.43 21.80 mean 6.94 5.28 3.85 3.31 1.23 8.40 sd 4.67 2.25 2.82 2.90 1.08 6.71 f. ornus 16 min 2.20 1.90 0.96 0.65 0.12 1.83 max 7.50 8.00 7.20 4.23 1.94 12.09 mean 4.48 4.55 3.24 2.23 0.61 6.07 sd 1.85 1.69 1.86 1.09 0.53 3.38 cas. sativa 7 min 3.20 2.74 1.10 0.46 0.30 1.85 max 12.00 9.60 16.72 9.22 5.15 31.09 mean 7.74 6.20 9.65 5.68 2.82 18.15 sd 2.76 2.07 4.90 2.89 1.59 9.30 table 3. summary statistics of aboveground and tree-components biomass of felled trees (dbh, diameter at breast height (cm), h, total tree height (m), dwst, stem dry weight (kg), dwbr branch dry weight (kg), dwf, foliage dry weight (kg), agb, aboveground tree biomass (kg)) table 2. summary of information of sampled trees used to develop the biomass functions from 0.65 to 4.23 kg, and foliage biomass from 0.12 to 1.94 kg. for f. ornus, the proportion of stem biomass increased proportionally with dbh from 53% to 60%, while the branch and foliage biomass proportions were inversely related to sampled tree dbh. the proportion of branch and foliage biomass in small diameter classes were 41.3 % and 7.4%, whereas in large diameter classes they were 33.2 % and 16.1 % respectively. for cas. sativa, the aboveground biomass per tree ranged from 1.85 to 31.09 kg, stem biomass from 1.10 to 16.72 kg, branch biomass from 0.46 to 9.22 kg and foliage biomass from 0.30 to 5.15 kg. for cas. sativa, the proportion of stem biomass was 54 % and 59 % for small and large diameter classes, respectively. the relative proportion of branch biomass increased proportionally with dbh from 25% to 30%, but the proportion of foliage biomass was equal (16% of the agb) for small and large diameter classes. istrefi et al. new zealand journal of forestry science (2019) 49:8 page 6 figure 2: box plots of biomass components (in percentages) by forest species estimated by field destructive sampling by dbh (1.3 m from the ground) and tree height (h) classes. frequency distribution of sampled tree variables all species exhibited different frequency distributions of diameter at breast height, where the biggest tree number occurred in the 2, 4 and 6 cm diameter classes (fig. 3). the sampled car. betulus trees had dbh values across the full range of diameter classes, whereas the frequency distribution of other species was noticeably concentrated in the smallest dbh classes which ranged from 2 to 10 cm. fifty-three trees comprising 94% of the total number of sampled trees in all species had a dbh ≤6 cm. regarding tree height, similar patterns in frequency distribution were observed for all species. a greater proportion of trees occurred in the mid-height range and a regular frequency decline of trees and a flattening of the curves towards the largest height classes was observed. we found a strong relationship between dbh and tree height (h) for all species (table 4). the r2 values of the applied linear models were always significant istrefi et al. new zealand journal of forestry science (2019) 49:8 page 7 (p<0.05), with a lower value in f. ornus, probably due to the characteristics of this species in young development phases. for all studied species, age distributions were produced using 10-year age classes (fig. 3c). in q. cerris, car. betulus and f. ornus we noticed a unimodal pattern in age distributions, typical for even-aged forest stands with few trees in young and old age classes. by contrast, in cas. sativa, was observed the lack of a typical age distribution as well as the absence of mature trees in the forest stand. relationship between agb and tree variables visual examination of the biomass data (fig. 4) revealed the existence of a strong relationship between dbh (and/or dbh2 × h) and agb on the logarithmic scale. the values of the correlation coefficient (r) between agb and dbh varied between 0.92 (p<0.05) for f. ornus and 0.96 (p<0.05) for the other species. in addition, correlation between agb and dbh2 × h ranged from r = 0.89 (p<0.05) for f. ornus to r = 0.97 (p<0.05) for car. betulus and remained in the same level for the two other species. the relationship between agb and tree height was clearly weaker for all species, where the values of the correlation coefficient varied from 0.78 (p<0.05) for f. ornus to 0.92 (p<0.05) for cas. sativa. there was also evidence of changing variance in agb values among species associated with an increasing value from cas. sativa to f. ornus. biomass model evaluation the coefficient estimates and goodness-of-fitstatistics (i.e. r2, rmse, rmspe, me) of all biomass models for the four species (after applying the logarithmic transformation) are shown in table 5. these statistics were significant at p<0.05 level, indicating that the fitted biomass equations performed well. the most reliable model fitting was obtained for aboveground and stem biomass, while the model fitting for foliage and branch biomass was less reliable with relatively lower r2, rmse and high me. in model 1, which uses dbh as predictor, the values of the coefficient of determination (r2) indicated that the fitted model explained between 75% and 97% of the observed biomass variance. we noted that r2 values were different between biomass models, tree-components and across species. the lowest values of the coefficient of determination (r2) were observed in model 3, which uses the product of squared dbh with tree height as predictors, whereas the highest values occurred in model 1 and model 4. the r2 values indicated that fitted models explained between 62% and 95% of the biomass variance per tree-component for all studied species with mean values of 75%, 79% and 82% of variance in foliage, branch and stem biomass, respectively, whereas the agb equation explained on average 84% of the total observed variance. the inclusion of tree height (h) as independent variable in the model 2, did not improve the goodness-of-fitstatistics for biomass prediction of tree-components in q. cerris, f. ornus and cas. sativa. the values of root mean square error (rmse) varied among species and ranged between 0.18 and 4.46. all -a -b -c -a -b -c -a -b -c figure 3: log-transformed frequency distribution (number per ha) of diameter at breast height (a), height (b) and age (c) of all sampled trees. the x-axis values are the central values of each class. a b c biomass models used in this study produced the most reliable estimates for agb and stem wood biomass, whereas the least accurate estimates were derived for branch and leaves biomass. most of the mean error (me) values obtained from all biomass models were close to zero, indicating that all allometric equations except model 2 provided accurate prediction of biomass. the values of this statistic obtained from biomass model 2 were far from zero, implying that allometric equations using tree height as a predictor performed the poorest. the dominance of positive values of me indicated that the biomass models provided an underestimate of the aboveground or tree-component biomass, whereas in those cases when this statistic was negative, the biomass was overestimated. in this study we found differences between species in brma values for relationships between agb versus dbh, h and their combination (dbh2 × h) (table 6). in the case of dbh, two out of four values for q. cerris and cas. sativa were significantly different from the others. in the allometric models using tree height (h) as biomass predictor, the brma values decreased from q. cerris to f. ornus. the lowest values of the scaling exponent of rma regression were found in model 3, which uses the predictor variable (dbh2 × h) for biomass estimate, whereas the highest values were obtained by model 2, where total height was the only biomass predictor. in contrast to the brma values, correction factor scores (cf) showed less variability across tree-components and species. the largest differences in cf values were found between foliage biomass and dbh in model 1 (table 6). when tree height was the predictor, we observed less variation in cf values compared to dbh table 2: confusion matrix and a wider range of cf values in stem (1.010–1.070) and agb (1.010–1.060). the largest values of cf were obtained when dbh2 and h were used as predictors in the allometric model 3, whereas the largest values at species level were found in f. ornus and cas. sativa. uncertainty of biomass models in order to assess the bias of allometric equations in biomass prediction, the quantitative estimates of uncertainty were applied. the uncertainty estimates of the biomass models developed in this study ranged from 1 to 24.6%, depending on the species and treecomponents (table 7). the highest bias was found for biomass model 4, while the best model based on uncertainty values was the allometric model 3, where dbh2 x h was the predictor variable. when considering the bias values of biomass models by forest species, the highest values were found in cas. sativa, and the lowest values in q. cerris. the uncertainty of biomass models increased from foliage to stem biomass indicating that stem biomass was estimated more accurately. discussion allometric equations four regression models developed to predict aboveground and tree-components biomass were evaluated for their performance and accuracy of the estimates. the multiple regression models which used dbh and total tree height (h) were the most accurate. the logarithmically transformed model using dbh alone as an independent variable performed better meeting the criteria of accuracy and biomass prediction istrefi et al. new zealand journal of forestry science (2019) 49:8 page 8 species dependent variable independent variable r2 regression coefficients ln β0 β1 q. cerris dbh h 0.802 0.326 0.467 car. betulus dbh h 0.970 0.689 0.474 f. ornus dbh h 0.752 0.018 0.788 cas. sativa dbh h 0.823 -0.091 0.683 table 4. relationship between dbh and h, r2 and values of parameters lnβ0 and β1 figure 4: relationship between aboveground biomass (agb) and diameter at breast height (dbh), tree height (h) and dbh2 × h on a logarithmic scale for all studied species. istrefi et al. new zealand journal of forestry science (2019) 49:8 page 9 species f. ornus cas. sativa model coefficients goodness-of-fit statistics model coefficients goodness-of-fit statistics ln β0 β1 β2 r² rmse rmspe me ln β0 β1 β2 r² rmse rmspe me 1 stem 0.983 1.418 0.88 1.02 0.99 1.00 -1.600 2.153 0.96 1.37 1.36 1.43 branch 0.013 1.405 0.83 1.02 1.00 1.00 -0.693 2.066 0.94 2.03 2.00 1.82 foliage -0.073 1.231 0.79 3.47 3.39 2.39 -1.131 2.328 0.87 0.81 2.21 2.18 agb -1.610 2.748 0.92 3.06 2.42 1.36 -1.253 2.143 0.97 0.23 0.18 0.00 2 stem 1.523 1.281 0.62 0.65 0.59 0.50 -0.262 2.253 0.81 1.24 1.20 1.54 branch -0.248 1.234 0.48 4.46 4.43 2.40 -0.588 2.164 0.72 1.83 1.78 1.80 foliage 1.539 2.459 0.43 3.27 3.12 3.10 -1.030 2.423 0.79 2.10 2.07 2.10 agb -1.540 2.459 0.66 3.46 2.84 1.34 -1.163 2.271 0.80 1.11 1.06 1.12 3 stem 0.285 0.031 0.70 0.78 0.72 0.58 0.920 0.021 0.60 1.31 1.29 0.97 branch 0.004 0.029 0.75 0.86 0.68 0.63 0.255 0.023 0.54 1.37 1.44 0.99 foliage -2.511 0.061 0.53 0.87 1.65 0.65 -0.412 0.023 0.61 1.29 1.35 0.94 agb 0.899 0.032 0.73 1.80 0.74 1.23 1.502 0.022 0.62 1.24 1.34 0.91 4 stem -0.948 1.503 -0.123 0.91 0.18 0.16 0.00 -3.516 1.517 0.732 0.94 0.19 0.16 0.01 branch -1.156 1.027 0.256 0.89 1.16 1.15 -1.15 -3.176 1.939 0.456 0.91 0.74 0.73 0.64 foliage -4.976 2.809 -0.078 0.67 0.76 0.63 0.00 -2.089 1.629 0.512 0.91 0.24 0.22 0.58 agb -0.376 1.413 0.006 0.91 0.18 0.15 0.00 -1.643 1.709 0.519 0.93 0.21 0.17 -0.04 table 5. the coefficients estimate and goodness-of-fit-statistics of log-transformed biomass equations (r2, coefficient of determination, rmse, root mean squared error, rmspe, root mean squared prediction error, me, mean error) for models 1–4. species q. cerris car. betulus model coefficients goodness-of-fit statistics model coefficients goodness-of-fit statistics ln β0 β1 β2 r² rmse rmspe me ln β0 β1 β2 r² rmse rmspe me 1 stem -0.303 1.901 0.85 1.26 1.22 1.22 -0.474 1.371 0.96 0.94 0.88 1.67 branch -0.599 1.918 0.82 1.69 1.65 1.26 -0.060 1.158 0.93 1.45 1.39 0.81 foliage -0.959 1.895 0.75 2.74 2.72 0.77 -0.773 1.636 0.93 2.05 2.00 0.54 agb -1.420 1.886 0.93 1.09 1.05 1.06 -1.148 1.633 0.97 0.33 0.26 0.34 2 stem -0.587 2.628 0.68 1.56 2.28 1.45 -0.617 2.263 0.95 1.33 1.31 1.31 branch -0.819 2.650 0.72 2.02 1.93 1.93 -0.696 1.927 0.92 1.80 0.57 3.71 foliage -1.113 2.615 0.71 3.13 3.07 1.12 -1.288 1.527 0.92 3.09 3.08 1.60 agb -1.525 2.618 0.72 1.34 1.23 1.21 -1.527 2.688 0.96 1.25 1.23 1.20 3 stem 0.610 0.027 0.65 1.33 1.06 0.76 0.217 0.017 0.74 1.59 0.95 1.03 branch -0.204 0.026 0.61 1.34 1.13 0.77 -0.499 0.024 0.71 1.15 1.36 0.75 foliage -1.751 0.026 0.62 1.34 1.12 0.77 -1.487 0.024 0.71 1.32 1.94 0.87 agb 1.056 0.026 0.65 1.32 1.08 0.76 0.767 0.020 0.75 2.29 1.11 -1.94 4 stem -1.933 1.798 0.224 0.93 0.72 0.64 0.64 -2.190 -0.084 2.162 0.90 0.26 0.18 -0.43 branch -2.715 1.781 0.213 0.88 0.38 0.36 0.00 -3.471 0.195 2.386 0.92 0.36 0.29 0.01 foliage -4.270 1.743 0.267 0.88 0.38 0.35 0.00 -4.433 0.216 2.346 0.91 0.36 0.29 0.00 agb -1.467 1.760 0.227 0.93 0.28 0.24 0.04 -1.787 0.084 2.130 0.95 0.22 0.18 0.00 biomass component biomass component quality. as suggested by other studies, dbh is more accurately measured and therefore, is relatively more reliable when is used as single independent variable to develop biomass equation (chave et al. 2005; pastor et al. 1984; mosseler et al. 2014) than other tree variables such as tree height (hosoda & lehara 2010; hunter et al. 2013). we observed that inclusion of tree height as single predictor variable did not improve the biomass model accuracy and this finding is consistent with that reported by johansson (1999). in contrast, other istrefi et al. new zealand journal of forestry science (2019) 49:8 page 10 model biomass component q. cerris car. betulus f. ornus cas. sativa brma cf brma cf brma cf brma cf 1 stem 1.971 1.280 1.392 1.020 1.478 1.170 2.197 1.056 branch 1.270 1.040 1.240 1.071 foliage 1.432 1.040 1.140 1.060 agb 1.030 1.030 1.210 1.060 2 stem 3.097 1.070 2.310 1.020 1.577 1.010 2.535 1.051 branch 1.060 1.040 1.030 1.065 foliage 1.060 1.040 1.050 1.057 agb 1.060 1.040 1.010 1.056 3 stem 1.194 1.255 0.860 1.295 0.733 1.063 1.140 1.215 branch 1.298 1.286 1.041 1.337 foliage 1.285 1.284 1.636 1.233 agb 1.243 1.159 1.060 1.247 4 stem 1.052 1.058 1.020 1.038 branch 1.093 1.081 1.020 1.070 foliage 1.089 1.082 1.424 1.060 agb 1.050 1.029 1.021 1.046 table 6. values of the scaling exponent (brma) of reduced major axis regression and the biomass correction factor (cf) by species and allometric models used in the study table 7. uncertainty estimates of biomass models by forest species and tree-components model biomass component uncertainty (%) of biomass models by species and tree components q. cerris car. betulus f. ornus cas. sativa 1 stem 3.70 2.70 2.40 5.00 branch 4.00 3.00 3.40 6.00 foliage 4.50 3.50 3.50 6.60 agb 2.33 1.75 1.73 3.51 2 stem 2.95 3.95 2.80 5.25 branch 3.24 4.25 3.40 5.70 foliage 4.15 4.70 3.70 6.80 agb 1.95 2.75 3.14 3.74 3 stem 1.00 1.40 2.30 1.50 branch 1.30 2.80 3.40 2.70 foliage 1.70 3.20 3.50 5.40 agb 1.26 1.24 3.03 3.41 4 stem 4.30 9.00 6.10 6.50 branch 6.10 17.00 7.80 10.30 foliage 9.00 17.36 22.80 24.60 agb 3.44 11.85 6.54 12.72 authors have found a significant improvement in model accuracy statistics when tree height was used as biomass predictor (reed & tome 1998), but this variable might be more useful for stand biomass estimates than for individual tree (wang et al. 2013). in general, allometric models were more robust for stem and aboveground biomass than for branch and foliage biomass. stem and aboveground biomass showed less variation than other tree components, and this may be related to the variation of local conditions, tree position in the canopy and sunlight availability. the coefficients of log-transformed allometric biomass equations differed between species. we found that the scaling exponent (brma) in regression equation varied among species and tree variables used as predictors, indicating that such models are species-specific and that the use of a common scaling coefficient for different species would lead to bias in biomass prediction. our values of brma (2.05 for dbh and 2.38 for tree height) differed from that used by west et al. (1999). these authors suggested that agb should be scaled with dbh according to a universal value of scaling exponent (brma = 8/3, i.e. 2.67), which depends on the tree architecture. west et al. (1999) assumed that using a universal value of b is not acceptable, especially when trees are growing under different environmental conditions. although our sampled trees were growing under similar site conditions, the brma value in allometric models developed in this study were different from those reported earlier by zianis and mencuccini (2004). therefore, we conclude that the scaling exponent value is probably mainly affected by species traits and their growth during the juvenile period (poorter et al. 2015). pilli et al. (2006) reported three different scaling exponent values for juvenile (b = 2.08), adult (b = 2.66) and mature (b = 2.51) trees, whereas niklas (2004) reported scaling exponents close to 1.0 in the case of young plants, while for large plants the exponents declined to below 0.75. in contrast, we found that tree age is associated with a decrease of b from 2.2 (age class 1–10 years) to 2.0 (age class 11–20 years). moreover, we observed that the variability of the scaling exponent was lower for stem biomass than for branch and foliage biomass. the scaling exponent values varied across species and were different from those reported in a previous study. thus, blujdea et al. (2012) reported for the same allometric model a higher value for f. excelsior (b = 3.04) and a lower value for quercus species (b = 1.22). the most reliable fitted model for estimation of aboveground biomass was the multiple regression function (model 4). based on the goodnessof-fit-statistics (r2; rmse; rmspe), all our biomass models produced the most reliable estimates for agb and stem biomass and the least accurate estimates for branch and foliage biomass. most of the mean error (me) values obtained from the biomass models were close to zero, indicating that all allometric equations except model 2 could provide accurate estimates. in albania, as in other european countries, biomass estimates based on allometric equations and befs have been applied without any quantitative estimates of uncertainty. therefore, information on the overall error occurring with the use of biomass models is missing. the uncertainty estimates of the biomass ranged from 1.0 to 24.6%, depending on the species, tree components and biomass model applied. the uncertainty in the biomass estimates of cas. sativa was high because the number of sampled trees used to generate allometric equations was low. by contrast, the lowest values of error were found in q. cerris where the number of trees used to develop the biomass models was the highest. the uncertainty results indicate that the applicability of biomass models needs to be carefully evaluated, especially for the presence of bias, before using these in other geographic areas or in other countries. contribution of tree-components to total agb for all species studied, stem wood was the main contributor to the total agb, followed by branches and foliage. the fact that the highest proportion of the agb is allocated in the stem has been documented in previous studies, with proportions ranging from 50% to 92% for different species (fonseca et al. 2011). our results are within the same range varying from 52% (car. betulus) to 65% (q. cerris). the proportions of biomass of each tree component relative to agb are consistent with those reported by blujdea et al. (2012) for several broadleaved species growing in romania. as expected, the proportion of biomass allocated to stems increased with tree size. as trees grow, age and size-related changes in tree shape and form alter the contribution of stem to whole-tree biomass increment (bartelink 1998). in contrast, the relative contribution of branches and foliage to agb decreased. we found that the biomass allocation to each tree-component differed among species, suggesting that their proportion relative to agb depends on tree species in a mixed stand and stem density. allometric equation comparison with previous models the literature and globallometree repository (http:// globallometree.org, henry et al. 2013) contain numerous allometric equations for different trees species from the mediterranean basin and other regions. such allometric equations to estimate tree biomass have been developed for various species growing under different site conditions. we noticed that information about biomass equations for q. cerris and f. ornus is missing in the globallometree web platform (table 8), and the equations developed in this study can enrich the database. the allometric equations we developed used variables from 58 sampled trees growing in specific sites representing a small region. some authors suggest that biomass regression equations developed for a site or region can be used to predict tree biomass in other places (wirth et al. 2004). to verify that, we compared the results provided by model 3 developed in our study with allometric models found in the globallometree platform for car. betulus and cas. sativa. we could not do the same comparison for other species because our models differed from those found in this platform. the average values of agb estimates from each allometric equation were compared using one-way anovas. means istrefi et al. new zealand journal of forestry science (2019) 49:8 page 11 were separated using tukey hsd post-hoc test. in the case of car. betulus we found that the mean values of biomass estimated from each model were significantly different (f=7.162; p=0.002), whereas in the case of cas. sativa this difference was not significant (f=1.739; p=0.204). the results of the post-hoc test revealed significant differences in the mean agb values estimated by our model for car. betulus and those of hoellinger (1987) (i.e. model 3 (p=0.002) and model 4 (p=0.022) in table 8). in contrast, the results of the post-hoc test for cas. sativa indicated no significant difference between the agb estimated by our model and those of hoellinger (1987) (i.e. models 5 (p=0.212) and 6 (p=0.352) in table 8).this suggests that the use of allometric equations needs to consider factors such as species characteristics, age, soil fertility and climate (madgwick & satoo 1975). another limitation in the use of biomass equations is the range of values of independent variables. extrapolation below or above these ranges could lead to substantial differences between the true and predicted values (zianis & mencuccini 2003). since allometric relationships between biomass and tree variables vary across species, tree size and age, the use of such models in other geographic areas is not suggested because this may lead to incorrect estimations (harding & grigal 1986; wang et al. 2002; zianis & mencuccini 2003; zabek & prescott 2006). finally, the biomass equations were developed on a relatively narrow dbh and h range and their application to large trees outside these ranges may be associated with larger errors in biomass prediction. conclusions the dataset from 58 young trees representing four broadleaved species were used to develop allometric equations to estimate aboveground and treecomponents biomass. the intention is to use these equations in biomass estimation for respective species in the framework of the national forest inventory. stem and agb biomass by dbh were the most accurately predicted, whereas the use of tree height as biomass predictor was associated with a decrease in prediction accuracy. the possibility of estimating young tree biomass by forest yield tables or forest inventory data is limited due to the lack of biomass expansion factors. since few biomass equations exist and there is a need for their development, further investigations in other ecological zones and other species should be conducted in albania. list of abbreviations agb, aboveground biomass; befs, biomass expansion factors; cf, correction factor; dbh, diameter at breast height; h, tree height; ipcc, intergovernmental panel on climate change; me, mean error; rmse, root mean squared error; rmspe, root mean squared prediction error; unfccc, united nations framework convention on climate change. competing interests the authors declare that they have no competing interests. authors’ contributions nç and bth undertook the fieldwork and data collection, et did the data processing, statistical analysis, and ei wrote the manuscript. all authors read and approved the manuscript. acknowledgements this research was conducted in the framework of phd studies. we acknowledge the help of ing. saimir beqo and ing. albert buzali with sample preparation and ing. edmir novaku with biomass data sampling. the authors thank the editor and two anonymous reviewers for their comments which helped to improve the quality and clarity of the manuscript. references anfi. (2004). albanian national inventory final report. tirana, albania: agrotec consortium, 140 pp. bevington, p. r., & robinson, d. k. (1992). data reduction and error analysis for the physical sciences. new york: mcgraw-hill. bartelink, h. h. (1998). a model of dry matter partitioning in trees. tree physiology, 18(2), 91–101. table 8. allometric equations found in the globallometree platform for species included in this study nr allometric equation number of trees r2 location source quercus spp. 1 log biomass = 0.883 + 2.14·log (dbh) 33 n.a. austria hochbichler, (2002). 2 log agb = 3.1009 + 2.6996 log(dbh) 18 1.0 england bunce (1968). car. betulus 3 agb = (0.0485×(dbh)2 × (h)) + 5.4 78 0.94 france hoellinger (1987) 4 agb = (0.0325×(dbh)2 × (h)) + 6.7 78 0.94 france hoellinger (1987) cas. sativa 5 agb = (0.045×(dbh)2 × (h)) 2.6 142 0.96 france hoellinger (1987) 6 agb= (0.038×(dbh)2 × (h)) 1.9 142 0.96 france hoellinger (1987) istrefi et al. new zealand journal of forestry science (2019) 49:8 page 12 blujdea, v. n. b., pilli, r., dutca, i., ciuvat, l., & abrudan, i. v. (2012). allometric biomass equations for young broadleaved trees in plantations in romania. forest ecology and management, 264, 172–184. bunce, r. g. h. (1968). biomass and production of trees in a mixed deciduous woodland. i. girth and height as parameters for the estimation of tree dry weight. journal of ecology, 56, 759–775. chave, j., andalo, c., brown, s., cairns, m. a., chambers, j. q., eamus, d., fölster, h., fromard, f., higuchi, n., kira, t., lescure, j. p., nelson, b. w., ogawa, h., puig, h., riéra, b., & yamakura, t. (2005). tree allometry and improved estimation of carbon stocks and balance in tropical forests. oecologia, 145(1), 87– 99. dutca, i., abrudan, i. v., stancioiu, p. t., & blujdea, v. (2010). biomass conversion and expansion factors for young norway spruce (picea abies (l.) karst) trees planted on non-forest lands in eastern carpathian. notulae botanica e horti agrobotanici cluj-napoca, 38(3), 286–292. fao. (2015). world reference base for soil resources 2014, update 2015 international soil classification system for naming soils and creating legends for soil maps. world soil resources reports no. 106. fao, rome. iuss working group wrb. http:// www.fao.org/3/i3794en/i3794en.pdf. fonseca, w., rey, b. j. m., & alice, f.e. (2011). carbon accumulation in the biomass and soil of different aged secondary forests in the humid tropics of costa rica. forest ecology management, 262, 1400–1408. harding, r. b., & grigal, d. f. (1986). site quality influences on biomass estimates for white spruce plantations. forest science, 32(2), 443–446. harris, i., jones, p. d., osborn, t. j., & lister, d. h. (2014). updated high-resolution grids of monthly climatic observations the cru ts3.10 dataset. international journal of climatology, 34(3), 623– 642. henry, h. a. l., & aarssen, l. w. (1999). the interpretation of stem diameter-height allometry in trees: biomechanical constraints, neighbour effects, or biased regressions? ecology letters, 2(2), 89–97. henry, m., bombelli, a., trotta, c., alessandrini, a., birigazzi, l., sola, g., ghislain vieilledent, g., santenoise, p., longuetaud, f., valentini, r., picard, n., & saint-andré, l. (2013). globallometree: international platform for tree allometric equations to support volume, biomass and carbon assessment. iforest, 6, 326–330. hochbichler, e. (2002). vorläufige ergebnisse von biomassen inventuren in buchenund mittel wald beständen. dietrich,h.-p., raspe, s. and preuhsler, t. (eds.): inventur von biomasse und nährstoffvorräten in waldbeständen. forstliche forschungsberichte, münchen, (186), 37–46 hoellinger, g. (1987). synthèse des expérimentations réalisées sur les différents chantiers. annales de mécanisation forestière del’ armef hosoda, k., & iehara t. (2010). aboveground biomass equations for individual trees of cryptomeria japonica, chamaecyparis obtusa and larix kaempferi in japan. journal of forest research, 15(5), 299–306. hunter, m. o., keller, m., victoria, d., & morton, d. c. (2013). tree height and tropical forest biomass estimation. biogeosciences, 10(12), 8385–8399. ibm corp. (2016). ibm spss statistics for windows, version 24.0. armonk, ny: ibm corp. (released 2016). ipcc. (2003). good practice guidance for land use, land use change and forestry. institute for global environmental strategies, hayama (japan). istrefi, e., toromani, e., & collaku, n. (2018). allometric relationships for estimation of above-ground biomass in young turkey oak (quercus cerris l.) stands in albania. acta silvatica et lignaria hungarica, 14(1), 65–81. johansson, t. (1999). biomass equations for determining fractions of pendula and pubescent birches growing on abandoned farmland and some practical implications. biomass and bioenergy, 16(3), 223–238. joosten, r., schumacher, j., wirth, c., & schulte, a. (2004). evaluating tree carbon predictions for beech (fagus sylvatica l.) in western germany. forest ecology and management, 189(1-3), 87–96. kaitaniemi, p. (2004). testing the allometric scaling laws. journal of theoretical biology, 228(2), 149–153. lehtonen, a. (2005). estimating foliage biomass in scots pine (pinus sylvestris) and norway spruce (picea abies) plots. tree physiology, 25(7), 803–811. madgwick, h.a.i. & satoo, t. (1975). on estimating the above ground weights of tree stands. ecology, 56(6), 1446–1450. mosseler, a., major, j.e., labrecque, m., & larocque, g. r. (2014). allometric relationships in coppice biomass production for two north american willows (salix spp.) across three different sites. forest ecology and management, 320, 190–196. niklas, k. j. (2004). plant allometry: is there a grand unifying theorem? biological reviews, 79(4), 871– 889. niklas, k. j. (2006). a phyletic perspective on the allometry of plant biomass-partitioning patterns and functionally equivalent organ-categories. new phytologist, 171(1), 27–40. omuri, i. (2006). studim mbi perllogaritjen e nivelit te sekuestrimit te karbonit ne komunen shishtavec istrefi et al. new zealand journal of forestry science (2019) 49:8 page 13 bazuar ne rigjenerimin e vegjetacionit autokton pyjor dhe menaxhimin e qendrueshem e shumfunksional te pyjeve. master thesis, universiteti bujqesor i tiranes, fakulteti i shkencave pyjore, 71 pp. pastor j., aber j. d., & melillo, j. m. (1984). biomass prediction using generalized allometric regressions for some northeast tree species. forest ecology and management, 7(4), 265–274. pilli, r., anfodillo, t., & carrer, m. (2006). towards a functional and simplified allometry for estimating forest biomass. forest ecology and management, 237(1), 583–593. poorter, h., jagodzinski, a. m., ruiz-peinado, r., kuyah, s., luo, y., oleksyn, j., usoltsev, v. a., buckley, t. h. a., reich, p. b., & sack, l. (2015). how does biomass distribution change with size and differ among species? an analysis for 1200 plant species from five continents. new phytologist, 208(3), 736–749. reed, d., & tomé, m. (1998). total aboveground biomass and net dry matter accumulation by plant component in young eucalyptus globulus in response to irrigation. forest ecology and management, 103(1), 21–32. sokal, r. r., & rohlf, f. j. (2012). biometry: the principles and practice of statistics in biological research. 4th edition. new york: w.h. freeman and co. 937 pp. isbn: 978-0-7167-8604-7. sprugel, d. g. (1983). correcting for bias in logtransformed allometric equations. ecology, 64(1), 209–210. toromani, e., sanxhaku, m., sallaku, m., & shaho, e. (2011). biomass production and carbon storage of populus × canadensis guinier i-214 plantations grown at pogradeci region in albania. southern forests: a journal of forest science, 73(2), 91–99. unfccc. (2009). assisted natural regeneration of degraded lands in albania. project design document form for afforestation and reforestation project activities (cdm-ar-pdd), 156 pp. van laar, a., & akça, a. (1997). forest mensuration. göttingen: cuvillier verlag, 418 pp. isbn 3-89588874-5. zabek, l. m., & prescott, c. (2006). biomass equations and carbon content of aboveground leafless biomass of hybrid poplar in coastal british columbia. forest ecology and management, 223(1-3), 291–302. zianis, d., muukkonen, p., mäkipää, r., & mencuccini, m. (2005). biomass and stem volume equations for tree species in europe. silva fennica monographs, 4, 5–63. zianis, d. & mencuccini, m. (2004). on simplifying allometric analyses of forest biomass. forest ecology and management, 187(2), 311–332. zianis, d., & mencuccini, m. (2003). aboveground biomass relationships for beech (fagus moesiaca cz.) trees in vermio mountain, northern greece, and generalised equations for fagus sp. annals of forest science, 60(5), 439–448. zeng, w. s., & tang, s. z. (2011). bias correction in logarithmic regression and comparison with weighted regression for nonlinear models. nature preceding, 24(2), 1–11. wang, x., ouyang, s., sun, o.j., & fang, j. (2013). forest biomass patterns across northeast china are strongly shaped by forest height. forest ecology and management, 293,149–160. wang, j. r., zhong, a. l., & kimmins, j. p. (2002). biomass estimations errors associated with the use of published regression equations of paper birch and trembling aspen. northern journal of applied forestry, 19(3), 128–136. west, g. b., brown, j. h., & enquist, b. (1999). a general model for the structure and allometry of plant vascular systems. nature, 400(6745), 664–667. wirth, c., schumacher, j., & schulze, e. d. (2004). generic biomass functions for norway spruce in central europe – a meta-analysis approach toward prediction and uncertainty estimation. tree physiology, 24(2), 121–139. istrefi et al. new zealand journal of forestry science (2019) 49:8 page 14 genetic variation in wood properties of mid-rotation age eucalyptus globoidea ebenezer a. iyiola, clemens m. altaner* and luis a. apiolaza school of forestry, university of canterbury, christchurch 8140, new zealand *corresponding author: clemens.altaner@canterbury.ac.nz (received for publication 13 july 2021; accepted in revised form 26 march 2022) abstract background: eucalyptus globoidea blakely produces ground-durable (class 2) and stiff wood and has the potential to be grown in new zealand to supply high-value environmentally-friendly timber for use as posts in the agricultural sector and stiff veneers for the lvl industry. the new zealand dryland forests initiative (nzdfi) has established a breeding programme for this species. the objective of this study was to identify trees with superior wood properties for commercial propagation enabling the establishment of a domestic plantation resource of ground-durable timber. methods: the genetic variation in wood properties at mid-rotation age (8-year-old) of 141 e. globoidea families was assessed for the following traits: heartwood diameter, diameter under bark at ~0.5 m height, combined sapwood diameter, heartwood collapse, sapwood collapse, standing tree acoustic velocity and extractive content in the heartwood. families were ranked and genotypes with large heartwood diameter, high extractive content and stiffness as well as low collapse were identified. results: heartwood diameter (h2 = 0.51) and extractive content (h2 = 1.16) showed good heritability, which in combination with high variation are promising traits for a breeding programme. the high heritability for extractive content indicated a closer relatedness within the population than the assumption of unrelated families of half-siblings. the unfavourable correlation between the heartwood diameter and extractive content (genetic correlation (rg) = −0.45) indicated that a compromise is required for simultaneous improvement of both traits. heritability estimates for heartwood collapse (h2 = 0.30) and acoustic velocity (h2 = 0.36) were moderate. conclusions: genetic selection for wood quality traits of e. globoidea is practically feasible. superior individuals with above average performance for multiple traits were present in the breeding populations, however, this was dependent on the intended end use of the timber. new zealand journal of forestry science iyiola et al. new zealand journal of forestry science (2022) 52:13 https://doi.org/10.33494/nzjfs522022x172x e-issn: 1179-5395 published on-line: 27/04/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access it has high stiffness and strength, and its heartwood is rated class 2 for in-ground durability, i.e. lasting 15 to 25 years in service (as5604 2005). the new zealand dryland forests initiative (nzdfi) is focusing on establishing a sustainable plantation resource of naturally durable eucalyptus timbers (millen et al. 2018). five eucalyptus species have been selected for their breeding programme, including e. globoidea, for the establishment of a domestic naturally ground-durable hardwood industry. such a sustainable domestic resource is an environmentally friendly highintroduction eucalyptus globoidea blakely, commonly known as white stringybark, belongs to the subgenus eucalyptus (brooker & kleinig 1983), and is native to south-eastern australia’s gentle undulating hill country near the coast and mountain slopes, extending to escarpments adjacent to tablelands but not inland of the ranges (boland et al 2006). it is a species that displays several appealing properties such as early heartwood formation, drought, frost and pest-tolerance, and grows well in the new zealand environment (millen et al. 2018). moreover, keywords: acoustic velocity; collapse; extractives; genetic gain; heartwood; natural durability; tree breeding; white stringybark http://creativecommons.org/licenses/by/4.0/), iyiola et al. new zealand journal of forestry science (2022) 52:13 page 2 value alternative not only to copper chromium arsenic (cca) treated radiata pine, which is difficult to dispose of (love 2007; townsend & solo-gabriele 2006) but also naturally durable timber imports, which often originate from endangered or unsustainably harvested tropical sources (unep 2012). envisaged uses for e. globoidea timber are primarily posts for the agricultural sectors (millen et al. 2018) but it can also be used for producing stiff veneers for laminated veneer lumber (lvl) (guo & altaner 2018) and sawn timber (jones et al. 2010). for the latter, stiffness and drying without collapse are relevant wood properties (bootle 2005; poynton 1979; somerville & gatenby 1996). collapse is wood shrinkage caused by the buckling of the cell walls and flattening of the cell lumen during drying due to negative pressure (chafe et al. 1992). it is different from normal shrinkage, which is caused by removal of bound water from the cell wall below the fibre saturation point (fsp). naturally durable timber has less pronounced and more variable durability than treated wood. the durability varies among trees due to a combination of genetic and environmental factors (zobel & jett 1995). genetic variation can be exploited in a breeding programme, while environmental variation can be addressed by appropriate siting, silviculture and timber processing. there is also variability within trees, since sapwood is never regarded as naturally durable, and typically durability of heartwood decreases towards the pith (as5604 2005). as a consequence, the heartwood of younger trees tends to have lower durability than heartwood from older trees (amusant et al. 2004). therefore, to produce ground-durable posts from young trees grown in short rotation plantations, it is necessary to deploy genetically improved tree stocks. young trees producing durable timber have been reported in breeding trials of e. cladocalyx (bush et al. 2011). assessing wood durability is time consuming and expensive. on the other hand, breeding programmes benefit from large sample sizes (raymond 2002). consequently, there has been little focus on natural wood durability in tree breeding programmes. natural durability assessments in breeding trials, including both sampling heartwood from the trees and assessing its durability, need to be rapid and cost effective. felling trees and sawing wood samples is destructive and too laborious for routine use in a breeding programme. a rapid means to obtain a heartwood sample without damaging the tree is taking increment cores (estopa et al. 2017; jones et al. 2008). direct assessments of natural durability by mass loss of wood samples when exposed to fungi (awpc 2007; cen 1989; en-350-1 1994) are impractical in a breeding context due to time and cost (bush & walker 2011; harju & venäläinen 2006; pâques et al. 2013). extractive content in the heartwood are responsible for the natural durability of wood (bush et al. 2011; hawley et al. 1924) and have been reported to be under genetic control (bush et al. 2011; li et al. 2018). nearinfrared (nir) spectroscopy has been proven effective in predicting the extractive content in wood (bush et al. 2011; li et al. 2018; stackpole et al. 2011). nir spectra can be acquired within seconds from a solid wood surface and are an inexpensive method for assessing the chemical components of wood (estopa et al. 2017; greaves et al. 1996; tsuchikawa & schwanninger 2013). nir calibrations for the extractive content were reported to have a root mean square error (rmse) of 0.91% to 1.16% for e. bosistoana f.muell. (li 2018), 0.62% for sweitenia marcrophylla king (da silva et al. 2013), and 0.66% for e. globoidea (kuwabara 2017). this study examined diameter under bark, heartwood diameter, combined sapwood diameter, extractive content (ec), sapwood collapse, heartwood collapse (all at 0.5 m stem height) and acoustic velocity (a proxy for wood stiffness) of e. globoidea. breeding values of these traits were calculated for 141 half-sib families of 8-yearold e. globoidea. the phenotypic and genetic correlations between the wood traits were estimated. materials and methods the atkinson trial established in november 2011 at wairarapa, new zealand (latitude 41° 35' 21" s and longitude 175° 24' 4" e), with 141 e. globoidea halfsib families was used in this study. the families in this trial were grown from seed collected from across the natural range of the species in australia and from three nz plantation sites with a known seedlot. the unknown demarcation of the seed collected in australia means that provenance is undeclared. the trial contained 8640 trees and was laid out in an incomplete block design with each family replicated 40 to 80 times. there were 36 trees per block, planted at a spacing of 2.4 m between rows and 1.8 m within rows. trees in the trial were assessed for diameter at breast height (dbh) and form in 2015 at age 3.4 years, pruned in august 2018 and thinned in april 2019 to remove trees with poorer stem form. all living trees not marked for removal and with a dbh larger than 30 mm were sampled with a purpose-built corer. a bark-to-bark 14 mm diameter core including the pith was extracted ~0.5 m above the ground from 2160 trees in march 2019. heartwood quantity and quality stem diameter under bark and heartwood diameter at 0.5 m height were measured on the green cores after highlighting heartwood with an aqueous 0.1% solution of methyl orange. combined sapwood diameter was calculated as the difference of the two measures. heartwood quality was assessed by predicting ethanol soluble extractive content (%) of dry cores from nir spectra (tensor 37, bruker, germany) obtained with a fibre optic probe at wavelengths from 9000 to 4000 cm-1 at 4 cm-1 intervals as described earlier (li & altaner 2019). tangential collapse after the nir measurements, the cores were equilibrated to a stable moisture content at 60% relative humidity and 25°c. ten cores were randomly selected, and the widest tangential diameter was measured with a vernier calliper. these 10 values were averaged and used as reference (d1) for the collapse assessment. collapse was assessed separately for heartwood and sapwood (equation 1). the narrowest tangential diameter of each core was measured in the sapwood (d2), as well as the two narrowest tangential diameters for the heartwood. two diameters were measured in the heartwood as collapse was more prominent in this region and heartwood is the main target product for durable timber production. the latter were subsequently averaged. tangential collapse (%) = ((d1–d2)/d1) × 100 (1) standing tree acoustic velocity ten trees were randomly sampled per family for those families with more than ten living trees. for families containing fewer than 11 standing trees, all trees in the family were sampled. acoustic velocity was measured on 1,147 trees with the standing tree time-of-flight treetap tool (university of canterbury, christchurch, new zealand) (toulmin & raymond 2007). a total of eight measurements were averaged per tree, which were acquired by placing the probes at one stem location; probes were spaced 1 m apart centered about breast height. statistical analyses the data were analysed using r statistical software (r core team 2019). a linear mixed model was used for the univariate analyses (heritability) using equation 2. yijk = μ + fi + bj + eijk (2) where yijk is a phenotypic observation of a single trait, µ is the overall intercept, fi is the random effect of the ith family, bj the random effect of the j th block and eijk the random residual for the kth individual of the ith family in the jth block. the residuals were assumed to be identically and independently normally distributed. the expected value for the phenotype is µ while the variances for each of the random terms are: v(f) = σ2f , v(b) = σ 2 b and v(e) = σ2e. the notation can be expanded to a bivariate scenario using vectors that have two phenotypes for traits 1 and 2 for each individual. the variances are then kronecker matrix products v(f) = f0 ⊗ zf , v(b) = b0 ⊗ zb , and v(e) = r0 ⊗ i where zf , zb and i are incidence matrices, connecting the observations to model terms and f0, b0 and r0 are illustrated below. (3) where σ2f1, σ 2 b1, and σ 2 e1 are the variances for family, block and residual; σf12, σb12, and σe12 represent the covariances for family, block and residual between traits 1 and 2. the model was fitted with the asreml-r package (butler et al. 2009) to estimate variance components and breeding values. the phenotypic and additive genetic variations were estimated to compute the narrow sense half-sib heritability (h2) of each trait using equation 4. h2 = var(a)/var(y) = 4σ2f /(σ 2 f +σ 2 b + σ 2 e) (4) the heritability estimated in this study assumed that families were true half-siblings with a relationship coefficient within families of one quarter. the coefficient genetic of variation (cgv) for each trait was determined using the equation below. cgv = (√4σ2f / population mean) × 100 (5) the coefficient of phenotypic variation (cpv) for each trait was determined using the equation below. cpv = (standard deviation/population mean) × 100 (6) the genetic correlation is a measure of the strength of the genetic association between the performance in one trait and performance in another trait (bourdon 2000). rg(m,n) = σ 2 fm / √ σ 2 fm σ 2 fn (7) is the covariance between traits m and n; σ2fm is the genetic variance for trait m; σ2fn is the genetic variance for trait n and rg(m,n) is the genetic correlation. the pearson correlation coefficient was used to express the phenotypic correlation (rp) between the traits. results and discussion collapse, acoustic velocity, and heartwood properties the descriptive statistics, heritability estimates, coefficients of phenotypic variation (cpvs) and coefficients of genetic variation (cgvs) for the assessed wood traits are presented in table 1. high coefficients of variation were observed for extractive content, combined sapwood diameter, heartwood diameter, diameter under bark and sapwood collapse. combined sapwood diameter exhibited larger variation than diameter under bark and heartwood diameter. the 8-year-old e. globoidea trees had produced more heartwood (mean heartwood diameter 91.0 mm) and had a lower coefficient of phenotypic variation (cpv = 28.9%) than has been reported for e. bosistoana (31.7 mm to 42.1 mm with a cpv of 51% to 61%) at 7 years of age (li et al. 2018). in 9-year-old e. globulus labill, heartwood iyiola et al. new zealand journal of forestry science (2022) 52:13 page 3 table 1: description of the study sites diameter ranged from 61 to 98 mm with a cpv of 6.8% to 45.1% (miranda et al. 2014). the mean heartwood diameter reported for 8-year-old e. cladocalyx f.muell ranged from 58 to 82 mm with a cpv of 29% based on samples collected from three different locations (bush et al. 2011). the mean combined sapwood diameter was 50.9 mm with a cpv of 31.8% (table 1). the average value reported in this study was lower than that reported for 7-year-old e. bosistoana (60.4 mm and 70.3 mm) (li et al. 2018). this confirmed the general observation that e. globoidea has a narrow sapwood band (bootle 2005). a narrow sapwood band is beneficial for post processing operations as less material needs to be removed. the cpv for e. globoidea (31.8%) was slightly higher than for e. bosistoana (26%) (li et al. 2018). the sapwood diameter reported for 9-year-old e. globulus ranged from 16 to 29 mm (miranda et al. 2014) and was 50 to 74 mm for 8-year-old e. cladocalyx (bush et al. 2011). the mean predicted extractive content was 9.4% with a cpv of 47.9%. this was comparable to 7-year-old e. bosistoana for which 7.7% and 9.6% extractive content was observed at two different sites, with a cpv of 46% (li et al. 2018). however, the extractive content of e. globoidea was lower than the average 12% observed for similarly-aged e. cladocalyx (bush et al. 2011). tangential collapse in the heartwood had a mean of 18.2%, similar to the value of 18.6% reported for the collapse prone e. nitens (h.deane & maiden) maiden (kube 2005) and e. dunnii maiden (14.9%) (arnold et al. 2004). however, care must be taken when comparing the mean values reported for collapse from different studies, as the measurement techniques differed. this study measured the maximum tangential dimensional change, a combination of normal shrinkage and collapse at the worst part of the sample, which should be related to checking. the coefficient of variation in collapse reported in this study was high (37.91%) and corresponded to the observed value (37%) for 6.5-year-old e. dunnii (arnold et al. 2004). collapse in the sapwood was lower (3.2%) but more variable (cpv 168.75%) compared to collapse iyiola et al. new zealand journal of forestry science (2022) 52:13 page 4 in heartwood (18.2%; cpv 37.91%). it is thought that collapse is caused by negative pressure generated over gas–liquid surfaces with high curvature when water evaporates from the wood. extractives can reduce pore size/permeability of cell walls and therefore increase the likelihood of collapse in heartwood (chafe et al. 1992). the mean standing tree acoustic velocity ranged from 2.13 to 4.27 km/s with a mean of 2.96 km/s and a cpv of 10.47% (table 1). it is not straightforward to compare standing tree acoustic velocity measurements with other studies, as commonly used acoustic tools appear to provide different absolute values (dungey et al. 2012). the standing tree acoustic velocity (treetap) of 2.96 km/s was similar to the 2.5 km/s (ranging from 2.2 to 2.8 km/s) reported for the species at age 25 years using the iml hammer (wiesloch, germany) (jones et al. 2010). this is comparable with 2.7 km/s in 10-yearold pinus radiata d.don in australia measured with treetap (toulmin & raymond 2007). it should also be acknowledged that age has a significant effect on the acoustic velocity of young trees, as the microfibril angle (mfa) decreases with cambial age (lachenbruch et al. 2011) and that the mean acoustic velocity is generally lower in softwoods than hardwoods due to their higher mfa (lindström et al. 2002). comparable cpvs for acoustic velocity were reported in the literature; 7.87% for 8-year-old e. fastigata h. deane & maiden (suontama et al. 2018), 8.1 to 10.2% for 20-year-old pseudotsuga menziesii (mirb) franco (klápště et al. 2019) and 9.05% for p. sylvestris l. (hong et al. 2015). an even lower cpv of 2.2% was reported for 10-year-old p. radiata (toulmin & raymond 2007). heritability and coefficient of genetic variation of studied wood properties this study used a coefficient of relatedness (rc) of halfsiblings (rc = 0.25), i.e. the families were assumed to be perfect half-siblings and unrelated to each other (table 1). heartwood diameter had a heritability estimate of 0.51 and a cgv of 20.7%. heritability reports for table 1: descriptive statistics, heritability (h2) with the 95% confidence interval in brackets for e. globoidea wood properties at age 8 years; coefficient of phenotypic variation (cpv) and coefficient of genetic variation (cgv); rc: coefficient of relatedness. trait mean standard deviation min max cpv (%) cgv (%) h2 (rc = 0.25) diameter under bark (mm) 141.9 33.7 60 255 23.75 19.59 0.67 (0.49, 0.85) heartwood diameter (mm) 91.0 26.3 0 190 28.90 20.70 0.51 (0.36, 0.66) combined sapwood diameter (mm) 50.9 16.2 6 150 31.83 25.44 0.63 (0.45, 0.80) predicted extractive content (%) 9.4 4.5 −4.4 31.7 47.87 51.58 1.16 (0.90, 1.39) acoustic velocity (km/s) 2.96 0.31 2.13 4.27 10.47 6.24 0.36 (0.18, 0.54) heartwood collapse (%) 18.2 6.9 −2.9 45.5 37.91 19.16 0.30 (0.17, 0.40) sapwood collapse (%) 3.2 5.4 −14.1 25.1 168.75 56.52 0.12 (0.03, 0.21) heartwood diameter ranged from 0.39 to 0.61 for e. grandis w.hill and e. cladoclayx (bush et al. 2011; santos et al. 2004), comparable with the result obtained in the current study. the observed heritability and variation in e. globoidea are suitable for increasing heartwood production through selective breeding; however, the coefficient of variation may decline with tree age (bush et al. 2011). the heritability estimates for combined sapwood diameter and diameter under bark were 0.63 and 0.67 with cgvs of 25.44% and 19.59%, respectively. similar heritability estimates (0.67 to 0.82) were reported for sapwood diameter in 7-year-old e. bosistoana (li et al. 2018) and larix kaempferi (lamb.) carrière (pâques 2001). the heritability of diameter under bark was comparable to the range (0.29 to 0.72) reported for diameter at breast height (dbh) in previous studies (kube 2005; whiteman 1992) and the cgv (19.59%) for diameter under bark was at the lower end of the range (15.1% to 87.1%) reported for growth of eucalyptus hybrid clones (wu et al. 2017). the heritability estimate for extractive content was 1.16 and had a cgv of 51.58%. the relatedness coefficient (0.25) used in this study pushed the heritability estimate above 1, potentially indicating deviations from halfsiblings and inbreeding effects (eldridge et al. 1994; elliott & byrne 2003). in general, heritability estimates reported for extractive content ranged from 0.19 to 0.56 for e. bosistoana and e. globulus (li et al. 2018; poke et al. 2006). the estimates of heritability in this species might be too high because of the unknown relatedness and the fact that the assessments were restricted to one site (white et al. 2007). the high cgv (51.58%) combined with strong genetic control indicated potential to improving natural durability of e. globoidea through selective breeding. collapse was moderately heritable in the heartwood (h2 = 0.30) but was not very variable (cgv of 19.16%), while it had low heritability in the sapwood (h2 = 0.12) where it was more variable (cgv = 56.52%). the narrow sapwood band in e. globoidea is unlikely to be converted into solid wood products. the heritability, although low, along with the coefficient of variation estimates indicated potential for improving tangential collapse in e. globoidea heartwood through selection. the heritability of collapse in the heartwood was similar to most values reported for other eucalypts; e. nitens (0.23 to 0.61) (hamilton et al. 2004; kube 2005) and e. grandis (0.23 to 0.31) (bandara 2006). the location where e. delegatensis f.muell. ex r.t.baker seed was collected was reported to influence the severity of the collapse, indicating some degree of genetic control (king et al. 1993). the heritability for standing tree acoustic velocity was 0.36 with a cgv of 6.24% for e. globoidea, similar to values reported in the literature: 0.26 to 0.75 for 11to 13-year-old p. menziesii (douglas fir) (dungey et al. 2012; klápště et al. 2019) and 0.35 for an e. fastigata progeny trial (suontama et al. 2018). the coefficient of genetic variation reported in this study was low. even lower values of genetic variability have been reported in the literature, for example 4.8% to 5.13% for p. menziesii and e. nitens, respectively (blackburn et al. 2014; klápště et al. 2019). the low genetic variability of acoustic velocity limits the potential for genetic improvement of stiffness. phenotypic and genetic correlations between traits of 8-year-old e. globoidea the phenotypic and genetic correlations between the wood traits are presented in tables 2 and 3. there were positive phenotypic correlations between diameter under bark and sapwood diameter (rp = 0.65) as well as heartwood diameter (rp = 0.88) (table 2). iyiola et al. new zealand journal of forestry science (2022) 52:13 page 5 table 2: phenotypic correlations between traits for 8-year-old e. globoidea (95% confidence interval in brackets). trait diameter under bark sapwood collapse heartwood diameter combined sapwood diameter extractive content acoustic velocity heartwood collapse −0.03 (−0.07, 0.02) 0.44 (0.41, 0.48) 0.03 (−0.03, 0.06) −0.09 (−0.13, −0.05) 0.11 (0.06, 0.15) −0.01 (−0.07, 0.05) diameter under bark −0.10 (−0.14, −0.06) 0.88 (0.87, 0.89) 0.65 (0.62, 0.67) −0.31 (−0.35, −0.27) 0.15 (0.10, 0.20) sapwood collapse −0.08 (−0.13, −0.05) −0.07 (−0.11, −0.02) −0.01 (−0.06, 0.03) −0.01 (−0.07, 0.05) heartwood diameter 0.20 (0.17, 0.25) −0.14 (−0.18, −0.10) 0.14 (0.09, 0.20) combined sapwood diameter −0.42 (−0.46, −0.39) 0.08 (0.02, 0.13) extractive content −0.07 (−0.13, −0.02) the strong positive correlations indicated that larger trees have wider heartwood as well as sapwood. similar phenotypic correlations were reported between dbh and heartwood diameter for 7-year-old e. bosistoana (rp = 0.59) (li et al. 2018), 22.5-year-old e. tereticornis sm. (rp = 0.79) (kumar & dhillon 2014), acacia melanoxylon r.br. (rp = 0.88) (knapic et al. 2006) and 30to 37-year-old plantation-grown pinus radiata d.don (rp = 0.71) (wilkes 1991). the genetic correlation between heartwood diameter and diameter under bark was stronger than the phenotypic correlation (rg = 0.91, ci95% 0.87, 0.95, table 3). strong positive genetic correlations between these traits were also reported for 9-year-old e. globulus (rg = 0.99) (miranda et al. 2014), 7-year-old e. bosistoana (rg = 0.89 to 0.98) (li et al. 2018), e. cladocalyx (rg = 0.44) (bush et al. 2011), l. kaempferi (rg = 0.87 to 0.92) (pâques 2001) and 35-year-old juglans nigra l. (rg = 0.98) (woeste 2002). no genetic correlation (rg = 0.022) was reported between the two traits for 25-year-old p. sylvestris (fries & ericsson 1998). the growth traits were favourably correlated with heartwood diameter being the most relevant measure when aiming to produce naturally durable timber. a negative phenotypic correlation was observed between heartwood diameter and extractive content at both the phenotypic (rp = −0.14, table 2) and genetic (rg = −0.45, table 3) level. a strong negative genetic correlation (rg = −0.86) was also reported between these traits for 7-year-old e. bosistoana (li et al. 2018). however, a positive genetic correlation between heartwood diameter and extractive content (rg = 0.32) was found for l. eurolepis (pâques & charpentier 2015). with the negative correlation observed for e. globoidea, trees with more heartwood tend to have lower amounts of extractives in the heartwood, resulting in lower decay resistance (li et al. 2020). selection for high durability will compromise heartwood quantity; however, as the correlation applies at the population level, individual trees that excel in both traits (i.e., so-called “correlation breakers”) may exist. the relationship between collapse in the sapwood and heartwood was positive at the phenotypic level (rp = 0.44, table 2), and the genetic level (rg = 0.64, table 3), suggesting that collapse is related to wood anatomy and amplified by heartwood extractives. no previously published results on the correlations between the two traits in other species were found for comparison. heartwood collapse was positively correlated with extractive content at both the phenotypic and genetic level (rp = 0.11; rg = 0.23). a positive phenotypic correlation was reported between the two traits for e. regnans f.muell. (chafe 1987). the positive correlation between the traits is in line with current understanding of the causes of drying collapse, as heartwood extractives reduce the permeability of the cell walls and consequently increase the negative pressure during evaporation of water. however, as these correlations were not strong other factors, e.g. density, contribute to collapse. there were only weak correlations between diameter under bark and acoustic velocity at both the phenotypic and genotypic level (rp = 0.15; rg = 0.18, table 2 and table 3). weak phenotypic relationships between these traits have also been reported for 8to 25-year-old p. radiata (0.04 to 0.18) (chauhan & walker 2006; toulmin & raymond 2007) and 25-year-old e. dunnii (0.14) (joe et al. 2004). in the current study, acoustic velocity was weakly correlated with heartwood diameter at the phenotypic level (rp = 0.14, ci95% 0.09, 0.20) but independent at the genetic level (rg = 0.10, table 3). therefore, our results indicate that the two traits need to be improved independently. furthermore, the weak unfavourable correlations between acoustic velocity and extractive content at the genetic level (rg = −0.15) and phenotypic level (rp = −0.07) (table 2) imply the need for compromise in selecting for the two traits simultaneously. iyiola et al. new zealand journal of forestry science (2022) 52:13 page 6 table 3: genetic correlations between the traits for 8-year-old e. globoidea (95% confidence interval in brackets). trait diameter under bark sapwood collapse heartwood diameter combined sapwood diameter extractive content acoustic velocity heartwood collapse −0.05 (−0.30, 0.22) 0.64 (0.36, 0.90) −0.09 (−0.36, 0.19) −0.02 (−0.25, 0.30) 0.23 (−0.01, 0.45) 0.05 (−0.39, 0.48) diameter under bark 0.18 (−0.15, 0.50) 0.91 (0.87, 0.95) 0.82 (0.75, 0.89) −0.67 (−0.81, −0.54) 0.18 (−0.09, 0.45) sapwood collapse 0.16 (−0.17, 0.51) 0.16 (−0.17, 0.52) −0.05 (−0.36, 0.29) 0.05 (−0.55, 0.62) heartwood diameter 0.53 (0.35, 0.71) −0.45 (−0.62, −0.28) 0.10 (−0.19, 0.43) combined sapwood diameter −0.79 (−0.89, −0.69) 0.24 (−0.01, 0.54) extractive content −0.15 (−0.41, 0.12) selection of superior genetics breeding programmes typically aim to improve multiple traits simultaneously, either by selecting the best or culling the poorest performing families. in the absence of economic weights for the traits, superior genotypes can be selected by choosing families with above average breeding values in several traits. approximately 15% of the 141 e. globoidea families showed both above average heartwood diameter and extractive content values (figure 1). however, in a situation where heartwood collapse as a third trait is considered, only four families met the criteria (highlighted in red in the top right quadrant). if the interest is in using wood for laminated veneer lumber (lvl), stiffness, growth (stem diameter) and collapse are important traits that need to be considered (figure 2). a total of 18 families highlighted in red in the top right corner of the quadrant met these criteria. heritability estimates and breeding values are dependent on the relatedness of the individuals and families. neither between family nor within family relatedness for the e. globoidea trees was known. the presented absolute values are likely overpredictions as no provenance effect was considered and families were not true half-siblings but a mixture of full-siblings and selfed individuals. nevertheless, the ranking of the families by their breeding values is independent of their relatedness and therefore better performing genotypes can be selected by the industry. conclusions the traits of e. globoidea investigated in this study were under varying degrees of genetic control and showed different degrees of variability. extractive content, an indicator of natural durability, and heartwood diameter, are the two key traits determining the value of the species for the intended use as ground-durable timber, and showed the highest coefficient of phenotypic and genetic variation (cpv = 47.87% and 28.90%). combined with the narrow sense heritability for heartwood diameter and extractive content of 0.51 and 1.16, respectively, significant genetic gain should be possible for these traits. iyiola et al. new zealand journal of forestry science (2022) 52:13 page 7 figure 1: relationship between family breeding values of heartwood diameter and ec for 141 e. globoidea families at age ~8 years. families performing above average are located in the top right quadrant. families with superior (red) and inferior (blue) average performance for heartwood collapse are highlighted. in accordance with the literature and the general understanding of the physical causes of collapse, heartwood was more prone to collapse than sapwood. collapse in the heartwood was under moderate genetic control and this study showed that maximal tangential collapse can be included at reasonable cost into a breeding programme if core samples are available. selection of superior genetics among the families would be practically feasible and could be used to increase utilisation of e. globoidea for solid timber products. acoustic velocity was under moderate genetic control (0.36) with low variability, implying improvement of this trait through breeding would be challenging. competing interests the authors declare no competing interests. authors’ contributions ei carried out experimental work, data analysis and drafted the manuscript. ca developed the measurement methods and conceived the study. la designed the trials and data analysis. all authors revised the manuscript. acknowledgements the nzdfi facilitated this study by establishing and maintaining the breeding programme. we appreciate m sharma, g hendriks, m holzenkämpfer, f guo, l nguyen, b disney, a millen, and r sheppard for their assistance in sample preparation. abbreviations cgv: coefficient of genetic variation cpv: coefficient of phenotypic variation dbh: diameter at breast height ec: extractive content lvl: laminated veneer lumber mfa: microfibril angle nir: near infrared spectroscopy nzdfi: new zealand dryland forests initiative funding this project was funded by the mbie specialty wood products partnership (ffrx1501 contract) and a university of canterbury phd scholarship awarded to ebenezer iyiola. iyiola et al. new zealand journal of forestry science (2022) 52:13 page 8 figure 2: relationship between breeding values of diameter under bark and acoustic velocity for 141 e. globoidea families at age ~8 years with the families superior for tangential collapse highlighted in red. references amusant, n., beauchene, j., fournier, m., janin, g., & thevenon, m.-f. (2004). decay resistance in dicorynia guianensis amsh.: analysis of inter-tree and intra-tree variability and relations with wood colour. annals of forest science, 61(4), 373-380. https://doi.org/10.1051/forest:2004030 arnold, r.j., johnson, i., & owen, j. (2004). genetic variation in growth, stem straightness and wood properties in eucalyptus dunnii trials in northern new south wales. forest genetics, 11(1), 1-12. as5604. (2005). timber natural durability ratings, australian standard, 29 p. awpc. (2007). protocols for assessment of wood preservatives. clayton, australia, the australian wood preservation committee. 30 p. bandara, k.m.a. (2006). genetic improvement of solid wood product value of subtropical eucalypts: a case study of eucalyptus grandis and e. dunnii. phd, australia national university, australia. blackburn, d., hamilton, m., williams, d., harwood, c., & potts, b. (2014). acoustic wave velocity as a selection trait in eucalyptus nitens. forests, 5(4), 744-762. https://doi.org/10.3390/f5040744 bootle, k. (2005). wood in australia types, properties and uses (2nd edition ed.). mcgraw-hill australia pty ltd. bourdon, r.m. (2000). understanding animal breeding (vol. 2). nj, usa: prentice hall. brooker, m.i.h., & kleinig, d.a. (1983). field guide to eucalypts. volume 1. south-eastern australia. melbourne, australia: inkata press pty ltd. bush, d., mccarthy, k., & meder, r. (2011). genetic variation of natural durability traits in eucalyptus cladocalyx (sugar gum). annals of forest science, 68(6), 1057-1066. https://doi.org/10.1007/ s13595-011-0121-z bush, d., & walker, j. (2011). selecting and breeding eucalypts for natural durability. developing a eucalypt resource: learning from australia and elsewhere. blenheim, nz: wood technology research centre, pp. 125-136. butler, d., cullis, b., gilmour, a., & gogel, b. (2009). asreml r package version 3.0. hemel hempstead, uk: vsn international ltd. cen. (1989). en 252. field test method for determining the relative protective effectiveness of a wood preservative in ground contact. brussels, belgium: european committee for standardisation. chafe, s. (1987). collapse, volumetric shrinkage, specific gravity and extractives in eucalyptus and other species. wood science and technology, 21(1), 2741. chafe, s., barnacle j.e, hunter a.j, llic r.l, northway, r., & roza a.n. (1992). collapse: an introduction. melbourne, australia: csiro division of forest products. chauhan, s.s., & walker, j.c. (2006). variations in acoustic velocity and density with age, and their interrelationships in radiata pine. forest ecology and management, 229(1-3), 388-394. https://doi. org/10.1016/j.foreco.2006.04.019 da silva, a.r., pastore, t.c.m., braga, j.w.b., davrieux, f., okino, e.y.a., coradin, v.t.r., camargos, j.a.a., & do prado, a.g.s. (2013). assessment of total phenols and extractives of mahogany wood by near infrared spectroscopy (nirs). holzforschung, 67(1), 1-8. https://doi.org/10.1515/hf-2011-0207 dungey, h., low, c., lee, j., miller, m., fleet, k., & yanchuk, a. (2012). developing breeding and deployment options for douglas-fir in new zealand: breeding for future forest conditions. silvae genetica, 61(16), 104-115. https://doi.org/10.1515/sg-20120013 eldridge, k., davidson, j., harwood, c., & wyk, g.v. (1994). eucalypt domestication and breeding. oxford, uk: clarendon press. elliott, c., & byrne, m. (2003). genetic diversity within and between natural populations of eucalyptus occidentalis (myrtaceae). silvae genetica, 52(3-4), 169-173. en-350-1. (1994). durability of wood and wood-based products natural durability of solid wood. part 1: guide to the principles of testing and classification of the natural durability of wood. brussels, belgium: european committee for standardization. estopa, r.a., milagres, f.r., oliveira, r.a., & hein, p.r.g. (2017). nir spectroscopic models for phenotyping wood traits in breeding programs of eucalyptus benthamii. cerne, 23(3), 367-375. https://doi.org/ 10.1590/01047760201723032319 fries, a., & ericsson, t. (1998). genetic parameters in diallel-crossed scots pine favor heartwood formation breeding objectives. canadian journal of forest research, 28(6), 937-941. https://doi. org/10.1139/x98-061 greaves, b.l., schimleck, l., borralho, n.m., michell, a.j., & raymond, c.a. (1996). genetic control and repeatability of near infrared reflectance from eucalyptus nitens woodmeal. appita journal, 49(6), 423-426. guo, f., & altaner, c. (2018). properties of rotary peeled veneer and laminated veneer lumber (lvl) from new zealand grown eucalyptus globoidea. new zealand journal of forestry science, 48:3. https:// doi.org/10.1186/s40490-018-0109-7 hamilton, m., potts, b., harwood, c., apiolaza, l., & gore, p. (2004). comparison of non-destructive iyiola et al. new zealand journal of forestry science (2022) 52:13 page 9 https://doi.org/10.1051/forest:2004030 https://doi.org/10.3390/f5040744 https://doi.org/10.1007/s13595-011-0121-z https://doi.org/10.1007/s13595-011-0121-z https://doi.org/10.1016/j.foreco.2006.04.019 https://doi.org/10.1016/j.foreco.2006.04.019 https://doi.org/10.1515/hf-2011-0207 https://doi.org/10.1515/sg-2012-0013 https://doi.org/10.1515/sg-2012-0013 https://doi.org/10.1590/01047760201723032319 https://doi.org/10.1590/01047760201723032319 https://doi.org/10.1139/x98-061 https://doi.org/10.1139/x98-061 https://doi.org/10.1186/s40490-018-0109-7 https://doi.org/10.1186/s40490-018-0109-7 iyiola et al. new zealand journal of forestry science (2022) 52:13 page 10 assessment techniques for shrinkage and collapse in eucalyptus nitens. paper presented at “eucalyptus in a changing world”. proceedings of iufro conference, 2 p. harju, a.m., & venäläinen, m. (2006). measuring the decay resistance of scots pine heartwood indirectly by the folin-ciocalteu assay. canadian journal of forest research, 36(7), 1797-1804. https://doi. org/10.1139/x06-074 hawley, l., fleck, l., & richards, c.a. (1924). the relation between durability and chemical composition in wood. industrial & engineering chemistry, 16(7), 699-700. https://doi.org/10.1021/ie50175a015 hong, z., fries, a., lundqvist, s.-o., andersson gull, b., & wu, h.x. (2015). measuring stiffness using acoustic tool for scots pine breeding selection. scandinavian journal of forest research, 30(4), 363-372. https:// doi.org/10.1080/02827581.2014.1001783 joe, w., dickson, r., raymond, c., ilic, j., & matheson, a. (2004). prediction of eucalyptus dunnii and pinus radiata timber stiffness using acoustics. a report for the rirdc/land & water australia/fwprdc/ mdbs joint venture agroforestry program. australia. https://www.agrifutures.com.au/wpcontent/uploads/publications/04-013.pdf jones, p., schimleck, l.r., daniels, r.f., clark, a., & purnell, r.c. (2008). comparison of pinus taeda l. whole-tree wood property calibrations using diffuse reflectance near infrared spectra obtained using a variety of sampling options. wood science and technology, 42(5), 385-400. https://doi. org/10.1007/s00226-008-0178-5 jones, t., mcconnochie, r.m., shelbourne, t., & low, c.b. (2010). sawing and grade recovery of 25-year-old eucalyptus fastigata, e. globoidea, e. muelleriana and e. pilularis. new zealand journal of forestry science, 40, 19-31. king, j.n., burdon, r., & young, g. (1993). provenance variation in new zealand-grown eucalyptus delegatensis. 2: internal checking and other wood properties. new zealand journal of forestry science, 23(3), 314-323. klápště, j., stovold, t., & dungey, h. (2019). evaluation of wood stiffness in douglas-fir progeny test fr280_2 and fr280_3. retrieved 25 april 2022 from https://www.nzffa.org.nz/system/assets/5213/ swp-t075.pdf knapic, s., tavares, f., & pereira, h. (2006). heartwood and sapwood variation in acacia melanoxylon r. br. trees in portugal. forestry, 79(4), 371-380. https:// doi.org/10.1093/forestry/cpl010 kube, p.d. (2005). genetic improvement of the wood properties of eucalyptus nitens: breeding to improve solid wood and pulp properties. doctor of philosophy. (phd), university of tasmania, australia. kumar, a., & dhillon, g. (2014). variation of sapwood and heartwood content in half-sib progenies of eucalyptus tereticornis sm. indian journal of natural products and resources, 5(4), 338-344. kuwabara, s. (2017). prediction of extractive content of eucalyptus globoidea heartwood using near infrared spectroscopy. bsc. hons, university of canterbury, new zealand. lachenbruch, b., moore, j.r., & evans, r. (2011). radial variation in wood structure and function in woody plants, and hypotheses for its occurrence. in: meinzer f.c., lachenbruch, b. & dawson, t.e. (eds.), size-and age-related changes in tree structure and function (pp. 121-164): springer. https://doi. org/10.1007/978-94-007-1242-3_5 li, y. (2018). use of near infrared spectroscopy to predict wood traits in eucalyptus species. phd, university of canterbury, new zealand. li, y., & altaner, c. (2019). calibration of near infrared spectroscopy (nirs) data of three eucalyptus species with extractive contents determined by ase extraction for rapid identification of species and high extractive contents. holzforschung, 73(6), 537-545. https://doi.org/10.1515/hf-2018-0166 li, y., apiolaza, l.a., & altaner, c. (2018). genetic variation in heartwood properties and growth traits of eucalyptus bosistoana. european journal of forest research, 137(4), 565-572. https://doi. org/10.1007/s10342-018-1125-0 li, y., sharma, m., altaner, c., & cookson, l.j. (2020). an approach to quantify natural durability of eucalyptus bosistoana by near infrared spectroscopy for genetic selection. industrial crops and products, 154, 112676. https://doi. org/10.1016/j.indcrop.2020.112676 lindström, h., harris, p., & nakada, r. (2002). methods for measuring stiffness of young trees. holz als roh-und werkstoff, 60(3), 165-174. https://doi. org/10.1007/s00107-002-0292-2 love, s. (2007). extended producer responsibility of treated timber waste. paper presented to sustainable building conference (sb07), auckland, 14-16 november. millen, p., altaner, c., & palmer, h. (2018). naturally durable timber posts performing well. new zealand tree grower 39(1), 24-26. millen, p., van ballekom, s., altaner, c., apiolaza, l., mason, e., mcconnochie, r., morgenroth, j., & murray, t. (2018). durable eucalypt forests-a multi-regional opportunity for investment in new zealand drylands. new zealand journal of forestry, 63, 11-23. miranda, i., gominho, j., araújo, c., & pereira, h. (2014). family effects in heartwood content of eucalyptus globulus l. european journal of forest research, https://doi.org/10.1139/x06-074 https://doi.org/10.1139/x06-074 https://doi.org/10.1021/ie50175a015 https://doi.org/10.1080/02827581.2014.1001783 https://doi.org/10.1080/02827581.2014.1001783 https://www.agrifutures.com.au/wp-content/uploads/publications/04-013.pdf https://www.agrifutures.com.au/wp-content/uploads/publications/04-013.pdf https://doi.org/10.1007/s00226-008-0178-5 https://doi.org/10.1007/s00226-008-0178-5 https://www.nzffa.org.nz/system/assets/5213/swp-t075.pdf https://www.nzffa.org.nz/system/assets/5213/swp-t075.pdf https://doi.org/10.1093/forestry/cpl010 https://doi.org/10.1093/forestry/cpl010 https://doi.org/10.1007/978-94-007-1242-3_5 https://doi.org/10.1007/978-94-007-1242-3_5 https://doi.org/10.1515/hf-2018-0166 https://doi.org/10.1007/s10342-018-1125-0 https://doi.org/10.1007/s10342-018-1125-0 https://doi.org/10.1016/j.indcrop.2020.112676 https://doi.org/10.1016/j.indcrop.2020.112676 https://doi.org/10.1007/s00107-002-0292-2 https://doi.org/10.1007/s00107-002-0292-2 iyiola et al. new zealand journal of forestry science (2022) 52:13 page 11 133(1), 81-87. https://doi.org/10.1007/s10342013-0741-y pâques, l. (2001). genetic control of heartwood content in larch. silvae genetica, 50(2), 69-74. pâques, l., & charpentier, j. (2015). perspectives for genetic improvement in heartwood size and extractive content in relation to natural durability and aesthetics in interspecific hybrid larch (larix × eurolepis). european journal of forest research, 134(5), 857-868. https://doi.org/10.1007/ s10342-015-0895-x pâques, l., garcía-casas, m.d.c., & charpentier, j. (2013). distribution of heartwood extractives in hybrid larches and in their related european and japanese larch parents: relationship with wood colour parameters. european journal of forest research, 132(1), 61-69. https://doi.org/10.1007/s10342012-0654-1 poke, f.s., potts, b.m., vaillancourt, r.e., & raymond, c.a. (2006). genetic parameters for lignin, extractives and decay in eucalyptus globulus. annals of forest science, 63(8), 813-821. https://doi.org/10.1051/ forest:2006080 r core team. (2019). r: a language and environment for statistical computing. r foundation for statistical computing, vienna, austria. url https://www.rproject.org/ raymond, c.a. (2002). genetics of eucalyptus wood properties. annals of forest science, 59(5-6), 525531. https://doi.org/10.1051/forest:2002037 santos, p., geraldi, i.o., & garcia, j.n. (2004). estimates of genetic parameters of wood traits for sawn timber production in eucalyptus grandis. genetics and molecular biology, 27(4), 567-573. https://doi. org/10.1590/s1415-47572004000400017 stackpole, d.j., vaillancourt, r.e., alves, a., rodrigues, j., & potts, b.m. (2011). genetic variation in the chemical components of eucalyptus globulus wood. g3: genes, genomes, genetics, 1(2), 151-159. https://doi.org/10.1534/g3.111.000372 suontama, m., miller, m., fleet, k., stovold, t., & dungey, h. (2018). genetic analysis of eucalyptus fastigata progeny trials and implications to selection. report for speciality wood products partnership. 10. retrieved 25 april 2022 from https://www.nzffa. org.nz/system/assets/5189/swp-t050.pdf toulmin, m.j., & raymond, c. (2007). developing a sampling strategy for measuring acoustic velocity in standing pinus radiata using the treetap time of flight tool. new zealand journal of forestry science, 37(1), 96-111. townsend, t.g., & solo-gabriele, h. (2006). environmental impacts of treated wood: crc press. https://doi. org/10.1201/9781420006216 tsuchikawa, s., & schwanninger, m. (2013). a review of recent near-infrared research for wood and paper (part 2). applied spectroscopy reviews, 48(7), 560587. https://doi.org/10.1080/05704928.2011.62 1079 unep. (2012). green carbon, black trade: illegal logging, tax fraud and laundering in the world’s tropical forests. united nations environment programme. https://wedocs.unep.org/20.500.11822/8030 white, t.l., adams, w.t., & neale, d.b. (2007). forest genetics. cabi. https://doi. org/10.1079/9781845932855.0000 whiteman, p. (1992). genetic parameters and selection strategies for eucalyptus nitens (dean and maiden) in victoria. silvae genetica, 41(2), 77-81. wilkes, j. (1991). heartwood development and its relationship to growth in pinus radiata. wood science and technology, 25(2), 85-90. https://doi. org/10.1007/bf00226808 woeste, k.e. (2002). heartwood production in a 35-yearold black walnut progeny test. canadian journal of forest research, 32(1), 177-181. https://doi. org/10.1139/x01-177 wu, s., zhu, y., xu, j., lu, z., chen, g., song, p., & guo, w. (2017). genetic variation and genetic gain for energy production, growth traits and wood properties in eucalyptus hybrid clones in china. australian forestry, 80(1), 57-65. https://doi.org/ 10.1080/00049158.2016.1275948 zobel, b., & jett, j.b. (1995). genetics of wood production. springer science & business media. https://doi. org/10.1007/978-3-642-79514-5 https://doi.org/10.1007/s10342-013-0741-y https://doi.org/10.1007/s10342-013-0741-y https://doi.org/10.1007/s10342-015-0895-x https://doi.org/10.1007/s10342-015-0895-x https://doi.org/10.1007/s10342-012-0654-1 https://doi.org/10.1007/s10342-012-0654-1 https://doi.org/10.1051/forest:2006080 https://doi.org/10.1051/forest:2006080 https://www.r-project.org/ https://www.r-project.org/ https://doi.org/10.1051/forest:2002037 https://doi.org/10.1590/s1415-47572004000400017 https://doi.org/10.1590/s1415-47572004000400017 https://doi.org/10.1534/g3.111.000372 https://www.nzffa.org.nz/system/assets/5189/swp-t050.pdf https://www.nzffa.org.nz/system/assets/5189/swp-t050.pdf https://doi.org/10.1201/9781420006216 https://doi.org/10.1201/9781420006216 https://doi.org/10.1080/05704928.2011.621079 https://doi.org/10.1080/05704928.2011.621079 https://wedocs.unep.org/20.500.11822/8030 https://doi.org/10.1079/9781845932855.0000 https://doi.org/10.1079/9781845932855.0000 https://doi.org/10.1007/bf00226808 https://doi.org/10.1007/bf00226808 https://doi.org/10.1139/x01-177 https://doi.org/10.1139/x01-177 https://doi.org/10.1080/00049158.2016.1275948 https://doi.org/10.1080/00049158.2016.1275948 https://doi.org/10.1007/978-3-642-79514-5 https://doi.org/10.1007/978-3-642-79514-5 performance of a whole tree mechanised timber harvesting system when clear-felling a 32-year-old pinus taeda l. stand natali de oliveira pitz1; jean alberto sampietro1*; erasmo luis tonett1; luis henrique ferrari1; philipe ricardo casemiro soares1; marcelo bonazza2; daiane alves de vargas1; marcos felipe nicolleti1; renato cesar gonçalves robert3 1department of forest engineering, santa catarina state university (udesc), lages, santa catarina, brazil. 2department of agriculture, biodiversity and forests, federal university of santa catarina, curitibanos, santa catarina, brazil. 3department of forest engineering and technology, federal university of paraná (ufpr), curitiba, paraná, brazil. *corresponding author: jean.sampietro@udesc.br (received for publication 27 january 2020; accepted in revised form 29 september 2021) abstract background: work studies are fundamental for the development and assessment of timber harvesting systems aimed at rationalising and improving forest management activities. methods: this study evaluated the operational performance of a mechanised whole-tree harvesting system in 32-yearold pinus taeda l. stands producing multiple timber products. a time and motion study at the cycle element level was conducted to evaluate the operational performance of each component of the harvesting system. equations were developed to estimate the productivity of tree extraction activity with a wheeled skidder and log loading with a mechanical loader. results: tree felling with an excavator-based harvester had the highest mean productivity (135 m3 per productive machine hour), followed by tree extraction with a wheeled skidder (117 m3 per productive machine hour), while manually processing larger logs with a chainsaw had the lowest productivity (25.7 m3 per productive machine hour). operator, extraction distance and mean log volume had a significant effect on the performance of different activities and were included in productivity models. conclusions: operational performance of equipment was variable and dependent on the effect of the operator, extraction distance and log volume. thus, the use of models to estimate productivity considering such factors, coupled with reduced delays to increase utilisation of equipment, will contribute to the better management and planning of forest harvesting operations under the evaluated conditions. new zealand journal of forestry science de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 https://doi.org/10.33494/nzjfs512021x96x e-issn: 1179-5395 published on-line: 19/10/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access diversification of timber products, the planted forests in brazil are usually managed on longer rotations (i.e. around 30 years), which requires the need for thinning, before the final cutting of the stand occurs. such a prescription affects the performance dynamics of forest operations, with several differences compared to those observed in short rotation planted forests. working conditions in final cutting operations tend to allow higher operational performance and lower unit costs introduction the management of planted pine forests is a consolidated activity in brazil, both by vertically integrated companies and independent producers. however, the forest management strategies adopted by vertically integrated companies differ from those adopted by independent producers, who typically aim to diversify forest production to market logs for different industrial segments. when the objective of forest production is keywords: forest operations and techniques; work study; forest mechanisation. http://creativecommons.org/licenses/by/4.0/), de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 page 2 compared to thinning operations. one of the key factors driving this is better access and mobility for machinery traffic due to the relatively small number of trees per unit area remaining at the final cutting. because these trees have larger dimensions and volumes, this results in lower specific time consumption and higher productivity of harvesting equipment (ghaffariyan et al. 2012; strandgard et al. 2013; walsh & strandgard 2014). however, the large size of the trees can lead to increased safety risks, which implies the use of specific techniques when performing cutting and extraction activities. in addition, a wide variety of log assortments are produced, which increases the complexity of operational aspects of pre-extraction, stacking and organisation of timber. in brazil, these forest operations are carried out predominantly using cut-to-length (ctl) or whole-tree (wt) harvesting systems (seixas & oliveira júnior 2001). in most cases the typical wt harvesting systems consist of a feller-buncher, skidders and processors, where only one machine perform all the tree bucking and processing (rocha et al. 2009; lopes et al. 2017; diniz et al. 2018a; rodrigues et al. 2019). however, there a very few studies analysing this system in brazilian pine plantations managed on longer rotations, (pereira et al. 2015; souza et al. 2018), especially when machinery configurations differ from the typical wt system. the evaluation of timber harvesting systems is essential for correcting and changing the production process to rationalise and optimise resources (magagnotti & spinelli 2012; ackerman et al. 2014; szewczyk et al. 2017). it is also an indispensable instrument for comparing different methods or equipment (spinelli et al. 2014; marčeta & košir 2016; pajkoš et al. 2018). our study aimed to: (i) evaluate the operational performance of a mechanised whole tree harvesting system in the final cutting of pinus taeda stands; (ii) verify the effect of operational factors on specific time consumption and productivity; and (iii) model the relationship between productivity and operational factors to provide information to improve management of these activities. methods study site and stand characteristics the study was conducted in a commercial forest stand in capão alto, santa catarina state, brazil. the terrain slope was level to gentle according to forestry commission uk (1996) (level=0°-6°, gentle= 6°-11°, moderate=11°-18°, steep=18°-27°, very steep=>27°) and the climate is classified as cfb according to köppen-geiger with no defined dry season, and mild summers (peel et al. 2007). the annual mean temperature ranges from 14 to 16°c and the annual precipitation is between 1600 to 1900 mm (alvares et al. 2013). the forest stand consisted of pinus taeda and its purpose was to produce wood for multiple uses, so it was subjected to four thinning interventions. our study was performed when the stand was undergoing the final felling, at the age of 32 years, with a stand density of 357 trees/ha, mean diameter at breast height of 45 cm, mean total height of 31 m and a mean individual tree volume of 2.46 m3. harvesting operations we evaluated a mechanised “whole tree” harvesting system configured to produce different demands of log assortments for different destinations. the system consisted of an excavator-based harvester (cat fm 320d) coupled to a 7000xt logmax head which felled the trees and a wheeled grapple skidder (john deere 748h), which extracted the trees from the cutting area to the roadside landing, with an extraction distance ranging from 30 to 310 m. at the roadside landing area, the trees were bucked and processed in three stages by different equipment. the first logs cut from the trees (large logs) were destined for sawmills and lamination plants, and had volumes ranging from 0.232 to 0.870 m3 log-1, smallend diameters ranging from 35 to 70 cm and, often had an irregular shape at the base. these were manually processed using a chainsaw (stihl ms 361) due to the limitations of other cutting equipment. the intermediate volume logs (medium logs), destined for sawmills with volumes between 0.157 to 0.227 m3 log-1 and diameter at the smaller end ranging from 25 to 35 cm, were processed using a mechanical slasher coupled to a caterpillar 320b. the lower volume logs (small logs), destined for pulp and mechanical processing with volumes between 0.087 to 0.132 m3 log1 and small-end diameters ranging from 8 to 25 cm, were processed by the same excavator-based harvester that was used for tree felling, but at a later point in time. the logs were stacked into product piles and organised in seven different log assortments according to small-end diameter and presence/absence of knots with lengths ranging from 1.90 to 3.10 m. after a period of between two to five days, the logs were loaded onto transport vehicles with a mechanical crawler loader (caterpillar 320b). the work schedule and utilisation of each piece of equipment within the harvesting system depended on commercial production needs and operational work restrictions. wood residues were not taken back into the stand. performance evaluation the operational performance of the activities was assessed by time and motion study at the cycle element level following the modelling approach (magagnotti & spinelli 2012). the work cycle of each piece of equipment was divided into elements (table 1) and then the time consumption was measured by the individual time clocking technique using a centesimal chronometer and specific forms. the number of trees felled, extracted or bucked at each working cycle was recorded. the volume produced at each working cycle (in cubic meters of solid wood over bark) was determined by multiplying the number of trees (or logs) by the mean individual tree (or log) volume for the stand. data on the volumes of individual trees and logs for the stand were obtained from the forestry company’s inventory records. de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 page 3 table 1: description of the study sites table 1: elements of the work cycle of each equipment and function/activity of harvesting system. equipment function/activity work cycle element description excavatorbased harvester tree felling movement (mv) equipment moving to the target tree boom movement and tree felling (bf) boom swings towards tree and executes felling drop and bunch organisation (db) felled trees dropped and organised into bunches wheeled skidder tree extraction travelling empty (te) movement of equipment from roadside landing to cutting area, close to the felled tree bunch manoeuvring and loading (ml) manoeuvring and loading of the tree bunch in the equipment’s grapple travelling loaded (tl) movement of equipment with tree bunch from cutting area to roadside landing area unloading and manoeuvring (um) manoeuvring and unloading the of tree bunch at the roadside manual chainsaw processing larger logs1 movement (mv) worker moves towards bunch of trees to execute log bucking log measurement (lm) worker measures the logs length with a stick and marks the location for cross-cutting tree bucking (tb) worker executes the crosscut and, if necessary, delimbs some branches mechanical slasher processing medium logs2 boom movement (bm) boom swings towards to the stem bunch accumulation and organisation (ao) stem accumulation and organisation in mechanical slasher stem bucking (sb) the mechanical slasher’s saw is activated and cuts the bunched stems swinging loaded grapple (sg) swinging loaded grapple with logs excavatorbased harvester processing small logs3 movement (mv) equipment moves to the stem bunch boom movement (bm) boom swings towards the bunched stems processing logs (pr) the processor head’s saw is activated and cuts the bunched stems mechanical loader loading logs onto the trucks swinging empty grapple (seg) empty grapple (unloaded) swings towards to the log pile grappling logs (gal) log bunch accumulation and organisation in equipment grapple swinging logs (slg) grapple loaded with logs swings towards to the trailer or semi-trailer of secondary transport vehicle bunking the logs in the truck (bat) bunking the logs in the trailer or semi-trailer of secondary transport vehicle 1 logs destined for sawmills and lamination with volume ranging from 0.232 to 0.870 m3 log-1. 2 logs destined for sawmills with volume ranging from 0.157 to 0.227 m3 log-1.. 3 logs destined for pulp and mechanical processing with volumes ranging from 0.087 to 0.132 m3 log-1. data referring to the operational performance factors were also measured for each working cycle. the operator (op) was considered a fixed-effect factor and different operators were only evaluated for the wheeled skidder and the mechanical loader. the slope (in degrees) was assessed with a trupulse 360 laser rangefinder. the extraction distance (ed, in meters) for a wheeled skidder, which corresponded to the distance between where the skidder stopped to load trees and then stopped to unload trees, was measured with the same device also used to assess the slope. the mean log volume (lv, in m3 log-1) for the mechanical loader was calculated by dividing the total loaded volume in a cycle by the number of logs loaded in the same cycle. the specific time consumption (s m-3) was calculated as the ratio between the time consumed for each element and the production in the respective work cycle. the productivity per productive machine hour without any delays (ppmh, m 3 pmh0 -1) was calculated as the ratio between the production in the work cycle and the total time consumed in the respective cycle (excluding delays). delay times were recorded and classified according to the iufro time model (björheden et al. 1995) so that the availability (ar) and utilisation rate (ur) could be calculated according to ackerman et al. (2014). data analysis specific time consumption and ppmh data were analysed by descriptive statistics and expressed as box and whisker plots. the estimation error for the ppmh variable was determined at the 95% level of probability significance, according to szewczyk et al. (2017). the de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 page 4 effect of influential factors for some activities was analysed using anova. prior to analysis, the data were subjected to a kolmogorov-smirnov normality test at 5% significance level and, in the case of non-normality, were mathematically transformed to achieve normality. for tree extraction activity with the wheeled skidder and log loading with the mechanical loader, multiple linear regression equations were fitted using a stepwise approach to test the effect of different independent variables on ppmh. goodness of fit for the models was evaluated by the adjusted coefficient of determination and absolute and relative standard errors of estimates. results mean values, estimation errors, and ratios of performance measures among the activities and equipment evaluated, the highest estimation error (ԑ) was found for processing medium logs with the mechanical slasher (ԑ = 9.80%), followed by tree extraction with a wheeled skidder (ԑ = 8.88%) (table 2). the activity of processing large logs with the manual chainsaw had the highest ar but the highest mean tcycle, lowest mean ppmh and ur (table 2). the highest operational performance was observed in the activity of tree felling with the excavator-based harvester (tcycle = 32.4 s m -3 and ppmh = 135 m 3 pmh0 -1), although this had the lowest ar (66.2%), and for the tree extraction with the wheeled skidder (tcycle = 45.8 s m -3 and ppmh = 117 m 3 pmh0 -1), which had the highest ur (61.0%). equipment function/ activity tcycle (s m-3) vcycle (m3 cycle-1) ppmh (m3 pmh0 -1) n (cycles) ԑ (%) ar (%) ur (%) excavator-based harvester tree felling 32.39 (±14.33) 2.46 (±1.28) 135.05 (±61.86) 223 5.95 66.2 44.5 wheeled skidder tree extraction 45.80 (±33.94) 6.60 (±2.47) 117.19 (±71.60) 276 8.88 93.2 61.0 manual chainsaw processing larger logs1 167.59 (±118.18) 0.498 (±0.158) 25.69 (±10.99) 416 4.35 94.7 49.3 mechanical slasher processing medium logs2 79.18 (±42.39) 0.937 (±0.301) 61.69 (±36.70) 246 9.80 90.1 50.0 excavator-based harvester processing small logs3 46.67 (±20.18) 1.743 (±0.932) 98.87 (±64.54) 513 3.58 89.7 34.5 mechanical loader loading logs onto the trucks 68.88 (±55.44) 0.960 (±0.421) 76.64 (±43.56) 502 6.32 86.7 29.9 1 logs destined for sawmills and lamination with volume ranging from 0.870 to 0.232 m3 log-1. 2 logs destined for sawmills with volume ranging from 0.227 to 0.157 m3 log-1. 3 logs destined for pulp and mechanical processing with volume ranging from 0.132 to 0.087 m3 log-1. table 2: mean values (± standard deviation) for total time taken per work cycle, volume per cycle, productivity per productive machine hour, estimation error, availability and utilisation rate for each piece of equipment and function/activity of the harvesting system. time consumption and effect of factors on performance for tree felling with an excavator-based harvester, bf was the element that consumed most time within the work cycle (figure 1a). there was also a significant effect of ground slope on the time consumed in this element (table 3). for tree extraction with a wheeled skidder (figure 1b), tl and te were the elements that consumed most time in the work cycle; they varied significantly between machine operators as indicated by anova (table 3). significant differences between machine operators were also observed on all other variables related to operational performance, except um (table 4). on average, operator 2 took more time and extracted 10.2% less volume per work cycle than operator 1, resulting in a 27.6% mean productivity difference (mean ppmh of 146 m 3 pmh0 -1 for operator 1 compared with 82.8 m3 pmh0 -1 for operator 2). ground slope had a significant effect on the tl, um and vcycle elements, which was due to the increased difficulty of working on steeper slopes. however, there was no significant effect of slope on tcycle and ppmh. extraction distance had a significant effect on all variables assessed for the skidder operation, except vcycle. it was also the single explanatory variable in the models for wheeled loader productivity (table 4). even though there was a significant difference in performance between the two operators, longer extraction distances resulted in more time being consumed which consequently reduced productivity (figure 2a). among the tree processing activities, more time per work cycle was consumed using a manual chainsaw (figure 1c) compared with a mechanical slasher and excavator-based harvester (figure 1d and 1e, respectively). most of the time consumed in the manual chainsaw work cycle occurred at the tb element due to the large size of the logs and, as already mentioned, because the activity was performed with manual equipment. ao was the element that consumed most of the work cycle time for processing medium logs with the mechanical slasher (figure 1d). in the case of processing small logs with the excavator-based harvester, most of the time consumed in the work cycle was with the pr element. there was a significant difference in the work cycle times between the two operators (figure 1f, table 3), but there was also a significant effect of mean log volume. in general, the bat element was responsible for most of the time taken during the work cycle (figure 1f ), which was due to the need to optimise the space occupied by the load on trucks. de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 page 5 figure 1: boxplot (showing quartiles, minimum, maximum and mean values) of specific time consumed for each work cycle element for each evaluated activity, equipment and operator (if applicable), where: (a) tree felling with the excavator-based harvester, (b) tree extraction with the wheeled skidder, (c) processing larger logs with the manual chainsaw, (d) processing medium logs with the mechanical slasher, (e) processing small logs with the excavator-based harvester and (f ) loading log with mechanical loader. discussion analysis of the operational performance of harvesting system equipment estimation errors (table 2) were due to variability of the operational performance of activities that, in turn, varied depending on interactions with factors such as the mean volume per tree, type of log assortment produced, extraction distance, slope, operator and among others. however, values of ԑ did not exceed 10% for any of the activities. under conditions of lower mean tree individual volume, pereira et al. (2015) reported slightly higher values of ppmh for tree felling with a tracked feller buncher than those observed in the present study and lower values for extraction activity with wheeled skidder. the operational performance for manually processing large logs with a chainsaw (table 2) was similar to that found by leite et al. (2014), although the latter study was conducted in eucalyptus plantations with lower mean individual-tree volume. de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 page 6 table 3: anova results showing the significance of different factors on aspects of operational performance. equipment function performance metric factors affecting performance slope operator extraction distance log volume excavator-based harvester tree felling mv 0.494 bf 0.030 db 0.978 tcycle 0.779 ppmh 0.648 wheeled skidder tree extraction te 0.181 <0.001 <0.001 ml 0.262 0.002 <0.001 tl 0.002 0.002 <0.001 um <0.001 0.277 0.018 tcycle 0.267 <0.001 <0.001 vcycle 0.035 <0.001 0.942 ppmh 0.262 <0.001 <0.001 mechanical loader loading logs onto the trucks seg <0.001 0.201 gal 0.459 <0.001 slg <0.001 0.001 bat <0.001 0.001 tcycle <0.001 <0.001 vcycle <0.001 <0.001 ppmh <0.001 <0.001 bold values indicate significant effect on a probability level of at least 5%. equipment: wheeled skidder function/activity: tree extraction op fitted equation adj. r2 se (m3 pmh0 -1) se (%) 1 ln(ppmh) = 5.482 0.006 ed 0.358 62.71 42.96 2 ln(ppmh) = 5.211 0.007 ed 0.430 31.79 38.41 equipment: mechanical loader function/activity: loading logs op fitted equation adj. r2 se (m3 pmh0 -1) se (%) 1 ln(ppmh) = 1.003 0.460 vcycle 2 + 3.320 √vcycle + 1.686 √lv 0.678 27.23 36.30 2 ln(ppmh) = 4.579 + 0.831 ln(vcycle) + 0.217 ln(lv) 0.595 28.59 31.73 2 ln(ppmh) = 4.542 + 0.675 ln(vcycle) + 0.304 ln(lv) 0.550 26.49 38.36 table 4: regression equations to estimate the productivity of the tree extraction with wheeled skidder and loading log with mechanical loader for each of the two operators. the ppmh of processing medium size logs with the mechanical slasher was higher than the value reported by conrad iv and dahlen (2019) and the ppmh of processing small logs with the excavator-based harvester was also higher compared with that reported by ghaffariyan et al. (2012) and scorupski et al. (2017), however, the operational conditions of these studies were different. the values of ppmh for log loading with a mechanical loader reported by ghaffariyan et al. (2012) were higher compared with present study, although in conditions of higher mean log volume. it should be noted that in most published studies, one piece of equipment performs all the tree bucking and processing operations in the wt harvesting system, which differs from the system studied here. this characteristic leads to a higher probability of occurrence of production bottlenecks, requiring attention in operational management to avoid this. in the current study, although the harvester had the highest ppmh, several delays occurred mainly due to corrective maintenance of the harvester head, resulting in the lowest ar value and, consequently, a relatively low ur. the wheeled skidder ar value was relatively high, and its ur was the highest, with delays due to auxiliary activities. in moments of “excess production time”, equipment performed other functions (support or production at another stage). therefore, better mechanical maintenance practices and use of the equipment according to the limits of technical capacity, can improve the excavator-based harvester availability and increase the overall system production. low operating performance was expected for manual processing with a chainsaw due to this being the only non-mechanised activity within the harvesting system. the need to wait for the trees on the roadside landing area to be organised at the end of each extraction cycle of the wheeled skidder caused most of the delays. hence, greater attention to operational management is required so that this activity does not become the bottleneck of the production system, especially because it is more susceptible to adverse weather conditions and, thus, subject to the risk of accidents and low ur (shrestha et al. 2005; silayo & migunga 2014; fulvio et al. 2017), as observed in present study. the ar of the mechanical slasher and excavatorbased harvester processing medium and small logs, respectively, was relatively high. the delays due to rework and organisation logs in product stacks at roadside for subsequent loading reduced considerably the equipment ur. the mechanical loader had the lowest ur, which was due to delays caused by waiting for transport vehicles or displacement between log piles or roadside landing areas. it is important to highlight that the ratios reported may not reflect the real proportion of availability and utilisation due to this study being short-term. longterm studies are recommended for more accurately determining the usage ratios, as well as for estimating delays (spinelli & visser 2008; magagnotti & spinelli 2012). factors affecting performance and modelling of productivity terrain slope had a significant effect on some elements of the tree felling work cycle with the excavator-based harvester and extraction with the wheeled skidder (table 3). it is expected that the increase in terrain slope increases the degree of work difficulty and, consequently, the operational safety risks. however, this factor had no significant effect on tcycle and ppmh, which possibly occurred because the maximum inclination observed in this study was only 9 degrees and, therefore, did not impose any major restrictions on equipment mobility. in clearcutting of a pinus plantation with a lower mean individual tree volume, diniz et al. (2018b) reported that the wheeler skidder performance only tended to be negatively affected when the slope was above 26 degrees. on lesser slopes, the operator was able to compensate for increased cycle times on steeper areas by working more quickly on the flatter areas, thus avoiding any productivity reduction. the operator had a significant effect on most of the operational performance variables of the wheeled skidder and the mechanical loader (table 3) and, therefore, regression equations were fitted to individual operators (table 4). for both types of equipment, the ppmh was greater for the more experienced operator (op. 1 in figure de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 page 7 figure 2: productivity per operator as a function of extraction distance (a) and log volume (b) for wheeled skidder and mechanical loader, respectively. 2a,b), which suggests that it is important to invest in people development and training in order to improve performance in forest operations. the predictability and effect of extraction distance on operational performance of the wheeled skidder is widely reported in the scientific literature for various equipment and operational conditions (behjou et al. 2008; rocha et al. 2009; ghaffariyan et al. 2012; walsh & strandgard 2014; strandgard et al. 2017). in the case of the mechanical loader, log volume had a significant effect on most of the operational performance variables (table 3) and, thus, was included as predictive factor for estimating the ppmh in all regression equations (table 4). there was a tendency to increase the ppmh as the log volume increases (figure 2b), similar to observations made by diniz et al. (2018c) for other operational conditions. conclusions the operational performance of the equipment in the harvesting system studied was variable and dependent on the effect of the operator, extraction distance and log volume. for this reason and because it has more equipment and a greater number of processing stages than most of the whole-tree systems that have been studies, there is greater likelihood of production bottlenecks, requiring attention in operational management to avoid or minimise this. the use of models to estimate productivity considering such mentioned factors and reduced delays to increase availability and utilisation of equipment will contribute to the better management and planning of forest harvesting operations under the evaluated conditions. competing interests the authors declare that they have no competing interests. acknowledgements the authors would like to thank staff from agroflorestal paquerê without whom this study would not have been realised, the foundation for research support of the santa catarina state (fapesc) case number 2019tr657 for the financial assistance to the research groups, and reviewers for the constructive and helpful feedback. authors' contributions conceptualisation: np, js and ps; methodology: js, mb, and rr; formal analysis: all authors; writing – original draft: np and js; writing – review & editing: all authors; project administration: js; funding acquisition: js. all authors read and approved the final manuscript. abbreviations adj. r2_adjusted determination coefficient ao_stems accumulation and organisation ar_availability rate bat_bunking the logs at the truck bf_boom swing towards a tree and felling bm_boom movement ctl_cut-to-length db_drop and tree bunch organisation ed_extraction distance gal_grappling logs glm_general linear model lm_log measurement ln_neperian logarithm lv_log volume m3_cubic meter over the bark ml_ maneuvering and loading mv_movement n_number of observations op_operator pmh0_productive machine hour without any delays ppmh_productivity per productive machine hour pr_processing logs s_seconds sb_stems bucking se_estimated standard error seg_swinging empty grapple sg_swinging loaded grapple with logs slg_swinging logs toward the truck tb_tree bucking tcycle_total time consumption per work cycle te_travelling empty tl_travelling loaded um_unloading and manoeuvring ur utilisation rate vcycle_volume produced per work cycle wt_whole tree ԑ_sampling error references ackerman p., gleasure e., ackerman s. & shuttleworth b. (2014). standards for time studies for the south african forest industry. icfr/fesa, south africa. 49 p. alvares a.c., stape, j.l., sentelhas, p.c., de moares gonçalves, j.l. & sparovek, g. (2013). köppen’s climate classification map for brazil. meteorologische zeitschrift. 22, 711-728. https:// doi.org/10.1127/0941-2948/2013/0507 behjou f.k., majnounian, b., namiranian, m. & dvořák, j. (2008). time study and skidding capacity of the wheeled skidder timberjack 450c in caspian forests. journal of forest science, 54(4),183-188. https://doi.org/10.17221/5/2008-jfs björheden r., apel k., shiba m. & thompson m.a. (1995). iufro forest work study nomenclature. department of operational efficiency, swedish de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 page 8 https://doi.org/10.1127/0941-2948/2013/0507 https://doi.org/10.1127/0941-2948/2013/0507 https://doi.org/10.17221/5/2008-jfs university of agricultural science, garpenberg, sweden. 16 p. conrad iv j.l. & dahlen j. (2019). productivity and cost of processors in whole-tree harvesting systems in southern pine stands. forest science, 65(6), 767775. https://doi.org/10.1093/forsci/fxz036 diniz c.c.c., robert r.c.g. & vargas m.b. (2018a). avaliação técnica de cabeçotes individual e múltiplo no processamento de madeira. advances in forestry science, 5(1), 253-258. diniz c.c.c., nakajima, n.y., robert r.c.g. & dolácio, c. (2018b). performance of grapple skidder in different ground inclinations. floresta, 49, 41-48. https://doi.org/10.5380/rf.v49i1.55744 diniz c.c.c., cerqueira c.l. & oliveira f.m. (2018c). influência do sortimento de toras na produtividade de um carregador florestal. agropecuária científica no semiárido, 14(3), 247-253. https://doi. org/10.30969/acsa.v14i3.1050 forestry commission uk. (1996). terrain classification. available: accessed 01 december 2020. http:// www.biomassenergycentre.org.uk fulvio f.d., abbas, d., spinelli, r., acuna, m., ackerman, p. & lindroos, o. (2017). benchmarking technical and cost factors in forest felling and processing operations in different global regions during the period 2013-2014. international journal of forest engineering, 28(2), 94-105. https://doi.org/10.10 80/14942119.2017.1311559 ghaffariyan m.r., sessions j. & brown m. (2012). machine productivity and residual harvesting residues associated with a cut-to-length harvest system in southern tasmania. southern forests, 74(4), 229235. https://doi.org/10.2989/20702620.2012.74 1770 leite e.s., carlos, f.h., guedes, i.l. & amaral, e.j. (2014). análise técnica e de custos do corte florestal semimecanizado em povoamentos de eucalipto em diferentes espaçamentos. cerne, 20, 637-643. https://doi.org/10.1590/010477602014200413 40 lopes, e.s., oliviera, d., rodrigues, c.k. & drinko, c.h. (2017). variables influencing working time and skidder productivity in wood extraction. nativa, 5(4), 298-302. https://doi.org/10.5935/23187670.v05n04a12 magagnotti n. & spinelli r. (2012). good practice guidelines for biomass production studies. sesto fiorentino, cnr ivalsa. marčeta d.& košir b. (2016). comparison of two felling and processing methods in beech forests. croatian journal of forest engineering, 37(1), 163-174. pajkoš m., klvač, r, neruda, j. & mishra, p.k. (2018). comparative time study of conventional cut-tolength and an integrated harvesting method-a case study. forests, 194(9), 1-10. https://doi. org/10.3390/f9040194 peel m.c., finlayson, b.a. & mcmahon, t.a. (2007). updated world map of the köppen-geiger climate classification, hydrology and earth system sciences, 11, 1633-1644. https://doi.org/10.5194/hess-111633-2007 pereira a.l.n. et al (2015). análise técnica e de custo do feller buncher e skidder na colheita de madeira em diferentes produtividades do povoamento. ciência florestal, 25, 981-989. https://doi. org/10.5902/1980509820659 rocha e.b., fiedler, n.c., alves, r.t. & lopes, e.s. (2009). produtividade e custos de um sistema de colheita de árvores inteiras. cerne, 15, 372-381. rodrigues c.k., lopes e.s., pereira a.l.n. & sampietro j.a. (2019). effect of individual tree volume on operational performance of harvester processor. floresta, 49(2), 345-352. https://doi.org/10.5380/ rf.v49i2.58233 seixas f. & oliveira júnior e.d. (2001). compactação do solo devido ao tráfego de máquinas de colheita de madeira. scientia forestalis, 60, 73-87. shrestha s.p., lanford, b.l., rummer, r.b. & dobois, m. (2005). utilization and cost of log production from animal logging operations. international journal of forest engineering, 16(2), 167-180. https://doi.org /10.1080/14942119.2005.10702524 silayo d.s.a. & migunga a. (2014). productivity and costs modeling for tree harvesting operations using chainsaws in plantation forests, tanzania. international journal of engineering & technology, 3(4), 464-472. https://doi.org/10.14419/ijet. v3i4.3407 souza f.l., sampietro, j.a., dacoregio, h.m., soares, p.r.c., da silva lopes, e. & quadros, d.s. (2018). densidade ótima e aceitável de estradas na colheita de pinus no sistema de toras curtas e árvores inteiras. scientia forestalis, 46(118), 189-198. https://doi. org/10.18671/scifor.v46n118.05 spinelli r., lombardini c. & magagnotti n. (2014). the effect of mechanization level and harvesting system on the thinning cost of mediterranean softwood plantations. silva fennica, 48(1), 1-15. https://doi. org/10.14214/sf.1003 spinelli r. & visser r. (2008). analyzing and estimating delays in harvester operations. international journal of forest engineering, 19(1), 36-41. https:// doi.org/10.1080/14942119.2008.10702558 strandgard m., mitchell r. & acuna m. (2017). time consumption and productivity of a forwarder operating on a slope in a cut-to-length harvest system in a pinus radiata d. don pine plantation. journal of forest science, 63(7), 324-330. https:// doi.org/10.17221/10/2017-jfs de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 page 9 https://doi.org/10.1093/forsci/fxz036 https://doi.org/10.5380/rf.v49i1.55744 https://doi.org/10.30969/acsa.v14i3.1050 https://doi.org/10.30969/acsa.v14i3.1050 http://www.biomassenergycentre.org.uk http://www.biomassenergycentre.org.uk https://doi.org/10.1080/14942119.2017.1311559 https://doi.org/10.1080/14942119.2017.1311559 https://doi.org/10.2989/20702620.2012.741770 https://doi.org/10.2989/20702620.2012.741770 https://doi.org/10.1590/01047760201420041340 https://doi.org/10.1590/01047760201420041340 https://doi.org/10.5935/2318-7670.v05n04a12 https://doi.org/10.5935/2318-7670.v05n04a12 https://doi.org/10.3390/f9040194 https://doi.org/10.3390/f9040194 https://doi.org/10.5194/hess-11-1633-2007 https://doi.org/10.5194/hess-11-1633-2007 https://doi.org/10.5902/1980509820659 https://doi.org/10.5902/1980509820659 https://doi.org/10.5380/rf.v49i2.58233 https://doi.org/10.5380/rf.v49i2.58233 https://doi.org/10.1080/14942119.2005.10702524 https://doi.org/10.1080/14942119.2005.10702524 https://doi.org/10.14419/ijet.v3i4.3407 https://doi.org/10.14419/ijet.v3i4.3407 https://doi.org/10.18671/scifor.v46n118.05 https://doi.org/10.18671/scifor.v46n118.05 https://doi.org/10.14214/sf.1003 https://doi.org/10.14214/sf.1003 https://doi.org/10.1080/14942119.2008.10702558 https://doi.org/10.1080/14942119.2008.10702558 https://doi.org/10.17221/10/2017-jfs https://doi.org/10.17221/10/2017-jfs de oliveira pitz et al. new zealand journal of forestry science (2021) 51:12 page 10 strandgard m., walsh d. & acuna m. (2013). estimating harvester productivity in pinus radiata plantations using stanford stem files. scandinavian journal of forest research, 28(1), 73-80. https://doi.org/10.1 080/02827581.2012.706633 szewczyk g., sowa, j.m., kamiński, k., kulak, d. & stańczykiewicz, a. (2017). selection of time study methods for forest operations. forestry letters, 110, 1-12. walsh d. & strandgard m. (2014). productivity and cost of harvesting a stemwood biomass product from integrated cut-to-length harvest operations in australian pinus radiata plantations. biomass and bioenergy, 66, 93-102. https://doi.org/10.1016/j. biombioe.2014.01.017 https://doi.org/10.1080/02827581.2012.706633 https://doi.org/10.1080/02827581.2012.706633 https://doi.org/10.1016/j.biombioe.2014.01.017 https://doi.org/10.1016/j.biombioe.2014.01.017 impacts of genetic selection on sequoia sempervirens mini-cutting rooting and initial growth in the field queli cristina lovatel, gabriel teixeira da rosa, alexandra cristina schatz sá, betel cavalcante lopes, erasmo luis tonett, romell alves ribeiro dias, mariane de oliveira pereira and marcio carlos navroski* university of the state of santa catarina, lages, santa catarina, brazil corresponding author: marcio.navroski@udesc.br (received for publication 26 november 2019; accepted in revised form 24 august 2021) abstract background: vegetative propagation from superior individuals allows multiple copies of plants that are genetically identical to the parent plant to be obtained. however, vegetative propagation success varies among individual genotypes, with some clones having more difficulty forming roots than others. the aim of this study was to evaluate the genetic gain in sequoia sempervirens (d.don) endl. clones using parameters describing vegetative propagation success and initial growth in field. methods: vegetative propagation success was quantified for 16 clones in a completely randomised design consisting of 10 replications, each containing 10 mini-cuttings. at 90 days, rooting (rt), survival (sv) and the number of new shoots (ns) were evaluated. performance after planting in the field was assessed using 13 clones from the previous experiment, arranged in linear parcels of 10 plants with 8 replicates. after 18 months, survival (sv), stem diameter (sd), height (h) and dominance breakdown (db) were assessed. estimates of variance components, heritability and genetic correlations were obtained using the selegen-reml/blup software. results: the mini cuttings of the 16 clones had a coefficient of genetic variation (cvgi%) of 32.32% for rt, 5.44% for sv and 5.35% for ns. the heritability of the total genetic effects (h2 g) for rt was 0.68. the clones with the best predicted genotypic classifications for the characteristics evaluated in the field were a116, a140 and a138 for sv, a126, a140 and a138 for sd, a138, a140 and a117 for h and a138, a228 and a116 for db. conclusions: in general, it was possible to obtain high genetic gain for rooting and medium gain for dendrometric variables in the field. new zealand journal of forestry science lovatel et al. new zealand journal of forestry science (2021) 51:11 https://doi.org/10.33494/nzjfs512021x84x e-issn: 1179-5395 published on-line: 04/09/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access generally less than 10% and seedling viability is very low (ozudogru et al. 2011). in addition, young plants have lower seed viability, with higher values being obtained from 250-year-old trees (olson et al. 1990). alternatively, in natural stands the species can reproduce by vegetative propagation. when a tree is cut or burned, sprouts appear from the cut trunk or already established roots and grow more vigorously than other understory plants. thus, an individual’s genetic information is the same as that of the trees that inhabited the site thousands of years ago (luna 2008). introduction the species sequoia sempervirens (d.don) endl. known worldwide as sequoia or redwood, is native to north america, mainly from the central and northern california coast, a region with moderate to high rainfall in winter and fog in summer (olson et al. 1990; cown and mckinley 2008). sequoia wood is widely used for the construction of decks, fences, windows, doors, shutters and interior applications where appearance and stability are important requirements (cown and mckinley 2008). sequoia is a species that mainly propagates through seed production, however, the seed germination rate is keywords: vegetative propagation; clonal test; selegen-reml/blup; selection gain. http://creativecommons.org/licenses/by/4.0/), lovatel et al. new zealand journal of forestry science (2021) 51:11 page 2 for commercial forestry, vegetative propagation from superior individuals allows the multiplication of the best genotypes and plants that are genetically identical to the parent plant to be obtained. it also allows the multiplication of selected trees that contain favourable genetic combinations and the production of genetically homogeneous material that can develop in a predictable and uniform manner (wendling et al. 2014). mini cuttings (cuttings taken from mini hedges not previously rejuvenated by in vitro techniques) is one of the most widely used vegetative propagation techniques. it allows the selection of those individuals with the best characteristics in tree breeding programmes and can reduce the length of the breeding cycle (freitas et al. 2016). compared with traditional cutting methods, which suffer from “physiological ageing” problems with the donor plants, mini-cuttings offer advantages in the small size of clonal mini-hedges, high productivity, lower cost and better rooting indices (majada et al, 2011). plant responses to vegetative propagation and their consequences have received little attention in the quantitative genetics literature. by contrast, studies in horticulture and silviculture have examined some aspects of this problem. genetic variation in rooting ability has been observed in several economically important species and environmental effects on rooting have been widely studied to find methods to optimise rooting of cuttings for commercial production (schwaegerle 2005). although rooting is an important factor in the selection of clones in forestry, few studies address the use of this characteristic (rooting) in the early selection of superior genotypes. our hypothesis is that characteristics such as the rooting of cuttings and initial growth in the field following out-planting can be used as selection criteria for sequoia clones. therefore, the objective of this work was to evaluate the genetic gain in sequoia sempervirens clones using parameters related to vegetative propagation success and initial growth in the field. methods vegetative propagation and rooting success of minicuttings the experiment compared the propagation success and rooting of mini-cuttings from sixteen different clones (a100, a113, a115, a116, a117, a126, a127, a130, a131, a133, a135, a136, a138, a140, a227 and a228). material for the experiment was sourced from stumps of selected 40-year-old sequoia parent trees. conventional cuttings were collected from the matrix trees 120 days after partial or complete annealing, as described by pereira et al. (2017). this material was transported to the forest nursery in lages, santa catarina state, brazil. from this material, 10 cm cuttings were made, each containing a pair of acicular leaves with the total leaf area reduced by 50%. these cuttings were treated with 6,000 mg l-1 of indole-3-butyric acid (iba), set in tubes (180 cm3) containing medium-sized (2–4 mm) vermiculite and commercial substrate (1:1 v/v) and placed in a mini-tunnel for rooting to occur. five months after these cuttings were set, rooted plants were acclimated in a shade house for 30 days and then transferred to a greenhouse to complete acclimatisation for another 30 days. subsequently, plants approximately 20 cm high were transferred to 5-l pots filled with a commercial substrate consisting of peat and decomposed pine bark (50%) and medium-sized (2–4 mm) vermiculite (50%). fifteen days after installation (the time needed to adapt the plants to the system) the apex of the plant’s main sprout was pruned to a height of 10 cm (± 2 cm), thus forming the ministumps for the establishment of the clonal mini-garden. pruning of the mini-stumps was performed each month over a four-month period. the nutrient solution used in the clonal mini-garden was based on commercial fertiliser comprising 10% n (water-soluble); 42% p2o5 (water soluble); 10% k2o (water soluble); 0.6% mg; 0.1% fe and 0.02% br. fertigation was carried out twice a week, with each mini-strain receiving 50 ml of the solution composed by diluting 1.5 g of fertiliser in 1l of water. after the formation of mini-stumps, shoots were collected for the experiments. from each clone, 8 to 10 cm long mini-cuttings were prepared, with the basal portion bevel cut and the upper portion incised transversely, and with a pair of acicular leaves cut in half. once prepared, the cuttings were placed to root in 180 cm³ polypropylene tubes containing an average size particle (2–4 mm) of vermiculite and commercial substrate (1:1 v/v) with the addition of 6 g l-1 of controlled release fertiliser, in which the insertion of approximately 2 cm from the base of the mini-cutting into the substrate was made. the commercial plant substrate, according to the manufacturer, is composed of peat, expanded vermiculite, pine bark and charcoal. the chemical and physical characteristics after substrate analysis are as follows: ph = 6.6; electrical conductivity = 0.13 ms cm-1; wet density = 450.8 kg m-³; dry density = 302.7 kg m-³; current humidity = 32.8 (%); total porosity = 72.2 (%); aeration space = 17.0 (%); easily available water = 12.2 (%); buffering water = 2.4 (%); remaining water = 40.5 (%); whc10 – water holding capacity at 10 cm = 55.1; whc50 = 42.9 and whc100 = 40.5. the trays containing the tubes with the minicuttings were placed in a mini-tunnel – plastic covered greenhouse structure (measuring 8.0 m long x 1.20 m wide x 0.9 m high). the temperature inside the greenhouse usually varied between 20–32°c and the relative humidity remained above 80%, being irrigated by micro-sprinkling for 5 minutes, five times a day. the mini-cuttings remained in this condition for 90 days, at which time the evaluation of the experiment was performed. the experiment followed a completely randomised design with 10 replications of 10 minicuttings from each clone. survival percentage (sv), percentage of rooted cuttings (rt) and the number of new shoots (ns) were evaluated. mini-cuttings with live wood, old leaves or young shoots, rooted or not, were considered survivors. the rooting percentage was considered over the total, not only the surviving mini-cuttings. the number of roots was not evaluated because the plants from the experiment were destined for field planting. counting the number of roots implies substrate disruption, causing serious damage to the seedling roots. initial growth in the field the clonal mini-cuttings test was installed in curitibanos – santa cristina (brazil) in october 2017. the region, according to koppen, has the mesothermal humid subtropical climate (cfb) (alvares et al. 2013). prior to planting, subsoiling and a rotary hoe were used to prepare the site. plants from the rooting experiment were used for this study. at the time of planting, they were nine months old. of the 16 clones used in the rooting experiment, four of them did not have sufficient rooted material for initial growth in the field (a113, a115, a127, a13). clone a110 was added to field planting experiment but was not used in the rooting experiment. at the time of planting, the mini-cuttings were approximately 30 cm tall and had a root-collar diameter greater than 4 mm. fifteen days before planting, they underwent an acclimatisation period, with reduced irrigation and maintenance in full sun. fifteen days after planting, 150 g of npk (5-20-20) fertiliser was applied to each plant. the growth study was undertaken using 13 clones (a100, a110, a116, a117, a126, a130, a131, a135, a136, a138, a140, a227 and a228) in lines of 10 plants per clone with 8 replicates for a total of 80 plants tested per clone. the planting spacing used was 3.0 x 3.0 m. all clonal stock was in the form of mini-cuttings produced from the clonal mini-garden. eighteen months after out-planting in the field, the following variables were measured: survival (sv), stem diameter (sd), height (h) and dominance breakdown (db) in relation to the early testing of sequoia clones in the field. statistical analysis estimates of variance components, heritability and genetic correlations were obtained from mixed models fitted using the selegen-reml/blup software (resende 2016). an alternative to be used in the construction of indexes that can lead to a more accurate selection process is the use of restricted maximum likelihood (reml) and best linear unbiased prediction (blup). this method is based on the assumptions that the smaller the standard deviation of genotypic behaviour between two sites, the greater the harmonic mean of their genotypic values between two sites (silva et al. 2011; rosado et al. 2012). the reml/blup procedure has several advantages as it considers genotypic effects as random, can deal with imbalance, non-orthogonality and heterogeneity of variances, outliers, correlated errors within locations, provides breeding values after discounting environmental effects, and can be applied to any number of environments (resende 2007). vegetative propagation and rooting success were evaluated assuming a completely randomised design, with testing of unrelated clones and more than one plant per parcel. the following statistical model was used for determining the genetic variance components: y= xu + zg + e [1] where: y = data vector; u = fixed effect vector of the general average; g = vector of genotypic effects assumed to be random; e = error vector or random residue. the variance components analysed for vegetative propagation and rooting were: σ2g= genotypic variance, σ2e= residual variance, σfi= individual phenotypic variance, h2g= heritability of individual parcels in the broad sense (i.e., effects of total genotypes), cvgi (%) = genotypic variation coefficient and cve (%) = residual variation coefficient and overall mean of the experiment. analysis of the different parameters related to the initial growth in the field was undertaken assuming a randomised block experimental design with several plants per parcel and testing of unrelated clones. the following statistical model was used for determining the genetic variance components: y = xr + zg + wp + e [2] where y = data vector; r = repeating effects vector (assumed to be random) plus the overall mean; g = genotypic effects vector assumed to be random; p = parcel effects vector; e = error vector or random residual. the variance components analysed in the field clone test were: σ2g= genotypic variance, σ 2 parc= environmental variance between parcels, σ2e= residual variance, σ2fi= individual phenotypic variance, h 2 g= heritability of individual parcels in the broad sense (i.e., the effects of total genotypes), h2aj= individual heritability in the broad sense (adjusted for parcel purposes), c²parc= coefficients for determining the effects of parcels, h2mc= heritability of the mean genotype, acclon(%) = the genotype selection accuracy, cvgi(%) = coefficient of genotypic variation, cve(%) = residual coefficient of variation, cvr= coefficient of relative variation, pev = prediction error variance of the genotypic values, sep = standard deviation of predicted genotypic value and overall mean of the experiment. results vegetative propagation and rooting of mini-cuttings results from the experiments with sequoia minicuttings showed that there was considerable potential to improve vegetative propagation success through genetic selection. the coefficient of genetic variation (cvgi%) for rt was 32.32% (table 1), which was high when compared with the same parameter obtained for sv (5.44) and ns (5.35). heritability estimates for the sv and ns variables were low, being 0.16 and 0.04, respectively. with this wide variation, genotypic means for rt ranged from 23.73% for clone a138 to 98.77% for clone a113 (table 2). mean sv values were above 85% for all clones and ns percentage was above 76%. the best results for rt were obtained for clones a113 (97.7%), a127 (97.8%), a136 (96.9%) and a115 (94.5%). these same clones exhibited good performance across the other traits that were assessed. the heritability of the total genetic effects (h2g) for the rooting variable was considered high (0.68), which has lovatel et al. new zealand journal of forestry science (2021) 51:11 page 3 table 1: description of the study sites important implications for the genetic selection of clones as this is the most limiting feature for the propagation of clones via mini-cuttings. initial growth in the field the clones with the highest survival when planted in the field were a116, a140 and a138; for stem diameter the best performing clones were a126, a140 and a138; for height: the best performing clones were a138, a140 and a117 and for dominance breakdown the best performing clones were a138, a228 and a116 (table 3). one of the best performing clones in the field (a138) had lovatel et al. new zealand journal of forestry science (2021) 51:11 page 4 the worst rooting score in the greenhouse experiment, and by contrast clone a136, which showed poorer field performance, had the third best rooting score (table 2). genotypic mean values (u+g) ranged from 0.33 to 0.81 (%) for sv; 12.02 to 27.25 (mm) for sd; 58.60 to 118.53 (cm) for h and 0.42 to 0.70 (%) for db. the worst performing clones in terms of survival, stem diameter, height and dominance breakdown were a117, a228, a228 and a136, respectively. high values of genotypic coefficients of variation (20.06 to 34.48%) were found, indicating that these traits, especially survival, have good potential to be selected for (table 4). the heritability of component sv (%) rt (%) ns σ2ga 26.70 580.00 19.45 σ2e 131.31 262.49 401.17 σ2fi 158.02 842.50 420.62 h2a 0.1689 ± (0.0841) 0.6884 ± (0.1698) 0.0462 ± (0.044) cvgi (%) 5.44 32.32 5.35 cve (%) 12.06 21.74 24.30 general average 94.98 74.49 82.41 table 1: variance components for the survival (sv), rooting (rt) and number of new shoots (ns) for 16 sequoia sempervirens clones measured 90 days after setting. clone rt (%) sv (%) ns rk¹ g u + g rk g u + g rk g u + g a100 15 -35.8052 38.6894 4 3.2440 98.2274 11 -0.902 81.5105 a113 1 24.2842 98.7788 7 2.0225 97.0060 7 0.786 83.1988 a115 4 20.0443 94.5389 6 2.3987 97.3822 13 -1.314 81.0987 a116 8 9.4153 83.9099 11 1.0166 96.0000 10 -0.687 81.7257 a117 13 -25.0068 49.4878 13 -4.7069 90.2765 3 2.621 85.0745 a126 14 -33.6137 40.8809 16 -9.7276 85.2559 14 -2.504 79.9084 a127 2 23.3836 97.8782 5 2.6909 97.6744 2 3.671 86.1796 a130 11 2.8962 77.3908 9 1.5486 96.5320 15 -3.831 78.5816 a131 12 -6.8018 67.6928 8 1.9007 96.8842 12 -0.942 81.4709 a133 9 9.3564 83.8510 1 3.7779 98.7614 1 4.094 86.5065 a135 10 4.8486 79.3433 12 0.1708 95.1543 9 -0.196 82.2168 a136 3 22.4858 96.9804 3 3.2440 98.2274 4 1.732 84.1442 a138 16 -50.7630 23.7316 2 3.7779 98.7614 16 -5.761 76.6514 a140 7 11.1684 85.6630 14 -6.4140 88.5695 5 1.707 84.1199 a227 5 12.5510 87.0457 10 1.5201 96.5035 8 0.185 82.5976 a228 6 11.5567 86.0513 15 -6.4642 88.5193 6 1.202 83.6145 table 2: genotype classification by predicted genotypic effect (g) and genotypic mean (u+g) for rooting (rt), survival (sv) and number of new shoots (ns) of sequoia sempervirens clones evaluated at 90 days after setting. ¹ genotype ranking parcels in relation to genotypes (0.093 to 0.157) was not as high as that for rooting (0.688), but values for the sd and h traits were close to those found for quantitative selection traits in other species (0.157 and 0.148 respectively). individual heritability values for parcel effects were very similar compared to the effects of total genotypes (0.103 to 0.157). discussion the coefficients of genetic variation, both at an individual level (cvgi) and progeny level (cvgp) are accepted as essential indicators of existing variation, allowing the estimation of genetic gains in provenance and progeny tests (sebbenn et al. 2009; rosado et al. 2012). the genotypic coefficient of variation values estimated for lovatel et al. new zealand journal of forestry science (2021) 51:11 page 5 table 3: genotype classification by predicted genotypic effect (g) and genotypic mean (u+g) for the survival (sv), stem diameter (sd), height (h) and dominance break (db) for sequoia sempervirens clones evaluated at 18 months after out-planting in the field. clone sv (%) sd (mm) h (cm) db rk¹ g u + g rk g u + g rk g u + g rk g u + g a100 4 0.1446 0.6680 9 -1.3003 19.8897 7 -0.4797 88.3223 7 0.0569 0.6393 a110 10 -0.1230 0.4005 12 -1.8618 19.3283 8 -4.1648 84.6372 4 0.1026 0.6850 a116 1 0.2866 0.8100 5 0.8082 21.9982 4 6.7229 95.5248 3 0.1031 0.6855 a117 13 -0.1947 0.3288 8 -0.7677 20.4224 3 7.5452 96.3472 6 0.0619 0.6443 a126 5 0.0568 0.5802 1 6.0628 27.2528 5 2.4033 91.2052 12 -0.1270 0.4554 a130 12 -0.1738 0.3496 7 -0.6574 20.5327 10 -5.3187 83.4832 10 -0.1183 0.4641 a131 11 -0.1428 0.3807 10 -1.5317 19.6584 6 1.1233 89.9253 8 0.0070 0.5894 a135 8 -0.0845 0.4389 4 1.9368 23.1268 9 -4.2276 84.5743 5 0.0792 0.6616 a136 9 -0.1031 0.4203 11 -1.5331 19.6570 12 -10.0620 78.7400 13 -0.1601 0.4223 a138 3 0.1700 0.6934 3 3.8569 25.0469 1 29.7256 118.5276 1 0.1226 0.7050 a140 2 0.1802 0.7037 2 4.4807 25.6707 2 14.7124 103.5143 9 -0.1142 0.4681 a227 7 -0.0485 0.4749 6 -0.3222 20.8678 11 -7.7769 81.0251 11 -0.1226 0.4597 a228 6 0.0325 0.5560 13 -9.1713 12.0187 13 -30.2030 58.5989 2 0.1089 0.6913 ¹ rk – genotype ranking table 4: variance components for survival (sv), stem diameter (sd), height (h) and dominance breakdown (db) of sequoia sempervirens clones evaluated at 18 months after out-planting in the field. component sv (%) sd (mm) h (cm) db σ2g 0.03 22.83 317.30 0.02 σ2parc 0.00 0.23 92.15 0.03 σ2e 0.21 122.13 1722.83 0.20 σ2fi 0.25 145.20 2132.28 0.25 h2g 0.1320 ± 0.0557 0.1572 ± 0.0799 0.1488 ± 0.0777 0.0929 ± 0.0614 h²aj 0.1321 0.1575 0.1555 0.1033 c2parc 0.00 0.00 0.04 0.10 h²mc 0.6628 0.7035 0.5854 0.3645 acclon (%) 0.81 0.84 0.77 0.60 cvgi (%) 34.48 22.55 20.06 26.19 cve (%) 24.59 14.64 16.88 34.58 cvr (%) 1.40 1.54 1.19 0.76 pev (%) 0.01 6.77 131.54 0.01 sep (%) 0.10 2.60 11.47 0.12 general average 0.52 21.19 88.80 0.58 the rt trait in mini-cuttings and for sv, sd, h and db traits in field trials were all above 20%. coefficients of genetic variation (cvg) above 10%, as found in the present study, indicate that there is considerable genetic variability to be explored and that there are genetically superior clones in the experiment (villacorta et al. 2015; stovall et al. 2011). this value demonstrates elevated experimental precision, meaning that the model was able to capture most of the variation in the test and, as such, the estimates can be trusted. these values are similar to those found by westbrook et al. (2015) and sykes et al. (2006) for loblolly pine (pinus taeda). based on resende (2007), the accuracy observed was high for survival (sv), stem diameter (sd) and height (h), and was similar to the accuracy levels reported in other studies on forestry species (gapare et al. 2015). the observed value of acclon indicates that selection based on sv, sd and h can be considered accurate, as the true values (which are unknown) and predicted values are very similar. estimates of genetic parameters are important for directing breeding programmes, as they aid the selection process and serve as a theoretical framework to support the recommendations of commercial materials (maia et al. 2009). however, it should be noted that this variation cannot be related only to the effects of heritability in the broad sense, i.e., the proportion of variability caused by the effects of genes (jung et al. 2008). this variation may also be linked to the clonal c-effects (pereira et al. 2018). c-effects can result in an artificial increase in clonal variation, which may increase the estimates of genetic gains in clonal selection (frampton & foster 1993). according to the same authors, this discrepancy may be due to the problems associated with vegetative propagation and the degree of environmental variation in the place where the clonal test is performed. overall, the best rooting clones were the ones with the poorest field performances, as was the case with clones a138 and a140, which had good rankings for the field variables, while in rooting their positions were intermediate to poor (16 and 7 respectively). by contrast, clone a136 which ranked third best for the rt trait demonstrated poorer field results for all four characteristics evaluated. differences in rooting may be associated with differences in subsequent growth. these differences may be related to carryover effects, or c-effects, which are intergenerational environmental effects that occur when the performance of vegetative propagules is influenced by the environment in which the propagules developed. c-effects can also bias estimates of genetic and environmental parameters in plant growth studies by inflating variation among genotypes. c-effects arise from environmental differences among plants used as the source for cuttings: plants of the previous clonal generation (schwaegerle 2005). although good levels of rooting were generally achieved, plants produced by sequoia mini-cuttings can be inefficient in the acclimatisation process and this can lead to high mortality, which requires careful consideration of environmental variables, particularly high temperatures (luna 2008). clones a113, a115 and a127, which were in the top four in the percentage of rt, were not field tested due to high levels of nursery mortality and the resulting lack of material for planting. the heritability of individual parcels of total genotypic effects found for rooting was high (0.68); values above 0.5 are considered high (resende 2002), indicating the possibility for satisfactory gains with selection for this propagation trait for sequoia mini-cuttings. however, vegetative propagation traits, such as rooting, must have a secondary and auxiliary character, the most important parameters for selection being those related to the survival, shape and productivity of trees in the field. for field evaluations, heritability was only found to be close to the normal range for quantitative traits (0.15– 0.5) for sd and h. polygenic traits such as diameter and height have complex gene interactions. they are highly influenced by the environment, making it difficult to identify superior genotypes based on the phenotype (chinelato et al. 2014). heritability cannot be fully assumed as part of the additive genetic variance, which is relative to the variance due to the additive effects of the genes and should only serve as an indication of the possibility of gain for future generations. one of the most important functions of this parameter in genetic breeding studies is its predictive role. heritability is the main indicator of success in the selection process, with higher values indicating greater genetic gains (braga et al. 2020). one of the reasons that the heritability values for sd (0.158) and h (0.149) are not higher in the present study may be due to the early evaluation period, since these traits have greater variation over time. variation in plant size can arise from genetic and environmental responses to vegetative propagation, in addition to variation caused by direct genetic and environmental effects on plant growth (schwaegerle 2005). the low heritability for db (0.09) was undesirable, as this characteristic is fundamental for stem quality, which is directly related to the final value of the product. therefore, the presence of apical dominance may be a limiting factor in the selection. in another field planting study reported by pereira (2018), sequoia plants lost their apical dominance due to attack by a grasshopper (chromacris speciosa), something that may have happened in this field test, where the presence of the insect was observed. although it did not cause high mortality in the current study, the occurrence of frost may cause the breakdown of apical dominance. very intense frosts can cause mortality, as observed in the study by pereira (2018). frosts were also responsible for the loss of trees planted in colder areas of new zealand (dean 2007). for each plant species there is a temperature at which protoplasm freezing will occur and may occur before the air temperature reaches 0°c. some species may have a higher resistance to freezing, remaining alive even after the phenomenon has occurred, but mortality may occur in younger tissues, such as the apex (perissato et al. 2013). kreyling et al. (2015) observed a cold tolerance of up to -9.2°c for sequoia sempervirens. in the same study, the authors also observed that the thermal amplitude lovatel et al. new zealand journal of forestry science (2021) 51:11 page 6 was one of the main factors responsible for high levels of damage to the plants. overall, considering genetic heritability, rt selection is easier. in addition to being a qualitative trait (influenced by fewer genes than quantitative traits), which generally has a higher heritability, it is an important variable determining the success of vegetative propagation of plants on a large scale for supplying commercial plantations. among the various genetic parameters estimated in progeny tests, one of the most important is the heritability coefficient, which measures the genetic control existing in a trait, and therefore the breeder’s potential to improve a trait through genetic selection. however, the phenotypic characteristics observed in the field, such as stem shape, growth and adaptation are the most important for the selection of superior genetic materials. therefore, consideration should be given to the development of clones in the field over time, in accordance with the desirable characteristics of the final product (e.g., volume, height, diameter, density, etc.) and thus to define other selection parameters. in general, the results from this study allow for different scenarios for the selection of genetic material (clones). good results for selection were obtained for rooting, but as already stated, this trait must be complementary to phenotypic traits in the field which are the most important for selection, particularly for species grown in commercial plantations in brazil. however, further studies and new introductions of genetic materials should be carried out, given that the origin of the seminal material that generated the rescued and propagated sequoia mother trees is unknown. there may be genetic proximity between the clones used in this study if the seeds were collected at the same site. conclusions genetic selection for rooting success in sequoia clones proved to be effective, with this trait showing a high heritability. all traits evaluated in field yielded satisfactory results for selection, indicating that there is potential for genetic gain through clonal selection. authors' contributions qcl and gtr evaluated the experiment, conducted statistical and genetic analysis, wrote the manuscript, and provided critical revisions of the manuscript. acss, bcl, elt and rard evaluated the experiment and provided critical revisions of the manuscript. mop and mcn designed the experiment and supervised the entire research. all authors read and approved the final manuscript. references alvares, c.a., stape, j.l., sentelhas, p.c., goncalves, j.l.m. & sparovek, g. (2013). köppen's climate classification map for brazil. meteorologische zeitschrift, 22(6): 711728. https://doi.org/10.1127/09412948/2013/0507 braga, r.c., paludeto, j.g.z., souza, b.m., aguiar, a.v., pollnow, m.f.m., carvalho, a.g.m. & tambarussi, e.v. (2020). genetic parameters and genotype × environment interaction in pinus taeda clonal tests. forest ecology and management, 474: 118342. https://doi.org/10.1016/j.foreco.2020.118342 chinelato, f.c.s., moraes, c.b., carignato, a., tambarussi, e.v., zimback, l., palomino, e.c. & mori, e.s. (2014). genetic variability in guapuruvu schizolobium parahyba progênies. scientia agropecuaria, 5(2), 71–76. https://doi.org/10.17268/sci. agropecu.2014.02.01 cown, d. & mckinley, r.b. (2009). wood properties of 38-year-old redwood from mangatu forest. new zealand journal of forestry, 54(2), 25–32. dean, m. (2007). coast redwood silviculture in new zealand. new zealand tree grower, 1, 6–7. frampton, l.j. & foster, g.s. field testing vegetative propagules. in: ahuja, m.r. & libby, w.j. (ed.). clonal forestry: genetics and biotechnology. budapest: springer-verlag, 1993. p. 110–134. https://doi. org/10.1007/978-3-642-84175-0_9 freitas, t. p., barroso, d.g., lamônica, k. r. & carvalho, g.c.m.w. (2016). aplicação de aib e tipo de miniestacas na produção de mudas de handroanthus heptaphyllus mattos. ciência florestal, 26(1), 313– 320. https://doi.org/10.5902/1980509821128 gapare, w.j., ivković, m., liepe, k.j., hamann, a. & low, c.b. (2015). drivers of genotype by environment interaction in radiata pine as indicated by multivariate regression trees. forest ecology and management, 353(1), 21–29. https://doi. org/10.1016/j.foreco.2015.05.027 jung, m. s., vieira, e. a., brancker, a. & nodari, r. o. (2008). herdabilidade e ganho genético em caracteres do fruto do maracujazeiro-doce. revista brasileira de fruticultura, 30(1), 209–214. https:// doi.org/10.1590/s0100-29452008000100038 kreyling, j., schmid, s. & aas, g. (2015). cold tolerance of tree species is related to the climate of their native ranges. journal of biogeography, 42(1), 156–166. https://doi.org/10.1111/jbi.12411 luna, t. (2008). vegetative propagation of coastal redwood (sequoia sempervirens (lamb. ex d. don) endl.). native plants journal, 9(1), 25–28. https:// doi.org/10.2979/npj.2008.9.1.25 majada, j., martínez-alonso, c., feito, i., kidelman, a., aranda, i., & alía, r. (2011). mini-cuttings: an effective technique for the propagation of pinus pinaster ait. new forests, 41(3), 399–412. https:// doi.org/10.1007/s11056-010-9232-x maia, m.c.c., resende, m.d.v., paiva, j.r., cavalcanti, j.j.v. & barros, l. m.b. (2009). seleção simultânea para produção, adaptabilidade e estabilidade genotípicas em clones de cajueiro, via modelos lovatel et al. new zealand journal of forestry science (2021) 51:11 page 7 https://doi.org/10.1016/j.foreco.2020.118342 https://doi.org/10.17268/sci.agropecu.2014.02.01 https://doi.org/10.17268/sci.agropecu.2014.02.01 https://doi.org/10.1007/978-3-642-84175-0_9 https://doi.org/10.1007/978-3-642-84175-0_9 https://doi.org/10.5902/1980509821128 https://doi.org/10.1016/j.foreco.2015.05.027 https://doi.org/10.1016/j.foreco.2015.05.027 https://doi.org/10.1590/s0100-29452008000100038 https://doi.org/10.1590/s0100-29452008000100038 https://doi.org/10.1111/jbi.12411 https://doi.org/10.2979/npj.2008.9.1.25 https://doi.org/10.2979/npj.2008.9.1.25 https://doi.org/10.1007/s11056-010-9232-x https://doi.org/10.1007/s11056-010-9232-x mistos. pesquisa agropecuária tropical, 39(1), 43-50. olson, d.f., roy, d.f. & walters, g.a. (1990). sequoia sempervirens (d. don) endl. redwood. in: burns, r.m and honkala, b,h. (technical coordinators) silvics of north america. agriculture handbook 654. washington dc, usa: us department of agriculture, forest service, pp. 541–551. ozudogru, e.a., kirdok, e., kaya, e., capuana, m., benelli, c. & engelmann, f. (2011). cryopreservation of redwood (sequoia sempervirens (d. don.) endl.) in vitro buds using vitrification-based techniques. cryoletters, 32(2), 99–110. pereira, m.o, ângelo, a.c., navroski, m.c., dobner júnior, m. & oliveira, l.m. (2017). vegetative rescue and rooting of cuttings of different stock plants of sequoia sempervirens. cerne, 23(4), 435–444. https://doi.org/10.1590/010477602017230424 52 pereira, m.o. (2018). resgate, propagação vegetativa e crescimento inicial de sequoia sempervirens (d. don.) endl., 225 p. tese de doutorado da universidade federal do paraná, curitiba. pereira, m.o., ângelo, a.c., navroski, m.c., nicoletti, m.f., nascimento, b., sá, a.c.s., gasparin, e. & lovatel, q.c. (2018). roots of mini-cutting of clones of sequoia sempervirens in different seasons of the year. cerne, 24(4), 452–460. https://doi.org/10.15 90/01047760201824042604 perissato, s. m., marcelino, w.l., acco, l.f., cabral, a.c., pinto, l.p. & frigo, j.p. (2013). efeito das geadas em culturas energéticas. revista brasileira de energias renováveis, 2(4), 49–58. https://doi.org/10.5380/ rber.v2i4.35793 resende, m.d.v. (2002). software selegen-reml/blup. colombo: embrapa florestas. 67 p. (embrapa florestas, documento, 77). resende, m.d.v. (2007). selegen-reml/blup: sistema estatístico e seleção genética computadorizada via modelos lineares mistos. colombo: embrapa florestas. resende, m.d.v. (2016). software selegen-reml/blup: a useful tool for plant breeding. crop breeding and applied biotechnology, 16(4), 330–339. https:// doi.org/10.1590/1984-70332016v16n4a49 rosado, a.m., rosado, t.b., alves, a.a., laviola, b.g.e. & bhering, l.l. (2012). seleção simultânea de clones de eucalipto de acordo com produtividade, estabilidade e adaptabilidade. pesquisa agropecuária brasileira, 47(7), 964–971. https:// doi.org/10.1590/s0100-204x2012000700013 schwaegerle, k.e. (2005). quantitative genetic analysis of plant growth: biases arising from vegetative propagation. evolution, 59(6), 1259–1267. https:// doi.org/10.1111/j.0014-3820.2005.tb01776.x sebbenn, a.m., freitas, m.l.m., zanatto, a.c.s. & morais, e. (2009). comportamento da variação genética entre e dentro de procedências e progênies de gallesia integrifolia vell. moq. para caracteres quantitativos. revista do instituto florestal, 21(2), 151–163. silva, g.o., carvalho, a.d.f., veira, j.v. & benin, g. (2011). verificação da adaptabilidade e estabilidade de populações de cenoura pelos métodos ammi, gge biplot e reml/blup. bragantia, 70(3), 494–501. https://doi.org/10.1590/s000687052011005000003 stovall, j.p., carlson, c.a., seiler, j.r., fox, t.r. & yanez, m.a. (2011). growth and stem quality responses to fertilizer application by 21 loblolly pine clones in the virginia piedmont. forest ecology and management, 261(3), 362–372. https://doi. org/10.1016/j.foreco.2010.10.018 sykes, r., li, b., isik, f., kadla, j. & chang, h.m. (2006). genetic variation and genotype by environment interactions of juvenile wood chemical properties in pinus taeda l. annals of forest science, 68(3), 897–904. https://doi.org/10.1051/ forest:2006073 villacorta, a.m.g., martin, t.a., jokela, e.j., cropper jr, w.p., gezan, s.a. (2015). variation in biomass distribution and nutrient content in loblolly pine (pinus taeda l.) clones having contrasting crown architecture and growth efficiency. forest ecology and management, 342(1), 84–92. https://doi. org/10.1016/j.foreco.2015.01.012 wendling, i., trueman, s. j. & xavier, a. (2014). maturation and related aspects in clonal forestrypart ii: reinvigoration, revitalization and juvenility maintenance. new forests, 45(1), 473–486. https://doi.org/10.1007/s11056-014-9415-y westbrook, j.w., walker, a.r., neves, l.g., munoz, p., resende jr, m.f., neale, d.b. & peter, g.f. (2015). discovering candidate genes that regulate resin canal number in pinus taeda stems by integrating genetic analysis across environments, ages, and populations. new phytologist, 205(2), 627–641. https://doi.org/10.1111/nph.13074 lovatel et al. new zealand journal of forestry science (2021) 51:11 page 8 https://doi.org/10.1590/01047760201723042452 https://doi.org/10.1590/01047760201723042452 https://doi.org/10.1590/01047760201824042604 https://doi.org/10.1590/01047760201824042604 https://doi.org/10.5380/rber.v2i4.35793 https://doi.org/10.5380/rber.v2i4.35793 https://doi.org/10.1590/1984-70332016v16n4a49 https://doi.org/10.1590/1984-70332016v16n4a49 https://doi.org/10.1590/s0100-204x2012000700013 https://doi.org/10.1590/s0100-204x2012000700013 https://doi.org/10.1111/j.0014-3820.2005.tb01776.x https://doi.org/10.1111/j.0014-3820.2005.tb01776.x https://doi.org/10.1590/s0006-87052011005000003 https://doi.org/10.1590/s0006-87052011005000003 https://doi.org/10.1016/j.foreco.2010.10.018 https://doi.org/10.1016/j.foreco.2010.10.018 https://doi.org/10.1051/forest:2006073 https://doi.org/10.1051/forest:2006073 https://doi.org/10.1016/j.foreco.2015.01.012 https://doi.org/10.1016/j.foreco.2015.01.012 https://doi.org/10.1007/s11056-014-9415-y https://doi.org/10.1111/nph.13074 a preliminary growth and yield model for eucalyptus globoidea blakely plantations in new zealand serajis salekin1,2,*, euan g. mason1, justin morgenroth1, dean f. meason2 1 new zealand school of forestry, university of canterbury, christchurch 8140, new zealand 2 scion, 49 sala street, private bag 3020, rotorua 3010, new zealand *corresponding author: serajis.salekin@canterbury.ac.nz (received for publication 14 may 2019; accepted in revised form 14 may 2020) abstract background: new zealand’s plantation forest industry is dominated by the exotic species radiata pine (pinus radiata d.don), which comprises approximately 90% of the net stocked area. however, there is interest in introducing new species to: (a) provide wood that is naturally decay-resistant as a substitute for wood treated with preservatives; (b) match species to the wide variety of environmental conditions in new zealand; and (c) reduce reliance on p. radiata. some eucalyptus species are considered as potential alternatives to p. radiata, specifically those that can survive in resource-limited conditions and produce high quality wood. while eucalyptus species are grown in plantations in many regions of the world, limited information is available on their growth in new zealand. eucalyptus globoidea blakley is of particular interest and has been planted in trials throughout new zealand. a complete set of preliminary growth and yield models for this species will satisfy the initial information requirements for diversifying new zealand’s plantation forest industry. methods: a set of growth and yield models was developed and validated, based on data from 29 e. globoidea permanent sample plots (psps) located mostly in north island and a few in south island of new zealand. trees were measured at different time intervals in these plots, with height and diameter at breast height (dbh) ranging from 0.1–39.8 m and 0.1–62.3 cm, respectively. an algebraic difference approach (ada) was applied to model mean top height, basal area, maximum diameter, and standard deviation of dbh. non-linear regression equations were used to project stand volume and height-diameter relationship, and reineke’s stand density index (sdi) approach was employed to model mortality. results: mean top height, maximum diameter, and standard deviation of dbh were best fitted by von bertalanffy-richards (se=1.1 m), hossfeld (se=2.4 cm), and schumacher polymorphic (se=1.6 cm) difference equations, respectively. basal area data were modelled with high precision (se=6.9 m2 ha-1) by the schumacher anamorphic difference equation. reineke’s sdi approach was able to explain the self-thinning as a reduction in the number of stems per hectare. stand-level volume per hectare and height-diameter relationship models were precise when including site-specific variables with standard errors of 40.5 m3 ha-1 and 3.1 m, respectively. conclusion: this study presents a set of preliminary growth and yield models for e. globoidea to project plot-level growth attributes. the models were path invariant and satisfied basic traditional mensurational-statistical growth and yield model assumptions. these models will provide forest growers and managers with important fundamental information about the growth and yield of e. globoidea. new zealand journal of forestry science salekin et al. new zealand journal of forestry science (2020) 50:2 https://doi.org/10.33494/nzjfs502020x55x e-issn: 1179-5395 published on-line: 27/05/2020 © the author(s). 2020 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access keywords: algebraic difference approach (ada), basal area, eucalyptus globoidea, growth and yield model, mean top height, mensurational-statistical model, stand density index. salekin et al. new zealand journal of forestry science (2020) 50:2 page 2 introduction a resilient forest economy would be diversified, with healthy forests of all ages producing a range of valuable products and services. the new zealand forestry industry is almost entirely based on radiata pine (pinus radiata d.don) plantations (new zealand forest owners association 2017) due to its rapid growth rate across a broad range of sites (turner et al. 2008) and established processing infrastructure and markets. however, there are opportunities to introduce new species and overcome some limitations of p. radiata (millen et al. 2018). for example, non-naturally durable wood, less diversified forest ecosystems and slow growth in drought-prone environments limit radiata pine’s potential. new zealand’s existing commercial forest sector could be complemented by introducing other species. this would help to reduce the reliance on largescale plantations containing pure stands of p. radiata which are at increased risk of pest and diseases attacks (chou 1991) and produce a relatively narrow range of forest products. species of eucalyptus have been considered as a commercial forestry alternative to p. radiata, especially those species that can grow well in dry conditions and produce high quality timber (menzies 1995). however, despite strong advocacy for alternative species from various groups, only 1.2% of the total plantation forest area in new zealand is comprised of various eucalyptus species (mpi 2019). growing eucalyptus in new zealand has, over the years, been challenging (berrill & hay 2005; berrill & hay 2006) due to specific site requirements such as sensitivity to soil moisture availability and frosty environments (bell & williams 1997; williams & woinarski 1997), pests and diseases that affect their health and productivity (lin 2017), and a lack of markets for eucalyptus wood products (apiolaza et al. 2011). recently, the situation has started to change, in part, because of the new zealand dryland forest initiative (nzdfi) and renewed consumer demand for eucalyptus timber (satchell & turner unpublished data). the nzdfi has catalysed research into several naturally durable eucalyptus species, chosen for their desirable properties (nicholas & millen 2012), for deployment on ex-pasture lands in relatively dry parts of the country (nzdfi 2013). despite these advances, little is known about the growth dynamics of many of these eucalyptus species in new zealand. managed forests are dynamic biological systems that change in response to environments and silvicultural practices. growth and yield models can support effective decision making by describing current and future forest dynamics (blake et al. 1990; blanco et al. 2005; castedo-dorado. et al. 2007; clutter et al. 1983). traditional time-based growth and yield models, called mensurational-statistical models, provide robust growth predictions but give little information about the mechanisms of forest dynamics (korzukhin et al. 1996). apart from being mathematically simple and biologically rational, clutter et al. (1983) noted several important features: i) representation of growth and yield should be compatible; ii) the functions should be consistent; iii) the functions should be path-invariant; and iv) the functions should rise to asymptotes. mensurational-statistical forest growth models are often based on large datasets, comprising repeated field measurements in permanent plots (castedo-dorado et al. 2007; pienaar & rheney 1995) or information obtained from remotely-sensed data (battaglia et al. 2004; landsberg et al. 2003). however, in scenarios where comprehensive data are not available, it may still be desirable to develop preliminary growth and yield models to forecast forest growth (vanclay 2010), especially for new species (berrill et al. 2007; kitikidou et al. 2016; palahí & grau 2003). such models are often imprecise, but can be useful (box 1976) to obtain an initial forecast in order to make decisions about establishment, tending, and potential log marketing. preliminary models are not only useful for characterising stand development but also provide insights into the yield potential of sites, a crucial factor for sound management of any forest stand (tewari & gadow 2003). moreover, preliminary mensurationalstatistical models can be easily implemented and used by forest managers to generate initial estimates of growth and yield. while preliminary models are available for eucalyptus fastigata, e. nitens, and overall stringy-bark groups in new zealand (berrill & hay 2005; berrill & hay 2006), no growth and yield models exist for the eucalyptus species within the nzdfi’s programme. despite being planted in trials and plantations around new zealand, managers and growers have limited knowledge of the expected growth and yield for these species. development of species-specific, stand-level preliminary models will not only give them more information but also guide them about species choice for planting and future management. the nzdfi selected a set of naturally durable eucalyptus species (see, page & sing 2014) based on australian timber durability standard (class 1 and 2). e. globoidea is one of the top ranked species in that list and is commonly classified into the stringybark group. it was sparsely planted around new zealand prior to the nzdfi programme. this species has naturally durable wood and is considered a highly durable timber (class 1 or 2) in the australian standards (as5606-2005) (nicholas & millen 2012a). e. globoidea is well adapted to dry parts of the new zealand. moreover, a strong consumer demand for naturally durable eucalyptus wood has been identified (kakitani 2017). growth and yield model functions must adequately describe the system at any point in time by allocating local transitions (garcia 1988), that is the rate of change of state as a function of the current state and of the current values of external control variables. therefore, the main objective of this study was to develop a preliminary stand-level e. globoidea growth and yield model. this model projects estimates of mean top height (mth), basal area/ha (g), maximum diameter at breast height (dmax), standard deviation of diameter (sdd), stand volume (v), self-thinning and height-diameter relationships (h-d) forward in time following measurements of stands. then dmax and sdd can be used to fit a reverse weibull function to describe the stand-level diameter distribution (garcía 1981; kuru et al. 1992). individual tree models were considered, but diameter distribution models have been shown to be superior to individual tree models when long-term projections are required by those planning harvests many years in the future (methol 2001). methods data preparation and description treeand plot-level e. globoidea data were available from a nationwide permanent sample plot system (pilaar & dunlop 1990). data from twenty nine permanent sample plots (psps) established in plantations at ten different localities were available (table 1 and figure 1). trees were measured in the psps at 1 to 10-year intervals with an irregular frequency. mean top height (average height of the 100 largest diameter stems per hectare) (mth) and maximum diameter (the largest diameter measured at right angles to the stem over the stubs) (dmax) of the trees were calculated from the individual tree measurements by following the procedure described by goulding (2005). the standard deviation of dbh (sdd) was calculated for each plot measurement. basal area (g) was calculated by summing the crosssectional area at breast height (1.4 m) of all trees in the plot, then dividing by plot size to provide a per hectare estimate. stand volume (v) was calculated within each sample plot by estimating and summing up individual tree volumes calculated using a simplified e. globoidea stem taper volume equation presented in lundgren (1995) (see the appendix). plot-level summary data were organised by representing all possible measurement time intervals and used to fit differential equations. simple time increment data from plot-level summaries were used to fit volume-per-hectare equations. the height-diameter relationship was modelled using individual tree measurement data from all plots. modelling and evaluation the algebraic difference approach (ada) (bailey & clutter 1974) was used to model mean top height (mth), basal area/ha (g), maximum diameter (dmax) and standard deviation of diameter (sdd). sixteen well-known and frequently used polymorphic and anamorphic forms of differential equations (bailey & clutter 1974; belli & ek 1988; ek 1974; vanclay 1994; zeide 1993) (table 2) were fitted to the data using non-linear least-squares (clutter 1963). volume per hectare yields (v) were modelled using various simple, established and commonly used functions (table 3). height-diameter (h-d) models were developed by fitting the näslund (1936) equation with a range of different exponent terms (zhao 1999): h = 1.4 + (α + β/d)-γ (17) where h is tree height (m), d is diameter (cm) at breast height (1.4 m), and α and β are model parameters. the exponent term (γ) here is variable. this function is widely used and can be expressed in a linear form: d/(h 1.4)0.4 = α × (d + β) (18) the linearity of the height-diameter relationship is a unique property at a plot-level or at a stand-level (curtis 1967; garcia 1974) or at a stand-level (zhao 1999) when a few plots are sampled from the same stand in a single site. however, fitting h-d relationships at a stand level results in underestimates of variability in mth (mason 2019), so sufficient heights (usually 12) were measured in each psp to allow the fitting of plot-level h-d relationships. a better height-diameter relationship can salekin et al. new zealand journal of forestry science (2020) 50:2 page 3 table 1: summary of the data used for modelling* variable unit statistical summary of variable mean min. max. sd plot size ha 0.06 0.04 0.10 0.02 age (t) years 14 3 25 6 individual tree height (h) m 12.90 0.10 39.80 9.05 mean top height (mth) m 18.98 3.50 28.80 7.05 diameter at breast height at 1.4 m (dbh) cm 22.90 0.10 62.30 14.49 max dbh (dmax) cm 39.79 5.40 62.30 13.66 standard deviation of dbh (sdd) cm 5.34 1.35 11.86 2.20 basal area (g) m2ha-1 30.59 0.54 77.88 18.84 volume (v) m3 ha-1 161.34 0.40 538.60 130.09 stocking (n) stems ha-1 496.99 141.09 1375 317.33 elevation (elv) m. asl 211.70 80 300 100.41 slope (°) 23.27 8 42 13.06 *a detailed individual psp description including silvicultural treatments is provided in the appendix. salekin et al. new zealand journal of forestry science (2020) 50:2 page 4 figure 1: permanent sample plot (psp) locations and topography. generic name expression no. po ly m or ph ic fo rm schumacher 1 1 schumacher 2 2 gompertz 1 3 gompertz 2 4 weibull 1 5 weibull 2 6 hossfeld 7 von bertalanffy-richards 1 8 von bertalanffy-richards 2 9 von bertalanffy-richards 3 10 a na m or ph ic fo rm schumacher a1 11 schumacher a2 12 gompertz 13 von bertalanffy-richards 14 weibull 15 hossfeld 16 y2 = e ln(y1)( t1 t2 )+α(1−t1t2 ) y2 = e ln(y1)( t1 t2 ) 𝛾𝛾 +α[1−(t1t2 )𝛾𝛾] 𝑌𝑌2 = 𝑒𝑒ln⁡(𝑌𝑌1)𝑒𝑒 −𝛽𝛽(𝑡𝑡2−𝑡𝑡1)𝑒𝑒𝛼𝛼[1−𝛽𝛽(𝑡𝑡2−𝑡𝑡1)] 𝑌𝑌2 = 𝑒𝑒ln⁡(𝑌𝑌1)𝑒𝑒 −𝛽𝛽(𝑡𝑡2−𝑡𝑡1)+𝛾𝛾(𝑡𝑡2 2−𝑡𝑡1 2)𝑒𝑒𝛼𝛼[1−𝑒𝑒−𝛽𝛽(𝑡𝑡2−𝑡𝑡1)+𝛾𝛾(𝑡𝑡2 2−𝑡𝑡1 2)] 𝑌𝑌2 = 𝑌𝑌1𝑒𝑒−𝛽𝛽(𝑡𝑡2 𝛾𝛾−𝑡𝑡1 𝛾𝛾) + 𝛼𝛼[1 − 𝛽𝛽(𝑡𝑡2 𝛾𝛾 − 𝑡𝑡1 𝛾𝛾)] 𝑌𝑌2 = 𝛼𝛼 − 𝛽𝛽( 𝛼𝛼 − 𝑌𝑌1 𝛽𝛽 ) (𝑡𝑡2𝑡𝑡1 )𝛾𝛾 𝑌𝑌2 = 1 1 𝑌𝑌1 (𝑡𝑡2𝑡𝑡1 )𝛽𝛽 + 𝛼𝛼[1 − (𝑡𝑡2𝑡𝑡1 )𝛽𝛽] 𝑌𝑌2 = 𝛼𝛼( 𝑌𝑌1 𝛼𝛼) ln⁡[1−𝑒𝑒(𝛽𝛽𝑡𝑡2)] ln⁡[1−𝑒𝑒(𝛽𝛽𝑡𝑡1)] 𝑌𝑌2 = 𝛼𝛼{1 − [1 − ( 𝑌𝑌1 𝛼𝛼) 1−𝜗𝜗] 𝑡𝑡2 𝑡𝑡1} 1 1−𝜗𝜗 𝑌𝑌2 = 𝛼𝛼{1 + [( 𝛼𝛼 𝑌𝑌1 ) 𝜗𝜗 − 1]𝑒𝑒[−𝛽𝛽(𝑡𝑡2−𝑡𝑡1)]} 1 𝜗𝜗 𝑌𝑌2 = 𝑌𝑌1𝑒𝑒 −𝛽𝛽( 1𝑡𝑡2 − 1𝑡𝑡1 ) 𝑌𝑌2 = 𝑌𝑌1𝑒𝑒 −𝛽𝛽[( 1𝑡𝑡2 ) 𝛾𝛾 −( 1𝑡𝑡2 )𝛾𝛾] 𝑌𝑌2 = 𝑌𝑌1 𝑒𝑒−𝛽𝛽𝑒𝑒−𝛾𝛾𝑡𝑡2 𝑒𝑒−𝛽𝛽𝑒𝑒−𝛾𝛾𝑡𝑡1 𝑌𝑌2 = 𝑌𝑌1[ 1 − 𝑒𝑒−𝛽𝛽𝑡𝑡2 1 − 𝑒𝑒−𝛽𝛽𝑡𝑡1] 𝛾𝛾 𝑌𝑌2 = 𝑌𝑌1 1 − 𝑒𝑒−𝛽𝛽𝑡𝑡2𝛾𝛾 1 − 𝑒𝑒−𝛽𝛽𝑡𝑡1𝛾𝛾 𝑌𝑌2 = 1 1 𝑌𝑌1 + 𝛽𝛽(1𝑡𝑡2 )𝛾𝛾 − (1𝑡𝑡1 )𝛾𝛾 table 2. different forms of differential equations. be obtained by identifying and incorporating relevant factors accounting for differences within stands. these could include tree species, stand age, site characteristics, genetics, stocking, and silvicultural treatment (zhao et al. 2006). this was achieved by separating and linearly expanding the regression coefficients described in woollons et al. (1997). generally, data on self-thinning or mortality follow a binomial or poisson distribution. therefore, development of models to describe this process needs larger datasets than for other models. due to the small number of plots, a conceptual self-thinning/mortality model was produced by applying reineke’s stand density index (sdi) (reineke 1933). this was done by estimating quadratic mean diameter at breast height (dbh) and basal area (g). the maximum sdi was assumed from the original range by selecting the highest stand density as there was no specific evidence of self-thinning in the dataset. for validation there was no independent dataset available for this study, nor was the dataset large enough to be subdivided into fitting and validation datasets. therefore, model validation was carried out by the ‘leave-one-out’ method of cross-validation (loocv), a method which is also called “jackknife” (arlot & celisse 2010). thus, the models were fitted times, leaving out each sample plot once, so that the number of model fits was equal to the number of plots (sánchez-gonzález et al. 2005), and residuals of predictions for the plots left out were compared with those of the overall model fit. all the models except the self-thinning model were evaluated through the validation procedure. validation included a visual analysis of graphs of the residuals, the calculation of root mean square error (rmse) (equation 23), mean absolute error (mae) (equation 24), bias (equation 25), and adjusted coefficient of determination (r2) (equation 27). adjusted r2 values were not considered for assessing differential equations as it is sensitive to grouped and repeated data (warren 1971). (23) (24) (25) (26) (27) where n = number of observations, o = observed value, = mean of observed values, p = predicted value, . k denotes the number of estimated parameters. the predictive ability of the models was evaluated using prediction errors or predictive residual error sum square (press) statistics (equation 28), oi p(i,-i) = e(i,-i) (i = 1,2,…..,n) (28) where oi is the observed value, p(i,-i) is the estimated value for observation i (where the latter is absent from the model fitting) and n is the number of observations. each model has n press residuals associated with it, and the press (prediction sum of square/p-square) statistic is defined as (myers & myers 1990): (29) the bias and precision of models were analysed by computing means of the press residuals. all statistical analysis was performed in the r statistical environment (r development core team 2017). different non-linear regressions were fitted using the “nls” function in the base package with appropriate significant variables. evaluation metrics “adj. r2”, and “rmse”, ”mae”, “bias” functions were used from the “metrics” package (hamner & frasco 2018). residuals were visually inspected for their normality and variance homogeneity. all graphical analyses were performed with the “ggplot2” package (wickham 2016). results mean top height (mth) model among the tested differential equations (table 2), the first von bertalanffy-richards polymorphic model (equation 8) exhibited the most precise fitting statistics based on goodness-of-fit. it minimised bias and standard error of prediction compared with the other models tested. however, the rmse and mae were higher in model fitting statistics, relative to validation, at which time they roughly halved to 3.8 m and 2.5 m, respectively (table 4). the model residuals were well distributed with minor heteroscedasticity at the beginning of the modelling period (figure 2a and 2b). the model predictions covered the entire range of measured mth values, except those for two stands (figure 2c). these salekin et al. new zealand journal of forestry science (2020) 50:2 page 5 expression reference no. (soalleiro 1995) 19 (jansen et al. 1996) 20 (burkhart 1977) 21 (candy 1989) 22 , and α, β, γ, and δ are model coefficients. v = α × g × mth v = g × mth(α+βt)e(γ+δt) v = g × (α + β mth) v = e(α+βlogmth)+γlogg table 3. volume yield equations. here, v is volume/ha, g is basal area/ha, mth is mean top height, t is age in years, and α, β, γ, and δ are model coefficients. rmse = √∑ (pi−oi) 2n i=1 n mae = ∑ |pi−oi| n i=1 n bias = ∑ (pi− n i=1 oi ) n r2 = ∑ p ′ i 2 ∑ o′ i 2 𝑅𝑅2 𝑎𝑎𝑎𝑎𝑎𝑎 = 1 − [ (1−𝑅𝑅2)(𝑁𝑁−1) 𝑁𝑁−𝐾𝐾−1 ] o o̅ p p′i = pi − o̅ o′i = oi − o̅ k oi − pi,−i = ei,−i (i = 1,2, … . . , n) rmse = √∑ (pi−oi) 2n i=1 n mae = ∑ |pi−oi| n i=1 n bias = ∑ (pi− n i=1 oi ) n r2 = ∑ p ′ i 2 ∑ o′ i 2 𝑅𝑅2 𝑎𝑎𝑎𝑎𝑎𝑎 = 1 − [ (1−𝑅𝑅2)(𝑁𝑁−1) 𝑁𝑁−𝐾𝐾−1 ] o o̅ p p′i = pi − o̅ o′i = oi − o̅ k oi − pi,−i = ei,−i (i = 1,2, … . . , n) o̅ p p′i = pi − o̅ o′i = oi − o̅ press = ∑ oi − (pi,−i)2ni=1 = ∑ (ei,−i)2ni=1 o̅ p p′i = pi − o̅ o′i = oi − o̅ press = ∑ oi − (pi,−i)2ni=1 = ∑ (ei,−i)2ni=1 o̅ p p′i = pi − o̅ o′i = oi − o̅ press = ∑ oi − (pi,−i)2ni=1 = ∑ (ei,−i)2ni=1 two stands contained outliers, which were left out from the initial model building procedure. model parameters are provided in the appendix (table a2). basal area per hectare (g) model among tested models (table 2), the first anamorphic schumacher model with a single parameter (equation 11) was found to be the best fit for basal area/ha projection. this model had the lowest error and greatest precision. precision increased during validation with much less error indicating stable model performance (table 5). the residual plot exhibited minor heteroscedasticity (figure 3a). the residual distribution was positively biased, which indicated a slight overprediction. moreover, the model predicted basal area values covering the measured range, except for two stands (figure 3). model parameters are provided in the appendix (table a2). maximum diameter (dmax) model the hossfeld polymorphic model (equation 7) predicted the maximum diameter (dmax) with greatest overall precision and least bias in comparison with other candidate model forms (table 2). in this case, rmse and mae increased from fitting to validation statistic and bias went from positive to negative (table 6), which indicated model under-prediction during validation. however, the standard error (se) reduced slightly in validation indicating higher precision. the low mpress and mapress values also indicated model goodnessof-fit (table 6). residuals were highly biased at the beginning and end of the modelling period (figure 4a), consistent with the limited availability of data. however, they were normally distributed (figure 4b). the function for predicting dmax enveloped all the measurements and followed a sigmoid shape (figure 4c), which ensures salekin et al. new zealand journal of forestry science (2020) 50:2 page 6 action rmse mae bias se mpress mapress fitting 7.185 5.467 -1.777 1.116 validation 3.852 2.512 0.066 1.112 0.009 0.946 table 4. mean top height (mth) m, von bertalanffy-richards polymorphic model fitting and validation statistics. figure 2. mean top height (mth) model results: a) residuals against prediction plot of first von bertalanffy-richards polymorphic equation, light blue points represent model fitting, red points indicate validation residuals, and model fit is shown by the black line; b) residuals frequency distribution, red dashed line shows the mean; and c) model fit (blue lines) over measured mth (thin black lines). biological rationality. model parameters are provided in the appendix (table a2). standard deviation of diameter (sdd) model among all the candidate models tested (table 2), the standard deviation of diameter (sdd) was best predicted by the second schumacher polymorphic model (equation 2). the model had the lowest prediction errors. the rmse (1.5 cm to 1.9 cm) and mae (1.2 cm to 1.5 cm) increased slightly from fitting to validation. minimal mpress and mapress values also confirmed precision of the model (table 7). model parameters are provided in the appendix (table a2). graphically, the model predicted values and residuals appeared to follow a normal distribution (figure 5a). the residuals plot shows overprediction and positive bias of the model with few outliers in the frequency salekin et al. new zealand journal of forestry science (2020) 50:2 page 7 distribution plot (figure 5b). the prediction plot shows that the model included the full range of measured sdd (figure 5c). volume per hectare (v) model the best performing model of all those tested for volume per hectare yield (table 3) was the four parameter jansen et al. (1996) model (equation 20). fitting statistics showed minimal prediction error and high precision, though validation statistics were greater in both cases (table 8), and these were confirmed by small mpress, small mapress, and high adjusted r2 value (table 8). these results are also confirmed graphically (figure 6), although there is a minor heteroscedastic tendency in the residuals (figure 6b). model parameters are provided in the appendix (table a2). action rmse mae bias se mpress mapress fitting 25.303 21.250 2.893 6.893 validation 13.431 9.988 0.653 6.800 1.054 0.841 table 5. basal area/ha (g), first anamorphic schumacher model fitting and validation statistics. figure 3. basal area (g) model results: a) residuals against prediction plot of first schumacher anamorphic equation, light blue points represent model fitting, the red points indicate validation residuals, and model fit is shown by the black line; b) residuals frequency distribution, red dashed line shows the mean; and c) model fit (blue lines) over measured g (thin black lines). height-diameter (h-d) model the stand-specific individual height-diameter (h-d) model gave precise predictions with an exponent of -2 (equation 30). stand-specific elevation and basal area (g) were found to influence the h-d relationship significantly (p<0.05) and adding them into the final model improved the fit. the goodness-of-fit values increased slightly from fitting to validation statistics, which indicated less precision in prediction (table 9). residuals were normally distributed, and the model fitted well (figure 7b and c). model parameters are provided in the appendix (table a3). (30) self-thinning model the self-thinning model based on reineke’s sdi fitted the limited available data well. stand density ranged from 150–1350 stems ha-1, with most plots having a stand table 2: confusion matrix density between 400 and 650 stems ha-1 (figure 8b). the maximum carrying capacity was calculated as 1350, 25 cm diameter trees per hectare. based on this value, a density management diagram was produced (figure 8a) with lines indicating understocking below 35% of maximum carrying capacity, full stocking between 35% to 55% of maximum carrying capacity, and over-stocking above 55% of maximum carrying capacity. natural mortality started to occur when stocking approached the maximum carrying capacity (figure 8a). discussion this study developed a preliminary set of stand-level growth and yield models for e. globoidea in new zealand using sparsely available data. the state of a plot was adequately described by the following state variables: mean top height, basal area/ha, volume/ha, stocking, maximum diameter, standard deviation of dbh and salekin et al. new zealand journal of forestry science (2020) 50:2 page 8 action rmse mae bias se mpress mapress fitting 2.400 1.759 0.054 2.411 validation 6.699 4.681 -0.061 2.388 0.059 0.932 table 6. maximum diameter cm (dmax) model fitting and validation statistics. figure 4. maximum diameter (dmax) model results: a) residuals against prediction plot of hossfeld polymorphic equation, light blue points represent model fitting, the red points indicate validation residuals, and model fit is shown by the black line; b) residuals frequency distribution, red dashed line shows the mean; and c) model fit (blue lines) over measured dmax (thin black lines). 𝐻𝐻 = 1.4 + ((𝛼𝛼0 + 𝛼𝛼1 × 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸) + (𝛽𝛽0+𝛽𝛽1×𝐺𝐺) 𝐷𝐷 )−2 the height-diameter relationship. the nature of the plot’s growth is described by the rate of change of these variables over time by their corresponding transition function. the final models were the best-fitted models, which generally had the highest accuracy among the tested set of equations from several differential forms. the best mth, dmax and sdd models took polymorphic forms, similar to earlier preliminary modelling studies in a range of species, including even-aged cupressus lusitanica mill. and c. macrocarpa hartw. plantations (berrill 2004), acacia melanoxylon r.br. (berrill et al. 2007), eucalyptus fastigata (berrill & hay 2005) in new zealand and pinus nigra arn. in catalonia, spain (palahí & grau 2003). however, basal area/ha (g) was best fitted with an anamorphic form, which is unusual but can be found in similar types of data-limited situations. for example, vanclay (2010) suggested one parameter anamorphic forms to deal with a similar small dataset. overall, model projections followed the growth pattern of this species. for example, the mean top height for a 15-year-old stand ranges from 15–22.5 m and basal area ranges from 14.5 to 62.5 m2 ha-1. apart from a few outlier stands, our results followed similar trends to those reported by nicholas and millen (2012b). similar to our study, their study and models developed in it were based on a very small number of measurement plots. furthermore, meason et al. (2016) reported that most of the small-scale plantation psps resided in new zealand’s north island. this may affect the model’s capability to perform over a wider range of environmental conditions. there were some errors in model prediction, which may be due to the irregular measurement intervals for the stands included in the study. lee (1998) reported that long measurement intervals can produce apparently larger errors than short measurement intervals, but longer intervals were vital for avoiding biased projections over long intervals. the measurement salekin et al. new zealand journal of forestry science (2020) 50:2 page 9 action rmse mae bias se mpress mapress fitting 1.571 1.224 0.412 1.577 validation 1.959 1.513 0.337 1.569 0.407 0.596 table 7. standard deviation of dbh cm (sdd) model fitting and validation statistics. figure 5. standard deviation of dbh (sdd) model results: a) residuals against prediction plot of hossfeld polymorphic equation, light blue points represent model fitting, the red points indicate validation residuals, and model fit is shown by the black line; b) residuals frequency distribution, red dashed line show the mean; and c) model fit (blue lines) over measured sdd (thin black lines). periods also differed among the psps, which may have caused bias and heteroscedasticity through the modelling period (lee 1998). furthermore, model precision could likely have been improved by reinforcing it with more biological, or silvicultural information, for example, thinning information or any kind of natural disturbance events (park & wilson 2007). in this study, such information was available for only a small number of plots (see table a1 in the appendix). borders et al. (1988) reported autocorrelation in data while using similar types of datasets to those used in this study, especially in a data-limited situation. when data covered all age classes as well as sites, such autocorrelation can be tested independently by separating each time interval (borders 1989), which was not possible in this case. however, this study aimed to use the available data to condition the shape of a previously-known process. also, all these models are based on mensurational equations and could receive further reinforcement from a biological perspective, by adding physiology into the modelling framework. the self-thinning model was based on the sdi concept of reineke (1933) with the “imminent competitionmortality” theory of drew and flewelling (1977). here, competition related mortality likely occurs within a zone defined by two lines: the maximum size-density relationship (100% relative density) and a second line paralleling the first at lower densities for the same mean salekin et al. new zealand journal of forestry science (2020) 50:2 page 10 action rmse mae bias se r2adj mpress mapress fitting 39.122 27.983 -1.102 40.5 0.91 validation 140.959 89.827 -0.582 39.413 0.95 -0.845 0.868 table 8. stand volume m3 (v), jansen et al. (1996) model fitting and validation statistics. figure 6. stand volume (v) model results: a) estimated stand volume from measured data; b) residuals against prediction plot, light blue points represent model fitting, red points indicate validation residuals, and model fit is shown by the black line; and c) residuals frequency distribution, red dashed line is shown the mean. salekin et al. new zealand journal of forestry science (2020) 50:2 page 11 figure 7. a) measured height-diameter (h-d), blue line shows the linear trend; b) residuals against prediction plot, light blue points represent model fitting, red points indicate validation residuals, model fit is shown by the blue line; and c) residuals frequency distribution, red dashed line is shown the mean. action rmse mae bias se r2adj mpress mapress fitting 3.08 2.41 -0.01 3.10 0.54 validation 3.80 2.91 0.03 3.05 0.53 -0.001 0.530 table 9. height-diameter relationship (h-d) model fitting and validation statistics. figure 8. a) reineke’s sdi curve represented with self-thinning lines and b) sdi distribution plot. tree size (55% relative density). according to this theory, a uniform stand may have imminent competitionmortality within the zone, but the probability of that could be lowered by substantially lowering the density. conversely, if a stand is allowed to grow for many years within the zone mortality will occur (drew & flewelling 1979). therefore, the sdi approach can only give an indication but not any causal explanation, so mortality cannot be precisely predicted on this basis (drew & flewelling 1979). the translation of specific management objectives into appropriate upper and lower levels of growing stock is the key and most critical step to design a density management regime (long 1985). moreover, stands in the dataset must exhibited self-thinning to fit these lines, which was not the case in this study. however, this study reported no live stems above the maximum line which indicated the sdi approach’s applicability. nevertheless, while the sdi approach can be easily estimated and applied (long 1985), it requires further testing and refinement with more data, particularly from older, higher-stocked plots where self-thinning is evident. a more comprehensive dataset would enable the slope (power term in the sdi function) of the size-density relationship to be investigated. both pretzsch and biber (2005), and saunders and puettmann (2000) reported that the sdi function’s power exponent term changed with species and site, but in this study the default value (1.605) was used due to a lack of data. these preliminary models offer an indication of how e. globoidea may grow in new zealand. they can be useful tools for forest managers to make initial management decisions for mature e. globoidea. for example, when applied in combination these set of models predict that 15-year-old stands will have mean top heights ranging from 15–22.5 m, basal area/ha ranging from 10–58 m2 ha-1 and volume ha-1 ranging from 52–252 m3. however, the set of models presented here did not cover all age classes, and so some interpolation or extrapolation may occur during projection. silvicultural and natural disturbances were not accounted in this study; therefore, the models’ performance can be altered once such effects are considered. the models are only strictly applicable to the site conditions specific to the plots used to develop them, hence need to be validated with data from plots in new sites. due to small amounts of data, especially the number of psps, the models must be regarded as preliminary and must not be used beyond the data range of the fitting dataset. conclusions the models developed in this study will provide valuable information and understanding about the growth patterns, stand density dynamics, and potential yield of e. globoidea stands in new zealand. this information substantially increases the limited knowledge base about this species, thus will help growers, managers and investors to make appropriate planting and management decisions, as well as the potential economic return at harvest. the growth patterns will vary at individual sites; therefore, caution must be exercised how these results are applied. moreover, more plot measurement data including site characteristics and silvicultural regimes may increase the precision of these models and reduce bias in future. this study also showed sparse dataset can be useful to make indicative prediction models, which could give a preliminary information about specific species. however, acquisition and maintaining longterm plot measurement data is indispensable to make comprehensive forecasting models needed to underpin management decisions. additional files additional file 1: appendix 1. list of abbreviations ada: algebric difference approach; dbh: diameter at breast height (1.4m); mth: mean top height; g: basal area; sdd: standard deviation of diameter; dmax: maximum diameter; v: stand volume; sdi: stand density index; rmse: root mean square error; se: standard error; press: predictive residual error sum square. software a web-based e. globoidea growth and yield software is available from the following link: http://www. treesandstars.com/models/egloboidea.htm competing interests the authors declare that they have no competing interests. authors’ contributions ss and egm conceived the study from the data provided by dfm. ss analysed the data and wrote the manuscript under supervision of egm and jm. all the authors read and approved the final version of the manuscript. funding funding was received from agricultural and marketing research and development trust (agmardt) and speciality wood product programme (swp), new zealand. acknowledgements the authors are grateful to agmardt and swp for providing financial support. they would like to thank scion ltd. for making the data available and professor horacio bown for providing comments on the initial draft. they also appreciate the suggestions and comments provided by two anonymous reviewers and co-editor dr john moore on earlier version of this manuscript. references apiolaza, l., mcconnochie, r., & millen, p. (2011). introducing durable species to new zealand drylands: genetics of early adaptation of eucalyptus bosistoana. paper presented at the developing a eucalypt resource: learning from salekin et al. new zealand journal of forestry science (2020) 50:2 page 12 australia and elsewhere symposium, wood technology research centre, christchurch, new zealand, 2 november 2011. https://luis.apiolaza. net/uploads/apiolaza_et_al_2011_eucalyptus_ bosistoana_in_nz_drylands.pdf arlot, s., & celisse, a. (2010). a survey of cross-validation procedures for model selection. statistics surveys, 4, 40-79. https://doi.org/10.1214/09-ss054 bailey, r.l., & clutter, j.l. (1974). base-age invariant polymorphic site curves. forest science, 20(2), 155159. battaglia, m., sands, p., white, d., & mummery, d. (2004). cabala: a linked carbon, water and nitrogen model of forest growth for silvicultural decision support. forest ecology and management, 193(1), 251-282. https://doi.org/10.1016/j.foreco.2004.01.033 bell, d.t., & williams, j.e. (1997). eucalypt ecophysiology. in williams, j., & woinarski, j. (eds.). eucalypt ecology: individuals to ecosystems. cambridge, uk: cambridge university press, pp. 168-196. belli, k.l., & ek, a.r. (1988). growth and survival modeling for planted conifers in the great lakes region. forest science, 34(2), 458-473. berrill, j., nicholas, i.d., & gifford, h.h. (2007). preliminary growth and yield models for evenaged acacia melanoxylon plantations in new zealand. new zealand journal of forestry science, 37(1), 37-56. berrill, j.p. (2004). preliminary growth and yield models for even-aged cupressus lusitanica and c. macrocarpa plantations in new zealand. new zealand journal of forestry science, 34(3), 272-292. berrill, j.p., & hay, a.e. (2005). indicative growth and yield models for even-aged eucalyptus fastigata plantations in new zealand. new zealand journal of forestry science, 35(2/3), 121-138. berrill, j.p., & hay, a.e. (2006). indicative growth and yield models for stringybark eucalypt plantations in northern new zealand. new zealand journal of forestry, 51(1), 19-22. blake, j., somers, g., & ruark, g. (1990). perspectives on process modeling of forest growth responses to environmental stress. in: r. dixon, r. meldahl, g. ruark, & w. warren (eds.), process modeling of forest growth responses to environmental stress (pp. 9-17). portland, or, usa: timber press. blanco, j.a., zavala, m.a., imbert, j.b., & castillo, f.j. (2005). sustainability of forest management practices: evaluation through a simulation model of nutrient cycling. forest ecology and management, 213(1), 209-228. https://doi.org/10.1016/j. foreco.2005.03.042 borders, b.e. (1989). systems of equations in forest stand modeling. forest science, 35(2), 548-556. borders, b.e., bailey, r.l., & clutter, m.l. (1988). forest growth models: parameter estimation using real growth series. usda forest service general technical report. minneapolis, mi, usa:nc-north central forest experiment station. box, g.e. (1976). science and statistics. journal of the american statistical association, 71(356), 791-799. https://doi.org/10.1080/01621459.1976.104809 49 castedo-dorado, f., diéguez-aranda, u., barrio-anta, m., & álvarez-gonzàlez, j.g. (2007). modelling stand basal area growth for radiata pine plantations in northwestern spain using the gada. annals of forest science, 64(6), 609-619. https://doi. org/10.1051/forest:2007039 castedo-dorado., f., diéguez-aranda, u., & álvarezgonzález, j.g. (2007). a growth model for pinus radiata d. don stands in north-western spain. annals of forest science, 64(4), 453-465. https:// doi.org/10.1051/forest:2007023 chou, c.k.s. (1991). perspectives of disease threat in large-scale pinus radiata monoculture the new zealand experience. european journal of forest pathology, 21(2), 71-81. https://doi. org/10.1111/j.1439-0329.1991.tb00946.x clutter, j.l. (1963). compatible growth and yield models for loblolly pine. forest science, 9(3), 354-371. clutter, j.l., fortson, j.c., pienaar, l.v., brister, g.h., & bailey, r.l. (1983). timber management: a quantitative approach. new york, usa: john wiley & sons, inc. curtis, r.o. (1967). height-diameter and heightdiameter-age equations for second growth douglas fir. forest science, 13, 365-375. drew, t.j., & flewelling, j.w. (1977). some recent japanese theories of yield-density relationships and their application to monterey pine plantations. forest science, 23(4), 517-534. drew, t.j., & flewelling, j.w. (1979). stand density management: an alternative approach and its application to douglas-fir plantations. forest science, 25(3), 518-532. ek, a.r. (1974). nonlinear models for stand table projection in northern hardwood stands. canadian journal of forest research, 4(1), 23-27. https://doi. org/10.1139/x74-004 garcia, o. (1974). height-diameter equations for pinus radiata. [nota tecnica, no. 19] instituto forestal, chile. garcia, o. (1988). growth modelling — a review. new zealand forestry, 33(3), 14-17. garcía, o. (1981). simplified method-of-moments estimation for the weibull distribution. new zealand journal of forestry science, 11(3), 304-306. goulding, c.j. (2005). measurement of trees. in m. colley (ed.), forestry handbook (4th ed., p. 316). christchurch, new zealand: the new zealand insitute of forestry (inc.). hamner, b., & frasco, m. (2018). metrics: evaluation metrics for machine learning. r package version 0.1.4. https://cran.r-project.org/ package=metrics jansen, j.j., sevenster, j., & faber, j. (1996). opbrengsttabellen voor belangrijke boomsoorten in nederland. wageningen, the netherlands: ibndlo kakitani, t. (2017). the global timberlization movement and the potential for durable eucalyts: downstream salekin et al. new zealand journal of forestry science (2020) 50:2 page 13 opportunities. paper presented at the durable eucalypts on drylands: protecting and enhancing value symposium, marlborough research centre, blenheim, new zealand 19 april 2017. kitikidou, k., papageorgiou, a., milios, e., & stampoulidis, a. (2016). preliminary individual tree growth model, site index model, mortality model for aleppo pine (pinus halepensis mill.) in chalkidiki (northern greece). journal ciência florestal, 26(4), 12471257. https://doi.org/10.5902/1980509825136 korzukhin, m.d., ter-mikaelian, m.t., & wagner, r.g. (1996). process versus empirical models: which approach for forest ecosystem management? canadian journal of forest research, 26(5), 879887. https://doi.org/10.1139/x26-096 kuru, g.a., whyte, a.g.d., & woollons, r.c. (1992). utility of reverse weibull and extreme value density functions to refine diameter distribution growth estimates. forest ecology and management, 48(1), 165-174. https://doi.org/10.1016/03781127(92)90128-v landsberg, j.j., waring, r.h., & coops, n. c. (2003). performance of the forest productivity model 3-pg applied to a wide range of forest types. forest ecology and management, 172(2), 199-214. https://doi.org/10.1016/s0378-1127(01)008040 lee, s.h. (1998). modelling growth and yield of douglasfir using different interval lengths in the south island of new zealand. (phd), university of canterbury, christchurch, new zealand. lin, h. (2017). risk and impact of insect herbivores on the development of dryland eucalyptus forestry in new zealand. (phd), university of canterbury, christchurch, new zealand. long, j.n. (1985). a practical approach to density management. the forestry chronicle, 61(1), 23-27. https://doi.org/10.5558/tfc61023-1 lundgren, a.c., gordon, a. & hay, e. (1995). taper equations for eucalyptus pilularis, e. globoidea and e. muelleriana in new zealand. rotorua, new zealand: management of eucalypts cooperative. mason, e.g. (2019). influences of mean top height definition and sampling method on errors of estimates in new zealand’s forest plantations. new zealand journal of forestry science, 49:1. https:// doi.org/10.33494/nzjfs492019x24x meason, d., bellè, p., höck, b., lumnitz, s., & todoroki, c. (2016). alternative species site mapping review and analysis. retrieved 20 july 2016 from new zealand farm forestry association https://www.nzffa.org. nz/system/assets/5143/swp-t004.pdf menzies, h. (1995). eucalypts show potential. farm forestry review, 33-34. methol, r.j. (2001). comparisons of approaches to modelling tree taper, stand structure and stand dynamics in forest plantations. (phd), university of canterbury, christchurch, new zealand. millen, p., van ballekom, s., altaner, c., apiolaza, l., mason, e., mcconnochie, r.,murray, t. (2018). durable eucalypt forests–a multi-regional opportunity for investment in new zealand drylands. new zealand journal of forestry, 63(1), 11-23. myers, r.h., & myers, r.h. (1990). classical and modern regression with applications (vol. 2): belmont, ca, usa: duxbury press. näslund, m. (1936). skogsförsöksanstaltens gallringsförsök i tallskog. meddelanden från statens skogsförsöksanstalt (reports of the swedish institute of experimental forestry) no. 29. stockholm, sweden: centraltrvckeriet, esselte h t t p s : / / p u b . e p s i l o n . s l u . s e / 1 0 1 5 9 / 1 / m e d d _ statens_skogsforskningsanst_029_01.pdf ministry for primary industries. (2019). national exotic forest description. wellington, new zealand. nicholas, i., & millen, p. (2012a). eucalyptus globoidea. [durable eucalypt leaflet series]. blenheim, new zealand: new zealand dryland forest initiative. https://nzdfi.org.nz/wpcontent/ uploads/2014/08/leaflet3e.globoideadec2012. pdf nicholas, i., & millen, p. (2012b). eucalyptus globoidea a durable hardwood for planting in the bay of plenty region. blenheim, new zealand: marlborough research centre. new zealand dryland forest initiative. (2013). introductory brochure. blenheim, new zealand: marlborough research centre. https://nzdfi.org. nz/wp-content/uploads/2015/04/eucalyptusgloboidea-a-durable-hardwood-for-planting-inthe-bop-region.pdf new zealand forest owners association. (2017). new zealand plantation forest industry: facts and figures 2016/17. wellington, new zealand. https:// www.nzfoa.org.nz/images/stories/pdfs/facts_ figures_2016_%c6%92a_web_version_v3.pdf page, d., & sing, t. (2014). durability of new zealand grown timbers. new zealand journal of forestry, 58(4), 26-30. palahí, m., & grau, j. (2003). preliminary site index model and individual-tree growth and mortality models for black pine (pinus nigra arn.) in catalonia (spain). journal of forest systems, 12(1), 137-148. park, a., & wilson, e.r. (2007). beautiful plantations: can intensive silviculture help canada to fulfil ecological and timber production objectives? the forestry chronicle, 83(6), 825-839. https://doi. org/10.5558/tfc83825-6 pienaar, l.v., & rheney, j.w. (1995). modeling stand level growth and yield response to silvicultural treatments. forest science, 41(3), 629-638. pilaar, c.h., & dunlop, j.d. (1990). the permanent sample plot system of the new zealand ministry of forestry. bulletin des recherches agronomiques de gembloux, 25(1), 5-17. pretzsch, h., & biber, p. (2005). a re-evaluation of reineke’s rule and stand density index. forest science, 51(4), 304-320. r development core team. (2017). r: a language and environment for statistical computing. vienna, austria: r foundation for statistical computing; 2016. http://www. r-project. org. salekin et al. new zealand journal of forestry science (2020) 50:2 page 14 reineke, l.h. (1933). perfecting a stand-density index for even-aged forest. journal of agricultural research, 4, 627-638. sánchez-gonzález, m., tomé, m., & montero, g. (2005). modelling height and diameter growth of dominant cork oak trees in spain. annals of forest science, 62(7), 633-643. https://doi.org/10.1051/ forest:2005065 saunders, m.r., & puettmann, k.j. (2000). a preliminary white spruce density management diagram for the lake states. [staff paper series no 145]. st paul, mi, usa: department of forest resources, university of minnesota. tewari, v.p., & von gadow, k. (2003). preliminary growth models for prosopis cineraria (l.) druce plantations in the hot arid region of india. forests, trees and livelihoods, 13(4), 361-373. https://doi.org/10.10 80/14728028.2003.9752471 turner, j.a., west, g., dungey, h., wakelin, s., maclaren, p., adams, t., & silcock, p.j. (2008). managing new zealand planted forests for carbon: a review of selected scenarios identification of knowledge gaps. wellington, new zealand: ministry of agriculture forestry. vanclay, j.k. (1994). modeling forest growth and yield. applications to mixed tropical forests. wallingford, uk: cab international. vanclay, j.k. (2010). robust relationships for simple plantation growth models based on sparse data. forest ecology and management, 259(5), 1050-1054. https://doi.org/10.1016/j. foreco.2009.12.026 warren, w.g. (1971). correlation or regression: bias or precision. journal of the royal statistical society. series c (applied statistics), 20(2), 148-164. https://doi.org/10.2307/2346463 wickham, h. (2016). ggplot2: elegant graphics for data analysis. new york, usa: springer-verlag. https:// doi.org/10.1007/978-3-319-24277-4 williams, j., & woinarski, j. (eds.). (1997). eucalypt ecology: individuals to ecosystems. cambridge, uk: cambridge university press. woollons, r.c., snowdon, p., & mitchell, n.d. (1997). augmenting empirical stand projection equations with edaphic and climatic variables. forest ecology and management, 98(3), 267-275. https://doi. org/10.1016/s0378-1127(97)00090-x zeide, b. (1993). analysis of growth equations. forest science, 39(3), 594-616. https://doi.org/10.1093/ forestscience/39.3.594 zhao, w. (1999). growth and yield of pinus radiata in canterbury, new zealand. (phd), university of canterbury, christchurch. (sd397.p6117.z63) zhao, w., mason, e.g., & brown, j. (2006). modelling height-diameter relationships of pinus radiata plantations in canterbury, new zealand. new zealand journal of forestry, 51(1), 23-27. salekin et al. new zealand journal of forestry science (2020) 50:2 page 15 new zealand journal of forestry science additional file 1: appendix a preliminary growth and yield model for eucalyptus globoidea (blakely) plantations in new zealand valg = π ∙ dbh2 ∙ h3 40000 ∙ (h − 1.4)2 ∙ (0.2134788 ∙ βc + 0.011344) 𝛽𝛽𝑐𝑐 βc = (1 − 1.4 h ) 0.280729 − 0.414429 (1 − 1.4 h ) 17.26155 eucalyptus globoidea age(years) residual stems ha-1 age(years) height (m) new zealand journal of forestry science . preliminary model’s parameter estimates. 𝛼𝛼 β 𝛾𝛾 δ mth p sig. g p sig. dmax p sig. sdd p sig. v p sig. α0 α1 β0 β1 h-d p sig. wood density estimates of standing trees by microdrilling and other non-destructive measures christine l. todoroki1*, eini c. lowell2 and cosmin n. filipescu3 1 scion, titokorangi drive (formerly, longmile road), whakarewarewa, rotorua 3010, new zealand 2 usda forest service, 620 sw main st. suite 502, portland, or 97205, usa 3 canadian wood fibre centre, 506 burnside road west, victoria, bc, v8z 1m5, canada *corresponding author: christine.todoroki@scionresearch.com (received for publication 11 october 2019; accepted in revised form 7 april 2021) abstract background: accurate estimates of wood density are needed by the forest sector to increase value along the tree-to-product value-chain. amongst tools supporting in-situ assessments, micro-drills and acoustic hammers have become increasingly popular. our objective was to use these tools, and other easily-obtained measures, to develop predictive wood density models for in-situ assessments of douglas-fir (pseudotsuga menziesii (mirb.) franco) trees in western north america. methods: wood density estimates of 133 trees, 60–75 years-old, were benchmarked against x-ray densitometry data using linear mixed-effects models. mean resistograph amplitude (unadjusted, adjusted, and standardised variants), and combinations of acoustic velocity, tree diameter, stand age, and site index were considered as fixed effects. plots, comprising differing treatments, and sites were considered as random effects. candidate models were selected based on fit statistics, and further evaluated with an independent external dataset comprising 37 douglas-fir trees. results: the optimal model comprised amplitude (adjusted), site index (transformed), and the quotient of velocity and age. it had a mean absolute percentage error, mape, of 4.1%, mean absolute error, mae, of 19.4 kg.m-3, a root-mean-squarederror, rmse of 25.0 kg.m-3, and marginal r2 for fixed effects, r2marg of 0.60. with external data, mape was 8.7%, mae 52.4 kg.m-3 and rmse 59.5 kg.m-3. fit statistics for a simpler two-variable model (standardised amplitude and transformed site index) were: mape 4.9%, mae 23.2 kg.m-3, rmse 28.0 kg.m-3, and r2marg, 0.48, and with external data mape was 8.5%, mae 51.6 kg.m-3 and rmse 59.3 kg.m-3. thus, with external data, the simpler model produced greater accuracy than the optimal model. amplitude, and all other single-variable models, recorded poorer levels of accuracy. conclusions: micro-drilling alone, though highly significant as a predictor, is insufficient for providing accurate wood density estimates of individual trees. site effects need to be considered too. standardisation of mean amplitudes to z-scores makes models highly portable across a range of resistance tools and operating speeds, and therefore more practical. as noted in the literature, optimal models are not necessarily best for predicting outcomes with other datasets, therefore model evaluation with external data is critical to determining how well a model will perform in practice. new zealand journal of forestry science todoroki et al. new zealand journal of forestry science (2021) 51:6 https://doi.org/10.33494/nzjfs512021x74x e-issn: 1179-5395 published on-line: 22/06/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access introduced tree species in new zealand (wang et al. 2001; maclaren 2009), france, germany, the uk, spain, belgium, and the czech republic (zeidler et al. 2017; spiecker et al. 2019). its popularity arises from its relatively fast growth, high productivity, and desirable wood properties. wood density is one such desirable property and an important indicator of the quality of solid wood products introduction douglas-fir (pseudotsuga menziesii (mirb.) franco) is widely recognised as being the most commercially significant species for structural applications in north america, particularly the pacific northwest of the united states, and the west coast of canada. outside of north america, douglas-fir is amongst the most commonly keywords: resistograph; acoustic velocity; douglas-fir; mixed-effects models http://creativecommons.org/licenses/by/4.0/), todoroki et al. new zealand journal of forestry science (2021) 51:6 page 2 (zobel & van buijtenen 2012). wood density directly influences and is correlated with stiffness and strength (zeidler et al. 2017), thus is used as a selection criterion in tree breeding programs for improving yield of high quality structural lumber (howe et al. 2006). because douglas-fir wood density varies considerably from one tree to another, from one site to another, and from one region to another (filipescu et al. 2014; kimberley et al. 2017), forest managers and wood processors need to be able to assess this wood property to best match the raw material to the final product. when assessed in-situ, there are greater opportunities to increase revenue to both land and mill owners, through improved selection, sorting, allocation to the correct processing pathway, and through reduced wastage and manufacturing costs. therefore, there is a need for rapid, accurate, and nondestructive methods for assessing wood density in standing trees. a range of non-destructive tools and methods have been developed for in-situ assessments of wood density (gao et al. 2017; schimleck et al. 2019). examples include wood increment borers, the pilodyn wood tester, torsiometers, and micro-drills (gao et al. 2012; wessels et al. 2011). acoustic tools, although developed primarily for assessing wood stiffness, have also been used in wood density studies (e.g. chauhan & walker 2006; el-kassaby et al. 2011; newton 2017). increment cores, when analysed using x-ray densitometry techniques (walker & dodd 1988; eberhardt & samuelson 2015), enable very accurate measurements of wood density. however, x-ray densitometry is expensive and time-consuming (chantre & rozenberg 1997), requires rigorous methods for preparing and processing cores, and therefore is neither a rapid technique nor can it be applied in-situ on standing trees. other tools such as the pilodyn and torsiometer, which penetrate only a small way through the bark, are less destructive than increment borers, but, according to gao et al. (2012) they cannot be considered as substitutes for the increment borer as they have had only limited success. in contrast, micro-drill resistance tools have demonstrated greater potential (isik & li 2003; gao et al. 2012), are less invasive than increment borers due to the smaller holes (3 mm) left following drilling (rinn 1988) and are used extensively in assessing progeny trials (bouffier et al. 2008; gwaze & stevenson 2008; elkassaby et al. 2011; desponts et al. 2017). micro-drill resistance tools record the amplitude of the resistance to turning (torque) experienced by a fine drill when driven through wood at a given forward speed (cm per min) and rotational frequency (rpm) (rinn 1988). as the bit progresses through the stem, resistance due to friction generally increases, thus creating an increasing trend. therefore, adjustments need to be made to the data prior to analyses, to remove any potential sources of bias. methods developed to adjust, or detrend, resistance profiles include trigonometric approaches (gantz 2002; fundova et al. 2018), smoothing functions, and translation functions that shift the baseline to correct for bias (isik & li 2003; eckard et al. 2010; fundova et al. 2018). due to the nature of these detrending methods, the full bark-to-bark profile is required. shorter profiles (e.g. the first 5 cm of the inside-bark profile) have also been evaluated (bouffier et al. 2008) and moderate relationships (i.e. r2 ≈ 0.41 and 0.48, based on correlations of 0.64, 0.69) reported between mean (adjusted) amplitude and mean wood density assessed in progeny trials of maritime pine (pinus pinaster ait.) at two sites. in general, only weak to moderate relationships have been found for individual trees. gwaze & stevenson (2008) reported an r2 of 0.23 for 25-year-old shortleaf pine (pinus echinata mill.) in missouri, usa, while walker et al. (2019) reported an r2 of 0.47 for 6-9-year-old loblolly pine growing in southeastern us and noted improved correlations with inclusion of site effects in their predictive models. isik and li (2003) obtained r2 values of 0.21, 0.24, 0.31, 0.44 for 11-year-old loblolly pine (pinus taeda l.) at four sites in south carolina, usa, and when all sites were combined, the phenotypic correlation was just 0.12. phenotypic correlations, often estimated using product-moment correlation statistics (also called pearson’s r) should not be confused with genetic correlations that, based on correlations between family means, will always be stronger than their phenotypic counterparts. it should also be noted that the coefficient of determination, r2, equal to the square of r, describes the explanatory power of a model with the dataset at hand, not the predictive ability or usefulness of the model to new data obtained from other settings. therefore, models that fit well to in-sample data (i.e. with a high r2 value), may not necessarily provide accurate predictions when applied to new data (mendenhall & sincich 2012). the same holds true for other variates of r2 including marginal and conditional r2 values, r2marg and r 2 cond respectively, (nakagawa & schielzeth 2013) where r2marg represents the proportion of variance explained by fixed effects alone (akin to r2 in a fixed-effects model) while r2cond represents the proportion of variance explained by both fixed and random effects. a low r2marg and high r2cond implies that the fixed effects (measured variables) explain little of the variance while random effects (unmeasured) explain a far greater proportion of the variance, indicating that there are other factors that the model may have failed to capture. in addition to examining goodness-of-fit statistics, the use of an independent dataset is considered the “gold standard” for assessing the predictive power of models (alexander et al. 2015). moreover, this new dataset adds a further level of scrutiny to the model construction process (snee 1977). when it is not possible to collect new data, techniques such as data-splitting, crossvalidating, or bootstrapping can be applied (snee 1977; dankers et al. 2019). despite the results of previous studies demonstrating the potential of micro-drilling to estimate wood density, a factor that has limited its practical use is the availability of accurate, reliable models that make predictions with new data from other settings. the objective, and key challenge of our research, was to develop individual tree-based predictive models and to access their accuracy with both in-sample and external datasets. using our predictive models, we wished to resolve three key questions: 1) can micro-drill resistance tools alone provide accurate assessments of wood density for a diverse set of trees? 2) when used in combination with acoustic velocity tools, and other easily measured variables (diameter, stand age, site index), what level of accuracy can be achieved? 3) how portable are the predictive models to new data? overall, we wanted to develop robust, portable models for rapid in-situ assessments of wood density in individual standing trees. methods study sites the douglas-fir trees of this study were located in coastal western north america. they ranged in age from 60 to 75 years old and were selected from six experimental installations, four of which were planted, and the remaining two of natural origin, both of which had regenerated naturally after wildfire. site index (at 50 years) ranged from 27 to 41 m, table 1. the installations were established between 1963 and 1970, on sites with elevations ranging from 274 to 823 m above sea level. measurements from five of the six stands were used to develop models for predicting wood density (known hereafter as the in-sample dataset). following model development, and candidate model selection, measurements from the sixth stand were obtained to test model portability and predictive ability with new observations (the external dataset). the external dataset was collected one year after the in-sample dataset. the five stands of the in-sample dataset were part of the levels-of-growing-stock (l.o.g.s) cooperative study in douglas-fir (williamson & staebler 1971; marshall et al. 1992). each of the l.o.g.s stands comprised plots with three differing treatments; a control (i.e. no treatment and plot density greater than 2450 stems per hectare), a light thinning treatment (for which 70% of basal area was retained), and a heavy thinning treatment (for which 30% of basal area was retained). plot densities were maintained over time, relative to the controls, through repeated thinning treatments. the sixth stand, though adjacent to one of the l.o.g.s sites, was in no way related to l.o.g.s. this completely independent stand was managed differently. it comprised four combinations of thinning and fertilisation treatments; 1) no thinning or fertilisation; 2) no thinning, but fertilisation with 448 kg n ha-1; 3) thinning with 1/3 of the basal area retained, but no fertilisation; and 4) thinning with 1/3 of the basal area retained and fertilisation with 448 kg n.ha-1. the sixth stand, with a different study design and evaluated with a different instrument, using different settings, provided an external dataset (i.e. completely separate data) and an opportunity to evaluate model portability and accuracy. tree measurements trees were randomly chosen from plots by stratification into three diameter at breast height (dbh) classes which differed with site and treatment. the three classes approximately represented boxplot statistics (i.e. minimum dbh to lower quartile, lower quartile to upper quartile, and upper quartile to maximum dbh) thus ensuring that dbh distributions were approximately normal. in total, 172 trees were sampled with 133 trees used for model development (five locations x three treatments x three replicates x three trees = 135, minus two trees for which data were missing due to issues with sample preparation and collection of field micro-drill data) and 37 trees for evaluation of model portability (one location x four treatments x two replicates x five trees = 40, minus three trees with unreliable microdrill data). the two datasets were collected during late summer / early autumn at the end of growing season. there was a one-year difference between collection of the two datasets, however weather conditions were very similar (dry and warm) and typical for the season in the pacific northwest. each tree was measured at breast height for diameter, by micro-drill resistance methods for amplitude, and by time-of-flight tools for acoustic velocity. a 5-mm core sample, also taken at breast height, was extracted from each tree for wood density assessment using x-ray todoroki et al. new zealand journal of forestry science (2021) 51:6 page 3 table 1: description of the study sites site latitude longitude elevation origin site index established* age in study (m, a.s.l.) (m, age 50) (years) hoskins 44°41’ 123°30’ 305 natural 41 1963 66 iron creek 46°24’ 121°59’ 762 planted 40 1966 60 sayward 50°04’ 125°35’ 274 planted 34 1969 62 shawnigan 48°38’ 123°43’ 335 planted 29 1970 64 stampede 42°53’ 122°49’ 823 natural 34 1968 75 shawnigan 48°38’ 123°43’ 335 planted 27 1971 64 *year of establishment indicates when experimental sites were established, not planted. table 1: location and description of study sites comprising in-sample and external datasets. densitometry. this approach was used to provide an accurate measure of wood quality, and as a benchmark against which the non-destructive measures could be evaluated. cores were initially frozen to prevent mold and stain, then dried at 50℃ for 24 hours. following conditioning (for at least 48 hours) to attain a uniform moisture content of 8%, x-ray densitometry profiles were generated at a resolution of 0.06 mm. profiles included ring width, earlywood and latewood densities, and earlywood/latewood proportions. ring density was calculated using the weighted areas of earlywood and latewood. mean wood density of each tree was calculated using the weighted average of ring density for the length of the core from inside bark to pith. micro-drill resistance methods were applied from bark to pith in close proximity to the location of the core samples. in-sample dataset trees were drilled using an iml resi f400-s tool at the maximum constant forward speed of 150 cm/min, while external dataset trees were drilled using a newer resi pd500 at a constant forward speed of 25 cm/min. the rotational frequency used for the latter tool was 1500 rpm while that for the former, while not recorded, was known to be at a value between 400-1200 rpm. together, the selection of rotational frequency and forward speed settings is important for preventing overloading the motor, while drilling speed selection can influence drilling resistance measurements and subsequent prediction of wood density (sharapov et al. 2019a). in general, resistance amplitudes tend to be lower at lower forward speeds, but the difference may be equal to several orders of magnitude (mattheck et al. 1997, rinn 2015). both instruments converted variations in torque into graphical and digital outputs of path length of the drill (measured at a resolution of 0.1 mm) and the relative resistance, given as an amplitude percentage, that the drilling bit encountered. before average amplitude was calculated, the initial portion of the profile through the bark was removed as was any portion that extended beyond the pith; identified by a bowl-like shape (rinn 2012; fundova et al. 2018). an average of the amplitude profile for each tree was calculated using six approaches. the first approach, the simplest, determined the arithmetic mean of the profile. since mean values are sensitive to outliers, with further bias introduced through increasing drill resistance, the second approach overcame these shortcomings and detrended the profiles through construction of trendlines that smoothed over fluctuations, and through correction of the baseline. the third approach was similar to the second, but after adjusting/detrending the profile, a value determined within the first 10 cm of the inside-bark profile was added to the mean. the third approach was motivated by the study of bouffier et al. (2008). however, rather than using 5 cm, as in their approach, we chose 10 cm due to our trees being considerably older, and hence larger. the remaining three approaches scaled the mean amplitudes to their respective z-scores to enable comparison of values from different samples (which may have different means and standard deviations, as can be the case when drills are todoroki et al. new zealand journal of forestry science (2021) 51:6 page 4 operated at different speeds, or different brands of drills are used). the mean amplitude values, x, were scaled using the usual method of calculation (draper & smith 1966) i.e. subtraction of the mean, μ, and dividing the result by the standard deviation, σ, i.e. z = (x – μ) / σ. details of the six approaches follow: a0: the arithmetic mean amplitude of unadjusted data. a1:the average difference between centered moving means and centered moving minimums (window widths of 10 and 100 respectively), plus the average centered moving minimum within 100 mm of the inside-bark profile (following isik & li 2003). a2: the average difference between centered moving means and centered moving minimums (window widths of 10 and 100 respectively), plus the average amplitude within 100 mm of the inside-bark profile. z0: the standardised equivalent of a0. z1: the standardised equivalent of a1. z2: the standardised equivalent of a2. with a0 being the unadjusted mean, a1 will be less than or equal to a0, and a2 greater than a0 (figure 1). data distributions for a0, a1, and a2 are preserved after standardisation to z0, z1, and z2 (figure 2), and therefore performance metrics arising from linear models with either standardised (z0, z1, z2) or nonstandardised data (a0, a1, a2) will be identical for insample datasets. trees were also acoustically assessed for timeof-flight using a hitman st300 (paradis et al. 2013). sensors were centered at breast height and placed approximately one metre apart. measurements were taken on opposite sides and oriented perpendicular to the slope orientation to avoid reaction wood. four measurements were taken per tree, and the mean of the measurements for each tree (automatically calculated by the hitman) used in the analyses. a summary of all data collected, for both in-sample and external datasets, is given in table 2. analyses to account for experimental variability and structure inherent in our data, linear mixed-effects models that allow for random group effects, were developed to estimate wood density. the models were developed using data from the 133 in-sample trees, and were formulated as: y = xβ + zu + ε where y is the response vector (mean wood density obtained by x-ray densitometry for each of the 133 trees), x and z are matrices of explanatory variables corresponding to fixed (observed/measured variables) todoroki et al. new zealand journal of forestry science (2021) 51:6 page 5 figure 1: example of amplitude profiles using an iml res f400-s (a) and a resistograph pd500 (b), before (red) and after (blue) adjustment. mean amplitudes for the unadjusted profile, a0, and adjusted variants, a1, a2, are indicated by horizontal solid red, and dashed blue and green lines respectively. figure 2: kernel density profiles of non-standardised mean resistograph amplitudes (a0, a1, a2), and standardised equivalents (z0, z1, z2) for in-sample and external datasets. sample size is indicated by n, and bandwidth by bw. and random effects (unobserved/unmeasured variables) respectively, β and u are the corresponding vectors of parameters for the respective fixed and random effects, and ε is a vector of random errors. explanatory variables included acoustic velocity, v (km.s-1), breast-height diameter, dbh (cm), stand age (years), site index, si (m, age 50), and micro-drill amplitude (% for a0, a1, a2, unitless otherwise). each of the six amplitude variants was sequentially evaluated. the quotient of v and age, v/age, was also evaluated as an explanatory variable. inverse and natural logarithmic transformations of variables were examined and selected by inspecting plots for linearity and constant variance with increasing mean values of the dependent variable. models developed with combinations of explanatory variables, including single variable models, were explored. for all models, site and plot (plot nested within site) were formulated as random effects. the models were fitted using the restricted maximum likelihood method (reml, searle et al. 1992) and developed using the linear and nonlinear mixed-effects models package, “nlme”, (pinheiro et al. 2019) within the r environment (r core team 2018). the significance of explanatory variables was evaluated with α = 0.05 using conditional t-tests and f-tests (pinheiro & bates 2006). moving (rolling, running) means were facilitated through the “catools” package (tuszynski 2019) and data frame manipulations through the “plyr” package (wickham 2011). performance of models that satisfied the conditional t and f tests, were evaluated using: aic (akaike 1974), r2marg and r 2 cond following johnson (2014), mean absolute error, mae (equation 1), mean absolute percentage error, mape (equation 2), and root-mean-squared error, rmse (equation 3). multiple metrics were applied because the use of a single metric could lead to an incorrect interpretation. though we report both r2marg and r 2 cond, our focus is primarily on the former metric, similar to the coefficient of determination, r2. diagnostics of the models included plots of residuals against fitted values, and plots of observed values versus fitted values. all predictions and subsequent calculations of mae, mape, and rmse were made at the population level, because, in practice, contributions due to random effects are unknown. (1) (2) (3) where yi and ŷi are observed and predicted values, and n the sample size. candidate models were selected using all but the r2cond performance metric. the models were first grouped by number of variables (as this influences performance metrics). within each model subgroup, any model having at least one performance metric in the top two of that metric was selected as a candidate model. this process essentially filtered out poorer-performing models. candidate models were then evaluated using external data and the two best candidate models selected based on mape, mae, rmse metrics and their model parameters presented. the simpler of the two models (since simple models are generally better in practice) was then selected for further analyses with overall performance demonstrated by the percentages of all trees within 25 kg.m-3, 50 kg.m-3, 5% and 10% of their true values. summary statistics are reported as means with variability indicated by standard errors (se). todoroki et al. new zealand journal of forestry science (2021) 51:6 page 6 site n density dbh v a0 a1 a2 z0 z1 z2 (kg.m-3) (cm) (km.s-1) (%) (%) (%) in-sample dataset hoskins 25 465±7 46±2 4.4±0.1 45±3 40±3 54±3 0.5±0.3 0.3±0.3 0.4±0.3 iron creek 25 450±7 42±2 4.5±0.1 35±2 32±2 45±2 -0.5±0.2 -0.5±0.2 -0.5±0.2 sayward 27 500±5 33±2 4.7±0.1 40±2 39±2 53±2 0.1±0.2 0.2±0.2 0.3±0.2 shawnigan 27 522±6 27±2 4.6±0.1 40±2 39±2 50±2 0.0±0.2 0.2±0.2 0.0±0.2 stampede 27 474±6 39±2 4.6±0.1 38±1 35±1 47±1 -0.1±0.1 -0.2±0.1 -0.3±0.1 all sites 133 483±4 37±1 4.6±0.1 40±1 37±1 50±1 0.0±0.1 0.0±0.1 0.0±0.1 external dataset shawnigan 37 590±6 24±1 4.6±0.1 6±0 6±0 8±0 0.0±0.2 0.0±0.2 0.0±0.2 table 2: summary data (mean ± se) of the in-sample and external datasets. dbh = diameter; v = acoustic velocity; a0 = mean amplitude; a1 = mean adjusted amplitude shifted by the mean rolling minimum amplitude within 100 mm of bark; a2 = mean adjusted amplitude shifted by the mean amplitude within 100 mm of bark; z0, z1, z2 are standardised equivalents of a0, a1, a2 respectively. results explanatory variables all variants of mean resistograph amplitude were highly significant as explanatory variables for predicting wood density, as was site index (transformed), acoustic velocity, v, and the quotient v/age. however, age was not significant as an explanatory variable. diameter was significant as a predictor of wood density only when in combination with v (table 3). model performance with in-sample data and candidate models the optimal model, i.e. the model with the best performance metrics (table 3), comprised three explanatory variables: 1/si, v/age, and a2 (or z2). for this model aic, mae, mape, and rmse were the lowest amongst all models, and r2marg (0.60) the highest. r2cond was 0.61, which indicated that the fixed effects accounted for the majority of the variance. mae was 19.4 kg.m-3, mape 4.1%, and rmse 25.0 kg.m-3. another three-variable model, with fixed effects x = [1/si, v, a2] (or similarly x = [1/si, v, z2]), also recorded very good performance statistics: r2marg 0.58, mae 19.9 kg.m -3, mape 4.2%, and rmse 25.4 kg.m-3. these two models were selected as candidate models for further evaluation with the external dataset. r2marg for all remaining threevariable models ranged 0.39 to 0.55, mae from 12.0 to 24.6 kg.m-3, mape from 4.5 to 5.2%, and rmse from 26.5 to 30.8 kg.m-3 (table 3). the poorest performance metrics were all attributed to one model with x = [1/ si, v, dbh]; the only model within this group without amplitude as an explanatory variable. amongst two-variable models, all models with transformed site index and amplitude (all variants) were selected as candidate models, as they all had at least one todoroki et al. new zealand journal of forestry science (2021) 51:6 page 7 explanatory variables 1, x r2marg r 2 cond aic mae (kg.m-3) mape (%) rmse (kg.m-3) 1/si 0.33 0.42 1307 26.0 5.5 32.2 v 0.03 0.48 1309 30.6 6.5 38.7 v/age 0.05 0.50 1308 30.6 6.5 38.9 a0 (z0) 0.12 0.56 1288 30.2 6.3 37.2 a1 (z1) 0.14 0.53 1286 28.6 6.0 35.6 a2 (z2) 0.18 0.59 1272 28.0 5.9 35.1 v, dbh 0.06 0.57 1304 33.0 7.0 41.0 1/si, v 0.36 0.46 1302 25.2 5.4 31.4 v/age, a0 (z0) 0.17 0.59 1277 28.9 6.1 35.7 v/age, a1 (z1) 0.18 0.56 1279 27.8 5.9 34.6 v/age, a2 (z2) 0.23 0.63 1264 27.2 5.7 34.1 1/si, v/age 0.38 0.46 1300 24.7 5.3 31.0 1/si, a0 (z0) 0.48 0.52 1278 23.2 4.9 28.0 1/si, a1 (z1) 0.49 0.52 1275 22.6 4.8 28.0 1/si, a2 (z2) 0.56 0.57 1258 20.9 4.4 26.2 1/si, v, dbh 0.39 0.49 1297 24.6 5.2 30.8 1/si, v, a0 (z0) 0.52 0.56 1268 21.8 4.6 27.1 1/si, v, a1 (z1) 0.52 0.56 1269 21.8 4.6 27.2 1/si, v, a2 (z2) 0.58 0.61 1251 19.9 4.2 25.4 1/si, v/age, a0 (z0) 0.55 0.56 1263 21.0 4.4 0.35 1/si. v/age, a1 (z1) 0.54 0.56 1265 21.1 4.5 26.7 1/si, v/age, a2 (z2) 0.60 0.61 1248 19.4 4.1 25.0 1 only models for which all explanatory variables, x, are significant are shown. si = site index; v = acoustic velocity; a0 = mean amplitude; a1 = mean adjusted amplitude shifted by the mean rolling minimum amplitude within 100 mm of bark; a2 = mean adjusted amplitude shifted by the mean amplitude within 100 mm of bark; z0, z1, z2 are standardised equivalents of a0, a1, a2 respectively (and can replace the non-standardised values with no change in fit statistics); dbh = diameter at breast height. table 3: performance metrics of mixed-effects models for estimating wood density, grouped by the number of explanatory variables. the two best metrics, used for candidate model selection, within each group, are shown in bold font. performance metric in the top two of that metric. for these candidate models r2marg ranged from 0.48 to 0.56, mae from 20.9 to 23.2 kg.m-3, mape from 4.4 to 4.9%, and rmse from 26.2 to 28.0 kg.m-3. one further candidate model with fixed effects x = [v/age, a2], had the secondbest aic, but other statistics were comparatively low (r2marg = 0.23). amongst single-variable models, transformed site index had the best r2marg (0.33), mae (26.0 kg.m -3), mape (5.5%) and rmse (32.2 kg.m-3). however, the best aic was associated with the model with amplitude (a2, or equivalently z2) as the sole predictor of wood density. though also having the second-best r2marg, mae and mape, these statistics, especially r2marg, were somewhat lower (0.18, 28.0 kg.m-3 and 5.9% respectively). candidate model performance with external data performance metrics of all candidate models evaluated with the external dataset are shown in table 4. metrics for models with non-standardised mean amplitudes (a0, a1, a2) are provided for comparative purposes only. standardised micro-drill amplitudes recorded substantially better performance statistics than their non-standardised counterparts. this was expected since the standardisation procedure converts values of the two tools to the same scale (whist retaining distribution properties). the comparison clearly demonstrates the need for standardisation. hereafter, the focus is on the standardised amplitude variants. the optimal model, x = [1/si, v/age, z2], had the best performance metrics amongst the three-variable candidate models, however was second best in terms of the performance metrics mae, mape, and rmse. the model with the best performance metrics comprised 1/ si and z0, the unadjusted amplitude. based on evaluation with the external dataset, the two candidate models selected were: • the optimal model in the in-sample dataset, with x = [1/si, v/age, z2] and • the simpler 2-variable model, with x = [1/si, z0]. parameters of the two selected candidate models are provided in table 5. random effects due to site were negligible (i.e. close to zero) for the optimal model, x = [1/si, v/age, z2], and ranged from -8.9 to 7.7 kg.m-3 for the simpler model, x = [1/si, z0]. in contrast, random effects due to plot ranged from -1.2 to 1.4 kg.m-3 for the optimal model and were negligible for the simpler model. residual plots of the two models shown in figure 3 appear to be reasonably random. though there is a small degree of asymmetry in both residual plots, overall there are no clear patterns. the simpler model with just two predictor variables x = [1/si, z0], though having an outlier with a residual of -82 kg.m-3, does not appear todoroki et al. new zealand journal of forestry science (2021) 51:6 page 8 explanatory variables, x mae (kg.m-3) mape (%) rmse (kg.m-3) 1/si 53.7 8.8 64.1 a2 182.0 30.6 185.0 z2 107.8 18.1 111.0 v/age, a2 179.4 30.2 183.0 v/age, z2 55.1 9.1 110.0 1/si, a0 100.8 16.8 107.0 1/si, z0 51.6 8.5 59.3 1/si, a1 112.5 18.8 117.8 1/si, z1 56.4 9.3 63.4 1/si, a2 130.4 21.8 135.0 1/si, z2 53.3 8.8 60.2 1/si, v, a2 131.5 22.0 136.0 1/si, v, z2 55.0 9.1 61.9 1/si, v/age, a2 130.3 21.8 135.0 1/si, v/age, z2 52.4 8.7 59.5 table 4: performance metrics for the external dataset, grouped by the number of explanatory variables in the model. the two best metrics within each group are shown in bold. grey font indicates the non-standardised amplitudes; shown for comparative purpose only. si = site index; v = acoustic velocity; a0 = mean amplitude; a1 = mean adjusted amplitude shifted by the mean rolling minimum amplitude within 100 mm of bark; a2 = mean adjusted amplitude shifted by the mean amplitude within 100 mm of bark; z0, z1, z2 are standardised equivalents of a0, a1, a2 respectively. model description explanatory variable value se df t-value p-value optimal intercept 240.8 24.1 86 10.0 0.0000 z2 18.5 2.2 86 8.3 0.0000 1/si 6346.0 613.8 3 10.3 0.0019 v/age 862.3 238.0 86 3.6 0.0005 best 2-var candidate intercept 294.3 35.0 87 8.4 0.0000 z0 14.8 2.5 87 5.8 0.0000 1/si 6587.9 1216.6 3 5.4 0.0124 table 5: parameters and statistics of two candidate models for predicting wood density in standing trees. z0, z2 = standardised mean resistograph amplitudes (unadjusted and adjusted respectively); si = site index; v = acoustic velocity. to be markedly worse than the more complex model. therefore, for operational purposes, the simpler model was selected as the model of choice. confidence and prediction intervals determined using the model of choice (with x = [1/si, z0] and parameters as in table 5) are demonstrated in figure 4 for each of the sites. both z0 and a0 are indicated on the figures, along with the measured densitometric values. the influence of site index is clear. in general, the model of choice estimated wood density with a good degree of accuracy (figure 5). nearly all trees in the in-sample dataset (93%) were within 50 kg.m-3 of their true values, and well over half (62%) were within 25 kg.m-3 of their true values. for external data, extrapolation of the model beyond 570 kg.m-3 resulted in wood density being under-estimated (figure 5). estimates beyond 600 kg.m-3 were amongst the worse, particularly for those trees on plots which had been fertilised and/or thinned. all external dataset trees on control plots which had neither been thinned nor fertilised were within 54 kg.m3 of their true values. for both in-sample and external datasets, 84% of all predictions were within 50 kg.m-3 of their true values, and 53% within 25 kg.m-3 of their true values. overall, 87% of all predictions were within 10% of their true values, and more than half (54%) within 5% of their true values. discussion the ability to non-destructively evaluate and predict wood quality is of great importance and has been reported by many authors for a variety of tools (cown 1978; wessels et al. 2011) and a variety of purposes such as assessing young trees for genetic heritability (gantz 2002; fundova et al. 2019), determining effects of silvicultural practices on product quality (wang 1999; briggs et al. 2008), and evaluating wood composites (winistorfer et al. 1995). many studies have done so in stands that are more homogenous in terms of age, geographic location, and genetic composition (desponts et al. 2017). because many tools and techniques are labour intensive, not field-based, or are too destructive, the micro-drill for non-destructively determining wood quality attributes has attracted attention. it meets the requirements of being portable, inexpensive to use, and has little impact on tested trees. so, can micro-drill resistance tools alone provide accurate assessments of wood density for a diverse set of trees? in our study, mean resistograph amplitude was highly significant as an explanatory variable for predicting wood density, however, the correlation between wood density and amplitude was weak (0.42, based on r2marg = 0.18). this correlation was a little lower than found in the literature for less diverse cohorts; e.g. r2 = 0.22, based on a correlation of 0.47 for 32-year-old douglas-fir trees from four comparable sites (el-kassaby et al. 2011), a range in r2 from 0.21 to 0.44 for 11 yearold loblolly pines (isik & li 2003), and r2 = 0.38, based on a correlation of 0.62, for a single stand of 25-yearold douglas-fir trees (chantre & rozenberg 1997). our performance statistics with in-sample data (table 3) indicate that estimates of wood density of a tree from within these stands (i.e. within the stands of the intodoroki et al. new zealand journal of forestry science (2021) 51:6 page 9 figure 3: residual plots of selected models for predicting wood density, with explanatory variables as indicated. z0, z2 = standardised mean resistograph amplitudes (unadjusted and adjusted respectively); si = site index; v = acoustic velocity. todoroki et al. new zealand journal of forestry science (2021) 51:6 page 10 sample dataset) will, on average, have an error of about 28 kg.m-3, or equivalently 6% (table 3). however, there will be a lot of scatter around this average, some trees will have greater error, others less. if trees are selected from other stands, i.e. new data, then larger errors could be expected. statistics for our external dataset (table 4; models shown in blank font) indicated an average error of about 61 kg.m-3, or equivalently 10%. this is quite large, and possibly too large for assessing wood density in individual trees. improvements in accuracy were obtained when site index was included in the predictive models. this is consistent with findings of walker et al. (2019). the best accuracy was achieved with a three-variable model, the optimal model, that in addition to adjusted amplitude, included site index and the quotient of acoustic velocity and age. r2marg was 0.60, equivalent to a correlation of 0.77. with this model we could expect accuracy to be within 20 kg.m-3 (mae was 19.4 kg.m-3), or equivalently 4.1%. rmse was 25.0 kg.m-3, indicating good proximity between estimated and true means. in comparison to the model in amplitude alone, statistics for our external dataset were greatly improved and approximately halved (mae 52.4 kg.m-3, mape 8.7%, rmse 59.5 kg.m-3). as noted in the literature, an optimal model for insample data may have a lower predictive ability for new/external data than a sub-optimal model (e.g. shmueli 2010). this was found to be true in our study, and a simpler model, recorded the best accuracy for the external dataset (mae 51.6 kg.m-3, mape 8.5%, rmse 59.3 kg.m-3). these statistics, while only marginally better than those for the optimal model, are based on a very simple model, x = [1/si, z0], requiring input of just site index and mean unadjusted standardised amplitude. we observe that differences in performance metrics within each a0, a1, a2 model triplet (table 3) are quite small; similarly, for the two-variable group in table 4. thus, with the simpler model we forgo the need to adjust the amplitude signal prior to determining the mean. parameters of the optimal and simpler models indicate that, as site index increases, wood density decreases (due to the inverse relationship, 1/si), and as amplitude increases wood density increases. wood density also increases as the ratio of velocity to age increases. therefore, within a given stand of trees of the same age, wood density increases as acoustic velocity increases. parameters of the simpler 2-variable model, x = [1/si, z0], indicate that for each unit increase in site index, a decrease in wood density of about 6 kg.m-3 could be expected, and for each 0.1 increase in the z-value associated with the standardised (unadjusted) microdrill amplitude, an increase in wood density of about 1.5 kg.m-3 could be expected. the simple procedure of standardisation provided large improvements in accuracy of estimates with our external dataset. in the scenario that another douglasfir dataset becomes available in the future, then all that is required is the computation of mean amplitude for each figure 4: confidence and prediction intervals of wood density by site index, augmented with data used in this study. mean amplitude of the inside-bark resistance profiles are shown with both unadjusted (a0) and standardised (z0) scales. todoroki et al. new zealand journal of forestry science (2021) 51:6 page 11 tree, and thereupon the mean and standard deviation of amplitudes to calculate the z-score. this data, together with site index, could then be evaluated as a further external dataset for the simpler model with x = [1/si, z0]. therefore, though we anticipate this model to be portable, more data from a greater range of site indices would be required, as are a greater range in tree ages to further test model portability. for all predictive models, evaluating performance with other datasets is a crucial step in gauging practical usefulness, yet despite its importance, the primary focus in predictive modelling studies has centred on how models were developed and on their explanatory accuracy. we suggest a greater focus be given to their predictive accuracy through external verification to assess the usefulness of the model in practice. with all tools, including micro-drills, many factors can cause measurement error. when using micro-drills, care needs to be taken to obtain accurate profiles. the angle of penetration of the bit determines the amplitude profile (rinn 2012) and due to the bit’s flexibility, can be influenced by operator movement, thereby altering the profile (ukrainetz & o’neill 2010). profiles in turn may be affected by moisture content (isik & li 2003; lin et al. 2003) which increases drilling resistance (kahl et al. 2009). however, in the case of small conditioned wood specimens, the effect of moisture content above fibre saturation (~32% mc) was not evident (sharapov et al. 2019b). air temperature (ukrainetz & o’neill 2010) and wood properties such as reaction wood, resin pockets, and branches/knots, can also affect the profiles (eckard et al. 2010). as forest management objectives shift, there is an increased need to better understand the resource for the products/services it can provide. the focus for industrial landowners has been primarily on volume yield for revenue generation. wood quality, which has a negative relationship with growth (jozsa & middleton 1994; kennedy 1995), is not often taken into consideration. while it can be challenging to predict future customer needs, land managers interested in marketing trees to mills manufacturing products that require a certain wood density level, the ability to plant, grow, manage, and harvest the trees at the optimal time will be of economic benefit. understanding the effect of land management regimes and integrating production of selected attributes along the value chain from raw materials to products promotes best use allocation of the forest resource in the future while providing information on how to grow and tend trees for specific end uses. conclusions we have provided evidence that wood density estimates of individual douglas-fir trees derived from microdrilling alone are insufficient for obtaining accurate figure 5: comparison of measured wood density with predictions. dark and light grey bands represent differences of up to 25 kg.m-3 and 50 kg.m-3 respectively. the vertical line indicates the maximum density within the insample dataset. wood density estimates from a diverse set of trees. site effects need to be considered too and with the simple inclusion of site index in models, wood density predictions improve considerably. another simple procedure, that of standardising mean amplitudes to z-scores, extends portability of models to future datasets that may use different micro-drills or may operate microdrills at different speeds. external data are critical to determining how well a model will perform in practice. competing interests the authors declare that they have no competing interests. acknowledgements we thank the field crew, laboratory technicians, and database managers who collected, processed, and organised the data, without which this study would not have been possible. and we extend our appreciation to the stand management cooperative, school of environmental and forest sciences, university of washington, seattle, wa, usa. author contributions ct and el conceived of the study and set overall objectives. ct participated in the design of the study, performed the statistical analyses, and drafted the manuscript. el and cf acquired and interpreted data and contributed to the writing of the manuscript. all authors have read and approved the final manuscript. funding this research was funded by an international joint venture agreement between the new zealand forest research institute limited, trading as scion, and the usda, forest service, pacific northwest research station. additional funding was provided by the forest innovation program, government of canada. references akaike, h. (1974). a new look at the statistical model identification. ieee transactions on automatic control, 19(6), 716-723. https://doi.org/10.1109/ tac.1974.1100705 alexander, d. l. j., tropsha, a., & winkler, d. a. (2015). beware of r2: simple, unambiguous assessment of the prediction accuracy of qsar and qspr models. journal of chemical information and modeling, 55, 1316-1322. https://doi.org/10.1021/acs. jcim.5b00206 bouffier, l., charlot, c., raffin, a., rozenberg, p., & kremer, a. (2008). can wood density be efficiently selected at early stage in maritime pine (pinus pinaster ait.)? annals of forest science, 65(1), 1-8. https:// doi.org/10.1051/forest:2007078 briggs, d.g., thienel, g., turnblom, e.c., lowell, e., dykstra, d., ross, r.j., wang, x., & carter, p. (2008). influence of thinning on acoustic velocity of douglas-fir trees in western washington and western oregon. paper presented at the 15th international symposium on nondestructive testing of wood, sept. 10-12, 2007, minnesota. chantre, g., & rozenberg, p. (1997). can drill resistance profiles (resistograph) lead to within-profile and within-ring density parameters in douglas fir wood. paper presented at the timber management toward wood quality and endproduct values. ctia/ iufro international wood quality workshop. chauhan, s.s., & walker, j.c.f. (2006). variations in acoustic velocity and density with age, and their interrelationships in radiata pine. forest ecology and management, 229(1-3), 388-394. https://doi. org/10.1016/j.foreco.2006.04.019 cown, d. (1978). comparison of the pilodyn and torsiometer methods for the rapid assessment of wood density in living trees. new zealand journal of forestry science, 8(3), 384-391. dankers, f.j.w.m., traverso, a., wee, l., & van kuijk, s.m.j. (2019). prediction modeling methodology. in p. kubben, m. dumontier, a. dekker (eds.), fundamentals of clinical data science (pp. 101120). springer international publishing. https:// doi.org/10.1007/978-3-319-99713-1_8 desponts, m., perron, m., & deblois, j. (2017). rapid assessment of wood traits for large-scale breeding selection in picea mariana [mill.] b.s.p. annals of forest science, 74(3): 53. https://doi.org/10.1007/ s13595-017-0646-x draper, n., & smith, h. (1966). applied linear regression: new york: wiley. eberhardt, t.l., & samuelson, l.j. (2015). collection of wood quality data by x-ray densitometry: a case study with three southern pines. wood science and technology, 49(4), 739-753. https://doi. org/10.1007/s00226-015-0732-x eckard, j.t., isik, f., bullock, b., li, b., & gumpertz, m. (2010). selection efficiency for solid wood traits in pinus taeda using time-of-flight acoustic and micro-drill resistance methods. forest science, 56(3), 233-241. el-kassaby, y.a., mansfield, s., isik, f., & stoehr, m. (2011). in situ wood quality assessment in douglas-fir. tree genetics and genomes, 7(3), 553-561. https:// doi.org/10.1007/s11295-010-0355-1 filipescu, c.n., lowell, e.c., koppenaal, r., & mitchell, a.k. (2014). modeling regional and climatic variation of wood density and ring width in intensively managed douglas-fir. canadian journal of forest research, 44(3), 220-229. https://doi. org/10.1139/cjfr-2013-0275 todoroki et al. new zealand journal of forestry science (2021) 51:6 page 12 https://doi.org/10.1109/tac.1974.1100705 https://doi.org/10.1109/tac.1974.1100705 https://doi.org/10.1051/forest:2007078 https://doi.org/10.1051/forest:2007078 https://doi.org/10.1016/j.foreco.2006.04.019 https://doi.org/10.1016/j.foreco.2006.04.019 https://doi.org/10.1007/s13595-017-0646-x https://doi.org/10.1007/s13595-017-0646-x https://doi.org/10.1007/s00226-015-0732-x https://doi.org/10.1007/s00226-015-0732-x https://doi.org/10.1007/s11295-010-0355-1 https://doi.org/10.1007/s11295-010-0355-1 https://doi.org/10.1139/cjfr-2013-0275 https://doi.org/10.1139/cjfr-2013-0275 fundova, i., funda, t., & wu, h.x. (2018). non-destructive wood density assessment of scots pine (pinus sylvestris l.) using resistograph and pilodyn. plos one, 13(9). https://doi.org/10.1371/journal. pone.0204518 fundova, i., funda, t., & wu, x.h. (2019). non-destructive assessment of wood stiffness in scots pine (pinus sylvestris l.) and its use in forest tree improvement. forests, 10(6). https://doi.org/10.3390/ f10060491 gantz, c.h. (2002). evaluating the efficiency of the resistograph to estimate genetic parameters for wood density in two softwood and two hardwood species. (msc), north carolina state university, raleigh, nc. gao, s., wang, x., brashaw, b.k., ross, r.j., & wang, l. (2012). rapid assessment of wood density of standing tree with nondestructive methods a review. paper presented at the international conference on biobase material science and engineering, bmse 2012. https://doi. org/10.1109/bmse.2012.6466226 gao, s., wang, x., wiemann, m.c., brashaw, b.k., ross, r.j., & wang, l. (2017). a critical analysis of methods for rapid and nondestructive determination of wood density in standing trees. annals of forest science, 74(2): 27. https://doi.org/10.1007/s13595-0170623-4 gwaze, d., & stevenson, a. (2008). genetic variation of wood density and its relationship with drill resistance in shortleaf pine. southern journal of applied forestry, 32(3), 130-133. https://doi. org/10.1093/sjaf/32.3.130 howe, g.t., jayawickrama, k., cherry, m., johnson, g., & wheeler, n.c. (2006). breeding douglas-fir plant breeding reviews (vol. 27, pp. 245-353). https:// doi.org/10.1002/9780470650349.ch6 isik, f., & li, b. (2003). rapid assessment of wood density of live trees using the resistograph for selection in tree improvement programs. canadian journal of forest research, 33(12), 2426-2435. https://doi. org/10.1139/x03-176 johnson, p.c.d. (2014). extension of nakagawa & schielzeth’s r2 glmm to random slopes models. methods in ecology and evolution, 5(9), 944-946. https://doi.org/10.1111/2041-210x.12225 jozsa, l., & middleton, g. (1994). a discussion of wood quality attributes and their practical implications. vancouver, bc: forintek canada corporation. accessed 4 october 2019 from: https://pdfs. semanticscholar.org/57b5/7386b0f5029e60812695c7 0ac34eb624069b.pdf kahl, t., wirth, c., mund, m., böhnisch, g., & schulze, e.d. (2009). using drill resistance to quantify the density in coarse woody debris of norway spruce. european journal of forest research, 128(5), 467473. https://doi.org/10.1007/s10342-009-0294-2 kennedy, r.w. (1995). coniferous wood quality in the future: concerns and strategies. wood science and technology, 29(5), 321-338. https://doi. org/10.1007/bf00202581 kimberley, m.o., mckinley, r.b., cown, d.j., & moore, j.r. (2017). modelling the variation in wood density of new zealand-grown douglas-fir. new zealand journal of forestry science, 47: 15. https://doi. org/10.1186/s40490-017-0096-0 lin, c.j., wang, s.y., lin, f.c., & chiu, c.m. (2003). effect of moisture content on the drill resistance value in taiwania plantation wood. wood and fiber science, 35(2), 234-238. maclaren, j.p. (2009). douglas-fir manual. [fri bulletin no. 237], 32 p. rotorua, new zealand: scion. marshall, d.d., bell, j.f., & tappeiner, j.c. (1992). levelsof-growing-stock cooperative study in douglasfir: report no. 10 – the hoskins study, 1963-83. research paper pnw-rp-448. portland, or, usa: usda forest service, pacific nortwest research station. https://doi.org/10.2737/pnw-rp-448 mattheck, c., bethge, k., & albrecht, w. (1997). how to read the results of resistograph m. arboricultural journal, 21(4), 331-346. https://doi.org/10.1080/ 03071375.1997.9747179 mendenhall, w., & sincich, t. (2012). principles of model building. in d. lynch (ed.), a second course in statistics: regression analysis (pp. 261-325). upper saddle river, nj: prentice hall. nakagawa, s., & schielzeth, h. (2013). a general and simple method for obtaining r2 from generalized linear mixed-effects models. methods in ecology and evolution, 4(2), 133–142. https://doi. org/10.1111/j.2041-210x.2012.00261.x newton, p.f. (2017). acoustic-based non-destructive estimation of wood quality attributes within standing red pine trees. forests, 8(10): 380. https://doi.org/10.3390/f8100380 paradis, n., auty, d., carter, p., & achim, a. (2013). using a standing-tree acoustic tool to identify forest stands for the production of mechanically-graded lumber. sensors (switzerland), 13(3), 3394-3408. https:// doi.org/10.3390/s130303394 pinheiro, j., & bates, d. (2006). mixed-effects models in s and s-plus: springer science & business media. pinheiro, j., bates, d., debroy, s., sarkar, d., & team, r.c. (2019). nlme: linear and nonlinear mixed effects models. r package version 3.1-141. https:// cran.r-project.org/package=nlme todoroki et al. new zealand journal of forestry science (2021) 51:6 page 13 https://doi.org/10.1371/journal.pone.0204518 https://doi.org/10.1371/journal.pone.0204518 https://doi.org/10.3390/f10060491 https://doi.org/10.3390/f10060491 https://doi.org/10.1109/bmse.2012.6466226 https://doi.org/10.1109/bmse.2012.6466226 https://doi.org/10.1007/s13595-017-0623-4 https://doi.org/10.1007/s13595-017-0623-4 https://doi.org/10.1093/sjaf/32.3.130 https://doi.org/10.1093/sjaf/32.3.130 https://doi.org/10.1002/9780470650349.ch6 https://doi.org/10.1002/9780470650349.ch6 https://doi.org/10.1139/x03-176 https://doi.org/10.1139/x03-176 https://doi.org/10.1111/2041-210x.12225 https://doi.org/10.1007/s10342-009-0294-2 https://doi.org/10.1007/bf00202581 https://doi.org/10.1007/bf00202581 https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf https://pdfs.semanticscholar.org/57b5/7386b0f5029e60812695c70ac34eb624069b.pdf r core team. (2018). r: a language and environment for statistical computing. vienna, austria. https:// www.r-project.org/ rinn, f. (1988). a new method for measuring treering density parameters. physics diploma thesis. institute for environmental physics, heidelberg university, 85 p. rinn, f. (2012). basics of typical resistance-drilling profiles. western arborist winter, 30-36. rinn, f. (2015). key to evaluating drilling resistance profiles. western arborist fall, 16-21. schimleck, l., dahlen, j., apiolaza, l. a., downes, g., emms, g., evans, r., moore, j., van den bulcke, j., & wang, x. (2019). non-destructive evaluation techniques and what they tell us about wood property variation. forests, 10(9): 728. https://doi. org/10.3390/f10090728 searle, s., casella, g., & mcculloch, c. (1992). variance components. new york: john wiley & sons, inc. https://doi.org/10.1002/9780470316856 sharapov, e., brischke, c., militz, h., & toropov, a. (2019a). impact of drill bit feed rate and rotational frequency on the evaluation of wood properties by drilling resistance measurements. international wood products journal, 10(4), 128-138. https:// doi.org/10.1080/20426445.2019.1688455 sharapov, e., brischke, c., militz, h., & smirnova, e. (2019b). prediction of modulus of elasticity in static bending and density of wood at different moisture contents and feed rates by drilling resistance measurements. european journal of wood and wood products, 77(5), 833-842. shmueli, g. (2010). to explain or to predict? statistical science, 25(3), 289-310. https://doi. org/10.1214/10-sts330 snee, r. (1977). validation of regression models: methods and examples. technometrics. 19, 415428. https://doi.org/10.1080/00401706.1977.10 489581 spiecker, h., lindner, m., & schuler, j. (2019). douglas-fir – an option for europe. joensuu, finland: european forest institute. accessed 10 october 2019 from: https://www.efi.int/sites/default/files/files/ publication-bank/2019/efi_wsctu9_2019.pdf tuszynski, j. (2019). catools: tools: moving window statistics, gif, base64, roc auc, etc.. r package version 1.17.1.2. https://cran.r-project.org/ package=catools ukrainetz, n.k., & o’neill, g.a. (2010). an analysis of sensitivities contributing measurement error to resistograph values. canadian journal of forest research, 40(4), 806-811. https://doi. org/10.1139/x10-019 walker, n.k., & dodd, r.s. (1988). calculation of wood density variation from x-ray densitometer data. wood and fiber science, 20(1), 35-43. walker, t.d., isik, f., & mckeand, s.e. (2019). genetic variation in acoustic time of flight and drill resistance of juvenile wood in a large loblolly pine breeding population. forest science, 65(4), 469482. https://doi.org/10.1093/forsci/fxz002 wang, x. (1999). stress wave-based nondestructive evaluation (nde) methods for wood quality of standing trees. (phd), michigan technological university, houghton, mi, usa. wang, x., ross, r.j., mcclellan, m., barbour, r.j., erickson, j.r., forsman, j.w., & mcginnis, g.d. (2001). nondestructive evaluation of standing trees with a stress wave method. wood and fiber science, 33(4), 522-533. wessels, c.b., malan, f.s., & rypstra, t. (2011). a review of measurement methods used on standing trees for the prediction of some mechanical properties of timber. european journal of forest research, 130(6), 881-893. https://doi.org/10.1007/ s10342-011-0484-6 wickham, h. (2011). the split-apply-combine strategy for data analysis. journal of statistical software, 40(1), 1-29. https://doi.org/10.18637/jss.v040. i01 williamson, r.l., & staebler, g.r. (1971). levels-ofgrowing-stock cooperative study on douglas-fir. rep. no. 1—description of study and existing study areas. research paper pnw-111. portland, or, usa: usda forest service, pacific nortwest forest and range experiment station. winistorfer, p.m., xu, w., & wimmer, r. (1995). application of a drill resistance technique for density profile measurement in wood composite panels. forest products journal, 45(6), 90. zeidler, a., borůvka, v., & schönfelder, o. (2017). comparison of wood quality of douglas fir and spruce from afforested agricultural land and permanent forest land in the czech republic. forests, 9(1), 13. https://doi.org/10.3390/ f9010013 zobel, b.j., & van buijtenen, j.p. (2012). wood variation: its causes and control: springer science & business media. todoroki et al. new zealand journal of forestry science (2021) 51:6 page 14 effects of site and tree size on wood density and bark properties of lebombo ironwood (androstachys johnsonii prain) tarquinio mateus magalhães departamento de engenharia florestal, universidade eduardo mondlane, campus universitário principal, edifício no. 1, 257 maputo, moçambique *corresponding author: tarquinio_magalhaes@uem.mz (received for publication 16 december 2018; accepted in revised form 15 march 2021) abstract background: wood and bark are important renewable natural resources. density is an important property that is used to describe wood and bark quality for a number of end uses. however, wood and bark density, bark proportion and dimensions vary with age and site, as well as among and within trees. the aim of this study was to investigate the effect of site, diameter class, and vertical position within the stem on the density of wood and bark, bark volume, bark dry-mass and thickness of lebombo ironwood (androstachys johnsonii prain). methods: the study was conducted on 93 lemombo ironwood trees growing in mozambique. eight discs were sampled from each selected tree and diameter over and under bark was measured. bark thickness, bark mass and bark density were determined along with the basic wood density of each disc. results: the overall average whole-stem properties were estimated at: 786 kg m–3 wood density, 586 kg m–3 bark density, 19% bark volume, 19% bark dry-mass, and 9 mm bark thickness. height level uniquely explained most of the variation in bark mass (97%), bark volume (95%) and wood density (86%). diameter class explained most of the variation in bark density (51%) and bark thickness (51%). site only explained a small proportion of the variation in all dependent variables. conclusions: overall, the patterns of variation of all wood and bark properties were highly dependent on tree diameter class and vertical position within the stem. site differences were not a significant source of variation in the properties studied. improved knowledge of the wood and bark properties of this species will aid its sustainable management and utilisation. new zealand journal of forestry science magalhães new zealand journal of forestry science (2021) 51:3 https://doi.org/10.33494/nzjfs512021x32x e-issn: 1179-5395 published on-line: 14/04/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access it is classified as a first-class timber and it is lawfully harvested at 30 cm dbh. this tree species rarely exceeds 35 cm in diameter at breast height (dbh) (magalhães and seifert 2015). lebombo ironwood has distinct growth ring boundaries and is diffuse-porous with vessels of two distinct diameter classes (bunster 2006; cardoso 1963). the growth rings are wavy causing a fine streaked appearance (bunster 2006). the fibres are very thickwalled with distinctly bordered pits common in both radial and tangential walls (ali 2008). the heartwood is yellowish-pale brown, sometimes slightly pink (bunster 2006). introduction mecrusse is a forest type in which the dominant canopy species is androstachys johnsonii prain, the relative cover of which varies from 80% to 100% (magalhães 2017a). androstachys johnsonii is known vernacularly as lebombo ironwood or cimbirre. the wood of a. johnsonii is also known as lebombo ironwood. lebombo ironwood is a tree species native to madagascar and africa, however, presently it is almost restricted to mozambique (cardoso 1963), where it is mainly found in the southmost part of the country, in inhambane and gaza province (magalhães 2015). in mozambique, keywords: bark traits; commonality analysis; regression effect; suppressor effect; wood traits http://creativecommons.org/licenses/by/4.0/), magalhães new zealand journal of forestry science (2021) 51:3 page 2 because of its durability (bunster 2006), lebombo ironwood is used as poles and stakes by the local community (magalhães 2017a) and in construction of houses and bridges (cardoso 1963). due to its favourable mechanical properties (cardoso 1963), lebombo ironwood is also suitable for flooring, turnery, marine uses, furniture and interiors (bunster 2006), stairs, laths, fences, railway sleepers, bridge piers, vehicle boards and draining boards (ali et al. 2008; cardoso 1963). these end uses are determined by the properties of this wood, hence the relevance of studying them. in the early 1960s, lebombo ironwood had already been reported to be almost completely restricted to mozambique (cardoso 1963), presumably due to overexploitation. five decades later, there is still a lack of studies on this species in any branch of forest science, especially studies related to wood and bark properties. this emphasises the need to study the wood properties of this species. basic wood density (hereafter referred simply as wood density) is an important wood quality characteristic. wood density is used to describe wood quality in construction and in the mechanical and chemical pulp industries (repola 2006). various solid wood utilization characteristics such us timber strength and stiffness, machinability and drying, and some manufacture and performance reconstituted products such as raw material consumption, uptake of chemicals, pulp yield, and paper properties are dependent on wood density (kimberley et al. 2015; tian et al. 1995). along with volume, wood density is a key determinant of tree biomass (bastin et al. 2015) and thus a key variable to make estimates of carbon pools in forests, and for studying other biochemical cycles and understanding the evolution and potential future changes of the climate system. lebombo ironwood is heavy, with a density of 754 and 880 kg/m³ at 0 and 12% moisture content, respectively (bunster 2006; magalhães 2015). its high density makes it attractive to the hardwood lumber industry. bark is also an economically important raw material. it is used in the production of cork, fibre, tannins, pallets, briquettes, insulation boards, fibreboards, hardboards, particleboards (pásztory et al. 2016). bark is also used in mulching and soil amendment (pásztory et al. 2016). specifically, due to its allelopathic effect (magalhães 2017a; molotja et al. 2011), the bark of lebombo ironwood is used to sterilise the soils in rural home gardens where only adult trees are desired (i.e. to avoid weed and seed germination). this highlights the importance of quantifying the properties and quantity of bark. except for chemical production, most of the uses of bark (e.g., production of boards) are influenced by its density. knowing the thickness of bark is critical to accurately estimate the relative volumes of solid wood and bark that are available (thomas and bennett 2014). stem bark volume and bark dry-mass are important because bark and wood mass are separated while processing logs, and accurate determination of volume is problematic (thomas and bennett 2014). wood and bark density, bark volume and mass, and bark thickness vary with age, diameter, site, tree and within tree (longitudinally and radially) (auty et al. 2014; cellini et al. 2012; kimberley et al. 2015; machado et al. 2014; murphy & cown 2015; paine et al. 2010; quilhó & pereira 2001; repola 2006; searle & owen 2005; sonmez et al. 2007). thus, the variations of these wood and bark properties are of particular interest to the forestry sector and the related industries (e.g. construction, fuel industries, etc.). although considerable studies have been carried out on the extent and sources of variation in wood and bark properties of coniferous species in europe, north america, south america and australasia, similar studies are lacking for tropical african broad-leaved species. the current study aimed to quantify the effects of site, diameter class, relative height within the stem and their interactions on wood and bark density, bark volume, mass, and thickness of lebombo ironwood. as these tree variables are often measured at a single specific sampling point, generally breast height (hernández & genes 2016; tian et al. 1995), whole-stem-based properties (e.g. whole-stem wood density) were compared to the equivalent breast height-based properties. methods study sites this study was carried out in mecrusse woodlands of the mandlakaze (24o 04´ – 25o 02´ s and 33o 47´ – 34o 39´ e), funhalouro (22o 09´ – 23o 42´ s and 33o 40´ – 34o 29´ e) and mabote (21o 18´ – 22o 54´ s and 33o 10´ – 34o 39´ e) districts of mozambique (10o 30´ – 26o 52´ s and 30o 15´ – 40o 45´ e), in southern africa. the physical and natural conditions of the study areas are shown in table 1. data collection ninety-three trees with dbh varying from 5 to 32 cm (average = 17.6 cm) and total height varying from 5 to 16 m (average = 12.3 m) were randomly selected and destructively measured. the trees were distributed across each site and dbh class as shown in table 2. felled trees were scaled up to a 2.5 cm top diameter. the stem was defined as the length of the trunk from the predefined stump height (20 cm) to the height that corresponded to a stem diameter of 2.5 cm. the trees had stem lengths varying from 4.25 to 14.85 m (average = 10.78 m). discs of approximately 5 cm thickness were taken from each tree at eight heights along the stem: breast height, 0, 10, 30, 50, 70, 90 and 100% of stem height. diameters over and under bark (dob and dub, respectively) were measured on each disc in the north– south direction (previously marked on the standing tree) using a ruler. the volume of the discs, before and after debarking, was determined using the water displacement method (brasil et al. 1994). therefore, it was possible to obtain disc volumes over and under bark. bark volume was determined as the difference between disc volume over and under bark. wood discs and associated bark were oven dried at 105°c to constant mass. wood and bark density were obtained by dividing the oven dry-mass of the disc and the bark by the relevant volume (magalhães & seifert 2015). bark volume proportion (hereafter referred simply as bark volume) was calculated as the percentage of bark volume relative to total disc volume. similarly, bark mass proportion (hereafter referred simply as bark mass) was calculated as the percentage of bark oven dry-mass relative to the total disc oven dry-mass. double bark thickness was determined as the difference between dob and dub. in total, 744 measurements of wood and bark density, bark volume and mass and double bark thickness were taken along the stems of the 93 sample trees. whole-stem wood and bark properties under study were calculated using data on the respective properties obtained at eight heights up the stem (i.e. breast height, 0, 10, 30, 50, 70, 90 and 100% of stem height). table 3 shows the correlation matrix between the wood and bark properties. data analysis three-way nested multivariate analysis of variance (nested manova) was performed to test the significance of site, diameter class, and height level on the wood and bark properties under study. commonality analysis was carried out to quantify how much of the variation in wood and bark properties was explained by the variance of all the predictors (site, dbh class, height level) and quantify the variance that was unique to each predictor and the variance that was common to groups of predictors. one-way nested analyses of variance (nested anova), with tukey’s hsd test, were carried out to verify the difference between whole-stem wood and bark properties and the relevant properties at different height levels within the stem. all statistical analyses and tests were carried out using r software (r core team 2019). one-way anova and three-way manova were run using the functions “aov” and “manova”, respectively. interactions plots were built using “ggplot2” package (wickham et al. 2018). commonality analyses were performed using “yhat” package (nimon et al. 2015). all tests were performed at significance level of α = 0.05. results basic wood density the average (± se) whole-stem wood density for all sites (whole population) was estimated at 786 (± 2.2) kg m–3, ranging among sites from 783 (± 3.0) to 790 (± 4.5) kg m–3. whole-disc wood density ranged from 729 (± 6.7) to 872 (± 4.2) kg m–3 within a tree (longitudinally) and from 760 (± 6.6) to 807 (± 4.2) kg m –3 among diameter classes. whole-disc wood density decreased considerably with height. it increased slightly with diameter class from mid dbh of 7.5 to 12.5 cm and then it remained constant to the mid dbh of 22.5 cm from where it increased slightly again (figure 1). however, it did not show a significant variation with site. for mandlakaze and mabote districts, the whole-stem wood density was only found to be different to the wood density at the base and top of the stem (relative heights of 0.0 and 1.0), but no significant differences were found between wholestem density and the densities of other height levels. for funhalouro district and for the whole population, significant differences were only observed with the densities at relative heights of 0.0, 0.9 and 1.0. diameter class and height level were found to be significant sources of variation of wood density (p< 0.0001; table 4). site was not a significant source of variation of wood density (p = 0.15). the interactions magalhães new zealand journal of forestry science (2021) 51:3 page 3 where n is number of lebombo ironwood trees per hectare (dbh ≥ 5 cm), g respective basal area, ahm annual heat moisture index, and ai aridity index. source: mae (2005a, 2005b, 2014), magalhães (2017b). table 1: description of the study sites district description climate and hydrology relief, topography and elevation n (ha–1) g (m2 ha–1) mandlakaze climate: dry and humid tropical; ahm: 34; ai: 0.7; hydrology: 63 lakes and 2 rivers. relief and topography: flat. elevation: 50 to 200 m a.s.l. 732 17 funhalouro climate: dry semi-arid; ahm: 52; ai: 0.4; hydrology: not crossed by any river, occurrence of meteorological droughts. relief and topography: flat. elevation: 100 to 200 m a.s.l. 1617 26 mabote climate: dry tropical; ahm: 57; ai: 0.4; hydrology: not crossed by any river, occurrence of meteorological droughts. relief and topography: flat. elevation: 100 to 200 m a.s.l. 1333 21 mid dbh (cm) mandlakaze funhalouro mabote total 7.5 3 6 9 18 12.5 3 7 8 18 17.5 4 6 8 18 22.5 4 5 8 17 27.5 + 6 7 9 22 total 20 31 42 93 table 2: number of harvested trees per site and diameter class figure 1: pattern of variation of wood and bark density with dbh class (a and b) and height level (c and d). magalhães new zealand journal of forestry science (2021) 51:3 page 4 wood density bark density bark volume bark content bark thickness wood density 1.0000 0.2577 – 0.4069 – 0.4846 0.3236 bark density 1.0000 – 0.3064 – 0.2339 0.4345 bark volume 1.0000 0.8136 – 0.2470 bark content 1.0000 – 0.4441 bark thickness 1.0000 table 3: person´s correlation matrix for wood and bark properties between site and dbh class, site and height level, dbh class and height level had statistically significant effects on wood density (p < 0.0001). however, the interaction between site, dbh class and height level had no significant effect (p = 0.95; table 4). from the commonality analysis (table 5) it was found that the unique effects and the common effects together explained 47.13% (sum of commonality coefficients) of the variation of wood density: i.e., the regression effect explained 47.13% of the variation of wood density. most of that regression effect (85.78%) was explained by variance that was unique to height level. dbh class and site accounted for 13.68 and 0.44% of the regression effect, respectively. together, the three independent variables uniquely accounted for 99.90% of the regression effect. the remaining 0.1% was due to the common effect of site and dbh class. basic bark density site, diameter class, and height level were found to be significant sources of variation in bark density (p< 0.0001; table 4). only the interactions between site and dbh class had a significant effect on bark density (p< 0.0001). higher values of bark density were observed in funhalouro district (with an average (± se) whole-stem density of 597 (± 4.6) kg m –3) and the lower ones in mandlakaze district (with a whole-stem density of 574 (± 9.1) kg m –3). mabote district had intermediate bark density values. whole-stem bark density increased significantly with increasing diameter class, from an average (± se) value of 522 (± 8.2) kg m–3 for the smallest diameter class to 634 (± 8.5) kg m–3 for the largest diameter class (figure 1). bark density decreased considerably with relative height up the stem, with an overall decrease of approximately 20% from the bottom to the top of the stem. magalhães new zealand journal of forestry science (2021) 51:3 page 5 factors and interactions df wood density bark density bark volume bark mass bark thickness p-value ω2 (%) p-value ω2 (%) p-value ω2 (%) p-value ω2 (%) p-value ω2 (%) site 2 0.1513 0.1 0.0085 0.7 0.0001 0.6 0.4742 0.0 0.0000 4.4 dbh class 4 0.0000 6.2 0.0000 14.2 0.0000 2.4 0.0000 2.4 0.0000 27.2 height level 7 0.0000 39.9 0.0000 12.4 0.0000 65.0 0.0000 70.0 0.0000 21.2 site × dbh class 8 0.0000 1.8 0.0014 1.7 0.2129 0.1 0.0000 1.2 0.0150 0.8 site × height level 14 0.0010 1.5 0.9429 0.0 0.0473 0.4 0.0044 0.4 0.2194 0.4 dbh class × height level 28 0.0005 2.2 0.8773 0.0 0.0000 2.3 0.0000 7.2 0.0015 2.1 site × dbh class × height level 56 0.9505 0.0 0.3753 0.3 0.9948 0.0 0.0014 0.9 0.8667 0.0 residuals 624 where df are degrees of freedom table 4: multivariate analysis of variance and omega-squared (ω2) for wood and bark properties factors and interactions wood density bark density commonality coefficient % of total variance commonality coefficient % of total variance site 0.0021 0.44 0.0157 5.50 dbh class 0.0645 13.68 0.1463 51.11 height level 0.4043 85.78 0.1306 45.63 site × dbh class 0.0005 0.10 – 0.0064 – 2.24 site × height level 0.0000 0.00 0.0000 0.00 dbh class × height level 0.0000 0.00 0.0000 0.00 site × dbh class × height level 0.0000 0.00 0.0000 0.00 total 0.4713 100.00 0.2862 100.00 table 5: commonality analyses for wood and bark density the regression effect explained 28.62% of the variation of bark density (table 5). of that regression effect, 51.11% was explained by the variance that was unique to dbh class and 45.63% was explained by the variance that was unique to height level (table 5). the variance unique to the site explained only 5.50% of the total regression effect. the interaction of site and height level had a negative effect on bark density, meaning that the predictor variables site and height level affected each other in the opposite direction. this suppression accounted for 2.24% of the regression effect. overall, the first-order effects (variance unique to the independent variables), second-order effects (variance common to pair of predictors), and third-order effects (variance common to all three predictors together) accounted for 102.24, –2.24, and 0.0% of the regression effect, respectively, totalling 100%. it was observed that in mandlakaze and mabote districts, bark density at different height levels was not significantly different from the whole-stem bark density. for funhalouro district and for the whole population, the whole-stem bark density was only significantly different from the bark densities at relative heights of 0.0 and 1.0. bark volume the average (± se) whole-stem bark volume for all sites was 19 (± 0.4) %, ranging from 18 (± 0.6) to 20 (± 0.7) % among sites. bark volume proportion ranged from 11 (± 0.4) to 33 (± 0.8) % within-tree, and from 18 (± 0.9) to 21 (± 0.6) % between dbh classes. overall, funhalouro district had the lowest values, followed in an increasing order by mabote and mandlakaze. in general, bark volume decreased slightly with dbh class and increased sharply with height level. the patterns of variation in bark volume at relative heights of 0.9 and 1.0 with site and dbh class were opposite to those observed at other height levels (figure 2). site, diameter class, and height level were significant sources of variation of bark volume (table 4). although statistically significant, site and dbh class contributed very little to the total variation in bark volume (tables 4 & 6). site and dbh class accounted for only 1.25 and 3.75% of the total regression effect (68.64%). thus, most of the regression effect was explained by the variance that was unique to height level (95.22%). the only significant common effect was between site and dbh class which, however, was a suppressor effect. although the secondand third-order interactions had a statistically significant effect on bark volume (table 4), none of them contributed to the total regression effect (table 6). whole-stem bark volume was statistically different from the bark volume at different height levels, except at relative height of 0.3, 0.5 and 0.7 for mandlakaze and mabote districts and only at relative height of 0.5 for funhalouro district and the whole population. bark mass the average (± se) whole-stem bark mass for all sites was estimated at 19 (± 0.3) %, and it was constant all over the sites. average bark mass ranged from 10 (± 0.3) to 32 (± 0.6) % within a tree (longitudinally), with the smallest values at the bottom of the stem. it ranged from 17 (± 0.8) to 21 (± 0.5) % among dbh classes. overall, bark mass decreased slightly with dbh class and increased sharply with increasing height level (figure 2). the pattern of variation of bark mass was independent of site (p = 0.47). however, it was dependent of other factors (dbh class and height level) and all interactions (p < 0.0001; table 4). the regression effect explained 72.72% of the variation in bark mass, of which 96.56% was explained uniquely by the variance of the height level. the variance of dbh class ranked next, accounting for 3.39% of the regression effect. site effects ranked last, accounting only for 0.05%. none of the common effects accounted for the regression effect. for funhalouro and mabote districts and for the whole population, whole-stem bark mass was significantly different from the bark mass at all height levels, except at a relative height of 0.5. for mandlakaze district, wholestem bark mass was only significantly different from bark mass at relative heights of 0.0 and 1.0. double bark thickness the average whole-stem double bark thickness and the double bark thickness at most height levels increased sharply from funhalouro to mabote district. however, from mabote to mandlakaze district it either showed a slight increase or decrease or remained constant. double bark thickness was found to increase sharply and rapidly with increasing dbh class (figure 3). average double bark thickness, for all sites and most dbh classes, increased slightly with increasing height level to the breast height or to the relative height of 0.3, and then it decreased non-linearly and rapidly to the top of the stem, resembling a quadratic parabola function (figure 3). the average (± se) whole-stem double bark thickness for all sites was estimated at 9 (± 0.2) mm, ranging between sites from 8 (± 0.2) to 10 (± 0.4) mm. average double bark thickness also ranged from 4 (± 0.2) to 11 (± 0.5) mm within an individual tree (higher values at the base of the stem) and from 5 (± 0.2) to 12 (± 0.4) mm among dbh classes. site, diameter class, and height level were significant sources of variation of double bark thickness (p < 0.0001; table 4). the interactions of site and dbh class, and dbh class and height level had a statistically significant effect on double bark thickness (p < 0.02). the regression effect explained 53.45% of the variation of double bark thickness (table 6). of that regression effect, 51.24% was explained by the variance that was unique to dbh class and 40.27% was explained by the variance that was unique to height level (table 6). the variance unique to the site explained 6.22% of the total regression effect. the only significant common effect was between site and dbh class, accounting for 2.27% of the regression effect. double bark thickness at most of the height levels was not significantly different from the whole-stem bark thickness. magalhães new zealand journal of forestry science (2021) 51:3 page 6 magalhães new zealand journal of forestry science (2021) 51:3 page 7 figure 2: pattern of variation of bark volume and bark mass with dbh class (a and b) and height level (c and d). factors and interactions bark volume bark mass bark thickness commonality coefficient % of total variance commonality coefficient % of total variance commonality coefficient % of total variance site 0.0086 1.25 0.0003 0.05 0.0332 6.22 dbh class 0.0257 3.75 0.0247 3.39 0.2739 51.24 height level 0.6536 95.22 0.7022 96.56 0.2152 40.27 site × dbh class – 0.0015 – 0.22 0.0000 0.00 0.0121 2.27 site × height level 0.0000 0.00 0.0000 0.00 0.0000 0.00 dbh class × height level 0.0000 0.00 0.0000 0.00 0.0000 0.00 site × dbh class × height level 0.0000 0.00 0.0000 0.00 0.0000 0.00 total 0.6864 100.00 0.7272 100.00 0.5345 100.00 table 6: commonality analyses for bark volume, bark mass and bark thickness discussion the overall average wood density at breast height found in this study (786 kg m–3) is higher than that reported by bunster (2006) and magalhães (2015) (754 kg m–3). the reason for this difference is because, as opposed to this study, these authors reported wood density over bark. however, calculations based on the wood and bark density and bark volume and mass revealed that the density over bark for this study (749 kg m–3) is close to the value reported by bunster (2006) and magalhães (2015). the overall density of bark at breast-height of lebombo ironwood (606 kg m–3) was higher than that of the wood of some miombo woodland species in mozambique (bunster 2006): pterocarpus angolensis (558 kg m–3), khaya nyasica (599 kg m–3) balanites maughamii (584 kg m–3), brachystegia spiciformis (588 kg m–3). overall, wood and bark density decreased with increasing height level of the stem and increased with increasing dbh class. conversely, bark volume and bark mass increased with increasing height level of the stem and decreased with increasing dbh class. this is supported by the person´s correlation matrix (table 3), which indicate a positive correlation between wood density and bark density, and a negative one between density (either of the wood or of the bark) with bark proportion (either on volume or on dry-mass basis). these results were consistent with the findings of nyg and elfving (2000) and deng et al. (2014). contrary to the findings of this study, deng et al. (2014) found statistical differences between whole-stem wood density and the density at different height levels up the stem. as in this study, no significant site effects on wood density were found by nyg and elfving (2000) for 57 african tree species. knapic et al. (2008) and miranda et al. (2001) also found similar results. wood density has been reported to be positively correlated to droughtprone areas or areas with drier climatic conditions (ibanez et al. 2017; nabais et al. 2018).although funhalouro and mabote districts are approximately twice as dry as mandlakaze district, as judged by annual heat moisture index (ahm) and aridity index (ai) (table 1), wood density did not differ among districts (sites) (table 4). this indicates that climatic conditions were not a source of variation of wood density of lebombo ironwood or that wood density of this species does not adapt in response to climatic variations. a decrease in wood density with increasing stem height has been reported by various authors (e.g. machado et al. 2014, kimberley et al. 2015). machado et al. (2014) suggested that the highest values of wood density at stem base might be the result of the root system and the lowest values at the top of stem are the result of youngest layers of the wood. it was presumed in this study that the lowest wood density for smallest trees in dbh (trees at dbh class of 7.5 cm) is also a result of younger layers of wood in these trees. the increase of wood density with increasing dbh class is a result of the age of the trees: holding all other factors constant, larger trees are often older than smaller ones. beets et al. (2012) reported that wood density at a given height up the stem increased with increasing age. because density measurements were made on unextracted samples (i.e. without the resins and other chemical constituents of heartwood removed), some of the trends in density may be an artefact of the heartwood content in a disc. heartwood is reported to decrease with height (miranda et al. 2015) and to be absent above 7.0 m in height up the stem (pérez et al. 2004). the decreased heartwood proportion with increasing magalhães new zealand journal of forestry science (2021) 51:3 page 8 figure 3: pattern of variation of bark thickness with dbh class (a) and height level (b). height and its absence in the upper portion of the stem may contribute to the trend of decreasing wood density with increasing height. this, in turn, may contribute to the increased proportions of bark volume and mass with increasing height. heartwood is also known to increase with increasing stem diameter (tewari & mariswamy 2013), which possibly also explains the increased wood density with dbh class observed in this study. therefore, the observed trend of decreased bark volume and mass with increasing dbh class may also be an artefact of increasing heartwood proportion with diameter. bark volume and bark mass estimated in this study are useful for estimating wood volume under bark and stem wood biomass. in mozambique, members of the local communities use lebombo ironwood trees for construction, as stakes or poles (magalhães 2017a; magalhães 2017b). the stakes and poles are debarked in the forest and the bark left in the forest floor and decompose (magalhães 2017a). thus, bark mass can be used to estimate the carbon dioxide that is released and the nutrients that are reclaimed by the site. bark volume values reported in this study (11 – 33%) are in line with those reported by murphy and cown (2015) for pinus radiata (3.4 – 31.3%) and pérez cordero and kanninen (2003) for tectona grandis (14 – 37%). however, in this study, bark volume was found to increase sharply with increasing height level as opposite to the findings by murphy and cown (2015) and antony et al. (2015). similarly to the findings by various studies (e.g. pérez cordero and kanninen 2003, pérez et al. 2004, antony et al. 2015, murphy & cown 2015), here bark volume percentage was higher for smaller diameter trees. the average bark mass reported in this study (19%) was larger than that reported by eberhardt et al. (2017) for pinus taeda (12.5%) and pinus elliottii (17%). the pattern of variation of bark mass and bark thickness were in agreement with that reported by quilhó and pereira (2001) and cellini et al. (2012): bark mass was site independent and increased with relative height up the stem. bark thickness, on the other hand, increased with dbh and decreased with relative height up the stem. bark thickness is known to increase with tree age (cellini et al. 2012; williams et al. 2007) and, consequently, with tree diameter (chowdhury et al. 2013; nefabas and gambiza 2007; sonmez et al. 2007; williams et al. 2007; zeibig-kichas et al. 2016). as in this study, williams et al. (2007) found that the bark was thicker at lower portion of the stem for six south african tree species. the average bark thickness of lebombo ironwood (9 mm, range: 4 – 12 mm) was two times higher than that reported by paine et al. (2010) for tropical rainforest trees in french guinea, western africa (4.5 mm, range: 0.5 – 29 mm), however with a narrower range. in semi-arid savannahs of zimbabwe, bark thickness was reported to vary from 4.3 to 15.2 mm (nefabas & gambiza 2007). paine et al. (2010) found that bark thickness was strongly positively correlated with dbh, consistent with the finding of this study, where dbh explained the majority of the regression effect. sonmez et al. (2007) also found that most of the regression effect on bark thickness was accounted by variation in dbh class. site was a significant source of variation of bark thickness. funhalouro and mabote districts share similarities with regard to site quality (precipitation, soils, hydrology, metrological droughts) but both differ in comparison to mandlakaze district (magalhães 2017b). nonetheless, although with marked differences in soil and climatic conditions, the average bark thickness of trees from mandlakaze did not differ from those sampled at mabote. on the other hand, although with no apparent differences in soil and climatic conditions, the bark of trees from mabote district was, on average, 25% thicker than that of trees from funhalouro district. this suggests that soil and climatic and/or environmental conditions were not the sources of variation of bark thickness between sites. a similar observation was made by rosell (2016) who found that environmental conditions were less important driver of bark thickness. in this study, it was found that dbh class explained most of the regression effect of bark thickness (51.24%), followed by height level (40.27%). site explained the minority of the regression effect (6.22%). these findings are consistent with the results by rosell (2016) who, based on 640 tree species from 153 angiosperm families, found that stem size was the main source of variation of bark thickness, with environmental conditions being less important. bark thickness is reported to be positively correlated with fire-prone habitats (hoffmann and solbrig 2003; nefabas and gambiza 2007; nieuwstadt and sheil 2005; uhl and kauffman 1990). compared with mandlakaze, mabote and funhalouro are more prone to forest fires due to low precipitation, intensive metrological draughts and intensive slash and burn agriculture. however, the differences in bark thickness are more marked between the fire-prone habitats (mabote and funhalouro) than otherwise (e.g. mabote and mandlakaze), suggesting that bark thickness was uncorrelated with tree association with fire-prone habitats. this is in agreement with the findings by paine et al. (2010). in addition, lebombo ironwood trees are known to be intolerant to fires. they die after being burnt and do not resprout (regrow) (magalhães 2017a). this is supported by nefabas and gambiza (2007) who stated that “although bark thickness contributes to fire tolerance of woody species, it can be misleading to rank species for fire tolerance based on a single variable such as bark thickness”. rosell (2016) found that stem diameter was 25 times more important than fire tolerance in explaining variations in bark thickness. wood density over and under bark, bark thickness and therefore bark proportions are often obtained at breast height by extracting wood cores (chowdhury et al. 2013; francis 1994; hietz et al. 2013; williamson & wiemann 2010). this study proved that the abovementioned whole-stem properties can be represented by the equivalent breast height properties, as no significant differences were observed between them. magalhães new zealand journal of forestry science (2021) 51:3 page 9 magalhães new zealand journal of forestry science (2021) 51:3 page 10 conclusions overall, the patterns of variation of all wood and bark properties of lebombo ironwood investigated in this study were found to be highly dependent on vertical position within the stem and tree diameter class. vertical position along the stem uniquely accounted for most of the regression effects for wood density, bark volume and bark mass. tree diameter class accounted uniquely for most of the regression effect for bark density and bark thickness. site differences did not account for a significant proportion of the variation in the wood and bark properties investigated in this study. overall, whole-stem average properties did not differ from the value measured at breast height. competing interests the author declares that he has no competing interests. acknowledgements thanks are addressed to the field team and to the anonymous reviewers. thanks are extended to professor ernesto uetimane jr. (mr chuck) for sharing the micrographs of lebombo ironwood and for all his advice. abbreviations dbh: diameter at breast height; ahm: annual heat moisture index; ai: aridity index. funding this work was supported by the swedish international development cooperation agency (sida). references ali, a.c., uetimane, e., lhate, i.a., & terziev, n. (2008). anatomical characteristics, properties and use of traditionally used and lesser-known wood species from mozambique: a literature review. wood sci ence technology, 453-472. https://doi. org/10.1007/s00226-008-0186-5 antony, f., schimleck, l.r., daniels, r.f., clark, a., borders, b.e., kane, m.b., & burkhart, h.e. (2015). whole-tree bark and wood properties of loblolly pine from intensively managed plantations. forest science, 61(1), 55-66. https://doi.org/10.5849/ forsci.12-030 auty, d., achim, a., macdonald, e., cameron, a.d., & gardiner, b.a. (2014). models for predicting wood density variation in scots pine. forestry, 87(3), 449458. https://doi.org/10.1093/forestry/cpu005 bastin, j.f., fayolle, a., tarelkin, y., van den bulcke, j., de haulleville, t., mortier, f., beeckman h, van acker j, serckx a, bogaert j, de cannière c. (2015). wood specific gravity variations and biomass of central african tree species: the simple choice of the outer wood. plos one, 10(11), e0142146. https://doi. org/10.1371/journal.pone.0142146 beets, p.n., kimberley, m.o., oliver, g.r., pearce, s.h., graham, j.d., & brandon, a. (2012). allometric equations for estimating carbon stocks in natural forest in new zealand. forests, 3(3), 818-839. https://doi.org/10.3390/f3030818 brasil, m.a.m., veiga, r.a. de a., & timoni, j.l. (1994). erros na determinação da densidade basica da madeira. cerne, 1(1), 55-57. bunster, j. (2006). commercial timbers of mozambique: technological catalogue. maputo: trafores lda. cardoso, g.a. (1963). madeiras de moçambique: androstachys johnsonii. maputo: serviços de agricultura e serviços de veterinária. cellini, j.m., galarza, m., burns, s.l., martínez pastur, g.j., & lencinas, m.v. (2012). equations of bark thickness and volume profiles at different heights with easymeasurement variables. forest systems, 21(1), 2330. https://doi.org/10.5424/fs/2112211-01963 chowdhury, m.q., khan, m.r., & mehedi, m.a.h. (2013). wood density variation in four plantation species growing in bangladesh. journal of the indian academy of wood science, 10(1), 32-38. https:// doi.org/10.1007/s13196-013-0090-y deng, x., zhang, l., lei, p., xiang, w., & yan, w. (2014). variations of wood basic density with tree age and social classes in the axial direction within pinus massoniana stems in southern china. annals of forest science, 71(4), 505-516. https://doi. org/10.1007/s13595-013-0356-y eberhardt, t.l., dahlen, j., & schimleck, l. (2017). species comparison of the physical properties of loblolly and slash pine wood and bark. canadian journal of forest research, 47(11), 1495-1505. https://doi. org/10.1139/cjfr-2017-0091 francis, j.k. (1994). simple and inexpensive method for extracting wood density samples from tropical hardwoods. tree planters notes, 45(1), 10-12. hernández, m.a., & genes, p.y. (2016). wood density at breast height as estimate of the whole tree density in grevillea robusta. australian forestry, 79(3), 168-170. https://doi.org/10.1080/00049158.201 6.1164578 hietz, p., valencia, r., & joseph wright, s. (2013). strong radial variation in wood density follows a uniform pattern in two neotropical rain forests. functional ecology, 27(3), 684-692. https://doi. org/10.1111/1365-2435.12085 hoffmann, w.a., & solbrig, o.t. (2003). the role of topkill in the differential response of savanna woody species to fire. forest ecology and management, 180(1-3), 273-286. https://doi.org/10.1016/ s0378-1127(02)00566-2 https://doi.org/10.1007/s00226-008-0186-5 https://doi.org/10.1007/s00226-008-0186-5 https://doi.org/10.5849/forsci.12-030 https://doi.org/10.5849/forsci.12-030 https://doi.org/10.1093/forestry/cpu005 https://doi.org/10.1371/journal.pone.0142146 https://doi.org/10.1371/journal.pone.0142146 https://doi.org/10.3390/f3030818 https://doi.org/10.5424/fs/2112211-01963 https://doi.org/10.1007/s13196-013-0090-y https://doi.org/10.1007/s13196-013-0090-y https://doi.org/10.1007/s13595-013-0356-y https://doi.org/10.1007/s13595-013-0356-y https://doi.org/10.1139/cjfr-2017-0091 https://doi.org/10.1139/cjfr-2017-0091 https://doi.org/10.1080/00049158.2016.1164578 https://doi.org/10.1080/00049158.2016.1164578 https://doi.org/10.1111/1365-2435.12085 https://doi.org/10.1111/1365-2435.12085 https://doi.org/10.1016/s0378-1127(02)00566-2 https://doi.org/10.1016/s0378-1127(02)00566-2 magalhães new zealand journal of forestry science (2021) 51:3 page 11 ibanez, t., chave, j., barrabé, l., elodie, b., boutreux, t., trueba, s., vandrot, h. & birnbaum, p. (2017). community variation in wood density along a bioclimatic gradient on a hyper-diverse tropical island. journal of vegetation science, 28(1), 19-33. https://doi.org/10.1111/jvs.12456 kimberley, m.o., cown, d.j., mckinley, r.b., moore, j.r., & dowling, l.j. (2015). modelling variation in wood density within and among trees in stands of new zealand-grown radiata pine. new zealand journal of forestry science, 45: 22. https://doi. org/10.1186/s40490-015-0053-8 knapic, s., louzada, j.l., leal, s., & pereira, h. (2008). within-tree and between-tree variation of wood density components in cork oak trees in two sites in portugal. forestry, 81(4), 465-473. https://doi. org/10.1093/forestry/cpn012 machado, j.s., louzada, j.l., santos, a. j.a., nunes, l., anjos, o., rodrigues, j., simões, r.m.s. & pereira, h. (2014). variation of wood density and mechanical properties of blackwood (acacia melanoxylon r. br.). materials and design, 56, 975-980. https:// doi.org/10.1016/j.matdes.2013.12.016 mae. (2005a). perfil do distrito de funhalouro: província de inhamabane. maputo: ministério de admnistração estatal. http://www.govnet.gov.mz/ mae. (2005b). perfil do distrito de mandlakaze: província de gaza. maputo: ministério de admnistração estatal. http://www.govnet.gov.mz/ mae. (2014). perfil do distrito de mabote: província de inhambane. maputo: ministério de admnistração estatal. http://www.govnet.gov.mz/ magalhães, t.m. (2017). carbon stocks in necromass and soil pools of a mozambican tropical dry forest under different disturbance regimes. biomass and bioenergy, 105. https://doi.org/10.1016/j. biombioe.2017.07.023 magalhães, t.m., & seifert, t. (2015). tree component biomass expansion factors and root-to-shoot ratio of lebombo ironwood: measurement uncertainty. carbon balance and management, 10(1). https:// doi.org/10.1186/s13021-015-0019-4 magalhães, t.m. (2015). live aboveand belowground biomass of a mozambican evergreen forest: a comparison of estimates based on regression equations and biomass expansion factors. forest ecosystems, 2(1), 28. https://doi.org/10.1186/ s40663-015-0053-4 magalhães, t.m. (2017). site-specific height-diameter and stem volume equations for lebombo-ironwood. annals of forest research, 60(2), 297-312. https:// doi.org/10.15287/afr.2017.838 miranda, i., almeida, m.h., & pereira, h. (2001). provenance and site variation of wood density in eucalyptus globulus labill. at harvest age and its relation to a non-destructive early assessment. forest ecology and management, 149(1-3), 235-240. https://doi.org/10.1016/s03781127(00)00560-0 miranda, i., gominho, j., & pereira, h. (2015). heartwood, sapwood and bark variation in coppiced eucalyptus globulus trees in 2nd rotation and comparison with the single-stem 1st rotation. silva fennica, 49(1), 1-13. https://doi.org/10.14214/sf.1141 molotja, g.m., ligavha-mbelengwa, m.h., & bhat, r.b. (2011). antifungal activity of root, bark , leaf and soil extracts of androstachys johnsonii prain. african journal of biotechnology, 10(30), 57255727. https://doi.org/10.5897/ajb10.2212 murphy, g., & cown, d. (2015). within-tree, betweentree, and geospatial variation in estimated pinus radiata bark volume and weight in new zealand. new zealand journal of forestry science, 45: 18. https://doi.org/10.1186/s40490-015-0048-5 nabais, c., hansen, j.k., david-schwartz, r., klisz, m., lópez, r., & rozenberg, p. (2018). the effect of climate on wood density: what provenance trials tell us? forest ecology and management, 408, 148-156. https://doi.org/10.1016/j. foreco.2017.10.040 nefabas, l.l., & gambiza, j. (2007). fire tolerance mechanisms of common woody plant species in a semiarid savanna in south western zimbabwe. african journal of ecology, 45(4), 550-556. https:// doi.org/10.1111/j.1365-2028.2007.00767.x nieuwstadt, m.g.l.v.a.n., & sheil, d. (2005). drought, fire and tree survival in borneo rain forest, east kalimantan, indonesia. journal of ecology, 93, 191-201. https://doi.org/10.1111/j.13652745.2004.00954.x nimon, k., oswald, f., & roberts, j.k. (2015). yhat: interpreting regression effects (r package version 2.0-0). vienna: r foundation for statistical computing. nyg, r., & elfving, b. (2000). stem basic density and bark proportion of 45 woody species in young savanna coppice in forests in burkina faso. annals of forest science, 57(2), 143-153. https://doi.org/10.1051/ forest:2000165 paine, c.e.t., stahl, c., courtois, e.a., patiño, s., sarmiento, c., & baraloto, c. (2010). functional explanations for variation in bark thickness in tropical rain forest trees. functional ecology, 24(6), 1202-1210. https://doi.org/10.1111/j.13652435.2010.01736.x pásztory, z., mohácsiné, i.r., gorbacheva, g., & börcsök, z. (2016). the utilization of tree bark. bioresources, 11(3), 7859-7888. https://doi.org/10.15376/ biores.11.3.pasztory https://doi.org/10.1111/jvs.12456 https://doi.org/10.1186/s40490-015-0053-8 https://doi.org/10.1186/s40490-015-0053-8 https://doi.org/10.1093/forestry/cpn012 https://doi.org/10.1093/forestry/cpn012 https://doi.org/10.1016/j.matdes.2013.12.016 https://doi.org/10.1016/j.matdes.2013.12.016 http://www.govnet.gov.mz/ http://www.govnet.gov.mz/ http://www.govnet.gov.mz/ https://doi.org/10.1016/j.biombioe.2017.07.023 https://doi.org/10.1016/j.biombioe.2017.07.023 https://doi.org/10.1186/s13021-015-0019-4 https://doi.org/10.1186/s13021-015-0019-4 https://doi.org/10.1186/s40663-015-0053-4 https://doi.org/10.1186/s40663-015-0053-4 https://doi.org/10.15287/afr.2017.838 https://doi.org/10.15287/afr.2017.838 https://doi.org/10.1016/s0378-1127(00)00560-0 https://doi.org/10.1016/s0378-1127(00)00560-0 https://doi.org/10.14214/sf.1141 https://doi.org/10.5897/ajb10.2212 https://doi.org/10.1186/s40490-015-0048-5 https://doi.org/10.1016/j.foreco.2017.10.040 https://doi.org/10.1016/j.foreco.2017.10.040 https://doi.org/10.1111/j.1365-2028.2007.00767.x https://doi.org/10.1111/j.1365-2028.2007.00767.x https://doi.org/10.1111/j.1365-2745.2004.00954.x https://doi.org/10.1111/j.1365-2745.2004.00954.x https://doi.org/10.1051/forest:2000165 https://doi.org/10.1051/forest:2000165 https://doi.org/10.1111/j.1365-2435.2010.01736.x https://doi.org/10.1111/j.1365-2435.2010.01736.x https://doi.org/10.15376/biores.11.3.pasztory https://doi.org/10.15376/biores.11.3.pasztory pérez cordero, l.d., & kanninen, m. (2003). heartwood, sapwood and bark content, and wood dry density of young and mature teak (tectona grandis) trees grown in costa rica. silva fennica, 37(1), 45-54. https://doi.org/10.14214/sf.511 pérez, d., kanninen, m., matamores, f., fonseca, w., & chaves, e. (2004). heartwood, sapwood and bark contents of bombacopsis quinata in costa rica. journal of tropical forest science, 16(3), 318-327. quilhó, t., & pereira, h. (2001). within and betweentree variation of bark content and wood density of eucalyptus globulus in commercial plantations. iawa journal, 22(3), 255-265. https://doi. org/10.1163/22941932-90000283 r core team. (2019). a language and environment for statistical computing. vienna: r foundation for statistical computing. repola, j. (2006). models for vertical wood density of scots pine, norway spruce and birch stems, and their application to determine average wood density. silva fennica, 40(4), 673-685. https://doi. org/10.14214/sf.322 rosell, j.a. (2016). bark thickness across the angiosperms: more than just fire. new phytologist, 211(1), 90102. https://doi.org/10.1111/nph.13889 searle, s.d., & owen, j.v. (2005). variation in basic wood density and percentage heartwood in temperate australian acacia species. australian forestry, 68(2), 126-136. https://doi.org/10.1080/000491 58.2005.10674956 sonmez, t., keles, s., & tilki, f. (2007). effect of aspect, tree age and tree diameter on bark thickness of picea orientalis. scandinavian journal of forest research, 22(3), 193-197. https://doi. org/10.1080/02827580701314716 tewari, v.p., & mariswamy, k.m. (2013). heartwood, sapwood and bark content of teak trees grown in karnataka, india. journal of forestry research, 24(4), 721-725. https://doi.org/10.1007/s11676013-0410-5 thomas, r.e., & bennett, n.d. (2014). estimating bark thicknesses of common appalachian hardwoods. in j. w. groninger, e. j. holzmueller, c. k. nielsen, & d. c. dey (eds.), proceedings of the 19th central hardwood forest conference (pp. 283-294). illinois, usa: u.s. forest service. tian, x., cown, d., & mcconchie, d. (1995). modelling of pinus radiata wood properties. part 2: basic density. new zealand journal of forestry science, 25(2), 214-230. uhl, c., & kauffman, j.b. (1990). deforestation, fire susceptibility, and potential tree responses to fire in the eastern amazon. ecology, 71(2), 437-449. https://doi.org/10.2307/1940299 wickham, h., chang, w., henry, l., pedersen, t.l., takahashi, k., wilke, c., & woo, k. (2018). ggplot2: create elegant data visualisations using the grammar of graphics description (r package version 3.1.0). vienna: r foundation for statistical computing. williams, v.l., witkowski, e.t.f., & balkwill, k. (2007). relationship between bark thickness and diameter at breast height for six tree species used medicinally in south africa. south african journal of botany, 73(3), 449-465. https://doi.org/10.1016/j. sajb.2007.04.001 williamson, g.b., & wiemann, m.c. (2010). measuring wood specific gravity correctly. american journal of botany, 97(3), 519-524. https://doi.org/10.3732/ ajb.0900243 zeibig-kichas, n.e., ardis, c.w., berrill, j.p., & king, j.p. (2016). bark thickness equations for mixedconifer forest type in klamath and sierra nevada mountains of california. international journal of forestry research, 2016, 28-31. https://doi. org/10.1155/2016/1864039 magalhães new zealand journal of forestry science (2021) 51:3 page 12 https://doi.org/10.14214/sf.511 https://doi.org/10.1163/22941932-90000283 https://doi.org/10.1163/22941932-90000283 https://doi.org/10.14214/sf.322 https://doi.org/10.14214/sf.322 https://doi.org/10.1111/nph.13889 https://doi.org/10.1080/00049158.2005.10674956 https://doi.org/10.1080/00049158.2005.10674956 https://doi.org/10.1080/02827580701314716 https://doi.org/10.1080/02827580701314716 https://doi.org/10.1007/s11676-013-0410-5 https://doi.org/10.1007/s11676-013-0410-5 https://doi.org/10.2307/1940299 https://doi.org/10.1016/j.sajb.2007.04.001 https://doi.org/10.1016/j.sajb.2007.04.001 https://doi.org/10.3732/ajb.0900243 https://doi.org/10.3732/ajb.0900243 https://doi.org/10.1155/2016/1864039 https://doi.org/10.1155/2016/1864039 comparison of cost, productivity and residue yield of cut-to-length and fuel-adapted harvesting in a pinus radiata d.don final harvest in western australia martin strandgard1,2,* and rick mitchell2 1 university of tasmania, churchill avenue, sandy bay, tasmania, 7005, australia 2 university of the sunshine coast, locked bag 4, maroochydore dc, queensland, 4558 australia *corresponding author: mstrandg@usc.edu.au (received for publication 26 september 2018; accepted in revised form 24 july 2019) abstract background: forest biomass is a major global source of biofuel. to compete with other energy sources its delivered costs need to be reduced. globally, logging residue (lr) is likely to be the cheapest, readily available forest biomass form. lr transport is a major cost component. methods: a harvester-forwarder harvest system was studied in two adjacent areas to compare swedish “fuel-adapted harvesting” with conventional cut-to-length harvesting at the stump in a mature pinus radiata d.don plantation in western australia to assess the impact of fuel-adapted harvesting on costs and productivity of a harvester and forwarder producing logs and extracting lr and on lr yield. results: harvester and forwarder productivities producing logs were significantly reduced in the fuel-adapted area compared with the conventional area which increased log production costs for the fuel-adapted site by 15%. forwarder productivity extracting lr and lr yield were significantly greater in the fuel-adapted area which reduced lr extraction costs by approximately 28%. this was due to the ease of loading lr from residue piles created during fueladapted harvesting compared with loading scattered residue from conventional harvesting. the cost reduction for lr extraction from the fuel-adapted area exceeded the increased log harvest and extraction costs. this resulted in the combined log and lr costs for the fuel-adapted area being approximately 12% lower than those for the conventional area. increased forwarder productivity through adoption of larger load bunks and residue-specific grapples combined with increased operator experience with fuel-adapted harvesting would be likely to further decrease log and lr production costs. conclusions: the results show that adoption of fuel-adapted harvesting could reduce lr delivered costs, thus increasing its viability as a biofuel. however, primary transport cost is only one component of lr delivered costs and needs to be considered in combination with the reduction of other supply chain costs, particularly secondary transport costs which can make up a large proportion of lr delivered costs. because removal of most lr from a site can reduce subsequent tree growth, guidelines specifying the proportion of lr retained should be considered. new zealand journal of forestry science strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 https://doi.org/10.33494/nzjfs492019x37x e-issn: 1179-5395 published on-line: 26/11/2019 © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access of worldwide forest biomass availability suggest that there is sufficient unused forest biomass to provide a substantial proportion of global energy needs (gregg & smith 2010). introduction forest biomass is a major energy source in many countries, supplying over 25% of the energy needs of sweden, finland and austria (aebiom 2013). estimates keywords: logging residue, bioenergy, fuel-adapted harvesting, pinus radiata, harvester, forwarder strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 2 for forest biomass to compete with conventional energy sources, its delivered costs must be minimised (caputo et al. 2005). sawmill residues are likely to be the cheapest form of forest biomass as they do not incur costs for collection and transport from the harvest site (mani et al. 2006). however, sawmill residues are often already used in pulp or particle board manufacture or as fuel on the sawmill site (parikka 2004). logging residue (lr) is the next cheapest form of forest biomass (rodriguez et al. 2011). globally it is estimated that there are several hundred million tonnes of lr potentially available annually after excluding lr retained to maintain site productivity or uneconomic to extract (gregg & smith 2010). the low bulk density, low energy density and high moisture content of fresh lr (wolfsmayr and rauch 2014) increase its transport costs relative to conventional energy sources (gold & seuring 2011). lr from harvest operations where trees are processed at roadside generally has the lowest delivered costs as primary transport costs are minimised (belart et al. 2017; yemshanov et al. 2014). however, primary transport can form a significant component of delivered costs for lr arising from cut-to-length (ctl) at the stump harvest operations, which is a commonly employed harvest system world-wide. during conventional ctl harvest operations trees are processed in front of the harvester causing lr to mainly fall in the harvester’s path which can help reduce soil compaction (cambi et al. 2015). however, this practice makes it difficult to collect the lr and increases its soil contamination (kizha and han 2016). this, in turn, increases wear and tear on equipment used to process lr into bioenergy and the ash content when this material is burned. in contrast, concentration of lr into piles or windrows can increase lr primary transport productivity (cuchet et al. 2004). one such method developed in sweden is called fuel-adapted harvesting; which involves the harvester processing trees at its side so that lr is placed between log piles. fuel-adapted harvesting has been shown to increase the proportion of lr extracted from a site (jacobson & filipsson 2013) and reduce contamination from attached soil (junginger et al. 2005). the objective of this trial was to compare the current harvesting method with the swedish ‘fuel-adapted’ harvesting method in terms of harvester and forwarder cost, productivity, and the yield of lr extracted from a cut-to-length, final harvest operation in a mature p. radiata d.don stand. methods study area the study was conducted in october/november 2017 in a 29-year-old pinus radiata plantation 80 km south-east of manjimup, western australia (34.47966s, 116.75446e) managed by the forest products commission (fpc). the plantation had been first thinned at age 15 years and had subsequently suffered sporadic windthrow over approximately 80% of its area. soil was a duplex sandy gravel and maximum slope was 5 degrees. weather during the study was generally fine with light rain falling on several days (<3 mm in total). the study site was divided into two adjacent areas: a 2.96 ha area harvested using a conventional harvester/ forwarder cut-to-length harvest system (the conventional area) and a 3.38 ha area harvested with a harvester and forwarder using the swedish “fuel-adapted” harvest system (the fuel-adapted area). diameter at breast height over bark (dbhob) was measured on 107 trees in the conventional area and on 104 trees in the fueladapted area. the total height of approximately one-third of these trees was measured using a vertex hypsometer (haglöf, sweden). each measured tree was numbered with paint for identification during the time and motion study. unmeasured tree heights were predicted using a height/diameter relationship developed from the trial site data. individual tree volumes were estimated using a tree volume function supplied by the plantation manager. stand and site characteristics are shown in table 1. retained lr quantities were measured using 1 m2 plots placed at randomly-selected intersection points on grids covering each area. estimated weights of retained lr (mg ha-1) were then calculated by multiplying the mean lr weight for the plots within each site by 10 to convert kg m-2 to mg ha-1. harvesting procedures in both trial areas the harvester travelled along the planting rows. in the conventional area trees were felled into the unharvested plantation to the right of the harvester and processed so that the logs were piled in table 1. stand and site characteristics for the conventional and fuel-adapted study sites parameter conventional site fueladapted site mean dbhob (mm) 396 389 mean height (m) 27 26.6 mean tree volume (m3) 1.2 1.2 stems per hectare (sph) 293 293 total log volume extracted (m3) 650 771 (m3 ha-1) 220 228 recovered residues green/oven dry (mg) 126 / 58 196 / 91 green/oven dry (mg ha-1)) 43 / 20 58 / 27 % of total 42 68 retained residues green/oven dry (mg) 181 / 83 95 / 44 green/oven dry (mg ha-1) 61 / 28 28 / 13 % of total 58 32 the harvested area to the harvester’s left (figure 1(a)). most lr fell in front of the harvester. in the fuel-adapted area trees were felled into the unharvested trees in front of the harvester and processed to its left side so that logs and residues were piled separately adjacent to the harvester’s travel path (figure 1(b)). approximately six to seven trees were processed to form each log and residue pile in the fuel-adapted area. logs were extracted to roadside by forwarder from each site and piled separately by log product. five log products were cut on each site using the same product definitions. the mean volumes of the different log products were estimated from stanford pri files (www. skogforsk.se). these were: short sawlogs (3.1m & 3.7m) 0.25m3, medium sawlogs (4.3 m & 4.9 m) 0.34 m3, long sawlogs (5.5 m & 6.1 m) 0.47 m3, export logs 0.14 m3, chiplogs 0.13 m3. of 42 forwarder cycles studied in each area, six loads had assortments of more than one product type in the conventional area and 18 loads in the fueladapted area. quantities of logs and lr extracted from each area are shown in table 1. log harvest time and motion study harvesting was performed by an experienced operator using a john deere 903kh single grip harvester (7600 engine hours) with a waratah 624c harvester head. a detailed time and motion study was performed in both areas to quantify any differences in harvester performance between them. harvester cycle and elemental times (table 2) were recorded on a tablet using umtplus time study software (www.laubrass. com/umtplus). delay times were excluded from cycle times. harvester productivity (m3 per productive machine hour delay free) (m3 pmh0 -1) was determined by dividing tree volume (m3) by delay-free cycle time (pmh0) for each tree. based on the method described by nurminen et al. (2006), the sum of the time for elements not assigned to a specific cycle (brushing/clearing and stacking/ bunching) in each area was divided by the number of trees studied in that area and this time was added to the cycle time for each tree. regression models with tree volume as the independent variable and harvester productivity as the dependent variable were fitted for each area. log and logging residue extraction time and motion study in both areas, log extraction was undertaken by an experienced operator using a john deere 1910e forwarder (load capacity 19 mg) (3400 engine hours). a different experienced operator, using the same model forwarder (5000 engine hours) carried out the lr extraction in both areas. lr weights were reported on both a green (53.5% mc) and oven dry basis. log extraction cycle and elemental times (table 3) were recorded on a tablet during the trial using umtplus time study software (www.laubrass.com/umtplus). lr extraction cycle and elemental times (table 3) were recorded from digital video recordings. delay times were excluded from cycle times. loading and unloading elemental times were expressed both as minutes and minutes per cubic metre (logs) or megagrams (lr) to remove the effect of load size variations. for log extraction, load volumes were estimated by multiplying the number of logs per load of each product type (counted during unloading) by the product type’s mean strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 3 figure 1. log and logging residue arrangements on the (a) conventional site; (b) fuel-adapted site. time element description moving/ positioning starts when tracks begin to move or when boom begins its swing towards next tree. ends when felling commences. felling starts when head clamps onto tree. ends when feed rollers are activated or tree is horizontal. processing starts when feed rollers are activated. delimbing and crosscutting of tree. ends when felling boom begins to swing to next tree or tracks begin to move. brushing/ clearing clearing of unmerchantable trees or processing debris/undergrowth. stacking/ bunching starts when the boom commences moving to retrieve, move or ‘stack’ any processed logs. ends when another element commences delay any interruption causing the harvester to cease working during a shift. table 2. harvester time element descriptions used in the study log volume. lr load weights were obtained using the forwarder’s crane scales. forwarder productivity was estimated by dividing load volume (m3) (log extraction) or load weight (mg oven dry and green) (lr extraction) by delay-free cycle time (pmh0) for each forwarder cycle. a global positioning system (gps) equipped multidat strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 4 onboard computer (brown et al. 2002) was installed in the forwarders to estimate their travel distances. extraction distance was defined as half the total cycle distance (tiernan et al. 2004). statistical analysis linear regression models were developed to understand the key factors associated with harvester and forwarder (logs and lr) cycle times and productivities in the conventional and fuel-adapted areas. these models were used to investigate potential relationships between different measures of forwarder productivity (cycle times (min), elemental times (min and min m-3) and productivity (m3 pmh0 -1)) and characteristics of the operation (extraction distance (m), load volume (m3) or load weight (mg (oven dry)). for log extraction only, potential relationships between productivity measures and either the number of logs per load or number of product types per load were examined. the goodness of fit of these regression models was assessed using r2, root mean square error (rmse) and the behaviour of residuals. all models were checked for compliance with the linear regression assumptions. for regressions with multiple independent variables, multi-collinearity was tested using a variance inflation factor (vif) threshold of five. when the dependent regression variable was logtransformed, the model was corrected for bias (snowdon 1991) and fit statistics were derived from backtransformed values. conventional and fuel-adapted area regression models for each machine were compared statistically using an f test (motulsky & christopoulos 2004). for each machine, mean cycle and elemental times were compared between the conventional and fueladapted areas using t-tests. all comparisons were made at p < 0.05. analysis was performed using ms excel 2016 and minitab v. 17. costs machine costs (m3 pmh0 -1 and mg (green) pmh0 -1) were calculated following the method developed by miyata (1980) using cost assumptions provided in table 4 and are presented in australian dollars (aud). results harvester time consumption and productivity for both sites, a natural log – linear model gave the best fit to the cycle time data and a natural log log model gave the best fit to the productivity data (table 5). mean harvester cycle times (conventional 1.00 min; fuel-adapted: 1.10 min) and cycle time regression models (figure 2) were significantly different. times for the moving/positioning and felling elements were significantly longer in the fuel-adapted area than in the conventional area (table 6). processing time accounted for the greatest proportion of cycle time in both areas (conventional: 65.8%; fuel-adapted: 68.9%). mean harvester productivities (conventional: 73.8 m3 pmh0 -1; fuel-adapted: 62.7 m3 pmh0 -1) and harvester productivity regression models (figure 3) were also significantly different. time element description travel empty starts when forwarder commences travel into the harvest area from the landing and ends when crane commences moving to collect logs/lr. loading starts when crane commences moving to collect logs/ lr and ends when another element commences. includes adjustments to the logs/lr on the bunk. moving during loading movement between logs/lr piles with no crane movement. starts when wheels begin rotating and ends when crane recommences movement. simultaneous crane and wheel movement is recorded as loading. travel loaded starts with travel to the landing with a load and ends when wheels cease to rotate or crane commences to move at the landing. unloading starts with commencement of crane movement, grapple empty, towards the forwarder’s bunk and ends when another element commences. includes adjustments to the log/lr stack. moving during unloading movement between log/lr stacks at the landing with no crane movement. starts when the wheels begin to rotate and ends when the crane recommences movement to the forwarder bunk. simultaneous crane and wheel movement is recorded as unloading. brushing/ clearing clearing of non-merchantable trees/undergrowth or processing debris. stacking/ bunching adjustment of logs/lr in a roadside stack not associated with unloading or loading. delay any interruption causing the forwarder to cease working during a shift. table 3. forwarder time element descriptions (log and lr extraction) used in the study strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 5 machine variable harvester forwarder purchase price (aud) 750,000 700,000 operating days per year 249 249 shifts per day 1 1 hours per shift 10.0 10.0 utilisation rate (%) 70 75 machine life (years) 5 7 salvage value (% of purchase price) 20 20 repair and maintenance (% of depreciation) 75 75 interest rate (% of average yearly investment) 9 9 insurance and tax rate (% of average yearly investment) 6 6 fuel cost (aud l-1)* 0.98 0.98 oil & lubricant (% of fuel cost) 50 50 labour costs (aud smh-1) 46.59 46.59 supervision (% of labour costs) 10 10 table 4. machine cost calculation assumptions used in the study * at time of study off-road vehicle use in australia was eligible for a tax rebate of aud0.403 litre-1 figure 2: harvester cycle time regression models for the conventional and fuel-adapted study sites. figure 3: harvester productivity regression models for the conventional and fuel-adapted study sites. table 5. harvester cycle time and productivity regression models and fit statistics site moving/positioning felling processing brushing/clearing stacking conventional 0.13* 0.11* 0.65 0.07 0.028 fuel-adapted 0.17* 0.13* 0.69 0.09 0.026 * significantly different elemental times between sites table 6. harvester mean elemental times (min) for the conventional and fuel-adapted study sites. site model sample size rmse r2 conventional cycle time = 0.0094* exp0.44 * tree volume 107 0.003 0.59 productivity = 68.1* tree volume0.47 107 14.2 0.53 fuel-adapted cycle time = 0.0114* exp0.39 * tree volume 104 0.004 0.41 productivity = 58.8* tree volume0.57 104 11.5 0.60 log extraction time consumption and productivity time elements not assigned to a specific cycle (brushing/clearing and stacking/bunching) were excluded from cycle times as they made up less than 1% of total forwarder study time at each site. in both areas, linear regression models gave the best fit to the forwarder cycle time and productivity data (table 7). interaction terms were not significant and vif values were less than 5. in both areas, load volume explained the greatest proportion of the variability in cycle time and productivity. mean forwarder cycle times (conventional: 24.1 min; fuel-adapted: 31.2 min) and the cycle time and productivity regression models were significantly different between areas. moving during loading, travel loaded, and moving during unloading elemental times were significantly longer in the fuel-adapted area (table 8). loading and unloading times were also significantly longer in the fuel-adapted area when expressed in minutes but were not significantly different when expressed in min m-3. mean extraction distance in the fuel-adapted area (298 m) was significantly longer than in the conventional area (236 m). mean load volume and number of logs were not significantly different between areas (conventional: 15.6 m3 and 64 logs; fuel-adapted: 17.7 m3 and 70 logs). for the pooled mean extraction distance (267 m), load volume (16.6 m3) and log number (67) across both area, strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 6 site sample size rmse r2 conventional cycle time = 1.27 + 0.77 × load volume + 0.092 × log number + 0.021 × extraction distance 42 2.21 0.85 productivity = 33.4 + 1.2 × load volume 0.035 × extraction distance 0.092 × log number 42 3.49 0.82 fuel-adapted cycle time = 5.0 + 0.69 × load volume + 0.085 × log number + 0.027 × extraction distance 42 4.01 0.52 productivity = 24.6 + 1.26 × load volume 0.032 × extraction distance 0.051 × log number 42 3.77 0.81 table 7. forwarder cycle time and productivity regression coefficients and fit statistics time element site regression r2 mean time (min) travel empty (min) conventional 0.0064 × extraction distance 0.74 1.51 fuel-adapted no significant relationship 1.90 loading (min m-3) conventional 0.83 0.0195 × load volume + 0.0042 × log number 0.61 0.80 fuel-adapted 1.42 – 0.024 × load volume – 0.081 × product number 0.58 0.85 loading (min) conventional 0.44 × load volume + 0.073 × log number 0.98 11.5* fuel-adapted 3.7 + 0.37 × load volume + 0.055 × log number 0.50 14.1* moving during loading (min) conventional 0.012 × extraction distance 0.76 2.82* fuel-adapted 0.064× log number 0.83 4.44* travel loaded (min) conventional 0.67 + 0.0058 × extraction distance 0.14 2.03* fuel-adapted 0.011 × extraction distance 0.88 3.30* unload (min m-3) conventional 0.54 – 0.0084 × load volume 0.37 0.41 fuel-adapted 0.48 0.0039 × load volume 0.10 0.41 unload (min) conventional 1.86 +0.27×load volume 0.79 6.02* fuel-adapted 1.25 +0.33× load volume 0.76 7.1* moving during unloading (min) conventional -0.17 + 0.18 × product number 0.48 0.042* fuel-adapted -0.37 + 0.37 × product number 0.74 0.25* * significantly different mean elemental times between sites table 8. regression models, associated r2 values and mean times for each forwarder time element (extracting logs) at the conventional and fuel-adapted study sites. forwarder cycle time was 25.8 min for the conventional operation and 29.3 min for the fuel-adapted operation, and forwarder productivity was 38 m3 pmh0 -1 for the conventional operation and 34 m3 pmh0 -1 for the fueladapted operation which represented a 14% increase in cycle time and an 11% productivity reduction for the fuel-adapted operation. regression models and mean times for each time element and area are shown in table 8. logging residue extraction time consumption and productivity for both areas a linear regression model form gave the best fit to the cycle time data (table 9). mean extraction distance and load weight were 245 m and 4.2 mg (oven dry) (9 mg (green)), and 226 m and 4.6 mg (oven dry) (9.9 mg (green)) in the conventional and fuel-adapted areas, respectively, and did not differ significantly between areas. mean forwarder cycle times (conventional: 38.9 min; fuel-adapted: 30.7 min) were strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 7 significantly different between areas. fuel-adapted area loading times were significantly longer than those in the conventional area (min and min m-3) (table 10). no other elemental times were significantly different. regression models and mean times for forwarder time elements are shown in table 10. mean forwarder productivity extracting lr was significantly greater in the fuel-adapted area (table 11). no significant relationships between forwarder lr productivity and either load weight or extraction distance were found for either area. there was little productivity variation among forwarder cycles as increases in load weights were accompanied by a corresponding increase in cycle times. costs costs to harvest and extract logs and lr to roadside were calculated from machine costs and machine productivity (table 12). site model sample size rmse r2 conventional cycle time = 8.0 + 7.4 × load weight 14 5.3 0.54 fuel-adapted cycle time = -0.55 + 6.86 × load weight 20 3.42 0.37 time element site regression r2 mean time (min) travel empty (min) conventional no significant relationship 1.91 fuel-adapted no significant relationship 2.26 loading (min mg (green)-1) conventional no significant relationship 6.06* fuel-adapted no significant relationship 3.81* loading (min) conventional 0.53 + 5.93 × load weight 0.54 25.3* fuel-adapted no significant relationship 17.4* moving during loading (min) conventional no significant relationship 0.62 fuel-adapted no significant relationship 0.44 travel loaded (min) conventional 0.77 + 0.0074 × extraction distance 0.37 2.58 fuel-adapted 0.58 + 0.0079 × extraction distance 0.51 2.36 unload (min mg (green)-1) conventional no significant relationship 2.05 fuel-adapted no significant relationship 1.81 unload (min) conventional 1.1 + 1.79 × load weight 0.44 8.52 fuel-adapted -6.2 + 3.19 × load weight 0.42 8.31 table 9. forwarder cycle time regression coefficients and fit statistics extracting logging residue * significantly different mean elemental times between sites table 10. regression models, associated r2 values and elemental times for forwarder time elements (extracting lr) at the conventional and fuel-adapted study sites. discussion harvester time consumption and productivity tree volume was the major factor determining harvester productivity in the study, as has been reported in many previous studies (e. g. jiroušek et al. 2007; nurminen et al. 2006; strandgard et al. 2013; walsh & strandgard 2014). harvester productivity in the conventional area was similar to that recorded by jiroušek et al. (2007) and walsh et al. (2014) in studies where the mean tree volume. was comparable with values observed here. harvester productivity was significantly less when harvesting the fuel-adapted area compared with the conventional area. the productivity reduction resulted from significantly longer cycle times in the fuel-adapted area caused by significantly longer moving/positioning and felling elemental times. the increased cycle and elemental times were largely related to lack of operator experience in felling trees using the fuel-adapted harvesting method. moving/positioning times increased because the operator took longer to select the next tree to be felled and to position the head to fell the tree in the correct direction. felling times increased because each tree was moved alongside the harvester for processing while falling, whereas in the conventional area trees were felled with little movement from where they stood. processing would also often commence in the conventional area while the tree was falling. forwarder time consumption and productivity extracting logs in both areas load volume, extraction distance and log number were significantly associated with the variation in forwarder productivity. forwarder productivity reported by tiernan et al. (2004) was similar to the current study for the same table 2: confusion matrix mean extraction distance, load volume and similar site conditions. greater mean tree volumes and shorter extraction distances than those in the current study resulted in considerably increased forwarder productivity for two trials in mature p. radiata final harvest operations under similar site conditions (ghaffariyan et al. 2015): 86m3 pmh0 -1 and (walsh and strandgard 2014): 93m3 pmh0 -1). forwarder productivity reported by strandgard et al. (2017) (44m3 pmh0 -1) was also greater than that in the current study due to greater mean log volumes but the difference in productivity was considerably less due to longer extraction distances in the strandgard et al. (2017) study. longer cycle times in the fuel-adapted area were associated with a significant reduction in forwarder productivity compared with the conventional area. the longer extraction distances in the fuel-adapted area increased travel loaded times, accounting for some of the difference in cycle times. other factors were the significantly longer loading (min), unloading (min) and moving during loading times in the fuel-adapted area. loading times in the fuel-adapted area were impacted by the poorer product separation which often required the operator to sort through log piles to select the desired product type(s). to reduce infield sorting, the forwarder operator increased the number of mixed-product loads. although products were predominantly kept separate on the forwarder bunk, some sorting was required at roadside resulting in increased unloading (min) and moving during unloading times in the fuel-adapted area. the longer moving during loading time in the fueladapted area resulted from differences in the loading method between the areas. in the fuel-adapted area, the forwarder was required to stop at each discrete log pile to load before moving to the next log pile. conversely, in the conventional area the continuous row of logs laid out alongside the extraction path predominantly allowed the strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 8 forwarder productivity (mg pmh0 -1) site oven dry weight green weight conventional 6.5 14.0 fuel-adapted 9.0 19.4 table 11. mean forwarder productivity extracting lr (mg pmh0 -1) at the conventional and fuel-adapted study sites. table 12. harvester and forwarder (logs and lr) costs for the conventional and fuel-adapted study sites. total cost is the cost to harvest and extract 1 cubic metre of logs and 1 mg (green) of lr to roadside. site harvester forwarder (logs) forwarder (lr) total cost aud pmh0 -1 aud m-3 aud pmh0 -1 aud m-3 aud mg (green)-1 aud m-3 + aud mg (green)-1 conventional 124.50 3.20 102.50 4.80 13.00 21.00 fuel-adapted 124.50 3.80 102.50 5.40 9.40 18.50 forwarder to travel slowly while loading logs. as expected, loading time (min) in both areas was positively related to load volume and number of logs. loading time (min m-3) also increased in both areas as the number of logs increased. as greater numbers of logs often resulted from loading smaller logs, the increased loading time reflected the greater time required to load smaller logs (danilović et al. 2014; nurminen et al. 2006). similarly, the apparently anomalous result that loading times (min m-3) and unloading times (min m-3) decreased with increasing load volume reflected the inverse relationship between number of logs and load volume (kellogg & bettinger 1994). unloading times (min) increased with increasing load volume, as would be expected. travel empty time was significantly associated with extraction distance in the conventional area only, whereas travel loaded time was significantly associated with extraction distance in both areas, though in all cases the relationships were weak. reduced harvester and forwarder productivity when producing logs from the fuel-adapted area was likely to reflect a learning effect as reported by purfürst (2010) and lapointe and robert (2000). unlike those studies, the operators in this study were experienced which suggested that tacit knowledge the operators had acquired was of limited use when learning the fueladapted harvesting approach and may have even been a hindrance as the operators had to “relearn” how to perform their tasks. forwarder time consumption and productivity extracting logging residue forwarder cycle times extracting lr were positively related to load weight in both areas. extraction distance was not a significant variable because travel times (loaded and empty) were a minor part of total cycle times (conventional 11.5%, fuel-adapted 15.1%). significantly shorter cycle times in the fuel-adapted area resulted in forwarder productivity being significantly greater in this area. the shorter cycle times in the fuel-adapted area resulted from significantly shorter loading times which reflected the relative ease of loading lr from piles compared with loading scattered lr from the ground. forwarder load weights in both areas were similar to those reported by nurmi (2007), however, forwarder productivity in the current study was considerably greater which was the result of longer extraction distances increasing cycle times in the nurmi (2007) study. lr load weights in the study reported here were approximately 50% of the forwarder’s weight capacity. forwarders with extended load bunks commonly used in sweden and finland are able to carry lr loads up to approximately 75% of the forwarder’s weight capacity (eliasson et al. 2011). modified forwarder grapples can also be used to load and unload lr faster than a conventional log grapple (eliasson & nordén 2010). logging residue yields total lr quantities at both sites in the current study were at the middle of the range of those measured at sixteen ctl final harvest operations in mature australian pinus radiata plantations (30 to 155 mg ha-1) (ghaffariyan 2013; ximenes et al. 2012). factors such as product specifications, stand density, site type and tree age (räisänen & nurmi 2011) were likely to have caused the variation in lr quantities between these sites. the lr recovery rate in the fuel-adapted area was similar to the mean recovery rate reported by thiffault et al. (2011) for studies in nordic countries (72%) while the rate in the conventional area was similar to that reported by thiffault et al. (2011) for studies in nonnordic countries where loose lr was gathered from the cutblock (32%). the higher lr recovery rate from the fuel-adapted area was likely to have resulted from piling of lr which allowed the forwarder operator to extract small residue pieces that would be impractical to pick up when scattered on the ground. this is consistent with findings reported in previous residue recovery studies (nurmi 2007; thiffault et al. 2015). lr can contain a significant proportion of the site’s nutrient capital (madgwick & webber 1987). a worldwide meta-study of harvest residue removal trials found that removal of harvest residues, particularly when foliage was removed, often reduced subsequent tree growth (achat et al. 2015). to mitigate the impact of lr removal on site productivity, lr foliage content can be reduced through infield drying (nilsson et al. 2013), higher levels of fertiliser can be applied (jones et al. 2011) or minimum levels of lr retention can be specified as is done in finland (abbas et al. 2011) and parts of the united states (briedis et al. 2011). costs harvester costs were comparable to those reported by walsh and strandgard (2014) and ghaffariyan et al. (2015) while forwarder extraction costs (logs) were approximately double those reported in these studies due to differences in productivity. the lr forwarder extraction cost reported by kärhä and vartiamäki (2006) was ~aud10 mg (green)-1 (exchange rate €1 = aud1.58) which was similar to that in the fuel-adapted area. bergström and di fulvio (2014) reported a cost of ~aud13 mg (green)-1 (exchange rate sek1 = aud0.15) which was similar to that in the conventional area. neither of these studies stated whether the lr was piled or scattered. total costs for log and lr harvest and extraction were less in the fuel-adapted area as the lower cost for lr extraction to roadside offset the higher harvester and forwarder (logs) costs. conclusions accumulating lr in piles in the fuel-adapted area enabled a greater yield of lr to be extracted and significantly increased forwarder productivity and decreased costs when extracting lr compared with extracting scattered lr in the conventional area. however, harvester and forwarder productivity when processing and extracting logs was significantly reduced and costs were increased in the fuel-adapted area compared with the conventional area. swedish experience found similar productivity strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 9 reductions to be the result of lack of operator experience with fuel-adapted harvesting. this suggested that continued use of fuel-adapted harvesting in australia is likely to see differences in machine productivity and costs between fuel-adapted and conventional harvesting operations reduce over time. overall the results show that adoption of fuel-adapted harvesting in australia could reduce the delivered costs of lr, increasing its viability as a biofuel. however, primary transport cost is only one component of the lr delivered costs and needs to be considered in combination with methods to reduce other supply chain costs, particularly secondary transport costs which can make up a large proportion of the delivered costs of lr. experience in nordic countries has shown that further significant forwarder productivity gains when extracting lr can be made through using forwarders with extended load bunks and residue specific grapples. investigation of these forwarder modifications is warranted in australia to determine their impact on delivered lr and logs costs with different species and larger harvest units. evidence from previous trials has shown that removing most or all of the lr from a logging site can significantly reduce subsequent tree growth. the high proportion of total lr removed from the fuel-adapted area in the current study highlights that standards or guidelines for minimum proportions of lr to be retained may need to be considered for regions or countries without existing rules to maintain site productivity. competing interests the authors declare that they have no competing interests. authors’ contributions ms was the primary author, undertook the analyses, and assisted with the field measurements. rm supervised the field measurements and provided critical revisions of the manuscript. both authors read and approved the final manuscript. funding the authors acknowledge the funding from the university of the sunshine coast and the industry partners of the australian forest operations research alliance. acknowledgements the authors would like to thank staff from wespine, the plantation logging company and the forest products commission without whom this study would not have been possible and julia bergamin, sao paulo state university, for her assistance in field data collection. availability of data and materials the data belongs in full to the university of the sunshine coast and this publication forms part of a phd thesis project developed by mr. martin strandgard at the university of tasmania. any request to use the original data for publication purposes requires a specific project proposal to be approved by the university of the sunshine coast for any interested scientist in making use of the datasets. references abbas, d., current, d., phillips, m., rossman, r., hoganson, h., & brooks, k. n. (2011). guidelines for harvesting forest biomass for energy: a synthesis of environmental considerations. biomass & bioenergy, 35(11), 4538-4546. achat, d., deleuze, c., landmann, g., pousse, n., ranger, j., & augusto, l. (2015). quantifying consequences of removing harvesting residues on forest soils and tree growth–a meta-analysis. forest ecology and management, 348, 124-141. aebiom (2013). european bioenergy outlook 2013. brussels: european biomass association. belart, f., sessions, j., leshchinsky, b., & murphy, g. (2017). economic implications of moisture content and logging system in forest harvest residue delivery for energy production: a case study. canadian journal of forest research, 47(4), 458466. bergström, d., & di fulvio, f. (2014). comparison of the cost and energy efficiencies of present and future biomass supply systems for young dense forests. scandinavian journal of forest research, 29(8), 793-812. doi:10.1080/02827581.2014.976590 briedis, j., wilson, j., benjamin, j., & wagner, r. (2011). logging residue volumes and characteristics following integrated roundwood and energy-wood whole-tree harvesting in central maine. northern journal of applied forestry, 28(2), 66–71. https:// doi.org/10.1093/njaf/28.2.66 brown, m., guimier, d., mercier, s., provencher, y., & turcotte, p. (2002). multidat and opti-grade: two knowledge-based electronic solutions for managing forestry operations more efficiently. in l kellogg, b spong, & p licht (eds.), woodfor africa forest engineering conference, pietermaritzburg, south africa, july 2-3; pp. 45-49. cambi, m., certini, g., neri, f., & marchi, e. (2015). the impact of heavy traffic on forest soils: a review. forest ecology and management, 338, 124-138. caputo, a. c., palumbo, m., pelagagge, p. m., & scacchia, f. (2005). economics of biomass energy utilization in combustion and gasification plants: effects of logistic variables. biomass & bioenergy, 28(1), 35-51. doi:http://dx.doi.org/10.1016/j. biombioe.2004.04.009 cuchet, e., roux, p., & spinelli, r. (2004). performance of a logging residue bundler in the temperate forests of france. biomass and bioenergy, 27(1), 31-39. doi:https://doi.org/10.1016/j. biombioe.2003.10.006 danilović, m., stojnić, d., karić, s., & sučević, m. (2014). transport of technical roundwood by forwarder and tractor assembly from poplar plantations. strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 10 nova mehanizacija šumarstva: časopis za teoriju i praksu šumarskoga inženjerstva, 35, 11-21. eliasson, l., brunberg, t., & lundström, h. (2011). productivity when forwarding fresh and dried logging residues. . paper presented at the formec 2011, graz, austria, october 9-13, 2011. eliasson, l., & nordén, b. (2010). a new grapple with potential to increase productivity. (efficient forest fuel supply systems. composite report from a four year r & d programme 2007-2010). uppsala: skogforsk. ghaffariyan, m. r. (2013). remaining slash in different harvesting operation sites in australian plantations. silva balcanica, 14(1), 83-93. ghaffariyan, m. r., spinelli, r., magagnotti, n., & brown, m. (2015). integrated harvesting for conventional log and energy wood assortments: a case study in a pine plantation in western australia. southern forests: a journal of forest science, 77(4), 249-254. doi:10.2989/20702620.2015.1052946 gold, s., & seuring, s. (2011). supply chain and logistics issues of bio-energy production. journal of cleaner production, 19(1), 32-42. doi:https://doi. org/10.1016/j.jclepro.2010.08.009 gregg, j. s., & smith, s. j. (2010). global and regional potential for bioenergy from agricultural and forestry residue biomass. mitigation and adaptation strategies for global change, 15(3), 241-262. jacobson, s., & filipsson, j. (2013). spatial distribution of logging residues after final felling – comparison between forest fuel adapted final felling and conventional final felling methods. arbetsrapport nr. 797, sweden: skogforsk. jiroušek, r., klvač, r., & skoupy, a. (2007). productivity and costs of the mechanised cut-to-length wood harvesting system in clear-felling operations. journal of forest science, 53, 476-482. jones, h., beets, p., kimberley, m., & garrett, l. (2011). harvest residue management and fertilisation effects on soil carbon and nitrogen in a 15-yearold pinus radiata plantation forest. forest ecology and management, 262(3), 339-347. https://doi. org/10.1016/j.foreco.2011.03.040 junginger, m., faaij, a., björheden, r., & turkenburg, w. c. (2005). technological learning and cost reductions in wood fuel supply chains in sweden. biomass and bioenergy, 29(6), 399-418. doi:http://dx.doi. org/10.1016/j.biombioe.2005.06.006 kärhä, k., & vartiamäki, t. (2006). productivity and costs of slash bundling in nordic conditions. biomass and bioenergy, 30(12), 1043-1052. doi:http:// dx.doi.org/10.1016/j.biombioe.2005.12.020 kellogg, l. d., & bettinger, p. (1994). thinning productivity and cost for a mechanized cut-to-length system in the northwest pacific coast region of the usa. journal of forest engineering, 5, 43-54. kizha, a. r., & han, h.-s. (2016). processing and sorting forest residues: cost, productivity and managerial impacts. biomass and bioenergy, 93, 97-106. lapointe, j.-f., & robert, j.-m. (2000). using vr for efficient training of forestry machine operators. education and information technologies, 5(4), 237250. madgwick, h., & webber, b. (1987). nutrient removal in harvesting mature pinus radiata. new zealand journal of forestry, 32(3), 15-18. mani, s., sokhansanj, s., bi, x., & turhollow, a. (2006). economics of producing fuel pellets from biomass. applied engineering in agriculture, 22(3), 421-426. doi:https://doi.org/10.13031/2013.20447 miyata, e. s. (1980). determining fixed and operating costs of logging equipment. general technical report nc-55. st. paul, mn, usa: u.s. dept. of agriculture, forest service, north central forest experiment station. motulsky, h., & christopoulos, a. (2004). fitting models to biological data using linear and nonlinear regression: a practical guide to curve fitting. oxford, uk: oxford university press. nilsson, b., blom, å., & thörnqvist, t. (2013). the influence of two different handling methods on the moisture content and composition of logging residues. biomass & bioenergy, 52, 34-42. doi:https://doi. org/10.1016/j.biombioe.2013.02.026 nurmi, j. (2007). recovery of logging residues for energy from spruce (pices abies) dominated stands. biomass & bioenergy, 31(6), 375-380. doi:http:// dx.doi.org/10.1016/j.biombioe.2007.01.011 nurminen, t., korpunen, h., & uusitalo, j. (2006). time consumption analysis of the mechanized cut-tolength harvesting system. silva fennica, 40(2), 335-363. parikka, m. (2004). global biomass fuel resources. biomass and bioenergy, 27(6), 613-620. purfürst, f. t. (2010). learning curves of harvester operators. croatian journal of forest engineering: journal for theory and application of forestry engineering, 31(2), 89-97. räisänen, t., & nurmi, j. (2011). impacts of changing the minimum diameter of roundwood on the accumulation of logging residue in first thinnings of scots pine and norway spruce. biomass & bioenergy, 35(7), 2674-2682. doi:http://dx.doi. org/10.1016/j.biombioe.2011.03.002 rodriguez, l. c., may, b., herr, a., & o’connell, d. (2011). biomass assessment and small scale biomass fired electricity generation in the green triangle, australia. biomass & bioenergy, 35(7), 2589-2599. doi:http://dx.doi.org/10.1016/j. biombioe.2011.02.030 snowdon, p. (1991). a ratio estimator for bias correction in logarithmic regressions. canadian journal of forest research, 21, 720-724. strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 11 strandgard, m., mitchell, r., & acuna, m. (2017). time consumption and productivity of a forwarder operating on a slope in a cut-to-length harvest system in a pinus radiata d.don pine plantation. journal of forest science, 63, 324-330. strandgard, m., walsh, d., & acuna, m. a. (2013). estimating harvester productivity in pinus radiata plantations using stanford stem files. scandinavian journal of forest research, 28, 73-80. thiffault, e., béchard, a., paré, d., & allen, d. (2015). recovery rate of harvest residues for bioenergy in boreal and temperate forests: a review. wiley interdisciplinary reviews: energy and environment, 4(5), 429-451. doi:10.1002/wene.157 thiffault, e., hannam, k. d., paré, d., titus, b. d., hazlett, p. w., maynard, d. g., & brais, s. (2011). effects of forest biomass harvesting on soil productivity in boreal and temperate forests—a review. environmental reviews, 19(1), 278-309. tiernan, d., zeleke, g., owende, p. m. o., kanali, c. l., lyons, j., & ward, s. m. (2004). effect of working conditions on forwarder productivity in cutto-length timber harvesting on sensitive forest sites in ireland. biosystems engineering, 87(2), 167-177. doi:http://dx.doi.org/10.1016/j. biosystemseng.2003.11.009 walsh, d., & strandgard, m. (2014). productivity and cost of harvesting a stemwood biomass product from integrated cut-to-length harvest operations in australian pinus radiata plantations. biomass & bioenergy, 66, 93-102. walsh, d., strandgard, m., & carter, p. (2014). evaluation of the hitman ph330 acoustic assessment system for harvesters. scandinavian journal of forest research, 29(6), 593-602. doi:10.1080/02827581 .2014.953198 wolfsmayr, u. j., & rauch, p. (2014). the primary forest fuel supply chain: a literature review. biomass & bioenergy, 60, 203-221. doi:https://doi. org/10.1016/j.biombioe.2013.10.025 ximenes, f., ramos, j., bi, h., cameron, n., singh, b. p., & blasi, m. (2012). determining biomass in residues following harvest in pinus radiata forests in new south wales. canberra, australia: rural industries research and development corporation. yemshanov, d., mckenney, d. w., fraleigh, s., mcconkey, b., huffman, t., & smith, s. (2014). cost estimates of post harvest forest biomass supply for canada. biomass & bioenergy, 69, 80-94. doi:http://dx.doi. org/10.1016/j.biombioe.2014.07.002 strandgard & mitchell. new zealand journal of forestry science (2019) 49:9 page 12 spectral quality influence on in vitro morphophysiological responses of eucalyptus dunnii maiden and eucalyptus grandis w.hill ex maiden × e. urophylla s.t.blake samira rangel do prado frade1, denys matheus santana costa souza1, sérgio bruno fernandes1, maria lopes martins avelar1, letícia vaz molinari1, douglas santos gonçalves1, thiago alves magalhães2 and gilvano ebling brondani1,* 1 laboratory of in vitro culture of forest species, department of forestry sciences, federal university of lavras, 37200-900, lavras, mg, brazil 2 laboratory of plant anatomy, department of biology, federal university of lavras, 37200-900, lavras, mg, brazil *corresponding author: gilvano.brondani@ufla.br (received for publication 17 january 2022; accepted in revised form 5 december 2022) abstract background: in vitro growth and development of plants in the micropropagation stages are influenced by several factors, including the light spectral quality, which has shown important effects on the photomorphogenesis. the work aimed to evaluate the photomorphogenic effect of spectral qualities on in vitro culture of eucalyptus dunnii and eucalyptus grandis × e. urophylla. methods: six light spectral qualities (i.e., red, white, blue, yellow, purple, and green) on in vitro multiplication, elongation, and adventitious rooting stages were evaluated through analysis of variance followed by a tukey’s test. results: white spectral quality was most adequate for in vitro multiplication of eucalyptus dunnii and eucalyptus grandis × e. urophylla, as it resulted in less tissue oxidation, longer shoot length, and more buds per explant. red, blue and yellow spectral qualities increased the chlorophyll a, chlorophyll b, and total chlorophyll (a+b) leaf contents of eucalyptus dunnii. to promote in vitro elongation, white spectral quality was most suitable for eucalyptus dunnii, and yellow for eucalyptus grandis × e. urophylla, as these resulted in more shoot length and shoots per explant. red, white, blue and purple spectral qualities increased the stomatal density of eucalyptus dunnii; while the white and yellow were the better for eucalyptus grandis × e. urophylla. to promote in vitro rooting, the white and yellow spectral qualities caused the best results for the eucalyptus dunnii and eucalyptus grandis × e. urophylla, with longer root length and more roots per explant. eucalyptus dunnii showed reduced adventitious rooting, regardless of spectral quality. conclusions: light quality influence the morphophysiological responses of eucalyptus in different stages of in vitro culture. our results contribute to maximise the in vitro cloning of important eucalypts species. new zealand journal of forestry science rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 https://doi.org/10.33494/nzjfs532023x218x e-issn: 1179-5395 published on-line: 29/01/2023 research article open access to adverse environmental conditions and are resistant to biotic factors, thus ensuring greater productivity and fulfilment of the demand for more wood and other products (abiri et al. 2020). among vegetative propagation techniques, micropropagation has been considered a tool in the production of clonal plants on a commercial scale, introduction species of eucalyptus have a great economic importance due to their potential to provide feedstock, fuelwood and wood for industrial purposes, being grown in largescale commercial forests (carrillo et al. 2018). in brazil, eucalypts forests are mainly made up of clonal varieties bred from superior genotypes that are better adapted keywords: in vitro propagation; plant cloning; light spectrum; pigment content; photomorphogenesis. © the author(s) 2023. open access. this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. mailto:gilvano.brondani@ufla.br http://creativecommons.org/licenses/by/4.0/), rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 2 since it enables the mass multiplication of vegetative propagules in a short time and in a small area (molinari et al. 2021; silveira et al. 2020; ashrafzadeh & leung 2021). nevertheless, in vitro growth and development of different species in micropropagation stages are influenced by several factors (batista et al. 2018; miranda et al. 2020; faria et al. 2021; faria et al. 2022). one of the important factors for the morphogenesis of different plant tissues is the availability of light as an energy source for photoassimilation, carbon partitioning (kölling et al. 2015), phenology (nord & linch 2009) and other important processes, therefore, variations in wavelength trigger different morphophysiological responses. activation is dependent on genetic and epigenetic factors intrinsic to the species, as well as on their interaction for phenotypic expression (gilmartin et al., 1990; he et al. 2011), however, this activation begins with the stimulation of photoreceptors that signal metabolic pathways causing direct effects on tissue cell division and differentiation (gilmartin et al. 1990). studies of different light spectra on the in vitro propagation of eucalypts species, provide important information to large-scale production of clones with high phytosanitary quality and high vigor (souza et al. 2020a; souza et al. 2020b). eucalypt micropropagation under different light qualities has been extensively studied showing positive results in plant health, inducing the increase of shoots and leaves (souza et al. 2020a; miranda et al. 2020), as well as seed germination (rokich & bell 1995). besides, the use of cellophane plastic has shown to be a low-cost viable alternative as a filter for incident fluorescent light and providing the desired wavelength in micropropagation protocols (souza et al. 2020b). consequently, we hypothesise that: (i) clones of eucalyptus dunnii and eucalyptus grandis × e. urophylla would respond similarly to light spectra; and (ii) different light spectra would trigger large differences in morphophysiological features. in this context, the study evaluated the effect of spectral quality on in vitro multiplication, elongation and adventitious rooting stages for both clones. methods plant material and in vitro multiplication selected plants of eucalyptus dunnii maiden and eucalyptus grandis w. hill ex maiden × eucalyptus urophylla s. t. blake (known as urograndis eucalypt) were used to obtain the explants. the plants were donated by the institute of forestry research and studies (ipef, brazil). clusters of buds (i.e., explants), were previously in vitro established and maintained for 60 d in wood plant medium (wpm) (lloyd & mccown 1981), supplemented with 30 g l-1 of sucrose. clusters with four buds (i.e., explant) were subcultured in glass flasks of 250 ml (6.0 × 7.0 cm, diameter × height) containing 50 ml of wpm supplemented with 0.5 mg l-1 of 6-benzylaminopurine (bap), 0.05 mg l-1 of α-naphthaleneacetic acid (naa), and 20 g l-1 of sucrose. explants were kept in a growth room at 24°c (± 1°c) with a 16-h photoperiod using two 0.60-m cool-white philips t10 fluorescent light bulbs with 20 w power each and 6400-6500 k colour temperature. the irradiance of the two light bulbs combined (40 μmol m-2 s-1) was measured with a photoradiometer model qso-s procheck + sensor-par photon flux (decagon devices, pullman, washington, usa). sixty days after inoculation, percentage of tissue oxidation (souza et al. 2020b), shoot length per explant (cm), number of buds per explant, and photosynthetic pigment content [carotenoids, chlorophyll a, chlorophyll b, and total chlorophyll (a+b)] were evaluated. the experiment was conducted as a completely randomised design with a factorial arrangement (2 × 6), testing two clones (i.e., eucalyptus dunnii and eucalyptus grandis × e. urophylla), and six spectral qualities (i.e., red, blue, green, purple, yellow, and white control). five explants were cultured per glass flask (experimental units), totaling twenty repetitions. photosynthetic pigments analysis was performed randomly in three repetitions per treatment. in vitro elongation the in vitro elongation experiment followed the same protocols as the multiplication experiment, only differing on the number of explants per glass flask (experimental units), being three in the former and five in the latter. the in vitro elongation experiment had twelve repetitions per treatment. shoots measuring 1-cm in length (i.e., explant), from the multiplication stage, were inoculated in glass flasks of 250 ml (6.0 × 7.0 cm) containing 50 ml of wpm culture medium supplemented with 0.5 mg l-1 naa, 0.05 mg l-1 bap, and 20 g l-1 of sucrose. thirty-five days after inoculation, tissue oxidation percentage, shoot length per explant (cm), number of shoots per explant, and leaf anatomical features [i.e., adaxial and abaxial epidermal thickness (µm), spongy parenchyma thickness (µm), palisade parenchyma thickness (µm), and stomatal density (mm2)] were evaluated. histological analysis was performed randomly in three repetitions per treatment. in vitro adventitious rooting microcuttings (i.e., elongated shoots 2-cm-long) were inoculated in glass flasks of 250 ml (6.0 × 7.0 cm) containing 50 ml of wpm culture medium supplemented with 1.0 mg l-1 naa, 0.5 mg l-1 of indole-3-butyric acid (iba), 0.05 mg l-1 bap, and 20 g l-1 of glucose and kept in a growth room. thirty-five days after inoculation, rooting percentage, root length per explant (cm), and number of roots per explant were evaluated. twenty explants considered elongated (> 3.0 cm) were selected and sectioned and kept per glass flask (experimental units), totaling twenty repetitions, and the experimental design was the same as in previous experiments (i.e., in vitro multiplication and elongation). culture medium preparation the culture media used in all experiments were prepared with deionised water, adding 6 g l-1 of agar, and the ph was adjusted to 5.80 ± 0.05 with naoh (0.1 m) or hcl (0.1 m) before autoclaving at 121°c (1.0 kgf cm-2) for 20 min. spectral quality spectral qualities red, blue, yellow, purple and green were provided, under all growth stages, by filtering the light output of the fluorescent light bulbs through double sheets of cellophane (souza et al. 2020a) replaced every cycle. the light spectra were measured using a spectra pen z850 portable spectrophotometer (qubit systems, kingston, ontario, usa). the spectra of each light source used in this experiment are shown in figures 1a-f. analysis of photosynthetic pigments contents of photosynthetic pigments [carotenoids, chlorophyll a, chlorophyll b, and total chlorophyll (a+b)] were evaluated in the in vitro multiplication stage. the extraction was performed according to the method adapted from lichtenthaler (1987). in a partially dark environment, 25 mg of leaf tissue was randomly collected in biological triplicate from explants from different flasks. a volume of 2.5 ml of dimethyl sulfoxide (dmso) solution was pipetted into each sample and stored in the dark for 48 h. the absorbance (a) of the pigment extracted from the leaves was read at 480 nm, 649 nm, and 665 nm in a spectrophotometer (thermoscientific, rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 3 table 1: description of the study sites figure 1: variations in the wavelength (nm) emitted by the fluorescent light bulbs and filtered through double sheets of cellophane (μw cm-2 nm-1). a = red; b = blue; c = green; d = purple; e = yellow; f = white (no cellophane control). usa). readings were performed in triplicate for each treatment. chlorophyll and carotenoid contents were calculated following the equations described by wellburn (1994): chlorophyll a = [(12.19 × a665) – (3.45 × a649)]; chlorophyll b = [(21.99 × a649) – (5.32 × a665)]; and carotenoids = (1000 × a480 – 2.14 × ca – 70.16 × cb)/220. the results were expressed in µg of pigment per mg of leaf tissue fresh weight (µg mg-1). histological analysis anatomical variables adaxial and abaxial epidermal thickness, palisade parenchyma thickness, spongy parenchyma thickness, and stomatal density were evaluated. leaf samples, randomly collected from different plants in the same flask, in biological triplicate, were fixed in 70% acetic acid formaldehyde (aaf) solution for 48 h and then kept in 70% ethanol. the samples were dehydrated using an increasing ethyl alcohol series (80, 90, and 100%) for 30 min in each solution, and then stored in a 100% ethanol:historesin (1:1) (leica®) solution in a hot oven overnight. after this, the samples were embedded in pure hydroxyethyl methacrylate resin (leica®). using a manual rotary microtome and a razor, the blocks containing the samples were cross-sectioned at 7-µm thickness. the sections were stained with toluidine blue solution, mounted on histological slides, and coated with stained-glass varnish. to obtain paradermal sections, the leaf samples were placed in a solution of sodium hypochlorite and distilled water (1:1) for 24 h so that tissue dissociation could occur. then the tissues of the abaxial side were removed, stained with 1% safranin solution, and mounted on histological slides. the histological slides of the crosssections and paradermal sections were observed under a zeiss® light microscope and photomicrographed with an axioncam erc5s digital camera at micrometric scale using 20× and 40× objective lenses. the thickness of the adaxial and abaxial epidermis, palisade parenchyma, spongy parenchyma, and total leaf blade tissues were measured at micrometric scale using the programs axiovision version 4.8 and imagej at three different points of three photomicrographs from each treatment, the means of which were taken for analysis. the thickness of the midrib was also measured in three photomicrographs from each treatment to obtain a final mean. the thickness of the polar and equatorial diameter of stomata was measured in the five most homogeneous stomata in three photomicrographs from each treatment, from which the mean was calculated. the stomatal density was calculated as the ratio between the mean number of stomata and the mean area of the image present in three photomicrographs from each treatment. experimental design and data analysis statistical analyses were performed in r software, version 3.0.3 (r core team 2018), using the expdes package, version 1.1.2 (ferreira et al. 2013). the variables that did not show a normal distribution according to the shapiro-wilk’s test (p > 0.05) were arcsine-transformed. heteroscedasticity was evaluated using bartlett’s test (p > 0.05). response variables to the treatments were rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 4 subjected to analysis of variance (anova, p < 0.05) followed by tukey’s test (p < 0.05). results effect of spectral quality on in vitro multiplication in the in vitro multiplication stage, there were significant effects of clones and spectral quality on morphophysiological features, 60 d after starting the in vitro culture (p < 0.05) (figures 2a-e). the results indicated generally high percentages of phenolic oxidation of tissues across all treatments, although this did not prevent the development of explants of eucalyptus dunnii (52.5% oxidation) or eucalyptus grandis × e. urophylla (56.6% oxidation) (figure 2a). the spectral qualities used had different effects on explant oxidation (there was not an interaction between clone and spectral quality), and the lowest tissue oxidation was observed with the white spectral quality (30.0%), which differed significantly from the other treatments (figure 2b). for shoot length, explants of eucalyptus grandis × e. urophylla (1.1 cm) (figure 2c) and explants from the white (1.2 cm) and blue (1.3 cm) spectral quality groups (figure 2d) had the highest means, which differed significantly from the other groups. explants of the eucalyptus grandis × e. urophylla clone showed a higher number of buds than eucalyptus dunnii under all the spectral qualities evaluated (figure 2e). regarding the best results obtained with the different spectral qualities, the white spectrum resulted in the most buds per explant (11.2 buds) for eucalyptus dunnii, differing significantly from the green light; while the blue for eucalyptus grandis × e. urophylla (25.7 buds), differing significantly from the red light (figure 2e). the appearance of the explants of eucalyptus dunnii and eucalyptus grandis × e. urophylla with regard to the morphological traits studied during their in vitro multiplication is shown in figure 3. regarding the levels of the photosynthetic pigments studied, there was a difference in the response between eucalyptus dunnii and the urograndis eucalypt with spectral qualities during the in vitro multiplication stage (figures 4a-e). only main effects were significant for carotenoid content, while for chlorophyll a, b, and total (a+b) there was clone and spectral quality interaction. the highest carotenoid content was observed in eucalyptus dunnii (0.24 µg mg-1), which differed significantly from eucalyptus grandis × e. urophylla (0.11 µg mg-1) (figure 4a). the blue and yellow spectral qualities resulted in the highest carotenoid content (0.20 µg mg-1), but there was no significant difference from the other treatments (figure 4b). for the chlorophyll a (figure 4c), chlorophyll b (figure 4d), and total chlorophyll (a+b) (figure 4e) contents, the eucalyptus dunnii also showed higher levels than eucalyptus grandis × e. urophylla. regarding the spectral qualities analysed, red, blue, and yellow resulted in the highest chlorophyll (a+b) content for eucalyptus dunnii (figure 4e), while for eucalyptus grandis × e. urophylla, no difference was observed between the light sources (figures 4c, 4d, and 4e). rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 5 figure 2: morphological features measured during the in vitro multiplication stage of eucalyptus dunnii and eucalyptus grandis × e. urophylla under different spectral qualities [i.e., red (re), white (wh), blue (bl), yellow (ye), purple (pu), and green (gr)] at 60 d. (a) oxidation percentage according to clone (n = 240, p-value = 0.512); (b) = oxidation percentage according to spectral quality (n = 240, p-value = 0.014); (c) shoot length per explant (cm) according to clone (n = 240, p-value = 0.001); (d) shoot length (cm) according to spectral quality (n = 240, p-value = 0.001); and (e) number of buds per explant (n = 240, p-value = 0.001). (a-d) means followed by the same letters do not differ significantly at p < 0.05. (e) means followed by the different uppercase letters represent significant differences when comparing different clones given the same spectral quality, and different lowercase letters represent significant differences when comparing spectral quality within the same clone at p < 0.05. error bars represent the standard error of the mean. effect of spectral quality on in vitro elongation under the experimental conditions used during the in vitro elongation stage, only the main effect of clone (figure 5a) but not light spectra (figure 5b) or their interaction were significant on tissue oxidation. however, the factors had different effects on shoot length (figure 5c) and number of shoots per explant (figure 5d). the oxidation percentage differed between eucalyptus dunnii and the urograndis eucalypt, being the lowest values observed in eucalyptus dunnii (38.0%) (figure 5a). there was no significant difference in oxidation percentage between the different spectral qualities (figure 5b). eucalyptus dunnii had the greatest shoot length when grown under white light (3.1 cm), differing significantly from the yellow, purple, and green groups (figure 5c). in the eucalyptus grandis × e. urophylla, the yellow spectral quality resulted in the highest mean shoot length (3.2 cm), significantly longer than that under the white, blue, and green lights (figure 5c). the highest number of shoots per explant (9.4 shoots) was observed in the white light for eucalyptus dunnii, only differing with the blue light (figure 5d). in contrast, in the eucalyptus grandis × e. urophylla clone, the best results were observed with the green (16.3 shoots), blue (13.6 shoots), and yellow (13.4 shoots) lights (figure 5d). the appearance of the explants of eucalyptus dunnii and eucalyptus grandis × e. urophylla with regard to morphological traits during in vitro elongation is shown in figure 6. the two clones showed statistical difference when analysing the anatomical features in the histological sections, in response to different spectral qualities during the elongation stage (figures 7a-h). the highest means for the adaxial (figure 7a) and abaxial (figure 7c) epidermal thickness and spongy parenchyma thickness (figure 7e) were observed for the eucalyptus dunnii clone (11.98, 10.69, and 64.63 µm, respectively), which differed significantly from the means in eucalyptus grandis × e. urophylla clone (11.19, 9.46, and 51.75 µm, respectively). there was no significant difference in the variables between the spectral qualities (figures 7b, 7d, and 7f). the palisade parenchyma thickness (figure 7g) and stomatal density (figure 7h) were affected by the interaction between the factors (clones and spectral quality). yellow light resulted in the thickest palisade parenchyma in eucalyptus dunnii (41.65 µm), differing significantly from the red and green qualities (figure 7g). in contrast, there was no significant difference in the palisade parenchyma thickness of the eucalyptus grandis × e. urophylla clone among light spectra (figure 7g). exposure to red, white, blue, and purple light resulted in the highest stomatal densities in eucalyptus dunnii, significantly differing from those grown under the yellow and green qualities (figure 7h). however, the highest stomatal density in eucalyptus grandis × e. urophylla was observed under white and yellow light, which differed significantly from red, blue, purple, and green (figure 7h). the stomatal density on the abaxial surface of leaves of eucalyptus dunnii and eucalyptus grandis × e. urophylla explants is shown in figures 8a-l. effect of spectral quality on in vitro adventitious rooting there was no interaction between clones and spectral quality on the in vitro adventitious rooting stage of the eucalyptus dunnii or eucalyptus grandis × e. urophylla clones after 35 d (figures 9a-f). the rooting percentage (figure 9a), root length (figure 9c), and number of roots per explant (figure 9e) were all influenced by clones. it was observed that 98.0% of e. grandis x e. urophylla clone exhibited roots (average of 2.5 roots per explant measuring 4.0 cm), while e. dunnii presented 3.0% of explant with roots (average of 0.03 roots per explant measuring 0.20 cm). regarding the effect of spectral quality on the adventitious rooting percentage (figure 9b) and number of roots per explant (figure 9f), the highest values were observed under the yellow light (53.0% of rooting and 1.5 roots per explant), but did not differ significantly from those under the other light sources. regarding root length (figure 9d), the use of white (2.4 cm) and yellow (2.7 cm) spectral qualities provided the highest rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 6 figure 3: visual characterisation of eucalyptus dunnii (1) and eucalyptus grandis × e. urophylla (2) explants on in vitro multiplication stage at 60 d according to spectral quality (i.e., red, white, blue, yellow, purple, and green). rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 7 figure 4: photosynthetic pigment contents measured after 60 d during the in vitro multiplication stage of eucalyptus dunnii and eucalyptus grandis × e. urophylla under different spectral qualities [i.e., red (re), white (wh), blue (bl), yellow (ye), purple (pu), and green (gr)]. (a) carotenoid content according to clone (n = 36, p-value = 0.001); (b) carotenoid content according to spectral quality (n = 36, p-value = 0.222); (c) chlorophyll a content (n = 36, p-value = 0.006); (d) chlorophyll b content (n = 36, p-value = 0.007); and (e) total chlorophyll (a+b) content (n = 36, p-value = 0.001). (a-b) means followed by the same letters do not differ significantly by tukey’s test at p < 0.05. (c-e) means followed by the different uppercase letters represent significant differences when comparing different clones given the same spectral quality, and different lowercase letters represent significant differences when comparing spectral quality within the same clone at p < 0.05. error bars represent the standard error of the mean. rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 8 figure 5: morphological features measured after 35 d during the in vitro elongation stage of eucalyptus dunnii and eucalyptus grandis × e. urophylla under different spectral qualities [i.e., red (re), white (wh), blue (bl), yellow (ye), purple (pu), and green (gr)]. (a) oxidation percentage according to clone (n = 144, p-value = 0.001); (b) oxidation percentage according to spectral quality (n = 144, p-value = 0.329); (c) shoot length per explant (cm) (n = 144, p-value = 0.003); and (d) number of shoots per explant (n = 144, p-value = 0.001). (a-b) means followed by the same letters do not differ significantly at p < 0.05. (c-d) means followed by the different uppercase letters represent significant differences when comparing different clones given the same spectral quality, and different lowercase letters represent significant differences when comparing spectral quality within the same clone at p < 0.05. error bars represent the standard error of the mean. figure 6: visual characterization of eucalyptus dunnii (1) and eucalyptus grandis × e. urophylla (2) explants on in vitro elongation stage at 35 d according to spectral quality (i.e., red, white, blue, yellow, purple, and green). rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 9 figure 7: anatomical features measured after 35 d during the in vitro elongation stage of eucalyptus dunnii and eucalyptus grandis × e. urophylla under different spectral qualities [i.e., red (re), white (wh), blue (bl), yellow (ye), purple (pu), and green (gr)]. (a) adaxial epidermis thickness according to clone (n = 36, p-value = 0.001); (b) adaxial epidermis thickness according to spectral quality (n = 36, p-value = 0.058); (c) abaxial epidermis thickness according to species (n = 36, p-value = 0.001); (d) abaxial epidermis thickness according to spectral quality (n = 36, p-value = 0.061); (e) spongy parenchyma thickness according to species (n = 36, p-value = 0.001); (f) spongy parenchyma thickness according to spectral quality (n = 36, p-value = 0.070); (g) palisade parenchyma thickness according to clone and spectral quality (n = 36, p-value = 0.001). (h) stomatal density according to species and spectral quality (n = 36, p-value = 0.001). (a-f) means followed by the same letters do not differ significantly by tukey’s test at p < 0.05. (g-h) means followed by the different uppercase letters represent significant differences when comparing different species given the same spectral quality, and different lowercase letters represent significant differences when comparing spectral quality within the same clone, by tukey’s test at p < 0.05. error bars represent the standard error of the mean. rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 10 mean, differing significantly from the red and blue lights (figure 9d). thus, methods that aim to maximise rooting may become important strategies to be adopted in propagation systems. the appearance of rooted plants of eucalyptus grandis × e. urophylla grown under the different light spectra and of eucalyptus dunnii under the red spectral quality is shown in figure 10. discussion effect of spectral quality on in vitro multiplication improving the micropropagation technique through the use for specific spectral qualities was studied in an attempt to establish efficient production systems of clonal plants for two eucalypts. considering the established hypotheses, the results showed that clones generally responded differently to different wavelengths. the light source used in the in vitro multiplication stage influenced the oxidation percentage, shoot length and number of buds per explant of both eucalyptus dunnii and eucalyptus grandis × e. urophylla clones, although with varying magnitudes (figures 2a-e). lowest percentage of phenolic oxidation was observed under the white spectral quality for both clones (figure 2b). in contrast, the use of white spectral quality resulted in greater oxidation in explants of corymbia torelliana × c. citriodora (souza et al. 2018). tissue oxidation is a problem with the micropropagation of woody species, as reported in other studies (oliveira et al. 2016; souza et al. 2019; souza et al. 2020b; faria et al. 2021; faria et al. 2022; souza et al. 2022). these results may be linked to internal environmental factors that affect the vigor of explants, where smaller flasks tend to have reduced carbon dioxide concentrations and high ethylene concentrations, and may also be affected by light irradiation, air temperature, and relative humidity (tisarum et al. 2018; chen et al. 2019; souza et al. 2021). the best result observed in eucalyptus dunnii and eucalyptus grandis × e. urophylla clones for shoot length was found under white and blue lights (figure 2d). due to the specificity of the wavelength, monochromatic light sources influence the photomorphogenic responses of explants grown in controlled environments, so they have emerged as a way to increase yield (batista et al. 2018; faria et al. 2019; souza et al. 2022). results described in the literature corroborate those found in this study, with spectral peaks of 450 nm in populus euramericana (kwon et al. 2015), acer saccharum (singh et al. 2017), corymbia citriodora × c. torelliana (souza et al. 2018), eucalyptus grandis × e. urophylla (souza et al. 2022) and zingiber officinale (gnasekaran et al. 2021) inducing more buds and greater shoot growth. for the number of buds per explant, the white spectral quality provided the best results in the eucalyptus dunnii clone, differing significantly from the green light (figure 2e). many studies have shown the superiority of white light sources for in vitro plant development (batista et al. 2018; abiri et al. 2020), although responses vary. genotypic differences in the in vitro development have also been reported in different clones of eucalyptus, such as eucalyptus grandis × e. urophylla (souza et al. 2020a; miranda et al. 2020), eucalyptus benthamii (brondani et al. 2018), and eucalyptus globulus (oliveira et al. 2016). thus, different results are observed for explants in the in vitro multiplication stage, the response varying according to the genotype and the culture conditions (e.g., spectral quality). eucalyptus dunnii had the highest levels of the photosynthetic pigments analysed [carotenoids, chlorophyll a and b, and total chlorophyll (a+b)] (figures 4a-e). the biosynthesis of carotenoids and chlorophyll a and b is influenced by the clones and cultivation conditions, and certain conditions may result in greater photosynthetic efficiency (gupta & karmakar 2017; oliveira et al. 2021). in addition, the effect of different wavelengths on the biosynthesis of photosynthetic pigments may depend on the tissues of the examined plants (tisarum et al. 2018; jung et al. 2021). the red, blue and yellow spectral qualities resulted in higher chlorophyll (a+b) content in eucalyptus dunnii, and no variation was observed for urograndis eucalypt (figure 4e). the primary photosynthetic pigments in higher plants, carotenoids and chlorophyll a and b, absorb figure 8: abaxial leaf epidermis of eucalyptus dunnii (1) and eucalyptus grandis × e. urophylla (2) explants after 35 d during the in vitro elongation stage according to spectral quality (i.e., red, white, blue, yellow, purple, and green). bar = 50.0 µm. rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 11 figure 9: morphological features measured on in vitro adventitious rooting of eucalyptus dunnii and eucalyptus grandis × e. urophylla under different spectral qualities [i.e., red (re), white (wh), blue (bl), yellow (ye), purple (pu), and green (gr)] at 35 d. (a) rooting percentage according to species (n = 240, p-value = 0.001); (b) rooting percentage according to spectral quality (n = 240, p-value = 0.067); (c) root length (cm) according to species (n = 240, p-value = 0.001); (d) root length (cm) according to spectral quality (n = 240, p-value = 0.001); (e) number of roots per explant according to species (n = 240, p-value = 0.001); number of roots per explant according to spectral quality (n = 240, p-value = 0.074). means followed by the same letters do not differ significantly by tukey’s test at p < 0.05. error bars represent the standard error of the mean. certain wavelengths, especially blue and red, playing key roles in morphology control (gupta & karmakar 2017; oliveira et al. 2021). the light absorption at specific wavelengths can promote a higher-energy state (faria et al. 2019; abiri et al. 2020). this excitation energy of the chlorophyll molecule can be used in the photochemical step, be lost in the form of heat, or cause damage to the photosynthetic apparatus, such as in the formation of blue free radicals (abiri et al. 2020). in gerbera jamesonii, higher chlorophyll a and b contents have been observed under peak wavelengths of 450 and 600 nm (pawłowska et al. 2018), and in stevia rebaudiana, the amount of carotenoids in shoots exposed to wavelengths of 450 and 600 nm were reportedly higher than those in shoots exposed to 450 nm (blue led) (ramírez-mosqueda et al. 2017). therefore, knowledge of the relationship between wavelengths and plant growth patterns in micropropagation can lead to a better understanding of the clones studied. effect of spectral quality on in vitro elongation the effects of specific wavelengths on each clone are reflected in the photomorphological responses of propagules grown in controlled environments. this technology is emerging as a useful way to optimise shoot development and yield. wavelength was found to be an important factor on in vitro elongation, having direct implications on oxidation, shoot length, and the number of shoots per explant in eucalyptus dunnii and eucalyptus grandis × e. urophylla (figures 5a-d). the control of spectral quality is essential for plant cultivation, as it can optimise the photosynthetic rate and shoot development (silva et al. 2017). such optimisation was observed with the white light for eucalyptus dunnii and the yellow light for eucalyptus grandis × e. urophylla (figure 5c). our findings are in line with the literature. greater absorption of a broad light spectrum (450-600 nm) has induced a greater number of shoots and higher shoot development in eucalyptus urophylla (miranda et al. 2020) and populus euramericana (kwon et al. 2015). the use of white or yellow light in crops can increase plant growth and yield, as it allows light to better penetrate the leaves, maximising the photosynthetic rate when compared to the use of blue and red monochromatic lights, as observed in lippia rotundifolia (hsie et al. 2019). these studies show that plants need a broad light spectrum to optimise their photosynthetic processes, although such requirements vary by species. on the analysed anatomical features, there was an influence of wavelength and clone after 35 d during the in vitro culture (figures 7a-h). the highest values found for adaxial (figure 7a) and abaxial (figure 7c) epidermis thickness, spongy parenchyma thickness (figure 7e), and palisade parenchyma thickness (figure 7g) were observed in eucalyptus dunnii. however, the highest stomatal density was observed in the eucalyptus grandis × e. urophylla (figure 7h). the morphological, anatomical, and physiological responses of plants vary by genotype, which are translated into changes in growth and development to adapt to changes in environmental conditions (cioć et al. 2019; miranda et al. 2020; souza et al. 2020a; nery et al. 2021). among the spectral qualities used, there was a significant interaction effect only on the palisade parenchyma thickness (figure 7g) and stomatal density (figure 7h). the use of yellow light resulted in the highest palisade parenchyma thickness in eucalyptus dunnii, differing from the red and green lights (figure 7g). similar results, using a light source with a broad light spectrum, have been observed in ajuga genevensis (sahakyan et al. 2016), abies × borisii-regis (smirnakou et al. 2017) and polygala paniculata (nery et al. 2021). the adverse effect of red leds on plant development has been reported in curculigo orchioides (gupta & sahoo 2015). kwon et al. (2015) observed a greater number of dividing parenchyma cells in populus euramericana when subjected to treatment with red/blue light, resulting in greater shoot and leaf development. in addition, macedo rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 12 figure 10: in vitro rooted explants of eucalyptus grandis × e. urophylla (left) under different spectral quality and eucalyptus dunnii (right) under red light at 35 d. bar = 1.0 cm. et al. (2011) reported that the blue wavelength (450 nm) led to an increase in epidermal, mesophyll, and palisade parenchymal cell thickness in the in vitro culture of alternanthera brasiliana. regarding the stomatal density in the analysed clones (figure 7h), it was found that the red, white, blue, and purple spectral qualities provided the best results for eucalyptus dunnii; and the white and yellow for eucalyptus grandis × e. urophylla. li et al. (2018) observed that yellow and blue leds were more advantageous than green leds for the morphology and anatomy of solanum tuberosum plants. in plants of capsicum chinense grown in pots under red light, thicker and differentiated epidermal and parenchymal cells were found (santanabuzzy et al. 2005). the wavelength of light influences the growth and development of plant cells, tissues, and organs, triggering different morphological and anatomical responses (cioć et al. 2019; miranda et al. 2020; nery et al. 2021). changes in the structural organisation of leaf cells can be observed at the anatomical level when plants are exposed to different environments, including different light spectra. these results are similar to those found by souza et al. (2020b) in eucalyptus grandis × e. urophylla explants under a broad light spectrum. each spectrum can alter the expression levels of specific genes involved in the structural organisation of plant cells and in photosynthetic performance (cioć et al. 2019; silveira et al. 2020). in addition, studies show that exposure to red and green light leads to low stomatal conductance and that this is an ecological adaptation to avoid excessive leaf water loss (aasamaa & aphalo 2016). our morphological and anatomical results have important implications for the optimisation of the production of clonal plants through the micropropagation technique. the broad light spectrum, through the yellow and white light qualities, was found to be suitable for the in vitro elongation of explants in controlled environments. effect of spectral quality on in vitro adventitious rooting eucalyptus grandis × e. urophylla clone showed the best results of adventitious rooting (figures 9a, 9c, and 9e). adventitious rooting of eucalyptus dunnii is a challenge in clonal propagation, varying from 3.0 to 46.5% (brondani et al. 2011; oberschelp et al. 2015; souza et al. 2019). in contrast, the eucalyptus grandis × e. urophylla has shown 80.0 to 100.0% (gallo et al. 2017; souza et al. 2020b; miranda et al. 2020). some species, such as hevea brasiliensis are considered difficult-toroot because of the presence of an almost continuous cylinder of lignified tissue that, together with chemical barriers, hinder root emission (almeida et al. 2017). under the different spectral qualities analysed, there was a significant difference between treatments only on the root length (figure 9d), with the best results observed under the white and yellow lights. in abies × borisii-regis explants, the best root development results were observed using a white light source (smirnakou et al. 2017). in contrast, the rooting and survival percentages and the number of roots in corymbia citriodora × c. torelliana and corymbia torelliana × c. citriodora microcuttings were greatest when using the red/blue spectral quality (souza et al. 2018). although light quality is an important factor for the in vitro culture, there are few studies on its effect in woody species. the absence of induction of adventitious roots is one of the main causes limiting cloning by the micropropagation technique (almeida et al. 2017; souza et al. 2022). thus, because the efficiency of cultivation conditions in plant production varies between species, a broad light spectrum is often needed. complementary studies are recommended to maximise the in vitro rooting of the eucalyptus dunnii clone using other spectral qualities. conclusions light spectra triggered large and differential responses in morphophysiological features of eucalyptus dunnii and eucalyptus grandis × e. urophylla clones. white spectral quality is most suitable for in vitro multiplication of eucalyptus dunnii and eucalyptus grandis × e. urophylla. white spectral quality is also the most suitable for in vitro elongation of eucalyptus dunnii, and the yellow for eucalyptus grandis × e. urophylla. white and yellow spectral qualities are the most suitable for in vitro adventitious rooting of the eucalyptus dunnii and eucalyptus grandis × e. urophylla clones. competing interests the authors declare that they have no competing interests. acknowledgements we thank the national council for scientific and technological development, brazil (“conselho nacional de desenvolvimento científico e tecnológico – cnpq”); coordination for improvement of higher education personnel, brazil (“coordenação de aperfeiçoamento de pessoal de nível superior – capes – código de financiamento 001”); foundation for research of the state of minas gerais, brazil (“fundação de amparo à pesquisa do estado de minas gerais – fapemig”); and institute of forestry research and studies, brazil (“instituto de pesquisa e estudos florestais – ipef”). abbreviations wpm wood plant medium (lloyd and mccown 1980) bap 6-benzylaminopurine naa α-naphthaleneacetic acid iba indole-3-butyric acid dmso dimethyl sulfoxide led light-emitting diode aaf acetic acid formaldehyde rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 13 authors’ contributions srpf was the primary author, conducted the experiments, analyses, review and discussion. dmscs, sbf, mlma, lvm geb and dsg contributed to the analyses, review and discussion; geb also provided supervision. tam contributed to the review and discussion. all authors contributed to writing the manuscript. references aasamaa, k., & aphalo, p.j. (2016). effect of vegetation shade and its components on stomatal responses to red, blue and green light in two deciduous tree species with different shade tolerance. environmental and experimental botany, 121, 94-101. https://doi.org/10.1016/j. envexpbot.2015.01.004 abiri, r., atabaki, n., abdul-hamid, h., sanusi, r., shukor, n.a.a., shaharuddin, n.a., ahmad, s.a., & malik, s. (2020). the prospect of physiological events associated with the micropropagation of eucalyptus sp. forests, 11(11), 1211. https://doi. org/10.3390/f11111211 almeida, m.r., aumond, jr.m., costa, c.t., schwambach, j., ruedell, c.m., correa, l.r., & fett-neto, a.g. (2017). environmental control of adventitious rooting in eucalyptus and populus cuttings. trees, 31, 13771390. https://doi.org/10.1007/s00468-0171550-6 ashrafzadeh, s., & leung, d.w.m. (2021). photoautotrophic micropropagation system (pam): a novel approach deserving increased uptake for commercial plant production. vegetos, 34, 13-18. https://doi. org/10.1007/s42535-020-00182-x batista, d.s., felipe, s.h.s., silva, t.d., castro, k.m., mamedes-rodrigues, t.c., miranda, n.a., ríos-ríos, a.m., faria, d.v., fortini, e.a., chagas, k., torressilva, g., xavier, a., arencibia, a.d., & otoni, w.c. (2018). light quality in plant tissue culture: does it matter? in vitro cellular & developmental biology – plant, 54, 195-215. https://doi.org/10.1007/ s11627-018-9902-5 brondani, g.e., dutra, l.f., wendling, i., grossi, f., hansel, f.a., & araujo, m.a. (2011). micropropagation of an eucalyptus hybrid (eucalyptus benthamii × eucalyptus dunnii). acta scientiarum. agronomy, 33(4), 655-663. https://doi.org/10.4025/ actasciagron.v33i4.8317 brondani, g.e., oliveira, l.s., konzen, e.r., silva, a.l.l., & costa, j.l. (2018). mini-incubators improve the adventitious rooting performance of corymbia and eucalyptus microcuttings according to the environment in which they are conditioned. anais da academia brasileira de ciências, 90(2; suppl. 1), 2409-2423. https://doi.org/10.1590/00013765201720170284 carrillo, i., mendonça, r.t., ago, m., & rojas, o.j. (2018). comparative study of cellulosic components isolated from different eucalyptus species. cellulose, 25, 1011-1029. https://doi.org/10.1007/s10570018-1653-2 chen, y.-m., huang, j.-z., hou, t.-w., & pan, i.-c. (2019). effects of light intensity and plant growth regulators on callus proliferation and shoot regeneration in the ornamental succulent haworthia. botanical studies, 60, 10. https://doi.org/10.1186/s40529019-0257-y cioć, m., kalisz, a., żupnik, m., & pawłowska, b. (2019). different led light intensities and 6-benzyladenine concentrations in relation to shoot development, leaf architecture, and photosynthetic pigments of gerbera jamesonii bolus in vitro. agronomy, 9(7), 358. https://doi.org/10.3390/agronomy9070358 faria, d.v., correia, l.n.f., souza, m.v.c., ríos-ríos, a.m., vital, c.e., batista, d.s., costa, m.g.c., & otoni, w.c. (2019). irradiance and light quality affect two annatto (bixa orellana l.) cultivars with contrasting bixin production. journal of photochemistry and photobiology b: biology, 197, 111549. https://doi. org/10.1016/j.jphotobiol.2019.111549 faria, j.c.t., ribeiro-kumara, c., costa, r.s.r., nieri, e.m., carvalho, d., pinto, j.e.b.p., sena neto, a.r., & brondani, g.e. (2022). use of biodegradable polyester-based microvessels for micropropagation of mature eucalyptus microcorys. new zealand journal of forestry science, 52, 10. https://doi. org/10.33494/nzjfs522022x139x faria, j.c.t., terra, j.a.p., molinari, l.v., delarmelina, w.m., ribeiro-kumara, c., sena neto, a.r., carvalho, d., & brondani, g.e. (2021). use of polylactic acid microvessel to obtain microplantlets of eucalyptus microcorys through indirect organogenesis. 3 biotech, 11, 364. https://doi.org/10.1007/s13205021-02822-8 ferreira, e.b., cavalcanti, p.p., & nogueira, d.a. (2013). expdes: experimental designs package. r package version 1.1.2. gallo, r., xavier, a., oliveira, b.a., moura, l.c., miranda, n.a., gatti, k.c., & otoni, w.c. (2017). micro-cutting propagation of eucalyptus grandis × eucalyptus urophylla through clumps of axillary buds using different containers and substrates. australian journal of crop science, 11(10), 1327-1333. https://doi.org/10.21475/ajcs.17.11.10.pne607 gilmartin, p.m., sarokin, l., memelink, j., & chua, n.h. (1990). molecular light switches for plant genes. the plant cell, 2(5), 369-378. https://doi. org/10.1105/tpc.2.5.369 gnasekaran, p., rahman, z.a., chew, b.l., appalasamy, s., mariappan, v., & subramaniam, s. (2021). development of micropropagation system of rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 14 https://doi.org/10.1016/j.envexpbot.2015.01.004 https://doi.org/10.1016/j.envexpbot.2015.01.004 https://doi.org/10.3390/f11111211 https://doi.org/10.3390/f11111211 https://doi.org/10.1007/s00468-017-1550-6 https://doi.org/10.1007/s00468-017-1550-6 https://doi.org/10.1007/s42535-020-00182-x https://doi.org/10.1007/s42535-020-00182-x https://doi.org/10.1007/s11627-018-9902-5 https://doi.org/10.1007/s11627-018-9902-5 https://doi.org/10.4025/actasciagron.v33i4.8317 https://doi.org/10.4025/actasciagron.v33i4.8317 https://doi.org/10.1590/0001-3765201720170284 https://doi.org/10.1590/0001-3765201720170284 https://doi.org/10.1007/s10570-018-1653-2 https://doi.org/10.1007/s10570-018-1653-2 https://doi.org/10.1186/s40529-019-0257-y https://doi.org/10.1186/s40529-019-0257-y https://doi.org/10.3390/agronomy9070358 https://doi.org/10.1016/j.jphotobiol.2019.111549 https://doi.org/10.1016/j.jphotobiol.2019.111549 https://doi.org/10.33494/nzjfs522022x139x https://doi.org/10.33494/nzjfs522022x139x https://doi.org/10.1007/s13205-021-02822-8 https://doi.org/10.1007/s13205-021-02822-8 https://doi.org/10.21475/ajcs.17.11.10.pne607 https://doi.org/10.1105/tpc.2.5.369 https://doi.org/10.1105/tpc.2.5.369 zingiber officinale var. rubrum theilade using different spectrum light-emitting diode (led) irradiation. industrial crops and products, 170, 113748. https://doi.org/10.1016/j. indcrop.2021.113748 gupta, s.d., & karmakar, a. (2017). machine vision based evaluation of impact of light emitting diodes (leds) on shoot regeneration and the effect of spectral quality on phenolic content and antioxidant capacity in swertia chirata. journal of photochemistry and photobiology b: biology, 174, 162-172. https://doi.org/10.1016/j. jphotochem.2017.09.033 gupta, s.d., & sahoo, t.k. (2015). light emitting diode (led)-induced alteration of oxidative events during in vitro shoot organogenesis of curculigo orchioides gaertn. acta physiologiae plantarum, 37, 233. https://doi.org/10.1007/s11738-015-19909 he, g., elling, a.a., & deng, x.w. (2011). the epigenome and plant development. annual review of plant biology, 62, 411-435. https://doi.org/10.1146/ annurev-arplant-042110-103806 hsie, b.s., bueno, a.i.s., bertolucci, s.k.v., carvalho, a.a., cunha, s.h.b., martins, e.r., & pinto, j.e.b.p. (2019). study of the influence of wavelengths and intensities of leds on the growth, photosynthetic pigment, and volatile compounds production of lippia rotundifolia cham in vitro. journal of photochemistry and photobiology b: biology, 198, 111577. https://doi.org/10.1016/j. jphotobiol.2019.111577 jung, w.-s., chung, i.-m., hwang, m.h., kim, s.-h., yu, c.y., & ghimire, b.k. (2021). application of lightemitting diodes for improving the nutritional quality and bioactive compound levels of some crops and medicinal plants. molecules, 26(5), 1477. https://doi.org/10.3390/molecules26051477 kölling, k., thalmann, m., müller, a., jenny, c., zeeman, s.c. (2015). carbon partitioning in arabidopsis thaliana is a dynamic process controlled by the plants metabolic status and its circadian clock. plant, cell and environment, 38, 1965-1979. https://doi.org/10.1111/pce.12512 kwon, a.-r., cui, h.-y., lee, h., shin, h., kang, k.-s., & park, s.-y. (2015). light quality affects shoot regeneration, cell division, and wood formation in elite clones of populus euramericana. acta physiologiae plantarum, 37, 65. https://doi. org/10.1007/s11738-015-1812-0 li, r., huang, w., wang, x., liu, x., & xu, z. (2018). effects of yellow, green, and different blue spectra on growth of potato plantlets in vitro. hortscience, 53(4), 541-546. https://doi.org/10.21273/ hortsci12848-18 lichtenthaler, h.k. (1987). chlorophylls and carotenoids: pigments of photosynthetic biomembranes. in: packer, l., & douce, r. (eds). methods in enzymology. vol 1. academic, london, pp 350-381. https://doi.org/10.1016/0076-6879(87)48036-1 lloyd, g., & mccown, b. (1980). commercially feasible micropropagation of mountain laurel, kalmia latifolia, by use of shoot-tip culture. combined proceedings of the international plant propagators society, 30, 421-427. macedo, a.f., leal-costa, m.v., tavares, e.s., lage, c.l.s., & esquibel, m.a. (2011). the effect of light quality on leaf production and development of in vitro-cultured plants of alternanthera brasiliana kuntze. environmental and experimental botany, 70(1), 43-50. https://doi.org/10.1016/j. envexpbot.2010.05.012 miranda, n.a., xavier, a., otoni, w.c., gallo, r., gatti, k.c., moura, l.c., souza, d.m.s.c., maggione, j.h., & santos, s.s.o. (2020). quality and intensity of light in the in vitro development of microstumps of eucalyptus urophylla in a photoautotrophic system. forest science, 66(6), 754-760. https://doi. org/10.1093/forsci/fxaa027 molinari, l.v., souza, d.m.s.c., avelar, m.l.m., fernandes, s.b., gonçalves, d.s., faria, j.c.t., carvalho, d.c., & brondani, g.e. (2021). effects of chemical sterilization of the culture media, porous membranes and luminosity on in vitro culture of eucalyptus grandis × eucalyptus urophylla. journal of forestry research, 32, 1587-1598. https://doi. org/10.1007/s11676-020-01240-5 nery, l.a., batista, d.s., rocha, d.i., felipe, s.h.s., queiroz, m.c., silva, p.o., ventrella, m.c., & otoni, w.c. (2021). leaf development and anatomy of in vitro-grown polygala paniculata l. are affected by light quality, gelling agents, and sucrose. vegetos, 34, 19-28. https://doi.org/10.1007/s42535-021-00192-3 nord, e.a., lynch, j.p. (2009). plant phenology: a critical controller of soil resource acquisition, journal of experimental botany, 60(7), 1927-1937. https:// doi.org/10.1093/jxb/erp018 oberschelp, g.p.j., gonçalves, a.n., meneghetti, e.c., graner, é., & almeida, m. (2015). eucalyptus dunnii maiden plant regeneration via shoot organogenesis on a new basal medium based on the mineral composition of young stump shoots. in vitro cellular & developmental biology – plant, 51, 626-636. https://doi.org/10.1007/s11627-0159715-8 oliveira, l.s., xavier, a., lopes, a.p., takahashi, e.k., & otoni, w.c. (2016). in vitro multiplication and elongation of hybrid clones of eucalyptus globulus. ciência florestal, 26(1), 235-247. https://doi. org/10.5902/1980509821524 rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 15 https://doi.org/10.1016/j.indcrop.2021.113748 https://doi.org/10.1016/j.indcrop.2021.113748 https://doi.org/10.1016/j.jphotochem.2017.09.033 https://doi.org/10.1016/j.jphotochem.2017.09.033 https://doi.org/10.1007/s11738-015-1990-9 https://doi.org/10.1007/s11738-015-1990-9 https://doi.org/10.1146/annurev-arplant-042110-103806 https://doi.org/10.1146/annurev-arplant-042110-103806 https://doi.org/10.1016/j.jphotobiol.2019.111577 https://doi.org/10.1016/j.jphotobiol.2019.111577 https://doi.org/10.3390/molecules26051477 https://doi.org/10.1111/pce.12512 https://doi.org/10.1007/s11738-015-1812-0 https://doi.org/10.1007/s11738-015-1812-0 https://doi.org/10.21273/hortsci12848-18 https://doi.org/10.21273/hortsci12848-18 https://doi.org/10.1016/0076-6879(87)48036-1 https://doi.org/10.1016/j.envexpbot.2010.05.012 https://doi.org/10.1016/j.envexpbot.2010.05.012 https://doi.org/10.1093/forsci/fxaa027 https://doi.org/10.1093/forsci/fxaa027 https://doi.org/10.1007/s11676-020-01240-5 https://doi.org/10.1007/s11676-020-01240-5 https://doi.org/10.1007/s42535-021-00192-3 https://doi.org/10.1093/jxb/erp018 https://doi.org/10.1093/jxb/erp018 https://doi.org/10.1007/s11627-015-9715-8 https://doi.org/10.1007/s11627-015-9715-8 https://doi.org/10.5902/1980509821524 https://doi.org/10.5902/1980509821524 oliveira, t., balduino, m.c.m., carvalho, a.a., bertolucci, s.k.v., cossa, m.c., coelho, a.d., leite, j.j.f., & pinto, j.e.b.p. (2021). the effect of alternative membrane system, sucrose, and culture methods under photosynthetic photon flux on growth and volatile compounds of mint in vitro. in vitro cellular & developmental biology – plant, 57, 529-540. https://doi.org/10.1007/s11627-020-10147-z pawłowska, b., żupnik, m., szewczyk-taranek, b., & cioć, m. (2018). impact of led light sources on morphogenesis and levels of photosynthetic pigments in gerbera jamesonii grown in vitro. horticulture, environment, and biotechnology, 59, 115-123. https://doi.org/10.1007/s13580-0180012-4 r core team (2014). r: a language and environment for statistical computing. vienna, austria: r foundation for statistical computing. ramírez-mosqueda, m.a., iglesias-andreu, g., & bautistaaguilar, j.r. (2017). the effect of light quality on growth and development of in vitro plantlet of stevia rebaudiana bertoni. sugar tech, 19, 331-336. https://doi.org/10.1007/s12355-016-0459-5 rokich, d.p., & bell, d.t. (1995). light quality and intensity effects on the germination of species from the jarrah (eucalyptus marginata) forest of western australia. australian journal of botany, 43(2), 169179. https://doi.org/10.1071/bt9950169 sahakyan, n., petrosyan, m., & trchounian, a. (2016). comparative analysis of chemical composition and biological activities of ajuga genevensis l. in in vitro culture and intact plants. international journal of biological, biomolecular, agricultural, food and biotechnological engineering, 10(7), 322-325. santana-buzzy, n., canto-flick, a., barahona-pérez, f., montalvo-peniche, m.c., zapata-castillo, p.y., solís-ruiz, a., zaldívar-collí, a., gutiérrez-alonso, o., & miranda-ham, m.l. (2005). regeneration of habanero pepper (capsicum chinense jacq.) via organogenesis. hortscience, 40(6), 1829-1831. https://doi.org/10.21273/hortsci.40.6.1829 silva, s.t., bertolucci, s.k.v., cunha, s.h.b., lazzarini, l.e.s., tavares, m.c., & pinto, j.e.b.p. (2017). effect of light and natural ventilation systems on the growth parameters and carvacrol content in the in vitro cultures of plectranthus amboinicus (lour.) spreng. plant cell, tissue and organ culture, 129, 501-510. https://doi.org/10.1007/s11240-0171195-6 silveira, a.a.c., lopes, f.j.f., & sibov, s.t. (2020). micropropagation of bambusa oldhamii munro in heterotrophic, mixotrophic and photomixotrophic systems. plant cell, tissue and organ culture, 141, 315-326. https://doi.org/10.1007/s11240-02001788-4 singh, a.s., jones, a.m.p., shukla, m.r., & saxena, p.k. (2017). high light intensity stress as the limiting factor in micropropagation of sugar maple (acer saccharum marsh.). plant cell, tissue and organ culture, 129, 209-221. https://doi.org/10.1007/ s11240-017-1170-2 smirnakou, s., ouzounis, t., & radoglou, k. (2017). effects of continuous spectrum leds used in indoor cultivation of two coniferous species pinus sylvestris l. and abies borisii-regis mattf. scandinavian journal of forest research, 32(2), 115-122. https://doi.org/10.1080/02827581.201 6.1227470 souza, d.m.s.c., avelar, m.l.m., fernandes, s.b., silva, e.o., duarte, v.p., molinari, l.v., & brondani, g.e. (2020a). spectral quality and temporary immersion bioreactor for in vitro multiplication of eucalytpus grandis × eucalyptus urophylla. 3 biotech, 10, 457. https://doi.org/10.1007/s13205-020-02447-3 souza, d.m.s.c., fernandes, s.b., avelar, m.l.m., frade, s.r.p., molinari, l.v., gonçalves, d.s., & brondani, g.e. (2019). mixotrophism effect on in vitro elongation and adventitious rooting of eucalyptus dunnii. cerne 25(4), 394-401. https://doi.org/10.1 590/01047760201925042638 souza, d.m.s.c., fernandes, s.b., avelar, m.l.m., frade, s.r.p., molinari, l.v., gonçalves, d.s., pinto, j.e.b.p., & brondani, g.e. (2020b). light quality in micropropagation of eucalyptus grandis × eucalyptus urophylla. scientia forestalis, 48(127), e3329. https://doi.org/10.18671/scifor. v48n127.03 souza, d.m.s.c., fernandes, s.b., molinari, l.v., avelar, m.l.m., gonçalves, d.s., faria, j.c.t., teixeira, g.l., & brondani, g.e. (2021). the influence of light quality in the in vitro cultivation of cattleya crispata (thunb.) van den berg. nativa, 9(5), 582588. https://doi.org/10.31413/nativa.v9i5.12541 souza, d.m.s.c., fernandes, s.b., silva, e.o., duarte, v.p., gonçalves, d.s., carvalho, d., teixeira, g.l., & brondani, g.e. (2022). effect of light intensity on in vitro introduction and multiplication of eucalyptus grandis × eucalyptus urophylla. in vitro cellular & developmental biology – plant, 58, 225-239. https://doi.org/10.1007/s11627-021-10237-6 souza, d.m.s.c., xavier, a., otoni, w.c., miranda, n.a., & maggioni, j.h. (2018). light quality in the in vitro introduction of corymbia hybrid clones. revista árvore, 42(6), e420604. https://doi. org/10.1590/1806-90882018000600004 tisarum, r., samphumphung, t., theerawitaya, c., prommee, w., & cha-um, s. (2018). in vitro photoautotrophic acclimatization direct transplantation and ex vitro adaptation of rubber tree (hevea brasiliensis). plant cell, tissue and organ culture, 133, 215-223. https://doi. org/10.1007/s11240-017-1374-5 rangel do prado frade et al. new zealand journal of forestry science (2023) 53:3 page 16 https://doi.org/10.1007/s11627-020-10147-z https://doi.org/10.1007/s13580-018-0012-4 https://doi.org/10.1007/s13580-018-0012-4 https://doi.org/10.1007/s12355-016-0459-5 https://doi.org/10.1071/bt9950169 https://doi.org/10.21273/hortsci.40.6.1829 https://doi.org/10.1007/s11240-017-1195-6 https://doi.org/10.1007/s11240-017-1195-6 https://doi.org/10.1007/s11240-020-01788-4 https://doi.org/10.1007/s11240-020-01788-4 https://doi.org/10.1007/s11240-017-1170-2 https://doi.org/10.1007/s11240-017-1170-2 https://doi.org/10.1080/02827581.2016.1227470 https://doi.org/10.1080/02827581.2016.1227470 https://doi.org/10.1007/s13205-020-02447-3 https://doi.org/10.1590/01047760201925042638 https://doi.org/10.1590/01047760201925042638 https://doi.org/10.18671/scifor.v48n127.03 https://doi.org/10.18671/scifor.v48n127.03 https://doi.org/10.31413/nativa.v9i5.12541 https://doi.org/10.1007/s11627-021-10237-6 https://doi.org/10.1590/1806-90882018000600004 https://doi.org/10.1590/1806-90882018000600004 https://doi.org/10.1007/s11240-017-1374-5 https://doi.org/10.1007/s11240-017-1374-5 effects of land cover type on carabid beetles (coleoptera: carabidae) of the canterbury foothills, new zealand lisa a. berndt1,2 and eckehard g. brockerhoff3,4,* 1 scion (new zealand forest research institute), private bag 3020, rotorua 3046, new zealand 2 current address: lifelab coaching, 29 iles rd, lynmore, rotorua 3010, new zealand 3 scion (new zealand forest research institute), po box 29 237, christchurch 8440, new zealand 4 swiss federal research institute wsl, zürcherstrasse 111, 8903 birmensdorf, switzerland *corresponding author: eckehard.brockerhoff@wsl.ch (received for publication 27 april 2019; accepted in revised form 23 december 2019) abstract background: land cover changes during the recent history of new zealand have had a major impact on its largely endemic and iconic biodiversity. as in many other countries, large areas of native forest have been replaced by other land cover and are now in exotic pasture grassland or plantation forest. ground beetles (carabidae) are often used as ecological indicators, they provide ecosystem services such as pest control, and some species are endangered. however, few studies in new zealand have assessed the habitat value for carabid beetles of natural forest, managed regenerating natural forest, pine plantation forest and pasture. methods: we compared the carabid beetle assemblages of natural forest of nothofagus solandri var solandri (also known as fuscospora solandri or black beech), regenerating n. solandri forest managed for timber production, exotic pine plantation forest and exotic pasture, using pitfall traps. the study was conducted at woodside forest in the foothills of the southern alps, north canterbury, new zealand, close to an area where the critically endangered carabid holcaspis brevicula was found. results: a total of 1192 carabid individuals from 23 species were caught during the study. all but two species were native to new zealand, with the exotic species present only in low numbers and one of these only in the pasture habitat. carabid relative abundance and the number of species was highest in the pine plantation, where a total of 15 species were caught; however, rarefied species richness did not differ significantly between habitats. the sampled carabid beetle assemblages were similar across the three forested habitat types but differed significantly from the pasture assemblages based on unconstrained and canonical analyses of principal coordinates. holcaspis brevicula was not detected in this area. conclusions: our results show that managed or exotic habitats may provide habitat to species-rich carabid assemblages although some native species occur only in natural, undisturbed vegetation. nevertheless, it is important to acknowledge the potential contribution of these land uses and land cover types to the conservation of native biodiversity and to consider how these can be managed to maximise conservation opportunities. new zealand journal of forestry science berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 https://doi.org/10.33494/nzjfs492019x54x e-issn: 1179-5395 published on-line: 30/12/2019 © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access arrival (mcglone 1989) ca. 750 yr ago (wilmshurst et al. 2008). lowland forests that have a particularly rich biodiversity have experienced the greatest losses and are under-represented in the conservation estate, and species of nothofagaceae (southern beech) are now introduction as in many other parts of the world, forest loss and fragmentation had a considerable impact on new zealand’s forests. approximately two thirds of the original area of native forest has been lost since human keywords: biodiversity, carabidae, community composition, exotic, forest, fuscospora, landscape ecology, native, nothofagaceae, pasture, pinus radiata berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 2 the most common canopy species, especially at higher elevation (wardle 1984; wiser et al. 2011). today, exotic grassland used for pastoral agriculture covers more than 50% of new zealand’s land area while plantation forests (largely based on pinus radiata d.don which is native to california) represent ca. 7% (leathwick et al. 2003). although there is no further decline in the area of natural forest in new zealand, globally natural forests are still declining, while the plantation forest area is increasing (brockerhoff et al. 2013; payn et al. 2015). new zealand’s plantation forests are the basis of an economically important industry (mpi 2019), they are an important carbon sink (beets & garrett 2018) and provide a range of other ecosystem services (e.g. brockerhoff et al. 2013). new zealand’s native forests are largely protected and part of a large conservation estate, although a small proportion is privately owned and in part managed for timber production (e.g. allen et al. 2012; ganivet et al. 2017). with increasing national and international focus on the protection of biodiversity on managed and private land (norton 2000; brockerhoff et al. 2001), a greater understanding is required of the role of production land, including managed native forest, plantation forest and agricultural land, in the conservation of native biodiversity. although the canopy tree species of new zealand’s plantation forests are exotic in origin, these forests can support a diverse suite of native flora and fauna, including rare species such as kiwi, new zealand falcon, long-tailed bat and various orchids (brockerhoff et al. 2003; pawson et al. 2010). plantation forests may also support various native invertebrates (hutcheson & jones 1999; berndt et al. 2008; pawson et al. 2008), including a critically endangered carabid beetle (brockerhoff et al. 2005). however, there is a paucity of studies that have compared invertebrate biodiversity in a variety of production land uses and managed and unmanaged natural vegetation. compared to plantation forests, exotic grassland used for pastoral agriculture appears to be less suitable for native species which are adapted to the previously dominant forest environment (e.g. kuschel 1990). some native invertebrates are present or even very abundant in pasture (crisp et al. 1998; berndt et al. 2008). in some cases species from adjacent native forest or shrubland may spill over into pasture (e.g. derraik et al. 2005; pawson et al. 2008). by contrast, native tussock grasslands support a large suite of native invertebrates, some of which also inhabit improved pasture (rufaut 2002). in a study in the central north island, native forest and adjacent pine plantations and exotic grassland were found to have approximately similar species richness of native beetles (including carabidae and two other families) but native forest had by far the lowest proportion of non-native species (pawson et al. 2008). carabid beetles are frequently used as indicators of environmental change and condition (butterfield et al. 1995; ferris & humphrey 1999; rainio & niemelä 2003) and they are known to provide ecosystem services such as pest and weed control (e.g. kulkarni et al. 2015). new zealand carabids are generally nocturnal and flightless, with 92% (391) of species being endemic to the country or certain regions of new zealand (larochelle & larivière 2001). furthermore, a number of new zealand carabids are threatened, including the critically endangered holcaspis brevicula butcher which is only known to occur in the former eyrewell forest on the canterbury plains ca. 20 km southeast from the area where the present study was carried out (brockerhoff et al. 2005). the carabids of the foothills of the southern alps are not well studied, and there was uncertainty whether the range of h. brevicula extends beyond the canterbury plains into the adjacent foot hills. the objective of this study was to compare the carabid beetle assemblages of natural nothofagaceae forest (here nothofagus solandri var. solandri (hook. f.) poole, also known as fuscospora solandri (hook.f.) heenan & smissen, or ‘black beech’), regenerating n. solandri forest managed for timber, exotic pine plantation forest and exotic grazed pasture to examine the response of these beetles to different land use and land cover types. in addition, this study also acted as a survey for the critically endangered holcaspis brevicula in an area of the canterbury foothills that is the closest larger native forest about 20 km from where h. brevicula has been found in the past. methods this study was conducted at woodside forest (lat. 43.26 s, long. 172.06 e; elevation 400–550 m) and adjacent parts of oxford forest conservation area (elevation ca. 550 m) in the canterbury foothills (fig. 1). mean annual rainfall in the area is ca. 1300 mm. woodside forest is a 121 ha privately owned property, managed primarily for timber production. seventy percent of the property is in black beech (novis et al. 2003), the natural native forest type of the area (wardle, 1984). much of this is managed for timber production under a government-approved sustainable forest management plan, with harvesting systems designed to mimic natural stand replacement (novis et al. 2003). another quarter of the property is in exotic conifer plantations of various species (novis et al. 2003). interspersed with the forest are clearings of grazed exotic pasture. the neighbouring oxford forest is an 11,000-ha natural forest managed by the department of conservation (doc). the forest type is mainly black beech at low altitudes, grading to mountain beech (n. solandri var. cliffortioides, also known as fuscospora cliffortioides (hook.f.) heenan & smissen) above 600 m (wardle 1984). the oxford forest area was logged for timber from 1851, with the woodside forest property logged between 1895 and 1909. a number of fires in the latter half of the 19th century destroyed most of the forest, and timber production in the area ceased in 1915 (novis et al. 2003). areas of unmanaged natural beech on the woodside forest property, as well as adjacent parts of oxford forest, regenerated after a major fire in 1898. other parts of the property were managed for sheep grazing from 1914. after the 1930s, grazing was reduced and more land reverted to beech. the present management was initiated in 1973, when the area of regenerating beech was increased and managed for wood production, and exotic plantations were established (novis et al. 2003). four habitat types were compared in this study: ‘natural beech’ forest (at least 100 years old, in woodside forest and adjacent parts of oxford forest); ‘managed beech’ (approximately 30 yr old, naturally regenerating and managed for timber production); ‘pine’ plantation (ca. 30-year-old pinus radiata, managed for timber production); and grazed ‘pasture’. five pitfall traps were installed in each of four replicates of each habitat on 2526 november 2004. pitfall traps were made from 750 ml polypropylene cups, installed such that the opening (diameter 110 mm) was level with the surrounding surface. to increase trap efficiency, two white intersecting guide panels (1.2 m long × 0.1 m high) were installed over the pitfall traps such that there was no gap between the panels and the ground. a white plastic rain cover (150 × 150 mm), held down with large pebbles, was placed on top of the guide panels above the trap opening. traps were filled with about 200 ml of trapping solution (70% water, 30% monoethylene glycol as preservative, with ca. 1 g salt and a large drop of soap added; salt acted as an additional preservative in case rain diluted the solution, and soap assisted with invertebrates sinking into the liquid). traps were cleared and reset monthly on 20–22 december 2004 and 25–26 january 2005, and the final collection was on 15–16 february 2005. all carabid specimens were transferred to 70% ethanol and later sorted to morphospecies for subsequent identification using various keys, named museum specimens, and specialist advice (see acknowledgements). the effect of habitat type and sampling date on log+1 transformed mean carabids per 100 trap days per plot were tested using anova. tukey’s hsd test was used for pairwise comparisons. the effect of habitat type on the relative proportions of the two most abundant species, holcaspis intermittens (chaudoir) and h. hudsoni britton, was analysed using a generalised linear model (glm) with a binomial error distribution, using backward stepwise selection. as this model was over-dispersed, significance testing was conducted using f–tests rather than χ2 tests. these analyses were performed using r version 1.9.1 (r development core team 2004). carabid species richness in each habitat was compared using sample-based rarefaction curves (estimates 7.0.0; colwell 2004), re-scaled to show individuals on the x-axis (gotelli & colwell 2001)). this comparison was made by bisecting the rarefaction curves at the smallest total number of individuals caught in any habitat (i.e. at the end point of the natural beech curve). species associated with each habitat were identified using the indicator species analysis of dufrêne and legendre (1997), carried out in pc-ord 4.01 (mccune & mefford 1999). this indicator species analysis method identifies indicator species that are characteristic of groups’ sites (such as habitat types), combining relative abundance data of species with their relative frequency of occurrence in the various groups of sites. statistical significance of species indicator values is calculated based on a randomisation method (dufrêne & legendre 1997). unconstrained principal coordinates analysis (pcoa) and constrained (or canonical) principal coordinates analysis (cap) were performed using the cap programme (anderson 2003) to explore relationships among the carabid assemblages of the four habitat types. chi-squared distances were used in the ordination to emphasise differences in composition (quinn & keough 2002; anderson & willis 2003). the unconstrained analysis was used to highlight overall patterns across the data cloud, whereas the constrained analysis imposed a priori habitat groupings, allowing location differences among groups to be seen more clearly because important ecological patterns can be masked in unconstrained pcoa, as explained in anderson and willis (2003). the null hypothesis of no difference in multivariate location among habitat types was tested by calculating the trace statistic of canonical discriminant analysis and obtaining a p-value (4999 permutations) (anderson & willis 2003). gaussian bivariate ellipses (probability 0.95) were fitted to illustrate groupings by habitat). results a total of 1192 carabids from 23 species were caught during this study. holcaspis brevicula was not detected, but four other species of holcaspis were found (table 1). berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 3 figure 1: map of the study area at woodside forest (latitude 43.26 s, longitude 172.06 e; elevation 400–550 m) and adjacent parts of oxford forest conservation area. the area is ca. 20 km northwest from the former eyrewell forest where holcaspis brevicula was found in previous studies. each dot represents the approximate location of one replicate group of five pitfall traps. berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 4 species origin** mean beetle abundance (indval) natural beech managed beech pine plantation pasture amarotypus edwardsii n 0.32 (25*) anchomenus ?integratus n 0.64 (3) 0.67 (2) cicindela parryi n 0.64 (0) 0.41 (3) 0.42 (3) 3.44 (55*) demetrida dieffenbachii n 0.67 (0) 1.25 (48*) dicrochile whitei n 0.59 (2) 0.67 (2) 0.59 (2) haplanister crypticus e 0.59 (2) 0.67 (2) 0.59 (2) holcaspis angustula n 0.54 (20*) holcaspis elongella n 0.32 (6) 0.59 (1) 0.12 (2) holcaspis hudsoni n 2.36 (4) 3.67 (8) 22.42 (78*) 0.24 (0) holcaspis intermittens n 7.55 (37) 9.59 (43*) 2.42 (8) 0.65 (1) hypharpax australis e 0.12 (10) mecodema fulgidum n 0.13 (1) 0.89 (16) 0.12 (1) mecodema ?rectolineatum n 0.32 (8) 0.30 (6) 0.12 (2) mecyclothorax rotundicollis n 6.11 (60*) megadromus antarcticus n 0.64 (0) 1.66 (26) 1.38 (17) 1.68 (11) megadromus n. sp. 1# n 0.32 (6) 0.72 (17) notagonum feredayi n 0.67 (3) 0.59 (2) pentagonica vittipennis n 0.64 (2) 0.59 (2) 0.67 (2) scopodes fossulatus n 0.12 (1) 0.67 (0) 0.89 (33*) selenochilus syntheticus n 0.64 (5) syllectus anomalus n 0.59 (5) unknown sp. ? 0.59 (5) zabronothus striatulus n 0.12 (10) total individuals 200 297 458 237 total species 13 12 15 13 exotic species 0 0 0 1-2 * indval (indicator values) significant at p < 0.01 ** origin: n = native, e = exotic, ? = origin unknown # an undescribed species known as “megadromus n. sp. 1” that is relatively common in upland nothofagaceae forest in this area (peter johns, pers. comm.). this is one of four species of megadromus known from nothofagaceae forests in the north canterbury foothills between mt. oxford and mt. grey, the others being m. antarcticus (widespread in canterbury), m. rectangulus (to the east from mt grey to motunau island, and the lowlands from ashley, rangiora, and waipara north), and another undescribed species “megadromus n. sp. 2” which occurs in nothofagaceae forest between ashley gorge and mt grey (peter johns, pers. comm.). table 1: species compostion and abundance (carabids per 100 trap days), indicator values (see footnote) for the four habitat types, and total number of individuals and species caught in each habitat. exotic species were present only in the pasture habitat, in very low numbers. two individuals of the australian species hypharpax australis (dejean) were caught in pasture, along with one individual of an unidentified species that is almost certainly exotic (peter johns pers. comm. 2005). carabid relative abundance in pitfall traps was significantly affected by habitat type (f = 3.368, df = 3, p = 0.022), and carabids were most abundant in the pine plantation habitat. tukey’s pairwise comparisons indicated a significant difference in abundance between pine plantation and pasture, but no significant differences between these habitats and the natural or managed beech habitats (fig. 2). sampling date had no significant effect on carabid relative abundance (f = 2.843, df = 2, p = 0.225), although more carabids appeared to be caught in january than in december or feburary. although the pine plantation had the highest number of species present of all habitats (table 1), rarefied species richness was higher in natural beech and pasture than managed beech or pine (fig. 3). however, 95% confidence intervals of all habitats overlapped, indicating a lack of significant differences in rarefied species richness between habitats. berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 5 indicator species analysis found two species, holcaspis angustula chaudoir and amarotypus edwardsii bates, to be significant indicators of the natural beech habitat (table 1, p < 0.05). holcaspis intermittens was a significant indicator of managed beech, although a high indicator value (37) was also found for this species in natural beech. holcaspis hudsoni was a significant indicator of the pine plantation habitat. four species, mecyclothorax rotundicollis (white), cicindela parryi (white), demetrida dieffenbachii (white) and scopodes fossulatus (blanchard), were significant indicators of pasture. holcaspis intermittens and h. hudsoni, were the dominant species in traps in the three forest habitats. the relative proportions of these two species differed across the habitats, with h. intermittens dominant in natural beech and managed beech, and h. hudsoni dominant in pine (f = 108.47, df = 3, p < 0.001). the unconstrained pcoa (fig. 4a) separated the carabid community of the pasture habitat from that of the three forest habitats, along both axis 1 (explaining 38.1% of the variation) and axis 2 (explaining 8.7% of the variation). constraining the pcoa using the cap procedure, according to the a priori habitat groupings, clarified the differences in carabid community composition (fig. 4b, trace = 1.612, p = 0.0002). cap axis 1 explained the most variation, separating pasture from the forested habitat types (canonical correlation δ = 0.912). the forest habitats were separated along axis 2 (canonical correlation δ = 0.737). all three forested habitats overlapped in multivariate space, with managed beech carabid communities sharing characteristics with those of both natural beech and pine. discussion differences and affinities between assemblages the perhaps most striking finding of this study is the strong separation between the carabid assemblages found in the three forest habitat types and those in the pasture grassland area. this was seen in the limited overlap between these habitat types in the principal coordinates analyses and in the large proportion of species that occurred only in one or the other habitat. ten species were found only in the forest habitat and four species only in pasture grassland. among the nine species that occurred in both forest and pasture, six species showed strong preferences for either forest or pasture in terms of their relative abundance between habitat types. therefore, most species were either forest specialists or open-habitat species, whereas only three species of low to moderate abundance appeared to be habitat generalists. the strongest separation between assemblages was between the largely undisturbed natural beech forest and the pasture grassland, both of which had several unique species (that did not occur in any other habitat types) and showed significant indicator values for these respective habitats. the assemblage in managed (and previously disturbed and regenerating) beech forest had several affinities with the pine forest assemblage which was apparent in both the principal coordinates analyses and the presence and abundance data (in table 1), suggesting that these are forest specialists that are tolerant of disturbance or more capable of recolonisation following disturbance. there was somewhat less overlap between the natural beech forest and pine forest assemblages than between the two beech forest assemblages. comparison with other studies the separation of assemblages in forest and grassland contrasts with the findings of a similar study on carabid assemblages on the canterbury plains nearby (i.e. ca. 20 km to the east of the present study). in that study, carabid assemblages in small native forest remnants, pine plantation forest, pasture grassland and gorse shrubland differed little in terms of species composition, and these assemblages largely overlapped in principal coordinates analyses (berndt et al. 2008). this difference is probably explained by the different land use history between these areas. woodside forest and the adjacent oxford forest conservation area cover a natural forest area of more than 11,000 ha with comparatively less disturbance, and consequently, this area has retained an assemblage of forest specialists. by contrast, the canterbury plains have experienced severe natural forest loss and disturbance, and today natural forest remants represent less than 1% of the land area there (ecroyd & brockerhoff 2005), so that mostly forest generalists and habitat generalists persist whereas forest specialists are rare (berndt et al. 2008). a study in the north island documented distinguishable assemblages of carabids and other beetles (mainly scarabaeidae and scolytinae) between native forest, plantation forest, clear-felled plantation forest and pastoral grassland (pawson et al. 2008). the species composition of beetles in mature plantation forest was most similar to that of native forest. apart from habitat type, the proportion of native vegetation (primarily forest) within the surrounding 500 m was the strongest predictor of variation in beetle species composition (pawson et al. 2008). our contrasting findings between the oxford forest and the canterbury plains assemblages are consistent with this observation about the importance of natural forest vegetation in the surrounding area. a comparison of herbivorous caterpillars and parasitoids in native nothofagaceae forest and adjacent pine plantations in the northern south island reported differences in insect species composition between forest types that appeared to be driven mostly by species variation in lower trophic-level taxa (peralta et al. 2018). this supports the notion that the understorey plants of plantation forests, and not only the canopy tree species, play an important role in the composition of insect assemblages. native vs. exotic species another noteworthy result of our study is the rarity of exotic species at oxford and woodside forests. only three individuals from two exotic carabid species were found in our samples. neither species occurred in the natural nothofagaceae forest while one occurred only in the pasture area. by contrast, five exotic beetle species occurred on the canterbury plains and several of these were abundant, especially in the open habitats of pasture and recently harvested and replanted plantation forest (berndt et al. 2008). the most common exotic species in that area was hypharpax australis which was also found in the grassland in the present study, albeit in very low numbers. this indicates that these particular exotic carabids are open-habitat specialists that colonise open habitats and cleared forest areas but either do not enter or do not survive in closed forest. this finding is consistent with other studies. for example, harris and burns (2000) and pawson et al. (2008) documented a substantially greater number and abundance of exotic beetle species in grassland than in natural forest remnants in the north island. collectively, these studies and our results suggest that new zealand’s natural forest is remarkably resistant to invasion by exotic beetles. this may be due to the high level of endemism in new zealand’s biota and to the relatively limited disturbance experienced by new zealand’s natural forests (brockerhoff et al. 2010). furthermore, older plantation forests (i.e. with trees older than ca. 20 years) share many characteristics with natural forests. they have similar species composition (i.e. beetle assemblages) and a small proportion of exotic beetles compared with pasture grassland vegetation. however, in other countries natural forests are by no means resistant to invasion by non-native beetles. this is particularly true in north america where more than 100 non-native beetle species have been recorded feeding on forest trees, including numerous important forest pests (aukema et al. 2010), and many other non-native insects are also found in forests (e.g. liebhold et al. 2016). berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 6 figure 2: mean relative abundance (± s.e.) of carabid beetles in each habitat type. bars sharing a letter do not differ significantly (α = 0.05). figure 3: sample-based rarefaction curves (± 95% ci) of carabid species richness in each habitat type, rescaled to show individuals on the x-axis. species richness is compared at the end point of the natural beech curve, indicated by the dashed line. characteristics of species in the different habitats three species (amarotypus edwardsii, holcaspis angustula and selenochilus syntheticus) were unique to the natural beech forest, the original forest type of the area. these species appear to be forest specialists as they were not found in our earlier survey of small forest remnants or plantation forests or other vegetation on the berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 7 canterbury plains nearby (berndt et al. 2008). holcaspis angustula has also been found in a native forest reserve on banks peninsula where it was typical of dense, moist native forest areas (bowie et al. 2018). it was the least abundant of four species of holcaspis in that study, and it appears to have declined since an earlier survey 30 years ago (bowie et al. 2018). according to larochelle and larivière (2001), h. angustula may also occur in plantation forest, tussock grassland and pasture. however, based on our findings and those of bowie et al. (2018), h. angustula shows clear preferences for dense forest. amarotypus edwardsii is a flightless forest species; it lives on tree trunks in wet forests and shrubland, feeding on invertebrates among mosses and lichens (johns 1980; larochelle & larivière 2001), although it must also be active on the ground at times to have been caught in the pitfall traps in this study. however, in the central north island, a. edwardsii was commonly found in pine plantations, and although it has been recorded in pasture grassland, this was restricted to traps placed no more than 25 m from a forest edge (pawson et al. 2008). therefore it cannot be considered an old forest specialist. selenochilus syntheticus is a flightless species of dry lowland nothofagaceae forests (larochelle & larivière 2001), but otherwise little is known about this species. two other species of holcaspis (h. intermittens and h. hudsoni) were the most abundant species in the forest habitats. these two species differed in dominance in each habitat, with a greater proportion of h. intermittens caught in beech and managed beech, and a greater proportion of h. hudsoni caught in pine. holcaspis intermittens is a dry forest and shrubland species (larochelle & larivière 2001), but it is thought to have a preference for more densely vegetated sites with higher humidity (johns 1986). holcaspis hudsoni has a broader habitat range than h. intermittens, being recorded from dry forests, shrublands and pine plantations, as well as tussock grasslands and pasture (larochelle & larivière, 2001). in this study, most beech forest plots had significant understorey vegetation, whereas the pine plantation had more limited undergrowth. this may have been due to greater canopy closure in the pine plantation than in the beech forest (brockerhoff et al. 2003), or to differences in soil moisture. beech forests on moist soils in the canterbury foothills where the study was located support a dense undergrowth of the fern species blechnum discolour and polystichum vestitum, whereas on drier slopes and ridges the shrubs cyathodes fasciculata and c. juniperina are often prominent in the understorey (wardle 1984). the pasture carabid assemblage was characterised by dominance of small species, and exotic species present in low numbers. mecyclothorax rotundicollis, a pasture habitat indicator species in our study, is a widespread, flighted species with a small body size (<10 mm). it tolerates or perhaps even prefers modified vegetation such as pasture and crop fields (johns 1986; larochelle & larivière 2001). habitats with greater disturbance, such as the grazed pasture in this study, tend to support carabid assemblages of a smaller average body size, as figure 4: (a) unconstrained principal coordinates analysis of pitfall trapped carabid beetles in four habitat types. two (out of 20) pasture traps were outliers and not included on the graph (axis 1 scores of –1.4 and –15.1). (b) constrained principal coordinates analysis of pitfall trapped carabid beetles in four habitat types. gaussian bivariate ellipses (probability 0.95) are shown on both graphs. (a) (b) -0.3 -0.2 -0.1 0.0 0.1 0.2 cap1 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 c a p 2 pine man. beech pasture beech legend -0.3 -0.2 -0.1 0.0 0.1 0.2 cap1 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 c a p 2 pine man. beech pasture beech legend -0.3 -0.2 -0.1 0.0 0.1 0.2 cap1 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 c a p 2 pine man. beech pasture beech legend 0.10 0.15 0.20 0.25 pca1 -2 -1 0 1 2 p c a 2 pine man. beech pasture beech legend was found in a study of carabids in grassland in england (blake et al. 1994). cicindela parryi, a tiger beetle, is typically found in open habitat, although usually in forest gaps, rather than in larger pasture areas (larochelle & larivière 2001). flighted species have the ability to disperse quickly which is an advantage if they rely on potentially ephemeral newly created open habitats such as forest gaps created by windthrow. absence of holcaspis brevicula one of the objectives of this study was to determine whether the range of the critically endangered holcaspis brevicula extends beyond the canterbury plains into the adjacent foot hills of the southern alps. our survey in the closest larger and unfragmented natural forest and adacent areas revealed the presence of four other species of holcaspis, but no h. brevicula. it is unlikely that we overlooked h. brevicula among any of the other holcaspis specimens because none of them are closely related to h. brevicula which is the only species of the algida complex that occurs in this area (butcher 1984). therefore, it is likely that its only occurrence is indeed in the area of the former eyrewell forest (brockerhoff et al. 2005), although it cannot be ruled out entirely that a population remains elsewhere. as eyrewell forest has been converted to a dairy farm in the last few years (hancock 2019), it is unlikely to have survived there. effects of land management and forest regeneration the four adjacent habitat types sampled for this project differed considerably in terms of their vegetation history and the current vegetation. the natural beech forest experienced relatively little disturbance although it is not clear to what extent this area was affected by the forest fire in the 1800s. the area of managed beech forest was affected by logging and fire between 1860 and 1910) and then grazed until the 1930s after which there was gradual re-establishment and regeneration of the beech forest (novis et al. 2003; allen et al. 2012). the period since then allowed ample opportunity for recolonisation by carabids from the adjacent forests and restoration of a relatively ‘natural’ forest carabid assemblage, although the species composition differed somewhat from the natural beech forest. the closeto-nature management of this managed beech forest, which aims to mimic the scale and intensity of natural disturbance events in the forest (novis et al. 2003; allen et al. 2012), does not appear to have any detrimental effects on the local ground beetle community. the pine forest area was planted in the 1970s on land that was previously pasture, so this area experienced disturbance more recently. in addition, the canopy tree, pinus radiata, is not native to new zealand; however, there was an understorey of mainly native plants of which many were shared with the adjacent native beech forest. despite these differences, the species composition of the pine forest area was almost identical to that of the managed beech forest, although there were differences in the relative abundance of several species. by contrast, the open pasture grassland area, which differed substantially from the three forest types in terms of vegetation table 2: confusion matrix structure and plant species composition, also revealed a distinct beetle assemblage, even though the pasture area is very small and surrounded by beech forest. it is likely that habitat connectivity plays an important role in maintaining native biodiversity in managed forests (e.g. norton 1998), and the different forest habitats at woodside are interconnected, with pasture present more as clearings than matrix habitat. the relatively small patches of the managed beech, pine forest and pasture habitat types created more edge or boundary habitat where assemblages partly overlap (koivula et al. 2004) with forest species and open-habitat species potentially ‘spilling over’ into adjacent habitats. however, the size and quality of habitats may be more important than edge effects and connectivity (e.g. hodgson et al. 2011), and the relatively large natural beech forest directly adjacent to the managed habitats probably plays a key role in the resemblance of these assemblages and the almost complete absence of non-native ground beetle species. native invertebrates can be abundant in exotic habitats (watts & gibbs 2002; harris et al. 2004; derraik et al. 2005; berndt et al. 2008; pawson et al. 2008), in some cases with similar species assemblages to equivalent native habitats. however, exotic habitats cannot and should not replace native ones, and the conversion and degradation of natural vegetation is a major cause of the worldwide loss and decline of biodiversity (newbold et al. 2015). furthermore, where possible, it would be preferable in terms of biodiversity conservation to restore natural vegetation and plant native trees. nevertheless, as new zealand is dominated by production land uses with crop plants of exotic origin, it is important to acknowledge the potential contribution of these land uses to the conservation of native biodiversity and to consider how these can be managed to maximise conservation opportunities. competing interests the authors declare that they have no competing interests. authors’ contributions lab and egb jointly conceived and planned the study and carried out much of the field work. lab identified most of the carabids and did most of the data analysis. lab and egb jointly wrote the manuscript. acknowledgements we would like to thank john and rosalie wardle for access to woodside forest and the department of conservation for access to mt. oxford forest (national permit number ca-13813-fau). we greatly appreciate the assistance of peter johns, andré larochelle and rowan emberson with carabid identification, janette beaton, chris berry and alan leckie in the field and laboratory, and steve pawson for advice on statistical analyses. funding for this study was provided by the foundation for research, science and technology (c04x0304). berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 8 references allen, r., hurst, j., wiser, s., & easdale, t. (2012). developing management systems for the production of beech timber. new zealand journal of forestry, 57(2), 38–44. anderson, m.j. (2003). cap: a fortran computer program for canonical analysis of principal coordinates. auckland, new zealand: department of statistics, university of auckland. anderson, m.j., & willis, t.j. (2003). canonical analysis of principal coordinates: a useful method of constrained ordination for ecology. ecology, 84, 511–525. aukema, j.e., mccullough, d.g., von holle, b., liebhold, a.m., britton, k., & frankel, s.j. (2010). historical accumulation of nonindigenous forest pests in the continental united states. bioscience, 60, 886–897. beets, p.n., & garrett, l.g. (2018). carbon fraction of pinus radiata biomass components within new zealand. new zealand journal of forestry science, 48, 14. berndt, l.a., brockerhoff, e.g., & jactel, h. (2008). relevance of exotic pine plantations as a surrogate habitat for ground beetles (carabidae) where native forest is rare. biodiversity and conservation, 17, 1171–1185. blake, s., foster, g.n., eyre, m.d., & luff, m.l. (1994). effects of habitat type and grassland management practices on the body size distribution of carabid beetles. pedobiologia, 38, 502–512. bowie, m.h., shields, m.w., he, s., ross, j.g., cruickshank, r.h., & hodge, s. (2018). a survey of ground beetles (coleoptera: carabidae) in ahuriri scenic reserve, banks peninsula, and comparisons with a previous survey performed 30 years earlier. new zealand journal of zoology, 46, 285–300. brockerhoff, e.g., barratt, b.i., beggs, j.r., fagan, l.l., malcolm, k., phillips, c.b., & vink, c.j. (2010). impacts of exotic invertebrates on new zealand’s indigenous species and ecosystems. new zealand journal of ecology, 34, 158–174. brockerhoff, e.g., berndt, l.a., & jactel, h. (2005) role of exotic pine forests in the conservation of the critically endangered new zealand ground beetle holcaspis brevicula (coleoptera: carabidae). new zealand journal of ecology, 29, 37–43. brockerhoff, e.g., ecroyd, c.e., & langer, e.r. (2001). biodiversity in new zealand plantation forests: policy trends, incentives and the state of our knowledge. new zealand journal of forestry, 46, 31–37. brockerhoff, e.g., ecroyd, c.e., leckie, a.c., & kimberley, m.o. (2003) diversity and succession of adventive and indigenous vascular understorey plants in pinus radiata plantation forests in new zealand. forest ecology and management, 185, 307–326. brockerhoff, e.g., jactel, h., parrotta, j.a., & ferraz, s.f. (2013). role of eucalypt and other planted forests in biodiversity conservation and the provision of biodiversity-related ecosystem services. forest ecology and management, 301, 43–50. butcher, m.r. (1984). a revision of the genus holcaspis (coleoptera: carabidae). journal of the royal society of new zealand, 14, 47–99. butterfield, j., luff, m.l., baines, m., & eyre, m.d. (1995). carabid beetle communities as indicators of conservation potential in upland forests. forest ecology and management, 79, 63–77. colwell, r.k. (2004) estimates: statistical estimation of species richness and shared species from samples. version 7. persistent url <purl.oclc.org/ estimates>. crisp, p.n., dickinson, k.j.m., & gibbs, g.w. (1998). does native invertebrate diversity reflect native plant diversity? a case study from new zealand and implications for conservation. biological conservation, 83, 209–220. derraik, k.j.d., derraik, j.g., rufaut, c.g., & closs, g.p. (2005). ground invertebrate fauna associated with native shrubs and exotic pasture in a modified rural landscape, otago, new zealand. new zealand journal of ecology, 29, 129–135. dufrêne, m., & legendre, p. (1997). species assemblages and indicator species: the need for a flexible asymmetrical approach. ecological monographs, 67, 345–366. ecroyd, c.e., & brockerhoff, e.g. (2005). floristic changes over 30 years in a canterbury plains kanuka forest remnant, and comparison with adjacent vegetation types. new zealand journal of ecology, 29, 279–290. ferris, r., & humphrey, j.w. (1999). a review of potential biodiversity indicators for application in british forests. forestry, 72, 313–328. ganivet, e., moltchanova, e., & bloomberg, m. (2017). developing a framework for growth modelling in a managed southern black beech forest. new zealand journal of forestry science, 47, 11. gotelli, n.j., & colwell, r.k. (2001). quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. ecology letters, 4, 379–391. hancock, f. (2019). hello cows, bye-bye rare beetle. newsroom, 13 february 2019. https://www. newsroom.co.nz/2019/02/12/440614/hellocows-bye-bye-rare-beetle (accessed 7 april 2019). harris, r.j., toft, r.j., dugdale, j.s., williams, p.a., & rees, j.s. (2004). insect assemblages in a native (kanuka kunzea ericoides) and an invasive (gorse ulex berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 9 europaeus) shrubland. new zealand journal of ecology, 28, 35–48. harris, r.j., & burns, b.r. (2000). beetle assemblages of kahikatea forest fragments in a pasture-dominated landscape. new zealand journal of ecology, 24, 57–67. hodgson, j.a., moilanen, a., wintle, b.a., & thomas, c.d. (2011). habitat area, quality and connectivity: striking the balance for efficient conservation. journal of applied ecology, 48, 148–152. hutcheson, j., & jones, d. (1999). spatial variability of insect communities in a homogenous system: measuring biodiversity using malaise trapped beetles in a pinus radiata plantation in new zealand. forest ecology and management, 118, 93–105. johns, p.m. (1980). ground beetles (carabidae) of arthurs pass. mauri ora, 8, 55–67. johns, p.m. (1986). arthropods of banks peninsula reserves. christchurch, new zealand: report to the commissioner of crown lands. koivula, m., hyyryläinen, v., & soininen, e. (2004). carabid beetles (coleoptera: carabidae) at forestfarmland edges in southern finland. journal of insect conservation, 8, 297–309. kulkarni, s.s., dosdall, l.m., & willenborg, c.j. (2015). the role of ground beetles (coleoptera: carabidae) in weed seed consumption: a review. weed science, 63, 355–376. kuschel, g. (1990). beetles in a suburban environment: a new zealand case study. auckland, new zealand: dsir plant protection report no.3. larochelle, a., & larivière, m.-c. (2001). carabidae (insecta: coleoptera): catalogue. fauna of new zealand, 43, 1–285. leathwick, j., wilson, g., rutledge, d., wardle, p., morgan, f., johnston, k., mcleod, m, & kirkpatrick, r. (2003). land environments of new zealand. auckland, new zealand: david bateman ltd. liebhold, a.m., yamanaka, t., roques, a., augustin, s., chown, s.l., brockerhoff, e.g., & pyšek, p. (2016). global compositional variation among native and non-native regional insect assemblages emphasizes the importance of pathways. biological invasions, 18, 893–905. mccune, b., & mefford, m.j. (1999). pc-ord. multivariate analysis of ecological data. version 4.01. gleneden beach, oregon: mjm software. mcglone, m.s. (1989). the polynesian settlement of new zealand in relation to environmental and biotic changes. new zealand journal of ecology, 12, 115– 129. mpi [ministry for primary industries] (2019). forestry overview. https://www.mpi.govt.nz/growingand-harvesting/forestry/overview/ (accessed 24 march 2019). newbold, t., hudson, l.n., hill, s.l., contu, s., lysenko, i., senior, r.a., börger, l., bennett, d.j., choimes, a., collen, b., & day, j. (2015). global effects of land use on local terrestrial biodiversity. nature, 520(7545), 45–50. norton, d.a. (1998). indigenous biodiversity conservation and plantation forestry: options for the future. new zealand forestry, 43, 34–39. norton, d.a. (2000). conservation biology and private land: shifting the focus. conservation biology, 14, 1221–1223. novis, j., platt, a., & griffiths, a. (2003). woodside forest. christchurch, new zealand: ministry of agriculture and forestry. also available at http://www.fao. org/3/ae542e/ae542e0a.htm (accessed 9 april 2019). pawson, s.m., brockerhoff, e.g., meenken, e.d., & didham, r.k. (2008). non-native plantation forests as alternative habitat for native forest beetles in a heavily modified landscape. biodiversity and conservation, 17(5), 1127–1148. pawson, s.m., ecroyd, c.e., seaton, r., shaw, w.b., & brockerhoff, e.g. (2010). new zealand’s exotic plantation forests as habitats for threatened indigenous species. new zealand journal of ecology, 34, 342–355. payn, t., carnus, j.m., freer-smith, p., kimberley, m., kollert, w., liu, s., orazio, c., rodriguez, l., silva, l.n., & wingfield, m.j. (2015). changes in planted forests and future global implications. forest ecology and management, 352, 57–67. peralta, g., frost, c.m., & didham, r.k. (2018). plant, herbivore and parasitoid community composition in native nothofagaceae forests vs. exotic pine plantations. journal of applied ecology, 55(3), 1265–1275. quinn, g.p., & keough, m.j. (2002). experimental design and analysis for biologists. cambridge, uk: cambridge university press. r development core team (2004). r: a language and environment for statistical computing. vienna, austria: r foundation for statistical computing. available at https://cran-archive.r-project.org/ bin/windows/base/old/1.9.1/ (accessed 21 dec. 2019) rainio, j., & niemelä, j. (2003). ground beetles (coleoptera: carabidae) as bioindicators. biodiversity and conservation, 12, 487–506. rufaut, c.g. (2002). biodiversity in fragmented indigenous grasslands: patterns of plant and invertebrate diversity in mid-altitude tall tussock grasslands, eastern otago. phd dissertation, university of otago, new zealand. berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 10 wardle, j.a. (1984). the new zealand beeches: ecology, utilisation and management. wellington, new zealand: new zealand forest service. watts, c.h., & gibbs, g.w. (2002). revegetation and its effect on the ground-dwelling beetle fauna of matiu-somes island, new zealand. restoration ecology, 10, 96–106. wilmshurst, j.m., anderson, a.j., higham, t.f., & worthy, t.h. (2008). dating the late prehistoric dispersal of polynesians to new zealand using the commensal pacific rat. proceedings of the national academy of sciences, 105, 7676–7680. wiser, s.k., hurst, j.m., wright, e.f., & allen, r.b. (2011). new zealand’s forest and shrubland communities: a quantitative classification based on a nationally representative plot network. applied vegetation science, 14, 506–523. berndt & brockerhoff. new zealand journal of forestry science (2019) 49:12 page 11 effects of selective-logging, litter and tree species on forests in the peruvian amazon: seed predation, seed pathogens, germination randall w. myster biology department, oklahoma state university, oklahoma city, ok 73107, usa mysterrwm1@gmail.com (received for publication 24 march 2021; accepted in revised form 18 july 2021) abstract background: the amazon basin contains mainly unflooded forests, and they are among the most important ecosystems in the world. field experiments on seed processes are very important in order to understand the structure, function and dynamics of these forests. methods: tree seeds of three species (cecropia latiloba, guarea macrophylla, socratea exorrhiza) were set out in amazon unlogged terra firme forest, in amazon selectively-logged terra firme forest, in amazon palm forest, and in amazon white sand forest either on top of or beneath the litter layer, and after two weeks scored for seeds taken by predators, seeds destroyed by pathogens and seeds that germinated. results: i found both terra firme forests (unlogged and selectively-logged) lost most of their seed to predators and the least of their seed to pathogens, white sand forests lost the least of their seed to predators and the most of their seed to pathogens, and the fewest seeds germinated in both terra firme forests and in palm forest. more specifically (1) within unlogged terra firme forest addition of litter reduced seed predation but increased seed losses to pathogens and germination, and c. latiloba lost the most seeds to pathogens, (2) within selectively-logged terra firme forest seeds showed the same trends as unlogged terra firme forest but without significant effects, (3) within palm forest addition of litter reduced predation but increased losses to pathogens, and s. exorrhiza lost the least seeds to pathogens, and (4) within white sand forests addition of litter increased germination. combining the results from all forests together, predators took most of the seeds, pathogens took most of the seeds that escaped predation, and most of the seeds that survived predation and pathogens germinated. conclusions: while such large losses of tree seed to predators and pathogens in these unflooded forests suggest limited recruitment, the variation demonstrated in these field experiments – among forest-types, among tree species, between litter situations on the forest floor – help to insure that recruitment does occur and that these unflooded forests continue to dominate the amazon basin. new zealand journal of forestry science myster new zealand journal of forestry science (2021) 51:9 https://doi.org/10.33494/nzjfs512021x153x e-issn: 1179-5395 published on-line: 2/08/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access need to know what controls the recruitment of their trees (grubb 1977). in particular the post-dispersal (myster 2017a) seed processes of seed predation, seed pathogens, and seed germination (myster 2003, myster 2015), play a critical role in determining the plant-plant replacements (myster 2018) of the unflooded forests in introduction the amazon basin contains some of the most important ecosystems in the world, significantly influencing the water, oxygen, carbon and other biogeochemical cycles for the entire planet. in order to understand the structure, function and dynamics of these key ecosystems, we keywords: allpahuayo-mishana national reserve, brosimum rubescens, palm, terra firme, white sand mailto:mysterrwm1@gmail.com http://creativecommons.org/licenses/by/4.0/), myster new zealand journal of forestry science (2021) 51:9 page 2 the amazon basin. research in the amazon has shown seed predation determines the fate of the majority of seeds (bodmer 1991, notman et al. 1996, russo 2005, paine & beck 2007) but even after suffering such large losses of seed, tree seedlings, tree saplings and mature trees do occur and regenerate these amazon forests. this must be due to critical sources of variation (sov) in how these mechanisms (seed predation, seed pathogens) and environmental tolerances (seed germination) work. while many of these sov have been investigated in amazon forests – for example variation among species, variation among microsites created by tree fall and conversion to agriculture, variation in the amount of litter, variation among different seasons, variation among different years and variation at different spatial scales (bodmer 1991, notman et al. 1996, notman & gorchov 2001, russo 2005, vieira & scariot 2006, paine and beck 2007, myster 2012, myster 2014, myster 2015, myster 2017b) – we need more information in how seed mechanisms and tolerances work. most of the amazon rainforest does not flood (kalliola et al. 1991, pitman et al. 2001) and terra firme forest (found on fertile clay or loam soils; daly & prance 1989) is the most common unflooded amazon forest-type. other common unflooded forest-types include palm forests, found on moderately fertile soils that can be waterlogged, and white sand forests, found on infertile soils with large amounts of quartz (tuomisto et al. 2003, honorio 2006). in addition to these soil differences – or perhaps because of them – these forests also vary in richness and physical structure (stropp et al. 2011) where terra firme forest has the most number of species and the most complex structure, white sand forest has the least number of species and relatively simplest structure, and palm forest is between those two foresttypes in species richness and structure (myster 2009). among these unflooded forests, terra firme forest is most often logged either by clear-cutting, where all trees are cut and removed above a certain minimum size (e.g. 5 cm diameter at breast height [dbh]; gorchov et al. 1993, notman et al. 1996, gorchov et al. 2004), or by selective-logging, where only trees of a select species (e.g. mahogany [swietenia macrophylla]; lambert et al. 2005, grogan & galvao 2006) above a critical size (e.g. 50 cm dbh; lambert et al. 2005) are cut and removed. while clear-cutting obviously changes terra firme forest dramatically, selective-logging can also change terra firme by, for example, creating gaps in the canopy, increasing road construction, and decreasing animal food resources and habitat which reduce pollination and seed dispersal (jansen & zuidema 2001). therefore because of the importance of these forests, and the need for more investigation into critical sov of their seed mechanisms and seed tolerances, i continue my past research into recruitment of amazon unflooded forests (myster 2012, myster 2014, myster 2017b) by conducting new field experiments on how selectivelogging, different tree seed species, and addition of litter interact to affect seed predation, seed pathogens, and germination in amazon unlogged terra firme forest, in amazon selectively-logged terra firme forest, in amazon palm forest, and in amazon white sand forest with litter interactions. i test these five hypotheses: (1) seed predators will take the majority of tree seeds put out in these amazonian forests with terra firme forests (both selectively-logged and unlogged) having the greatest losses, white sand forests having the least, and palm forests between the two in losses (bodmer 1991, notman et al. 1996, russo 2005, paine & beck 2007, myster 2009, myster 2014, myster 2017b). (2) seed pathogens will take the majority of tree seeds that predators do not take with palm forests having the greatest losses, white sand forests having the least, and terra firme forests (both selectively-logged and unlogged) between the two in losses (myster 2017b). (3) most seeds that survive predators and pathogens will germinate, in all forests-types (notman et al. 1996, myster & everham 1999, myster 2014). (4) adding litter on top of seeds will reduce predation but increase pathogenic effects and germination, again in all forest-types (fenner 1985, cintra 1997). (5) larger, heavier tree seeds will be taken by predators more than smaller, lighter tree seeds but there will be no seed-size trends for pathogens or germination (pringle et al. 2007). materials and methods the study site was the allpahuayo-mishana national reserve (amnr) located 23 km from iquitos, peru in the loreto region of maynas province (3.9° s, 73.6° w; hice et al. 2004; saaksjarvi et al. 2004). amnr was established in 1999, and is managed by servicio nacional de áreas naturales protegidas por el estado (http://www.sernanp.gob.pe/allpahuayo-mishana) and the instituto de investigación de la amazonía peruana (http://www.iiap.org.pe) with no hunting allowed. the reserve covers 57,667 ha and lies between 110 and 180 m above sea level. the substrate is composed of alluvial and fluvial holocene sediments from the eastern slopes of the andes. annual precipitation is approximately 2800 mm per year, and the rainy season is between october and may (johnson 1976). the average temperature is relatively steady at 26 oc. terra firme forest, palm forest and white sand forest are common within the reserve, and terra firme forest has also been selectively-logged there. in may and early june 2017, 20 forest stands intermingled within amnr were chosen – on the advice of my field assistant – consisting of: (1) five terra firme forest stands; (2) five palm forest stands; (3) five white sand forest stands; and (4) five terra firme forest stands that had been selectively-logged in 2010 by cutting and removing trees of brosimum rubescens (locally called “palo de sangre”; shirota et al. 1997) that were at least 60 cm in diameter (jorge chávez, pers. comm.). in each http://www.sernanp.gob.pe/allpahuayo-mishana http://www.iiap.org.pe of the 20 forest stands a 100 m transect was set up with study microsites marked off every 20 m, creating five microsites per stand and 100 microsites total. the bottoms of six plastic petri dishes (9 cm in diameter; hulme 1994) were randomly placed in each microsite, three of those six dishes on top of the natural litter layer and three of those six dishes beneath the natural litter layer but on top of the soil. these litter treatments realise observed field conditions: (1) when a seed disperses on top of the litter layer and stays there; and (2) when a seed disperses on top of the litter layer but then falls down through the litter layer due to gravity over time which is more likely the heavier the seed. bare areas without litter were very rarely observed in these forests, and so were not included in the treatments. in each group of three dishes (three on top, and three beneath, the litter layer) 100 seeds in a 1-g seed pulp mass of cecropia latiloba (urticaceae: 0.002 g per seed; http://data.kew.org/sid, also see myster 2015), five seeds of guarea macrophylla (meliacea: 0.6 g per seed; http://data.kew.org/sid) and five seeds of socratea exorrhiza (arecaceae: 3.4 g per seed; http://data.kew. org/sid) were placed in a separate dish which was randomly chosen. these three study tree species are frequent and abundant in terra firme, palm and white sand forests (myster 2009; myster 2017c). five plastic seed mimics – made of the same size, shape and color of the real seeds or seed pulp mass – were also placed in each dish with holes in the bottom of the dishes to permit drainage in order to better understand seed removal. the seeds or seed pulp mass were collected, using gloves, locally from one individual tree of that species the same day they were put out. they were then visually inspected for damage or infestation, cleaned of fruit by hand (except for cecropia latiloba) again using gloves and then floated to exclude nonviable seeds (except for cecropia latiloba). after two weeks in the field the percentage of seeds, either taken out of each petri dish or still there but mainly eaten, were scored as eaten by predators. this scoring is justified because evidence of animal activity – such as chewed seeds, seed husks, and small mammal feces – was observed in the dishes (as in blaney & kotanen 2001) and because the plastic seed mimics were not taken (notman et al. 1996), both strongly suggesting that removal by abiotic agents – such as wind or rain – could be discounted. thus i am assuming that seeds were removed by animals, and then either eaten or made nonviable in some other way by the animals that removed them (see myster 2015). the remaining seeds or seed pulp mass, while still in their dishes, were then taken to an on-site shade house with sufficient light for germination (as in nature only those seeds that survive predation may attempt to germinate) and incubated in pots where litter was added on top of those dishes that were under litter in the field of the same type and density as in the field where they were taken from (myster 1994). pots were watered daily in amounts similar to natural rainfall and after five weeks seeds were examined under a dissecting microscope and scored as germinated, scored as destroyed by pathogens (where the seed did not germinate and had extensive fungal damage; myster 2014; myster 2017b), or other. a single three-way analysis of variance (anova) for seed loss due to predation, another one for seed loss due to pathogens and another one for seeds that germinated, with main effects of: (1) forest-type; (2) seed species; and (3) top/bottom of the litter layer, was not appropriate with this experimental design because treatments were not independently available and thus could not be randomly assigned in the field, i.e. foresttypes occur only in large stands, not in the small patches that would be necessary for a complete randomised experimental design. and so three separate one-way anovas were first performed with forest-type (terra firme, selectively-logged terra firme, palm, white sand) as the only main effect where data were pooled across tree seed species and litter treatment within each foresttype. one of these anovas used percent seed losses to predation as the response variable, one of these anovas used percent seed losses to pathogens as the response variable, and one of these anovas used percent seeds that germinated as the response variable. furthermore because seeds may not have survived predation and/ or pathogens in some of the petri dishes, which would have resulted in empty dishes and thus an unbalanced design in the field, anovas were performed within the more robust general linear model (glm with a binomial errors model:sas 1985). transformation of the data by arcsine is not needed to address normality when using glm (wilson et al. 2013). then a two-way anova was performed for each forest-type with a main effect of litter (dish placed on top of the litter layer, dish placed on the soil underneath the litter layer with the litter layer then replaced on top of the dish), a main effect of tree species (cecropia latiloba, guarea macrophylla, socratea exorrhiza) and an interaction effect of litter x species. this anova was performed three times – once for seeds lost to predators, once for seeds lost to pathogens, and once for seeds that germinated – for each of the four forest-types for a total of 12 two-way anovas. for all anovas if significance was found, means tests were conducted with the tukey procedure (sas 1985) to find which levels within treatments were most important in determining significance (bolded in the results). results forest-types differed significantly for seed losses to predation (table 1: df = 3, f = 5.4, p = 0.03) and white sand forests were lower than the other forest-types (46.1±2.3% [mean±standard error] of seeds taken, 90.2±3.5% for unlogged terra firme, 73.3±1.9% for selectively logged terra firme, 81.8±4.4% for palm). forest-types differed significantly for seed losses to pathogens (table 2: df = 3, f = 9.9, p = 0.005) and unlogged terra firme forests (4.1±1.1%) and white sand forests (43.1±5.2%) were most different from the other forestmyster new zealand journal of forestry science (2021) 51:9 page 3 table 1: description of the study sites http://data.kew.org/sid http://data.kew.org/sid http://data.kew.org/sid http://data.kew.org/sid types (logged 20.6±2.2%, palm 16.9±1.8%). foresttypes differed significantly for seeds that germinated (table 3: df = 3, f = 7.1, p = 0.01) and unlogged terra firme (1.2±0.2%) and palm (2.5±0.8%) were lower than selectively logged terra firme (5.2±1.1%) and white sand (7.7±1.7%). in unlogged terra firme forests seed predation was significantly different (df = 1, f = 2.5, p = 0.05) between litter treatments (96.1±7.2% no litter vs. 84.4±3.8% litter), seed pathogenic attack was significantly different (df = 2, f = 3.5, p = 0.02) among tree species where cecropia latiloba (19.2±1.2%) was higher than guarea macrophylla (4.4±0.9%) or socratea exorrhiza (1.5±0.3%) and significantly different (df = 1, f = 4.2, p = 0.01) between litter treatments (4.1±1.1% no litter vs. 13.4±2.1% litter), and seed germination was significantly different (df = 1, f = 2.6, p = 0.05) between litter treatments (0.6±0.1% no litter vs. 3.7±1.1% litter). selectively-logged terra firme forests showed the same results as unlogged terra firme forests. in palm forests seed predation was significantly different (df = 1, f = 2.6, p = 0.05) between litter treatments (93.1±5.9% no litter vs. 69.3±6.2% litter), seed pathogenic attack was significantly different (df = 2, f = 3.3, p = 0.02) among tree species where socratea exorrhiza (6.7±3.3%) was lower than cecropia latiloba (25.4±3.8%) and guarea macrophylla (16.2±2.3%), and significantly different (df = 1, f = 4.8, p = 0.01) between litter treatments (5.4±1.7% no litter vs. 27.7±3.7% litter). in white sand forests seed germination was significantly different (df = 1, f = 2.5, p = 0.05) between litter treatments (4.9±1.2% no litter vs. 8.4±2.1% litter). myster new zealand journal of forestry science (2021) 51:9 page 4 discussion among the four forest-types: both unlogged and selectively-logged terra firme forests lost the most seed to predators but lost the least seed to pathogens, white sand forests lost the least seed to predators but lost the most seed to pathogens, and the fewest seeds germinated in both terra firme forests and in palm forest. within the two terra firme forests, litter reduced seed predation but increased both seed losses to pathogens and germination, and the smallest-seeded cecropia latiloba lost the most seeds to pathogens. within palm forests addition of litter again reduced predation but increased losses to pathogens, and socratea exorrhiza lost the least seeds to pathogens. within white sand forests litter increased germination. selectively-logged terra firme forests followed the same trends as unlogged terra firme forests but without any significant effects. hypothesis one was supported by the results (most tree seeds were also taken by predators in a singapore tropical forest; wong et al. 1998) and selectively-logged forests had less predation than palm forests. seed predation was greatest in unlogged terra firme forest and least in white sand forest, in medium intensity palm forest and in selectively-logged terra firme (close to 100% loss in africa selectively-logged terra firme forest; hall 2008). the terra firme results could be due to more predator species and abundance in those forests and/ or to a more specific search image for those predators (gripenberg et al. 2019). because most seeds were lost to predation (myster 2014) the variation in how it operates – here for example among species, and on top of/under the litter layer – may be very important in determining table 1: the percentage (mean ± standard error) of the total seeds taken by seed predation among forest-types, among tree seed species, and in the no litter/litter treatment. species terra firme logged terra firme palm white sand cecropia sp. 94.1±1.5/60.3±2.6 90.2±1.6/51.2±2.2 91.7±1.5/55.4±2.3 55.5±2.3/47.3±1.7 guarea macrophylla 96.2±1.2/95.5±1.7 89.6±1.8/45.3±1.2 93.1±2.4/65.8±2.2 51.6±2.2/43.5±2.2 socratea exorrhiza 99.3±3.4/98.6±2.2 85.3±3.3/80.5±1.3 95.4±1.2/88.7±2.8 40.5±2.3/39.2±1.5 table 2: the percentage (mean ± standard error) of the total seeds taken by seed pathogens among forest-types, among tree seed species, and in the no litter/litter treatment. species terra firme logged terra firme palm white sand cecropia sp. 6.2±1.4/33.5±1.2 8.4±2.2/40.1±1.3 9.2±3.1/40.6±3.4 39.7±1.5/43.2±1.1 guarea macrophylla 4.4±3.3/5.2±1.4 5.5±2.4/42.4±3.7 4.4±1.3/30.1±1.8 44.4±3.3/37.7±1.8 socratea exorrhiza 1.1±2.7/1.4±1.9 11.9±1.7/15.4±3.2 3.2±1.7/9.5±2.4 51.3±2.6/45.2±1.8 table 3: the percentage (mean ± standard error) of the total seeds germinated among foresttypes, among tree seed species, and in the no litter/litter treatment. species terra firme logged terra firme palm white sand cecropia sp. 0.3±3.6/7.1±2.2 2.2±1.4/6.5±2.1 0.4±1.5/3.1±2.9 3.4±3.1/10.9±2.4 guarea macrophylla 0.1±2.6/0.4±2.2 5.1±2.4/10.3±1.7 3.1±2.2/2.4±1.5 3.3±2.4/15.2±2.2 socratea exorrhiza 0.5±1.3/1.2±1.6 4.5±2.2/3.8±3.4 1.1±2.4/2.7±2.5 5.1±1.4/8.8±1.3 recruitment in these unflooded amazon forests. indeed because predation on top of the litter layer is mainly by vertebrates and predation below the litter layer is mainly by invertebrates (grogan & galvão 2006), with time as seeds (especially dense, heavy seeds) fall down the litter layer due to gravity, predation may shift from being due primarily to vertebrates to being primarily due to invertebrates. predation may also be due to invertebrates more than vertebrates for seeds of short dispersal (notman & villegas 2005). finally, low levels of seed predation in white sand forests may be related to its low animal richness and abundance (myster 2009). hypothesis two was also supported by the results, with the caveat that it operated only on the seeds that survived predation. most seeds germinated if they could escape predators and pathogens supporting hypothesis three (common in tropical forests world-wide; vazquezyanes & oroza-segovia 1983) and had complex interactions with forest-type, litter addition and tree seed species. hypothesis four was supported mainly for the two terra firme forests and palm forest. litter effects dominated seed losses to pathogens. litter addition increased losses to pathogens in unlogged terra firme, in selectively-logged terra firme, and in palm forest. the species of these pathogens (which may have included the fungal pathogens colletotrichum sp., pythium sp. and fusarium sp. found on tree seeds on landslides in puerto rico; see myster 1997) may have been influenced by the species of neighboring trees (grogan & galvão 2006). there were several interactive effects with tree seed species, that both supported and did not support hypothesis five, as was seen in an africa unflooded tropical forest (hart 1995; norghaner & newbery 2011). outside of these hypotheses, results could also have been influenced by factors such as the ecological characteristics of the seed species used which are often associated with successional status, the biology and available of predators and pathogens, and the quality of seeds as it relates to fruiting phenology. results generally agree with other amazon unlogged terra firme studies that found up to 90% of large seeds were lost to predators (60% after 16 days; russo 2005), mainly to mammals (paine & beck 2007) with at most 12% scatter-hoarded but even those were eaten later. further interactions were found in another unlogged amazon terra firme forest where (1) seed predation rates were higher when monkey dung was present (andresen 2002) and (2) invertebrate seed predation showed distance effects more than vertebrate seed predation (terborgh et al. 1993). in amazon unlogged terra firme studies using palm seeds and seedlings (1) seeds and seedlings of astrocaryum murumera and dipteryx micrantha survived better in gaps than in the understory (cintra & horna 1997), (2) white-lipped peccaries decreased the density of astrocaryum murumera seedlings (silman et al. 2003), (3) attalea maripa seed survivorship was unrelated to distance from individual fruiting palms (salm 2006), (4) the removal of the exocarp and the mesocarp by large mammals increased attalea maripa seed predation by beetles (silvins and fragoso 2002), and (5) litter increased seed and seedling survivorship for astrocaryum murumera but only seed survivorship for dipteryx micrantha (cintra 1997). other amazon unlogged terra firme studies showed that (1) larger seeds were taken by pathogens more than smaller seeds, germination was approximately 43% and pathogen loss was up to 75% depending on species (pringle et al. 2007), (2) secondary dispersal was low (culot et al. 2009), (3) post-dispersal palm seed predation was reduced under litter, but increased under thicker litter (cintra 1997), and (4) after bat defecation seeds were eaten 8% per week with possible satiation (romo et al. 2004). in another study where terra firme forests were selectively-logged for mahogany (sweitenia macrophylla) (1) 40% of seeds were taken by predators and pathogens, and more seeds germinated (36%) compared to an area where mahogany was not logged (grogan & galvão 2006) and (2) intensity of logging did not correlate with seed predation rates (lambert et al. 2005). in a clearcut terra firme forest close to iquitos (all trees 5 cm dbh or greater were cut) (1) dispersal from the surrounding forest was rare, and regeneration came mainly from the seed bank and stumps with multiple sprouts (gorchov et al. 2004) except for cecropia sp. and alchornea triplinervia (gorchov et al. 1993), and (2) predation levels were the same as the primary forest 2-3 years after felling (notman et al. 1996) with most predation occurring at the edge of forest and clear-cut and most seeds germinated (notman et al. 1996). in an amazon study comparing both unflooded and flooded forest-types (myster 2017b) (1) unlogged terra firme/white sand, and várzea (flooded by whitewater)/igapó (flooded by black-water for one month per year) were significantly different for seed predation, seed pathogens and germination, (2) in unlogged terra firme forest seed predators took most seeds regardless of species, (3) in palm forest species were different regardless of seed process, (4) in white sand forest seed predators took most seeds regardless of species, and (5) in várzea forest seed predators took most seeds but with some species differences. looking at these foresttypes together, seed predation losses decreased as the forest became more stressed – perhaps by loss of soil fertility and/or by flooding with nutrient-poor water – while seed pathogens become more important with waterlogged soils and flooding. and seed loss variation among species was always a secondary factor for all effects. the higher plant species richness and complexity of unlogged terra firme (valencia et al. 2004), and thus more seed predators, may explain the increase in seed predation. likewise in species-rich várzea forest, there was more seed predation than in species-poor igapó forest at the same inundation time period. within igapó forests more flooding lead to less predation, just as the wet palm forest had less predation than terra firme. for pathogens, standing water in the palm forest lead to the greatest losses, but increased flooding in igapó forests also lead to increasing loss of seeds to pathogens. myster new zealand journal of forestry science (2021) 51:9 page 5 conclusions seed predation dominates in these unflooded forests (bodmer 1991; notman et al. 1996), and those losses decrease as forests become more stressed with loss of soil fertility (and/or with selective-logging; paine & beck 2007). litter effects dominate seed losses to pathogens and germination interactions with study area, litter and species were complex. most seeds were lost to seed predators over all microsites. while addition of litter reduced seed predation, it increased losses of seed to pathogens (fenner 1985). in addition, most seeds germinated if they could survive predators and pathogens (myster 2014). with such intense predation and losses to pathogens in these forests, seed survival and germination is difficult. differences in forest-type (perhaps related to soil fertility), as well as distributions of litter on the forest floor and variation in tree seed species may, nevertheless, facilitate recruitment. the complexities of forest recruitment, however, continue to be a major challenge for modelers who wish to predict the plant-plant replacements that cause amazon forest community patterns, such as biodiversity. these results and other experiments in unlogged terra firme (myster 2014) suggest other mechanisms or other sov in the workings of these seed processes may play significant roles (muller-landau et al. 2008). towards that goal, i continue to sample and conduct experiments in a one ha plot in igapó flooded forest in peru for plant-plant replacements and recruitment processes (author, unpub. data). competing interests the author declares that he has no competing interests. acknowledgements i thank luis guerra for his help in the field and dr. juan celedonio ruiz macedo, curator of the universidad nacional de la amazonia peruana herbarium, for his help in tree seed identification. i also thank t. fahey, h. asselin, s. mikheyev, j. powers, j. dalling, h. makinen, p. campanello, c. keller, c. lusk, v. pillar and c. lacroix for commenting on a past draft of the manuscript. references andresen, e. ( 2002). primary seed dispersal by red howler monkeys and the effect of defecation patterns on the fate of dispersed seeds. biotropica, 34, 261272. https://doi.org/10.1111/j.1744-7429.2002. tb00537.x blaney, c.s., & kotanen, p.m. (2001). post-dispersal losses to seed predators: an experimental comparison of native and exotic old field plants. canadian journal of botany, 79, 284–292. https://doi.org/10.1139/ b01-003 bodmer, r.e. (1991). strategies of seed dispersal and seed predation in amazonian ungulates. biotropica, 23, 255-261. https://doi.org/10.2307/2388202 cintra, r. (1997). leaf litter effects on seed and seedling predators of the palm astrocaryum murumura and the legume tree dipteryx micrantha in amazonian forest. journal of tropical ecology, 13, 709-725. https://doi.org/10.1017/s0266467400010889 cintra, r., & horna, v. (1997). seed and seedling survival of the palm astrocaryum murumura and the legume tree dipteryx micrantha in gaps in amazonian forest. journal of tropical ecology, 13, 257-277. https://doi.org/10.1017/s0266467400010440 culot, l, huuynen, m, gerard, p & heymann, ew. (2009). short-term post-dispersal fate of seeds defecated by two small primate species (saguinus mystax and s. fuscicollis) in the amazonian forest of peru. journal of tropical ecology, 25, 229-238. https:// doi.org/10.1017/s0266467409005860 daly, d.c., & prance, g.t. (1989). brazilian amazon. in d.g. campbell & h.d. hammond (eds.). floristic inventory of tropical countries: the status of plant systematics, collections, and vegetation, plus recommendations for the future (pp. 401-426). new york: new york botanical garden. fenner, m. (1985). seed ecology. new york, usa: chapman and hall. https://doi.org/10.1007/97894-009-4844-0 gentry, a.a. (1993). field guide to woody plants of northwest south america (colombia, ecuador, peru). washington, dc: conservation international. gorchov, d.l., cornejo, f., ascorra, c. & jaramillo, m. (1993). the role of seed dispersal in the natural regeneration of rain forest after strip-cutting in the peruvian amazon. vegetatio, 107/108, 339-349. https://doi.org/10.1007/978-94-011-1749-4_23 gorchov, d.l., palmeirim, j.m., jaramillo, m. & ascorra, c.f. (2004). dispersal of seeds of hymenaea courbaril (fabaceae) in a logged rain forest in the peruvian amazonian. acta amazonia, 34, 251-259. https:// doi.org/10.1590/s0044-59672004000200014 gripenberg, s., basset, y., lewis, o.t., christopher, j., terry, d., wright, s.j., simon, i., fernandez, d.c., cedeno-sanchez, m., rivera, m., barrios, h., brown, j.w., calderon, o., cognato, a.i., kim, j., miller, s.e., morse, g.e., pinzon-navarro, s., quicke, d.l.j., robbins, r.k., salminen j., & vesterinen, e. (2019). a highly resolved food web for insect seed predators in a species-rich tropical forest. ecology letters, 22, 1638–1649. https://doi.org/10.1111/ele.13359 grogan, j., & galvão, j. (2006). factors limiting postlogging seedling regeneration by big-leaf mahogany (swietenia macrophylla) in southeastern amazonian brazil, and implications for sustainable management. biotropica, 38, 219-228. https://doi. org/10.1111/j.1744-7429.2006.00121.x grubb, p.j. (1977). the maintenance of species richness in plant communities: the importance of the regeneration niche. biological review of the myster new zealand journal of forestry science (2021) 51:9 page 6 https://doi.org/10.1111/j.1744-7429.2002.tb00537.x https://doi.org/10.1111/j.1744-7429.2002.tb00537.x https://doi.org/10.1139/b01-003 https://doi.org/10.1139/b01-003 https://doi.org/10.2307/2388202 https://doi.org/10.1017/s0266467400010889 https://doi.org/10.1017/s0266467400010440 https://doi.org/10.1017/s0266467409005860 https://doi.org/10.1017/s0266467409005860 https://doi.org/10.1007/978-94-009-4844-0 https://doi.org/10.1007/978-94-009-4844-0 https://doi.org/10.1007/978-94-011-1749-4_23 https://doi.org/10.1590/s0044-59672004000200014 https://doi.org/10.1590/s0044-59672004000200014 https://doi.org/10.1111/ele.13359 https://doi.org/10.1111/j.1744-7429.2006.00121.x https://doi.org/10.1111/j.1744-7429.2006.00121.x cambridge philosophical society, 52, 107-111. https://doi.org /10.1111/j .1469-185x.1977. tb01347.x hall, j.s. (2008). seed and seedling survival of african mahogany (entandrophragma spp.) in the central african republic: implications for forest management. forest ecology and management, 255, 292-299. https://doi.org/10.1016/j. foreco.2007.09.050 hart, t.b. (1995). seed, seedling and sub-canopy survived in monodominant and mixed forests of the ituri forest, africa. journal of tropical ecology, 11, 443-459. https://doi.org/10.1017/ s0266467400008919 hice, c.l., velazco, p.m., &willig, m.r. (2004). bats of the reserva nacional allpahuayo -mishana, northeastern peru, with notes on community structure. acta chiromete, 6, 319-334. https://doi. org/10.3161/001.006.0210 honorio, e.n. (2006). floristic relationships of the tree flora of jenaro herrera, an unusual area of the peruvian amazon. m.s. thesis, university of edinburgh, edinburgh, uk. hulme, p.e. (1994). post-dispersal seed predation in grassland: its magnitude and sources of variation. journal of ecology, 82, 645-652. https://doi. org/10.2307/2261271 jansen, p.a., & zuidema, p.a. (2001). logging, seed dispersal by vertebrates, and natural regeneration of tropical timber trees. in r. fimbel, a. grajal, & j. robinson (eds.), the cutting edge. conserving wildlife in logged tropical forests (pp 35-59). new york: columbia university press. https://doi. org/10.7312/fimb11454-006 johnson, a.m. (1976). the climate of peru, bolivia, and ecuador. in w. schwerdtfeger (ed.), climates of central and south america, world survey of climatology, (vol. 12; pp. 147–210). amsterdam: elsevier scientific. kalliola, rs, jukka, m, puhakkaa, m, &rajasilta, m. (1991). new site formation and colonizing vegetation in primary succession on the western amazon floodplains. journal of ecology, 9, 877-901. https:// doi.org/10.2307/2261087 lambert, t.d., malcolm, j.r., & zimmerman, b.l. (2005). effects of mahogany (swietenia macrophylla) logging on small mammal communities, habit structure, and seed predation in the southeastern amazon basin. forest ecology and management, 206, 381-398. https://doi.org/10.1016/j. foreco.2004.11.012 muller-landau, h.c., wright, s.j., calderon, o., condit, r., & hubbell, s.p. (2008). interspecific variation in primary seed dispersal in a tropical forest. journal of ecology, 96, 653-667. https://doi.org/10.1111/ j.1365-2745.2008.01399.x myster, r.w. (1994). contrasting litter effects on oldfield tree germination and emergence. vegetatio, 114, 169-174. myster, r.w. (1997). seed predation, disease and germination on landslides in neotropical lower montane wet forest. journal of vegetation science, 8, 55-64. https://doi.org/10.2307/3237242 myster, r.w. (2003). seed regeneration mechanisms over fine spatial scales on recovering coffee plantation and pasture in puerto rico. plant ecology, 166, 199205. https://doi.org/10.1023/a:1023262109867 myster, r.w. (2009). plant communities of western amazonia. the botanical review, 75, 271-291. https://doi.org/10.1007/s12229-009-9032-1 myster, r.w. (2012). spatial and temporal heterogeneity of light and soil water along a terra firme transect in the ecuadorian amazon. canadian journal of forest research, 42, 1-4. https://doi.org/10.1139/ x11-168 myster, rw. (2014). interactive effects of flooding and tree fall gap formation on terra firme forest seed and seedling mechanisms and tolerances in the amazon. community ecology, 15, 212-221. https:// doi.org/10.1556/comec.15.2014.2.10 myster, r.w. (2015). seed predation, pathogens and germination in primary vs. secondary cloud forest at maquipucuna reserve, ecuador. journal of tropical ecology, 31, 375-378. https://doi. org/10.1017/s0266467415000127 myster, rw. (2017a). does the seed rain limit recruitment, regeneration and plant community dynamics? ideas in ecology and evolution, 10, 1-12. https:// doi.org/10.4033/iee.2017.10.3.c myster, r.w. (2017b). comparing and contrasting flooded and unflooded forests in western amazonia: seed predation, seed pathogens, germination. community ecology, 18, 169-174. https://doi. org/10.1556/168.2017.18.2.6 myster, rw. (2017c). forest structure, function and dynamics in western amazonia. oxford, uk: wiley-blackwell. https://doi. org/10.1002/9781119090670 myster, r.w. (2018). the nine classes of plant-plant replacement. ideas in ecology and evolution, 11, 29-34. https://doi.org/10.4033/iee.2018.11.4.n myster, r.w., & everham iii, e.m. (1999). germination cues across the disturbance regime in the puerto rican rainforest. tropical ecology, 40, 1-10. norghauer, j.m., & newbury, d.m. (2011). seed fate and seedling dynamics after masting in two african rain forest trees. ecological monographs, 81, 443469. https://doi.org/10.1890/10-2268.1 notman, e., & gorchov, d.l. (2001). variation in postdispersal seed predation in mature peruvian myster new zealand journal of forestry science (2021) 51:9 page 7 https://doi.org/10.1111/j.1469-185x.1977.tb01347.x https://doi.org/10.1111/j.1469-185x.1977.tb01347.x https://doi.org/10.1016/j.foreco.2007.09.050 https://doi.org/10.1016/j.foreco.2007.09.050 https://doi.org/10.1017/s0266467400008919 https://doi.org/10.1017/s0266467400008919 https://doi.org/10.3161/001.006.0210 https://doi.org/10.3161/001.006.0210 https://doi.org/10.2307/2261271 https://doi.org/10.2307/2261271 https://doi.org/10.7312/fimb11454-006 https://doi.org/10.7312/fimb11454-006 https://doi.org/10.2307/2261087 https://doi.org/10.2307/2261087 https://doi.org/10.1016/j.foreco.2004.11.012 https://doi.org/10.1016/j.foreco.2004.11.012 https://doi.org/10.1111/j.1365-2745.2008.01399.x https://doi.org/10.1111/j.1365-2745.2008.01399.x https://doi.org/10.2307/3237242 https://doi.org/10.1023/a:1023262109867 https://doi.org/10.1007/s12229-009-9032-1 https://doi.org/10.1139/x11-168 https://doi.org/10.1139/x11-168 https://doi.org/10.1556/comec.15.2014.2.10 https://doi.org/10.1556/comec.15.2014.2.10 https://doi.org/10.1017/s0266467415000127 https://doi.org/10.1017/s0266467415000127 https://doi.org/10.4033/iee.2017.10.3.c https://doi.org/10.4033/iee.2017.10.3.c https://doi.org/10.1556/168.2017.18.2.6 https://doi.org/10.1556/168.2017.18.2.6 https://doi.org/10.1002/9781119090670 https://doi.org/10.1002/9781119090670 https://doi.org/10.4033/iee.2018.11.4.n https://doi.org/10.1890/10-2268.1 low-land tropical forest and fallow agricultural sites. biotropica, 33, 621-636. https://doi. org/10.1111/j.1744-7429.2001.tb00220.x notman, e., gorchov, d.l., & cornejo, f. (1996). effect of distance, aggregation and habitat on levels of seed predation for two mammal-dispersed neotropical rain forest tree species. oecologia, 106, 221-227. https://doi.org/10.1007/bf00328602 notman, e.m., & villegas, a.c. (2005). patterns of seed predation buy vertebrate versus invertebrate seed predators among different plant species, seasons and spatial distributions. in p.m. forget, j.e. lambert, p.e. hulme, & s.b. vander wall (eds), seed fate. (pp. 55-75). cab international. https:// doi.org/10.1079/9780851998060.0055 paine, c.e.t., & beck, h. (2007). seed predation by neotropical rainforest mammals increases diversity in seedling recruitment. ecology, 88, 3076-3087. https://doi.org/10.1890/06-1835.1 pitman, n.c., terborgh, j.w., silman, m.r., nunez, p.v., neill, d.a., ceron, c.e., palacios, w.a. & aulestia, m. (2001). dominance and distribution of tree species in upper amazonian terra firme forests. ecology, 82, 2101-2117. https://doi.org/10.1890/00129658(2001)082[2101:dadots]2.0.co;2 pringle, e.g., alvarez-loayzu, p. & terborgh, j. (2007). seed characteristics and susceptibility to pathogen attack in tree seeds of the peruvian amazon. plant ecology, 193, 211-222. https://doi.org/10.1007/ s11258-006-9259-4 romo, m., tuomisto, h. & loiselle, b.a. (2004). on the density-dependence of seed predators in dipteryx micrantha, a bat-dispersed rain forest tree. oecologia, 140, 76-85. https://doi.org/10.1007/ s00442-004-1502-x romoleroux, k., foster, r., valencia, r., condit, r., balslev, h. & losos, e. (1997). especies lenosas (dap => 1 cm) encontradas en dos hectáreas de un bosque de la amazonia ecuatoriana. in r. valencia, & h.r. balslev (eds.), estudios sobre diversidad y ecologia de plantas (pp. 189–215). quito, ecuador: pontificia universidad católica del ecuador. russo, s.e. (2005). linking seed fate to natural dispersal patterns: factors affecting predation and scatterhoarding of virola calophylla seeds in peru. journal of tropical ecology, 21, 243-253. https://doi. org/10.1017/s0266467405002312 saaksjarvi, i., haataja, s., neuvonen, s., gauld, i.d., jussila, r., salo, j., & burgos, a. (2004). high local species richness of parasitic wasps (hymenoptera: ichneumonidae; pimplinae and rhyssinae) from the lowland rainforests of peruvian amazonia. ecological entomology, 29, 735–743. https://doi. org/10.1111/j.0307-6946.2004.00656.x salm, r. (2006). invertebrate and vertebrate seed predation in the amazonian palm attalea maripa. biotropica, 38, 558-560. https://doi.org/10.1111/ j.1744-7429.2006.00163.x sas (1985). user’s guide: statistics, ver. 5. sas institute inc. cary, nc. shirota, o, takizawa, k, sekita, s, satake, m, hirayama, y., hakamata, y., hayashi, t. & yanagawa, t. (1997). antiandrogenic natural diels−alder-type adducts from brosimum rubescens. journal of natural products, 60, 997–1002. https://doi.org/10.1021/ np9607215 silman, m.r., terborgh, j.w. & kiltie, r.a. (2003). white-lipped peccaries decreased the density of astrocaryum murumra seedlings. ecology, 84, 431-438. https://doi.org/10.1890/00129658(2003)084[0431:proadr]2.0.co;2 silvins, k.m., & fragoso, .jm.v. (2002). pulp handling by vertebrate seed dispersers increases palm seed predation by bruchid beetles in the northern amazon. journal of ecology, 90, 1024-1032. https:// doi.org/10.1046/j.1365-2745.2002.00728.x stropp, j., van der sleen, p., assuncao, a.a., silva, a.l., & ter steege, h. (2011). tree communities of whitesand and terra-firme forests of the upper rio negro. acta amazonica, 41, 521-544. https://doi. org/10.1590/s0044-59672011000400010 terborgh, j., losos, e., riley, m.p., & bolanos-riley, m. (1993). predation by vertebrates and invertebrates on the seeds of five canopy tree species of an amazonian forest. vegetatio, 107/108, 375-386. https://doi.org/10.1007/978-94-011-1749-4_26 tuomisto, h., ruokolainen, k., & yli-halla, m. (2003). dispersal, environment and floristic variation of western amazonian forests. science, 299, 241-244. https://doi.org/10.1126/science.1078037 valencia, r., foster, r.b., villa, g., condit, t.r., svenning, j., hernandez, c., romoleroux, k., losos, e., magard, e., & balslev, s.l. (2004). tree species distributions and local habitat variation in the amazon: large forest plot in eastern ecuador. journal of ecology, 92, 214-229. https://doi.org/10.1111/j.00220477.2004.00876.x vazquez-yanes, c., & orozco-segovia, a. (1983). ecophysiology of seed germination in the tropical humid forests of the world: a review. in e. medina, h.a. mooney, & c. vazquez-yanes. (eds.), physiological ecology of plants of the wet tropics (pp. 37-50) proceedings of an international symposium, mexico. https://doi.org/10.1007/978-94-0097299-5_5 vieira, d.l.m., & scariot, a. (2006). effects of logging, liana tangles and pasture on seed fate of dry forest tree species in central brazil. forest ecology myster new zealand journal of forestry science (2021) 51:9 page 8 https://doi.org/10.1111/j.1744-7429.2001.tb00220.x https://doi.org/10.1111/j.1744-7429.2001.tb00220.x https://doi.org/10.1007/bf00328602 https://doi.org/10.1079/9780851998060.0055 https://doi.org/10.1079/9780851998060.0055 https://doi.org/10.1890/06-1835.1 https://doi.org/10.1890/0012-9658(2001)082%5b2101:dadots%5d2.0.co;2 https://doi.org/10.1890/0012-9658(2001)082%5b2101:dadots%5d2.0.co;2 https://doi.org/10.1007/s11258-006-9259-4 https://doi.org/10.1007/s11258-006-9259-4 https://doi.org/10.1007/s00442-004-1502-x https://doi.org/10.1007/s00442-004-1502-x https://doi.org/10.1017/s0266467405002312 https://doi.org/10.1017/s0266467405002312 https://doi.org/10.1111/j.0307-6946.2004.00656.x https://doi.org/10.1111/j.0307-6946.2004.00656.x https://doi.org/10.1111/j.1744-7429.2006.00163.x https://doi.org/10.1111/j.1744-7429.2006.00163.x http://pubs.acs.org/author/shirota%2c+osamu http://pubs.acs.org/author/takizawa%2c+kenji http://pubs.acs.org/author/sekita%2c+setsuko http://pubs.acs.org/author/satake%2c+motoyoshi http://pubs.acs.org/author/hirayama%2c+yutaka http://pubs.acs.org/author/hakamata%2c+yusuke http://pubs.acs.org/author/hayashi%2c+tatsuo http://pubs.acs.org/author/yanagawa%2c+takuma https://doi.org/10.1021/np9607215 https://doi.org/10.1021/np9607215 https://doi.org/10.1890/0012-9658(2003)084%5b0431:proadr%5d2.0.co;2 https://doi.org/10.1890/0012-9658(2003)084%5b0431:proadr%5d2.0.co;2 https://doi.org/10.1046/j.1365-2745.2002.00728.x https://doi.org/10.1046/j.1365-2745.2002.00728.x https://doi.org/10.1590/s0044-59672011000400010 https://doi.org/10.1590/s0044-59672011000400010 https://doi.org/10.1007/978-94-011-1749-4_26 https://doi.org/10.1126/science.1078037 https://doi.org/10.1111/j.0022-0477.2004.00876.x https://doi.org/10.1111/j.0022-0477.2004.00876.x https://doi.org/10.1007/978-94-009-7299-5_5 https://doi.org/10.1007/978-94-009-7299-5_5 and management, 230, 197-205. https://doi. org/10.1016/j.foreco.2006.05.002 wilson, e., underwood, m., puckrin, o., letto, k., doyle, r., caravan, h., camus, s., & bassett, k. (2013). the arcsine transformation: has the time come for retirement? http://www.mun.ca/biology/ dschneider/b7932/b7932finall0dec2010 wong, t.c.m., sodhi n.s., & turner, i.m. (1998). artificial nest and seed predation experiments in tropical lowland rainforest remnants of singapore. biological conservation, 85, 97-104. https://doi. org/10.1016/s0006-3207(97)00145-6 myster new zealand journal of forestry science (2021) 51:9 page 9 https://doi.org/10.1016/j.foreco.2006.05.002 https://doi.org/10.1016/j.foreco.2006.05.002 http://www.mun.ca/biology/dschneider/b7932/b7932finall0dec2010 http://www.mun.ca/biology/dschneider/b7932/b7932finall0dec2010 https://doi.org/10.1016/s0006-3207(97)00145-6 https://doi.org/10.1016/s0006-3207(97)00145-6 characterisation of harvest residues on new zealand’s steepland plantation cutovers campbell harvey* and rien visser school of forestry, university of canterbury, new zealand *corresponding author: campbell.harvey@pg.canterbury.ac.nz (received for publication 23 july 2021; accepted in revised form 25 february 2022) abstract background: timber harvesting in new zealand’s plantation forests results in relatively large volumes of woody residues being generated. while a proportion of these residues are concentrated at the landings where the trees are processed, the majority of residues are distributed throughout the cutover. harvest residues present a biomass market opportunity, however managing un-merchantable residues remains essential as the material can present a mass mobilisation risk. quantifying cutover residues in terms of volume provides an important step for marketing and for improving post-harvest management. methods: a refined line intersect sampling (lis) method was used to measure the cutover residues at 17 recently harvested steepland sites. these covered a range of whole tree harvesting systems, silviculture and geographical locations. the harvesting sites varied in size from 2.3 to 41.1 ha, with an average of 11x 60 m lis transect plots completed at each site. woody harvest residues >25 mm in diameter were measured. results: the median volume of woody residues was 88 m3/ha, ranging from 0 m3/ha in an area swept bare, up to 580 m3/ ha in an area severely impacted by windthrow prior to harvest. a distribution of volumes by plot showed a positive skew with an interquartile range of 87 m3/ha. timber that was considered merchantable as a log at the time of harvest, being >10 cm in small end diameter and >4 m in length, accounted for a median of 11 m3/ha. residues >10 cm in small end diameter and >80 cm in length that could make a viable biomass product, described as ‘binwood’, accounted for a further 19 m3/ha at the median. cutovers harvested with cable-based systems had greater median total residue volumes than those harvested with ground-based systems (110 m3/ha versus 68 m3/ha) however the felling method employed made no significant difference to total residue volumes. conclusions: this study provides cutover residue measurements that can be used to improve post-harvest management, as both a substantial opportunity for improved crop utilisation and also for reducing mobilisation risk. it also provides a contemporary benchmark against which to measure change as harvesting technology or methodology develops. new zealand journal of forestry science harvey & visser new zealand journal of forestry science (2022) 52:7 https://doi.org/10.33494/nzjfs522022x174x e-issn: 1179-5395 published on-line: 25/03/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access or woody debris (mpi 2017). there are two broad categories of above ground biomass relating to clearfell, whole-tree harvesting (wth). the first is on or near the landing; that is residues that are discarded from the processing operation once the trees are extracted. the other is residues in the cutover as a result of processing or breakage during harvesting and also natural attrition during the growing cycle. while stumps and belowground root systems are also residual biomass, they are not typically regarded as harvesting residues. introduction the volume of timber harvested and sold from new zealand production forests has increased rapidly from 20m  m3 in 2010 to over 30m  m3 since 2018 (nzfoa 2019/2020). the commonly-accepted goal at the time of harvest is to maximise the value of the products that can be processed from the extracted trees, which are typically transported from the forest as logs (murphy 2005). unmerchantable materials left behind after harvest are known as residues, but can also be referred to as slash keywords: slash; biomass; environmental impacts; harvesting operations http://creativecommons.org/licenses/by/4.0/), harvey & visser new zealand journal of forestry science (2022) 52:7 page 2 residue volumes and distribution can vary in the cutover; depending on a range of factors including: felling method, terrain, crop type and extraction method (hall 1999b). there are a range of generalisations in new zealand-based literature of the typical volume of residual biomass from harvesting plantation-grown radiata pine. hall (2001) noted that a typical pine tree harvested in new zealand yields 0.10  m3 of aboveground residual biomass. extrapolating hall’s figure, typical crops that range from 250-350 stems per hectare at economic maturity (mead 2013) could be expected to yield residue volumes ranging from 25–35  m3/ha. goulding (2005) presents experiential evidence, stating harvest residue volumes may range from 5 30% of total standing volume (tsv), depending on terrain, silviculture and degree of malformation. in a review of literature for residue mobilisation risk, visser et al (2018) reported a range of study values from new zealand and international studies, but indicated an expected average of 75 m3/ha for new zealand plantation conditions. harvest residues have been studied intermittently in new zealand since the 1960’s. manual surveys of residues are inherently difficult to complete, and it is not practicable to measure all residue volumes over large areas. early work focussed on quantifying only the residual, merchantable logs on a cutover, and warren and olsen (1964) produced the first recognised method for economically quantifying those residues, referred as the ‘line intersect sampling’ (lis) method. lis is based on the principles underpinning buffon’s needle problem (buffon 1777) and has been refined for applications such as forest fire research for quantifying ground fuels, or log waste fields (fraver et al. 2018; sikkink & keane 2008; van wagner 1968). messinger (1974) published the first new zealandbased literature review of total residual biomass (merchantable and un-merchantable) which appears to have been prompted by “public criticism due the obvious waste and cluttered appearance” of harvesting residues on forest cutovers. messinger reported new zealand radiata pine cutover residue volumes ranging from 298 to 1068  ft3/ac (21 – 75 m3/ha) in the years spanning 1971–1974. more recently, the new zealand logging industry research organisation published a series of studies on the volume and management of harvesting residues (hall 1994, 1995, 1996, 1998, 1999a, 1999b, 2000a, 2000b, 2013; hall & mcmahon 1997). this spanned multiple facets with a particular focus on reducing cost and risk associated with the long-term storage of the material post-harvest. hall (1999b) used a detailed lis survey to demonstrate the spatial variation of residue volumes across a cutover. the study found residue volumes ranging from 1-280  m3/ha across six sites with three different harvest systems. notably, the cable harvests resulted in large accumulations of residue in the lowest point of the cutover (the gullies) whereas the groundbased systems resulted in a more uniform distribution with a trend of higher residue accumulations at greater distance from the landing. there has been significant variation in methodology and results from biomass assessments in radiata pine cutovers across the world. cut-to-length (ctl) harvest operations in australian radiata pine plantations have been reported to retain 52 oven-dry tonnes of residues (needles, cones bark and wood) per hectare on the cutover (smethurst & nambiar 1990) and more recently, a range from 43–151 green tonnes per hectare (ghaffariyan 2013) before post-harvest biomass extraction. many studies report the extracted volume from biomass harvesting operations. in chile, harvested radiata pine plantations are expected to yield 12–14 tonnes of dry biomass per hectare on average (acuña et al. 2017), where earlier estimates were in the range of 45–80 m3/ha (guzmán 1984). in spain, a case study of the oka river basin estimated that the 8764 hectare radiata pine resource may be able to supply 0.72 tonnes of residual biomass per hectare per year to an energy market (mateos & edeso 2015); implying a yield of approximately 22 tonnes per hectare at an average harvest age of 30 years (mateos & ormaetxea 2018). merino et al. (2004) argue that a ‘lack’ of decomposing pinus radiata slash on a sensitive spanish cutover site may be 35 tonnes per hectare. the variation is reflective of the various climates, topographies and biomes in which radiata pine is grown along with the silvicultural and harvesting methods employed on individual crops. volumes of residue on the cutover following harvest can be of importance to forest owners if the material represents waste or unrealised value. harvesting system selection is critical to the economic viability of the operation. most harvesting operations choose to split stem-wood extraction and residual biomass extraction into two separate harvesting processes. mcmahon et al. (1998) demonstrated that a cable harvesting operation could be 26% more productive if only extracting stems >30 cm diameter at the large end (led) & >3.7 m long, when compared to extracting all material >10 cm in diameter at the small end (sed) & >3.7 m long. it highlights that if the removal of residue material from the cutover is integrated into a cable harvesting operation, the added inefficiencies may result in increased harvesting cost. biomass for bioenergy has surged in popularity in parts of the european union since the 1970’s oil crisis (telenius 2006); for example sweden has recently invested over 1.68 billion euros in larger scale combined heat and power using forest biomass, with further projects also planned (haaker 2017). emergence of new bioenergy markets has renewed interest in largescale harvest residue recovery alongside traditional harvesting in new zealand plantations (hall 2013; hall & evanson 2007; visser et al. 2019; visser et al. 2009, 2010). these new bioenergy markets include fuel for industrial heat, domestic heat, transport and more (east harbour management services & scion n.d.) and involve processing raw harvest residues into different forms for specific applications. to date there has been limited market opportunities for harvest residues in many new zealand regions (visser et al. 2018). transport costs, due to long cart distances, low material density and inconsistent quality (hall & evanson 2007) have been suggested as reasons for underutilisation to date – importantly viewed in context with alternative energy sources (e.g. coal). hall (2001) suggests that energy density (as a function of moisture content) has little effect on transport cost in modelling biomass distribution networks, yet kent et al. (2011) advise the contrary. critical to the question of optimising transport efficiency is the nature of the load (raw/bundled/comminuted, wet/seasoned/dry), local standards for heavy vehicle design and permissible loads on road networks. it also needs to be acknowledged that forest residues play an important part in the nutrient cycle in living soils (bray & gorham 1964) and also aquatic ecosystems. needles, flowers and woody biomass form the nutrientrich layer of litter (or ‘duff ’) (ballard & will 1981a) that decomposes and releases nutrients which sustain soils. silviculture, windthrow and harvesting largely dictate the timing of large changes of litter volume on the forest floor and hence nutrient input. harvest residues in the form of large woody debris (lwd) in streambeds provide stability to highly mobile stream beds such as pumice substrates, and complexity to flow, allowing higher populations of invertebrates to be sustained (baillie et al. 1998). however high concentrations of lwd contribute to high measures of dissolved organic carbon which can promote bacterial slime growth and exacerbate oxygen depletion (collier & bowman 2003). harvesting and also stream cleaning practices can therefore be shown to have positive or negative impacts on soils and the life of waterways (baillie 1999; froehlich et al. 1972; swanson et al. 1976). the role of residues in soil nutrient recycling is presently given little regard in new zealand planning frameworks or forestry best practice guidelines, for example the environmental code of practice (nzfoa 2007) only highlights the protection of soil. a case study where 100% of harvest residues and litter fall were removed over 16 years on a pumice site in kaingaroa forest showed that residues on the forest floor do contribute to forest productivity (ballard & will 1981b). in the eastern usa, incorporating residues into the top soil layers improved early site productivity with diminishing benefits as the stand aged (maier et al. 2012). some plantation forests where soil fertility has traditionally been poor are carefully managed for nutrient loss (beets et al. 2001; wilks & wang 2009). developments in tree breeding and mycorrhizal fungi have improved growth and nutrient availability for plantation radiata pine (theodorou & bowen 1970), potentially reducing the dependency on harvesting residues for crop yield. plantation forestry land in new zealand’s most erosion-prone regions is susceptible to slumping, landslides, debris flows and debris avalanches following harvest (phillips et al. 2012). these erosion processes can mobilise and deposit harvest residues far from their source on the cutover (cave et al. 2017). landslip risk increases when soil moisture levels exceed a sitespecific critical water content (crozier 1999) amongst other factors such as the declining strength given to soil by root networks as they decay (phillips et al. 2015). forest harvesting of any type (clearfell, coup, selective etc.) decreases rainfall interception (phillips et al. 2012), increasing the volume of rainfall hitting the forest floor, contributing to soil moisture levels and therefore increasing landslip risk during extreme rainfall events. while mass movements are part of natural erosion processes (bloomberg & davies 2012; phillips et al. 2012), the increased frequency and entrainment of harvest residues as a result of cyclical growing and harvesting fuels debate about production forestry as an appropriate land use in erosion-prone catchments (phillips et al. 1996). steepland forest harvesting is almost exclusively done as clearfell in new zealand (visser 2018), which is where large, contiguous areas of similar age-class forest are felled, leaving forest land temporarily un-stocked with trees. it is predicted that steepland forests will yield 40 60% of the annual harvest volume over the coming years (ffr 2010) and it is expected that most or all of this volume is clear-felled. the word steepland is not officially recognised however (oxford university press 2018) , nor is it universal (gomez et al. 2010) yet steepland is used frequently throughout published literature on forestry in both new zealand and overseas. this manuscript adopts the definition of steepland as ‘an area of land generally unsuitable for ground-based logging systems to operate without significant earthworks or traction assistance’. the breakpoint is typically where rolling hill country, which can be traversed by wheeled or tracked harvesting machinery with little trail construction, transitions to steepland – where significant construction, traction assistance or cable-based systems must be used to extract timber from the cutover. notably however, ground-based harvesting systems are frequently used on steep terrain (berkett 2012) but with increased earthworks requirements. while technically, biomass refers to the mass of living organisms, including plants, animals, and fungi, for the purpose of managing the residual woody biomass as either potential resource, or a mobilisation risk, it is important to define a lower bound for the size of biomass being considered. various studies have set different limiting diameters between coarse woody debris (cwd) and fine woody debris (fwd). the usfs down woody materials field guide sets the limit at 3 inches (usfs 2011), and hall (1999b) refers to branches 0–25 mm as “small”; not defining a diameter boundary between cwd and fwd. in other studies, the cwd-fwd diameter boundaries have been set to 25 mm (wei et al. 1997), 70 mm (manies et al. 2005) and 100 mm (harmon et al. 1995). while there is no clear precedent set in literature, hall (1999b) showed that for the pine plantations, approximately 90% of lis transect intersections occur with woody biomass <25 mm in diameter, while only contributing on average 17% to the volume on site. as such, 25 mm is considered to be a reasonable lower-bound diameter for the purposes of this study and allows direct comparison with hall’s previous work. harvey & visser new zealand journal of forestry science (2022) 52:7 page 3 table 1: description of the study sites the purpose of this research is to establish a current estimate for harvest residue volumes remaining on steepland cutovers, and also provide a detailed characterisation of size (both diameter and length) of ‘potentially merchantable’ residual timber. such detailed information serves to improve our understanding of cutover harvest residues as a resource, and also gauge for risk if mobilised by erosion processes. it is recognised that both silviculture and harvesting practices change over time, so this is a snapshot that reflects current practices and sets a benchmark to measure future performance against. methods the sampling procedure was based on the us forest service (usfs) method for measuring ‘down woody materials’ (dwm) (usfs 2011). harvesting boundaries of each site were reconstructed in a geographic information system (gis) with a grid of plot centres overlaid. the approximate coverage was one lis plot per 1.8 ha of harvest area. each lis plot consisted of three transect lines, the first oriented in a random direction (random number from 1–360°), and the following two lines oriented at 120° to the first, making a trigonal planar shape when viewed from above. the shape reduces orientation bias in sampling compared to a transect in a single direction, or one with a rightangle (e.g. van wagner 1968) while remaining relatively straightforward to establish on a steep site. each transect line is length-corrected to 20 m on the horizontal plane by measuring the average terrain slope along the axis of the transect with a handheld clinometer and adjusting by the cosine of the slope. harvey & visser new zealand journal of forestry science (2022) 52:7 page 4 van wagner’s (1968) governing equation (1) for the volume per area on a flat surface relates the diameters of residues and length of transect line to volume (per hectare). the equation is used on the plot scale, rather than on individual transects to reduce orientation bias impacting the results. v = π2σd2/8l (1) where: v is volume per hectare (m3/ha), d is the diameter of the intersected particle (cm) and l is the horizontal length of transect line (m). three preliminary sites (site codes: gt, mh and tp – see table 1) were measured with the plot dimensions described, capturing the mid-length diameter of all ‘sound’ pieces of harvesting residue that were >25 mm in diameter where they intersected the transect. results from the three initial sites were used to refine the method, including a reduced transect length for residues with diameters <50 mm at the intersection point. the refined method for the remaining 14 sites involved measuring mid-length diameters of all residues >25 mm in diameter at the intersection point, from the plot centre to 5 m (horizontal) along each transect. from 5–20 m along each transect line, only residues with a diameter >50 mm at the intersection point had their mid-length diameters recorded. the effect of the refined method is a reduced workload due to the relative abundance of material <50 mm in diameter on a cutover. sample plot centres were moved or transects shortened (with the actual length recorded) due to safety concerns on occasion; usually due to terrain features such as bluffs. transects that extended beyond site code region study area (ha.) no. plots approx. trv (m3/ha) felling extraction gj canterbury/ waitaha 8.7 7 472 mechanised ground-based gt 31.0 17 546 mechanised cable mh 12.6 7 motor-manual ground-based gn tasman/ te tai-o-aorere 9.5 8 611 mechanised cable mg 36.8 20 392 mechanised ground-based ht gisborne/ te tai rāwhiti 25.3 18 553 motor-manual cable pk 23.0 10 507 motor-manual cable ma 16.7 7 594 mechanised cable mc 8.3 6 866 mechanised cable pe 6.9 9 507 motor-manual cable hf 13.9 11 553 motor-manual cable mo marlborough/ te tauihu-o-te-waka 41.1 18 mechanised ground-based tp 21.2 13 407 mechanised ground-based pg wellington/ te whanga-nui-a-tara 2.3 2 746 mechanised ground-based pc 5.9 6 746 mechanised cable rk 6.1 8 795 motor-manual cable tk otago/ōtākou 33.5 18 841 mechanised cable table 1: harvesting site details. the harvesting boundary also had length to the boundary measured to ensure the results were not impacted by apparent low volume. high spatial variance of individual plot volumes was expected in this study; therefore, each plot was also classified by its location; either ‘spur’, ‘gully’ or ‘face’. average terrain slope, aspect, measure of terrain shape, distance from the landing and the nearest track were either collected at the plot or measured in a gis post-visit. other data collected, where available from the hosting forest company, included stand age, silviculture regime, harvesting system (felling/extraction/processing), expected merchantable volume and actual volume recovered. from these supplied data, log grade outturn could be aggregated, for example ‘large industrial’ was one category, being ‘korean industrial’ (ki) grade logs, and another was ‘large structural’, being export a-grade logs and domestic structural-grade logs. aggregated log grade outturn can indicate the ‘quality’ of a particular stand of trees, with certain characteristics about the stand inferred from the relative proportions of each aggregated classification. a total of 17 recently harvested sites were measured across new zealand as a part of this study. sites were selected by the supporting forestry companies, with all being steepland and of typical silviculture and harvesting practices (see table 1). the dataset was analysed to describe the influence of key variables on the total volume of woody residues found in a location on the cutover. generalised linear regression was used, iterating over all continuous and categorical variables to find a linear regression model that minimised the mallow’s cp value. results the summary data of all 185 lis plots shows the median value for total residue volume (>25 mm diameter) on the cutover was 88 m3/ha, with 11 and 19 m3/ha for merchantable logs (≥4 m long, >10 cm in sed & of reasonable quality) and binwood (≥0.8 m long, >10 cm in sed & of reasonable quality) respectively (table 2). figure 1 details how the average total residue volume (all material >25mm in diameter) varied from site to site, and also the variation of volumes found on each site, expressed as standard deviation to the mean. given that the average reported total harvest volume was 599 m3/ ha, total residue volume is 15% of the trv and 2% and 3% for merchantable logs and binwood respectively. the distribution of the total volume showed positive skew due to a significant number of plots returning high residue volumes (figure 2). the minimum volume was 0 m3/ha on one plot and 23 plots returned residue volumes greater than 200 m3/ha (maximum was 580 m3/ha, see table 2). the distribution of total residue volumes is best described by the bounded johnson function with the parameters: γ = 3.78, δ = 1.35, λ = 1670 & ξ = -9.46 (see figure 2). the volume of potentially merchantable residues on the cutover yields similar distributions (figure 3) to that of total residue volume (figure 2) with a positive skew. for the 185 plots, the 5th and 95th percentiles values were 0 and 63 m3/ha for merchantable, and 0 and 88 m3/ha for binwood. seventeen and 31 plots had volumes >50 m3/ ha of merchantable pulp (or higher specification) and binwood respectively, which might be considered a high volume, approximately equivalent to two truckloads per hectare. however, 52 and 29 plots recorded no volume at all for merchantable pulp (or higher specification) and binwood respectively, highlighting the nature of the materials’ distribution. in addition to the volumes of merchantable logs and binwood, table 3 provides a summary of the diameters and lengths of the material. a total of 1000 pieces of ‘potentially merchantable’ material were measured across the 17 sites. the median merchantable log was 6.4 m long, with a mid-point diameter of 180 mm which indicates much of the material may be suitable as a ‘small industrial’ export log, subject to quality. binwood on average had a smaller diameter (160 mm) than merchantable logs with the average closer to the minimum specification of 100 mm at the small end. of the 17 sites, 11 were harvested with cable yarders (118 lis plots) and 6 with ground-based systems (67 lis plots). the median total residue volume for cable yarder sites was 110 m3/ha, and 68 m3/ha for groundbased. one-way anova demonstrates that the effect of harvesting system on total residue volume is significant (p<0.01). six sites (63 lis plots) were felled motormanually (chainsaws) and 11 sites (122 lis plots) with mechanised systems. the median total residue volume for motor-manual sites was 94 m3/ha; against 86 m3/ ha for mechanised felling (which was not significantly different: p>0.05). while these results show some comparative differences, it should be cautioned that these are based on a relatively small sample of sites. harvesting systems and felling methods are also not always interchangeable. general linear regression was used to establish which variables contribute to the total volume of residues at a given location on the cutover. seven model variables minimise the mallow’s cp factor for the dataset tested (see table 4). those meeting the p<0.001 significance harvey & visser new zealand journal of forestry science (2022) 52:7 page 5 parameter average (m3/ha) median (m3/ha) interquartile range (m3/ha) 5th/95th percentile (m3/ha) min/max (m3/ha) total volume 109 88 87 17 / 269 0 / 580 merchantable logs 17 11 23 0 / 63 0 / 144 binwood 27 19 32 0 / 88 0 / 160 dead wood 25 7 24 0 / 93 0 / 539 table 2: summary of volumes of harvest residue components measured across 17 steepland cutovers. level as predictors of the box-cox transformed total residue volume (λ = 0.297) were the proportion of the log production from the stand meeting large structural log specifications (scale 0-100) and terrain slope (in degrees), followed by planform curvature (x 10) – an objective measure of the ‘sharpness’ of a spur or gully derived from an 8 metre resolution digital terrain model (esri 2019). at lower significance level and contributing little to residue volume were categorical variables describing whether the area of interest was on a spur or not, whether pulp and/or binwood was on the cutting instructions, felling method and profile curvature – a continuous variable and an objective measure of the concavity/convexity of a slope in profile (esri 2019). all seven variables in the regression model present a low likelihood of multicollinearity, with variance inflation factors <10. isolating the 118 cable-harvested lis plots and applying general linear regression to the box-cox transformed total residue volume dataset (λ = 0.324) reveals that predicted residue volume on at any given location on a cable-harvested sites is most influenced by the terrain slope, proportion of large structural timber in the stand and also the planform curvature measure of the terrain. four additional variables contribute to the regression model (see fig. 4), satisfying the minimised mallow’s cp criteria (see table 5). likelihood of multicollinearity is low, with variance inflation factors <10 for all contributing variables. harvey & visser new zealand journal of forestry science (2022) 52:7 page 6 figure 1: average total volume of woody residues on each cutover site, including the standard deviation. figure 2: distribution of total volume of residues >25 mm in diameter as measured across the 17 sites and 185 lis plots. parameter merchantable logs binwood n pieces measured 365 635 le ng th ( m ) average 7.4 2.9 median 6.4 2.6 interquartile range 3.6 1.7 m id -p oi nt d ia m et er (m m ) average 195 163 median 180 145 interquartile range 90 85 table 3: summary statistics of potentially merchantable residues measured on the 17 sites. figure 3: distributions of residue volumes with merchantable potential across the 17 sites and 185 lis plots. discussion hall (1999b) represents the most recent study on residue volumes on new zealand’s steepland cutovers. two intensively surveyed hauler cutovers yielded an average volume of 61 m3/ha, ranging from 1 to >200 m3/ha. the sites additionally indicated significant accumulations of residues in the lowest point of the harvest setting. harvesting practice around gullies may have improved, but this study has showed that some concentration of residues in low-points continues to occur. for hauler sites, the significance of the planform curvature variable in the regression analysis confirms the observation of a difference in residue volume between gullies and spurs. despite a step-change in harvesting mechanisation over the previous decade (raymond 2018), cutover residue volumes on harvesting sites appear to remain similar. increasing demand for wood fibre could lead to greater residue recovery from the cutover where the market conditions allow – however there remains competition from residues accumulating at landings, super-skids harvey & visser new zealand journal of forestry science (2022) 52:7 page 7 table 4: effect of significant variables on the box-cox transformed volume of residues remaining on the cutover (λ = 0.297) post-harvest (adjusted r2 = 0.32 from 17 sites and n = 185 lis plots). effect regression coefficient f value p value variance inflation factor intercept 23.7 48 0.000 large structural logs proportion (0-100) -0.252 11 0.001 1.9 terrain slope (degrees) -0.0955 11 0.001 1.1 planform curvature (x10) -0.117 8.9 0.003 1.4 spur (1 = yes, 0 = no) 0.677 5.1 0.03 1.1 pulp &/or binwood on cutting instructions (1 = yes, 0 = no) 1.32 4.5 0.04 6.6 motor-manual falling (1 = yes, 0 = no) 0.886 2.7 0.1 5.1 profile curvature (x 10) 0.0891 2.7 0.1 1.3 figure 4: results of multiple regression for the box-cox transformed total residue volume, including a 95% confidence interval. and central processing yards (processing sites) as a more readily available and equivalent resource. having accounted for residues at processing sites, this study enables a forest owner to understand some of the drivers for woody residue accumulations on cutovers, assisting decision-making for specific cutover residue management interventions or discussions with potential residue customers. in steepland forests, the resource remains an opportunity for greater recovery and utilization until there is a significant shift in the market for the product. this study has demonstrated that tree breakage during either or both of the felling and extraction phases of harvesting on steepland sites remains a key opportunity for improved value recovery. murphy (1982) investigated how value loss due to felling breakage could be minimised and revealed that trees with larger diameters were more susceptible (than small diameter trees) to multiple breaks along the stem during felling. this implies that older stands or stands with low stocking may be more likely to produce high residue volumes. this hypothesis was not directly evidenced by the regression models in this study, though a stand with a high proportion of large structural logs may lead to lower residue volumes in the cutover. importantly, data was collected post-extraction, where further breakage can occur. therefore, the residue volumes measured cannot be definitively attributed to either felling or extraction. innovations in machinery design and harvesting methods, aimed at reducing stem breakage will result in lower volumes of large woody biomass remaining on steepland cutovers following harvest. whilst this progress would be undoubtedly beneficial for forest owners and contractors if considering value recovery from the stem, increasing the volume of branches extracted to landings may transform what is currently a distributed ‘problem’, into a concentrated one at processing sites, if the product is un-merchantable. on sites prone to landslips and debris flows, concentration at processing areas may be beneficial by ensuring debris can be piled/stabilised on low-risk landforms. development of harvesting technology and methods is important and will continue, but a strong biomass market is key to avoid harvest residue piles at the processing site becoming a more significant constraint or cost associated with harvest operations. this study relied on tried and tested in-field line intersect survey methods. an emerging method for measuring harvest residue volumes is via passive and active remote sensing technologies (davis 2015; joyce et al. 2019). the benefits of this technology promise to be substantial when considering the increased speed, safety, and control of spatial and temporal resolutions (tang & shao 2015). early applications of deep learning on aerial imagery show significant promise for capturing merchantable volume measurements on steepland cutovers (herries 2021). conclusions this study sets the latest benchmark for measuring progress on stem breakage and value recovery in new zealand’s steepland plantations. plantations continue to offer potential for greater utilisation with a median total volume of residues remaining on the cutovers of 17 sites measuring 88 m3/ha. total volume follows a welldefined right-skewed distribution showing that small areas of harvesting sites contain high volumes of harvest residues. extraction system appears to impact total residue volume, with cable yarder operations leaving behind more cutover residues than ground-based operations. prediction of harvest residue concentrations on steepland sites ahead of harvest is possible using the regression coefficients presented. variables collected in this study can account for 32% of the variation observed on the 17 sites measured. markets for harvest residues are developing and innovations to harvest systems are promising to reduce the production/distribution of residue material. harvey & visser new zealand journal of forestry science (2022) 52:7 page 8 effect regression coefficient f value p value variance inflation factor intercept 25.8 43 0.000 terrain slope (degrees) -0.104 10 0.002 1.1 large structural proportion of trv (0-100) -0.274 9.9 0.002 1.9 planform curvature (x10) -0.128 8.4 0.004 1.4 pulp &/or binwood on cutting instructions (1 = yes, 0 = no) 1.46 4.2 0.04 6.6 spur (1 = yes, 0 = no) 0.692 4.0 0.05 1.1 profile curvature (x10) 0.106 2.9 0.09 1.4 motor-manual falling (1 = yes, 0 = no) 1.02 2.8 0.10 5.0 table 5: effect of significant variables on the box-cox transformed volume of residues remaining on hauler-harvested cutovers (λ = 0.324) post-harvest (adjusted r2 = 0.31 from 11 sites and n = 118 lis plots). competing interests the authors have no competing interests to declare. acknowledgements we gratefully acknowledge the support and consultation of peter hall (scion) and keith raymond from forest growers research (fgr). authors’ contributions the project idea and methodology was developed by ch within the scope of his phd, under the guidance and direction of rv. ch prepared the literature review and completed all of the field work, with the support of rv for the data analyses. the manuscript was prepared jointly. funding this research project is supported by forest growers research limited and the new zealand institute of forestry foundation. references acuña, e., cancino, j., rubilar, r., & silva, l. (2017). volume, physical characteristics and costs of harvest residue utilization of pinus radiata as an energy source. custos e agronegócio, 13(1), 442463. baillie, b., collier, k., & halliday, j. (1998). woody debris in pumice-bed streams how long does it last and its use by aquatic insects. liro report vol. 23 no. 10, 5 p. rotorua, nz: liro forestry solutions. retrieved from https://fgr.nz/documents/ download/4336?1328479511 baillie, b. (1999). management of logging slash in streams of new zealand results from a survey. project report 85, 32 p. rotorua, nz: liro forestry solutions. retrieved from https://fgr.nz/ documents/download/4077?2132325627 ballard, r., & will, g.m. (1981a). accumulation of organic matter and mineral nutrients under a pinus radiata stand. new zealand journal of forestry science, 11(2), 145-151. ballard, r., & will, g.m. (1981b). removal of logging waste, thinning debris, and litter from a pinusradiata pumice soil site. new zealand journal of forestry science, 11(2), 152-163. beets, p.n., gilchrist, k., & jeffreys, m.p. (2001). wood density of radiata pine: effect of nitrogen supply. forest ecology and management, 145(3), 173-180. https://doi.org/10.1016/s0378-1127(00)004059 berkett, h. (2012). an examination of the current slope gradients being experienced by ground based forest machines in new zealand plantation forests. (masters in forestry science), university of canterbury. retrieved from https://go.exlibris. link/rw94d4by bloomberg, m., & davies, t. (2012). do forests reduce erosion? the answer is not as simple as you may think. new zealand journal of forestry, 56(4), 1620. bray, j.r., & gorham, e. (1964). litter production in forests of the world. in j.b. cragg (ed.), advances in ecological research (vol. 2, pp. 101-157): academic press. https://doi.org/10.1016/s00652504(08)60331-1 buffon, g.-l.l.d. (1777). essai d’arithmétique morale. in g.-l.l. de buffon (ed.), supplément à l’histoire naturelle (vol. 4, pp. 46-148). paris: de l’imprimerie royale. cave, m., davies, n., & langford, j. (2017). cyclone cook slash investigation. 121. retrieved from h t t p s : / / w w w. g d c . g o v t . n z / _ _ d a t a / a s s e t s / p d f _ f i l e / 0 0 1 3 / 1 0 4 0 8 / c y c l o n e c o o ks l a s h investigation-2017-report.pdf collier, k.j., & bowman, e.j. (2003). role of wood in pumice-bed streams i: impacts of post-harvest management on water quality, habitat and benthic invertebrates. forest ecology and management, 177(1-3), 243-259. https://doi.org/10.1016/ s0378-1127(02)00447-4 crozier, m.j. (1999). prediction of rainfall-triggered landslides: a test of the antecedent water status model. earth surface processes and landforms, 24(9), 825-833. https://doi.org/10.1002/ ( s i c i ) 1 0 9 6 9 8 3 7 ( 1 9 9 9 0 8 ) 2 4 : 9 < 8 2 5 : : a i d esp14>3.0.co;2-m davis, b. (2015). refinement of a drone-based method for estimating coarse woody debris and biomass residue following forest harvest. (msc.), university of georgia, georgia, usa. east harbour management services, & scion. (n.d.). assessment of possible renewable energy targets direct use: woody biomass. energy efficiency and conservation authority. esri. (2019). curvature function. retrieved from https:// desktop.arcgis.com/en/arcmap/10.7/managedata/raster-and-images/curvature-function.htm ffr (future forest research). (2010). ffr welcomes funding for forest harvesting research. retrieved from http://fgr.nz/2010/05/26/ffr-welcomesfunding-forest-harvesting-research/ fraver, s., ducey, m.j., woodall, c.w., d’amato, a.w., milo, a.m., & palik, b.j. (2018). influence of transect length and downed woody debris abundance on precision of the line-intersect sampling method. forest ecosystems, 5(1). https://doi.org/10.1186/ s40663-018-0156-9 froehlich, h., mcgreer, d., & sedell, j. (1972). natural debris within the stream environment. coniferous forest biome. us international biological program, internal report, 96. retrieved from https:// harvey & visser new zealand journal of forestry science (2022) 52:7 page 9 https://fgr.nz/documents/download/4336?1328479511 https://fgr.nz/documents/download/4336?1328479511 https://fgr.nz/documents/download/4077?2132325627 https://fgr.nz/documents/download/4077?2132325627 https://doi.org/10.1016/s0378-1127(00)00405-9 https://doi.org/10.1016/s0378-1127(00)00405-9 https://go.exlibris.link/rw94d4by https://go.exlibris.link/rw94d4by https://doi.org/10.1016/s0065-2504(08)60331-1 https://doi.org/10.1016/s0065-2504(08)60331-1 https://www.gdc.govt.nz/__data/assets/pdf_file/0013/10408/cyclone-cook-slash-investigation-2017-report.pdf https://www.gdc.govt.nz/__data/assets/pdf_file/0013/10408/cyclone-cook-slash-investigation-2017-report.pdf https://www.gdc.govt.nz/__data/assets/pdf_file/0013/10408/cyclone-cook-slash-investigation-2017-report.pdf https://doi.org/10.1016/s0378-1127(02)00447-4 https://doi.org/10.1016/s0378-1127(02)00447-4 https://doi.org/10.1002/(sici)1096-9837(199908)24:9%3c825::aid-esp14%3e3.0.co;2-m https://doi.org/10.1002/(sici)1096-9837(199908)24:9%3c825::aid-esp14%3e3.0.co;2-m https://doi.org/10.1002/(sici)1096-9837(199908)24:9%3c825::aid-esp14%3e3.0.co;2-m https://desktop.arcgis.com/en/arcmap/10.7/manage-data/raster-and-images/curvature-function.htm https://desktop.arcgis.com/en/arcmap/10.7/manage-data/raster-and-images/curvature-function.htm https://desktop.arcgis.com/en/arcmap/10.7/manage-data/raster-and-images/curvature-function.htm http://fgr.nz/2010/05/26/ffr-welcomes-funding-forest-harvesting-research/ http://fgr.nz/2010/05/26/ffr-welcomes-funding-forest-harvesting-research/ https://doi.org/10.1186/s40663-018-0156-9 https://doi.org/10.1186/s40663-018-0156-9 https://andrewsforest.oregonstate.edu/sites/default/files/lter/pubs/pdf/pub1665.pdf#page=16 harvey & visser new zealand journal of forestry science (2022) 52:7 page 10 andrewsforest.oregonstate.edu/sites/default/ files/lter/pubs/pdf/pub1665.pdf#page=16 ghaffariyan, m.r. (2013). remaining slash in different harvesting operation sites in australian plantations. silva balcanica, 14(1), 83-93. gomez, b., page, m.j., & trustrum, n.a. (2010). sediment cascades: an integrated approach. in t.p. burt & r.j. allison (eds.), (pp. 117-152): wiley-blackwell. goulding, c.j. (2005). measurement of trees. in m. colley (ed.), forestry handbook (4 ed., pp. 145-148). tauranga, new zealand: the new zealand institute of forestry (inc.). guzmán, j.a. (1984). study of wood chip production from forest residues in chile. biomass, 5(3), 167-179. https://doi.org/10.1016/0144-4565(84)90021-0 haaker, a. (2017). over eur 1.68 billion worth of biomass power projects in sweden. retrieved from bioenergy international website: https:// bioenergyinternational.com/heat-power/eur-168-billion-worth-biomass-power-projects-sweden hall, p. (1994). waste wood at logging landings. liro report vol. 19, no. 15, 13 p. rotorua, nz: logging industry research organisation new zealand (liro). retrieved from https://fgr.nz/documents/ download/4514?102200476 hall, p. (1995). collection and transportation of logging residues. liro report vol. 20, no. 16, 8 p. rotorua, nz: logging industry research organisation new zealand (liro). retrieved from https://fgr.nz/ documents/download/4480?861344768 hall, p. (1996). cutover waste assessment a comparison of sampling techniques and intensities. liro project report 60, 38 p. rotorua, nz: logging industry research organisation new zealand (liro). retrieved from https://fgr.nz/documents/ download/4128?1077031238 hall, p. (1998). logging residue at landings. new zealand forestry, 43(1), 30-32. hall, p. (1999a). effect of logmaking on production of stem wood residue at landings. liro report vol. 24, no. 4, 4 p. rotorua, nz: liro forestry solutions. retrieved from https://fgr.nz/documents/ download/4310?879758742 hall, p. (1999b). logging residue distribution. liro report vol. 24, no. 9, 6 p. rotorua, nz: liro forestry solutions. retrieved from https://fgr.nz/ documents/download/4300?1527567485 hall, p. (2000a). effects of removing logging residue for bioenergy on some potentially available cutover nutrients. liro report vol. 25, no. 2, 6 p. rotorua, nz: liro forestry solutions. retrieved from https:// fgr.nz/documents/download/4282?935627279 hall, p. (2000b). skid site rehabilitation tree growth trials, results to age 7. liro report vol. 25, no. 1, 10 p. rotorua, nz: liro forestry solutions. retrieved from https://fgr.nz/documents/ download/4284?696343852 hall, p. gigler, j.k., & sims, r.e.h. (2001). delivery systems of forest arisings for energy production in new zealand. biomass & bioenergy, 21(6), 391-399. https://doi.org/10.1016/s0961-9534(01)000472 hall, p. (2013). bioenergy options for new zealand: key findings from five studies. wiley interdisciplinary reviews: energy and environment, 2(6), 587-601. https://doi.org/10.1002/wene.60 hall, p., & evanson, t. (2007). forest residue harvesting for bioenergy fuels: may 2007-phase 1. scion report, 28 p. energy efficiency and conservation authority (eeca). retrieved from https://www. bioenergy.org.nz/documents/resource/reportforest-residue-harvesting-fuels-part-1-scion-eheeeca-may2007.pdf hall, p., & mcmahon, s. (1997). logging residue at hauler landings results from an industry survey. liro report vol. 22, no. 2, 8 p. rotorua, nz: liro forestry solutions. retrieved from https://fgr.nz/ documents/download/4394?565924820 harmon, m.e., whigham, d.f., sexton, j., & olmsted, i. (1995). decomposition and mass of woody detritus in the dry tropical forests of the northeastern yucatan peninsula, mexico. biotropica, 27(3), 305316. https://doi.org/10.2307/2388916 herries, d. (2021, april 2021). drone based cutover merchantable volume assessment. paper presented at the harvesttech, rotorua, nz. joyce, m.j., erb, j.d., sampson, b.a., & moen, r.a. (2019). detection of coarse woody debris using airborne light detection and ranging (lidar). forest ecology and management, 433, 678-689. https://doi. org/10.1016/j.foreco.2018.11.049 kent, t., kofman, p.d., & coates, e. (2011). harvesting wood for energy. cost-effective woodfuel supply chains in irish forestry. 118. retrieved from http://www.coford.ie/media/coford/content/ publications/projectreports/harvesting_wood_ low_res_for_web.pdf maier, c.a., johnsen, k.h., dougherty, p., mcinnis, d., anderson, p., & patterson, s. (2012). effect of harvest residue management on tree productivity and carbon pools during early stand development in a loblolly pine plantation. forest science, 58(5), 430-445. https://doi.org/10.5849/forsci.11-069 manies, k.l., harden, j.w., bond-lamberty, b.p., & o’neill, k.p. (2005). woody debris along an upland chronosequence in boreal manitoba and its impact on long-term carbon storage. canadian journal of forest research, 35(2), 472-482. https://doi. org/10.1139/x04-179 mateos, e., & edeso, j.m. (2015). estimation of residue biomass in the oka river basin (spain). in c. vilarinho, f. castro & m. russo (eds.), wastes: solutions, treatments and opportunities https://andrewsforest.oregonstate.edu/sites/default/files/lter/pubs/pdf/pub1665.pdf#page=16 https://andrewsforest.oregonstate.edu/sites/default/files/lter/pubs/pdf/pub1665.pdf#page=16 https://doi.org/10.1016/0144-4565(84)90021-0 https://bioenergyinternational.com/heat-power/eur-1-68-billion-worth-biomass-power-projects-sweden https://bioenergyinternational.com/heat-power/eur-1-68-billion-worth-biomass-power-projects-sweden https://bioenergyinternational.com/heat-power/eur-1-68-billion-worth-biomass-power-projects-sweden https://fgr.nz/documents/download/4514?102200476 https://fgr.nz/documents/download/4514?102200476 https://fgr.nz/documents/download/4480?861344768 https://fgr.nz/documents/download/4480?861344768 https://fgr.nz/documents/download/4128?1077031238 https://fgr.nz/documents/download/4128?1077031238 https://fgr.nz/documents/download/4310?879758742 https://fgr.nz/documents/download/4310?879758742 https://fgr.nz/documents/download/4300?1527567485 https://fgr.nz/documents/download/4300?1527567485 https://fgr.nz/documents/download/4282?935627279 https://fgr.nz/documents/download/4282?935627279 https://fgr.nz/documents/download/4284?696343852 https://fgr.nz/documents/download/4284?696343852 https://doi.org/10.1016/s0961-9534(01)00047-2 https://doi.org/10.1016/s0961-9534(01)00047-2 https://doi.org/10.1002/wene.60 https://www.bioenergy.org.nz/documents/resource/report-forest-residue-harvesting-fuels-part-1-scion-ehe-eeca-may2007.pdf https://www.bioenergy.org.nz/documents/resource/report-forest-residue-harvesting-fuels-part-1-scion-ehe-eeca-may2007.pdf https://www.bioenergy.org.nz/documents/resource/report-forest-residue-harvesting-fuels-part-1-scion-ehe-eeca-may2007.pdf https://www.bioenergy.org.nz/documents/resource/report-forest-residue-harvesting-fuels-part-1-scion-ehe-eeca-may2007.pdf https://fgr.nz/documents/download/4394?565924820 https://fgr.nz/documents/download/4394?565924820 https://doi.org/10.2307/2388916 https://doi.org/10.1016/j.foreco.2018.11.049 https://doi.org/10.1016/j.foreco.2018.11.049 http://www.coford.ie/media/coford/content/publications/projectreports/harvesting_wood_low_res_for_web.pdf http://www.coford.ie/media/coford/content/publications/projectreports/harvesting_wood_low_res_for_web.pdf http://www.coford.ie/media/coford/content/publications/projectreports/harvesting_wood_low_res_for_web.pdf https://doi.org/10.5849/forsci.11-069 https://doi.org/10.1139/x04-179 https://doi.org/10.1139/x04-179 harvey & visser new zealand journal of forestry science (2022) 52:7 page 11 (p. 133-138). the netherlands: crc press/ balkema. retrieved from https://books.google. c o . n z / b o o k s ? i d = w d c yc g a aq b a j & l p g = pa 1 3 3 & o t s = 0 f i o 2 4 g v t t & d q = % 2 2 s p a i n % 2 2 % 2 0 and%20%22radiata%22%20and%20%22re sidue%22&lr&pg=pa133#v=onepage&q=%22 s p a i n % 2 2 % 2 0 a n d % 2 0 % 2 2 ra d i a t a % 2 2 % 2 0 and%20%22residue%22&f=false mateos, e., & ormaetxea, l. (2018). spatial distribution of biomass and woody litter for bio-energy in biscay (spain). forests, 9(5), 253. https://doi. org/10.3390/f9050253 mcmahon, s., evanson, t., hall, p., & baillie, b. (1998). cable extraction of pulp: effect of minimum extracted piece size on productivity. liro report vol. 23, no. 14, 5 p. rotorua, nz: liro forestry solutions. retrieved from https://fgr.nz/ documents/download/4328?829207617 mead, d.j. (2013). sustainable management of pinus radiata plantations. in fao (series ed.) fao forestry paper no. 170 (pp. 265). retrieved from https://www.fao.org/3/i3274e/i3274e.pdf merino, a.n., fernández-lópez, a., solla-gullón, f., & edeso, j.m. (2004). soil changes and tree growth in intensively managed pinus radiata in northern spain. forest ecology and management, 196(2), 393-404. https://doi.org/10.1016/j. foreco.2004.04.002 messinger, r.c. (1974). logging residue management: a literature review of new zealand and north american methods with management implications. a project submitted in partial fulfilment of the requirements for the degree of master of forestry science in the university of canterbury. murphy, g. (1982). value savings from alternative felling patterns on steep country. lira report vol. 7, no. 8, 4 p. rotorua, nz: logging industry research association inc. (lira). retrieved from https://fgr. nz/documents/download/5053?973453426 murphy, g. (2005). quality and value recovery. in m. colley (ed.), forestry handbook (4 ed., pp. 181). tauranga, new zealand: the new zealand institute of forestry (inc.). mpi (2017). resource management (national environmental standards for plantation forestry) regulations, 75 p. wellington, nz: ministry for primary industries (mpi). retrieved from h t t p : / / w w w. l e g i s l a t i o n . g ov t . n z / re g u l a t i o n / public/2017/0174/latest/whole.html nzfoa (2007). new zeland environmental code of practice for plantation forestry. (version 1 ed., pp. 168). new zealand forest owners association. nzfoa (2019/20). facts & figures 2019/20. new zealand forest owners association. oxford university press. (2018). oxford english dictionary. retrieved from https://www-oed-com. ezproxy.canterbury.ac.nz/noresults?browsetype= sortalpha&noresults=true&page=1&pagesize=20 &q=steepland&scope=entry&sort=entry&type= dictionarysearch phillips, c., marden, m., & basher, l. (2012). plantation forest harvesting and landscape response what we know and what we need to know. new zealand journal of forestry, 56(4), 4-12. phillips, c., marden, m., & basher, l.r. (2015). forests and erosion protection getting to the root of the matter. new zealand journal of forestry, 60(2), 1115. phillips, c., pruden, c., & coker, r. (1996). forest harvesting in the marlborough sounds flying in the face of a storm? new zealand forestry, 41(1), 27-31. raymond, k. (2018). pgp steepland harvesting – a collaborative research and development programme. new zealand journal of forestry, 63(3), 18-21. sikkink, p.g., & keane, r.e. (2008). a comparison of five sampling techniques to estimate surface fuel loading in montane forests. international journal of wildland fire, 17(3), 363-379. https://doi. org/10.1071/wf07003 smethurst, p.j., & nambiar, e.k.s. (1990). distribution of carbon and nutrients and fluxes of mineral nitrogen after clear-felling a pinus radiata plantation. canadian journal of forest research, 20(9), 14901497. https://doi.org/10.1139/x90-197 swanson, f.j., lienkaemper, g.w., & sedell, j.r. (1976). history, physical effects, and management implications of large organic debris in western oregon streams (vol. 56). department of agriculture, forest service, pacific northwest forest and range experiment station. tang, l., & shao, g. (2015). drone remote sensing for forestry research and practices. journal of forestry research, 26(4), 791-797. https://doi. org/10.1007/s11676-015-0088-y telenius, b. (2006). bioenergy in sweden, editorial, iea bioenergy news, p. 8. retrieved from https://www.ieabioenergy.com/wp-content/ uploads/2013/10/iea-bioenergy-news-182.pdf theodorou, c., & bowen, g.d. (1970). mycorrhizal responses of radiata pine in experiments with different fungi. australian forestry, 34(3), 183-191. https://doi.org/10.1080/00049158.1970.106755 25 usfs (2011). phase 3 field guide down woody materials (version 5.1 ed.): usda forest service. van wagner, c.e. (1968). line intersect method in forest fuel sampling. forest science, 14(1), 20-20. visser, r. (2018). benchmarking harvesting cost and productivity – 2017 update. (htn10-02), 4. retrieved from https://fgr.nz/documents/ download/7091?936289211 https://doi.org/10.3390/f9050253 https://doi.org/10.3390/f9050253 https://fgr.nz/documents/download/4328?829207617 https://fgr.nz/documents/download/4328?829207617 https://www.fao.org/3/i3274e/i3274e.pdf https://doi.org/10.1016/j.foreco.2004.04.002 https://doi.org/10.1016/j.foreco.2004.04.002 https://fgr.nz/documents/download/5053?973453426 https://fgr.nz/documents/download/5053?973453426 http://www.legislation.govt.nz/regulation/public/2017/0174/latest/whole.html http://www.legislation.govt.nz/regulation/public/2017/0174/latest/whole.html https://www-oed-com.ezproxy.canterbury.ac.nz/noresults?browsetype=sortalpha&noresults=true&page=1&pagesize=20&q=steepland&scope=entry&sort=entry&type=dictionarysearch https://www-oed-com.ezproxy.canterbury.ac.nz/noresults?browsetype=sortalpha&noresults=true&page=1&pagesize=20&q=steepland&scope=entry&sort=entry&type=dictionarysearch https://www-oed-com.ezproxy.canterbury.ac.nz/noresults?browsetype=sortalpha&noresults=true&page=1&pagesize=20&q=steepland&scope=entry&sort=entry&type=dictionarysearch https://www-oed-com.ezproxy.canterbury.ac.nz/noresults?browsetype=sortalpha&noresults=true&page=1&pagesize=20&q=steepland&scope=entry&sort=entry&type=dictionarysearch https://www-oed-com.ezproxy.canterbury.ac.nz/noresults?browsetype=sortalpha&noresults=true&page=1&pagesize=20&q=steepland&scope=entry&sort=entry&type=dictionarysearch https://doi.org/10.1071/wf07003 https://doi.org/10.1071/wf07003 https://doi.org/10.1139/x90-197 https://doi.org/10.1007/s11676-015-0088-y https://doi.org/10.1007/s11676-015-0088-y https://www.ieabioenergy.com/wp-content/uploads/2013/10/iea-bioenergy-news-182.pdf https://www.ieabioenergy.com/wp-content/uploads/2013/10/iea-bioenergy-news-182.pdf https://doi.org/10.1080/00049158.1970.10675525 https://doi.org/10.1080/00049158.1970.10675525 https://fgr.nz/documents/download/7091?936289211 https://fgr.nz/documents/download/7091?936289211 visser, r., harrill, h., & baek, k. (2019). biomass recovery operations in new zealand: a review of the literature. fgr report h041, 32 p. rotorua, nz: forest growers research (fgr). retrieved from https://fgr.nz/documents/download/8199 visser, r., spinelli, r., & brown, k. (2018). best practices for reducing harvest residues and mitigating mobilisation of harvest residues in steepland plantation forests., 53. retrieved from http:// www.nzffa.org.nz/system/assets/3046/1879gsdc152-best-practices-for-reducing-harvestresidues-a.pdf visser, r., spinelli, r., & stampfer, k. (2009). integrating biomass recovery operations into commercial timber harvesting: the new zealand situation. paper presented at the 32nd annual meeting of the council on forest engineering (cofe 09), kings beach, ca, usa. http://hdl.handle. net/10092/2623 visser, r., spinelli, r., & stampfer, k. (2010). four landing biomass recovery case studies in new zealand clear-cut pine plantations. paper presented at the formec, padova, italy. warren, w.g., & olsen, p.f. (1964). a line intersect technique for assessing logging waste. forest science, 10(3), 267-276. wei, x., kimmins, j.p., peel, k., & steen, o. (1997). mass and nutrients in woody debris in harvested and wildfire-killed lodgepole pine forests in the central interior of british columbia. canadian journal of forest research, 27(2), 148-155. https://doi. org/10.1139/x96-169 wilks, p., & wang, h. (2009). the rabbit island biosolids project. new zealand journal of forestry, 54(2), 3336. harvey & visser new zealand journal of forestry science (2022) 52:7 page 12 https://fgr.nz/documents/download/8199 http://www.nzffa.org.nz/system/assets/3046/1879-gsdc152-best-practices-for-reducing-harvest-residues-a.pdf http://www.nzffa.org.nz/system/assets/3046/1879-gsdc152-best-practices-for-reducing-harvest-residues-a.pdf http://www.nzffa.org.nz/system/assets/3046/1879-gsdc152-best-practices-for-reducing-harvest-residues-a.pdf http://www.nzffa.org.nz/system/assets/3046/1879-gsdc152-best-practices-for-reducing-harvest-residues-a.pdf https://doi.org/10.1139/x96-169 https://doi.org/10.1139/x96-169 cs f.r.i. rotorua production forestry division internal f{eport no. 187 title: first assessment of field progeny trial of selections of pinus radiata for resistance to diplodia infection. project: gti 24 work plan: 96 date: 1980 r.d. burdon and by: c.b. low summary wind-pollinated progenies of pinus radiata trees which had been selected intensively for resistance to shoot dieback associated with diplodia infection were assessed in tarawera and kaingaroa forests 6~ years after planting. separate records were made of dieback on leaders and laterals, while stem diameters were measured, and stem straightness and desirability of branching habit were scored. growth was faster, and dieback more prevalent at tarawera, where there was quite good resolution of progeny differences (repeatabilities of progeny means 0.51-0.73, p < 0.05-< 0.001) in the amount of dieback, irrespective of the measure used. resolution of progeny differences in the incidence of dieback at kaingaroa was poor (repeatability of progeny means~ 0.35, p > 0.05), presumably because of a very low disease incidence. there was no convincing evidence that progeny rankings differed materially between the two sites. the select material did not show significantly less shoot dieback in the field than two control lots (seed orchard and unselected bulk in this case), even though it had performed significantly (p < 0.05) better than controls in a glasshouse inoculation trial. more definitely, diepack incidence of individual progenies in the field was effectively uncorrelated with infection response in the glasshouse. this suggests' that genetic resistance, if present, may be highly specific to the circumstances of infection. transformation of data, in an attempt to overcome strongly asymmetric distributions of dieback counts, had little effect on results of analysis of variance. no single feature of monoterpene composition of the parent clones could be correlated convincingly with disease incidence among the progenies, either in the field or in the glasshouse. some ~· muricata, which grew more slowly, showed more dieback, especially at tarawera where animal damage was a complicating factor. however, it showed much less needle cast. introduction selection of p. radiata for resistance to shoot dieback associated with infection by diplodia pinea has been done on a pilot scale. this was on sites where the incidence of dieback was very high and where, perforce, the occurrence of chance escapes from disease was least likely. seed from the selections has been used for establishing a field progeny trial, to test the effectiveness of the field selection and potentially for reselection of the parents. 2 since the establishment of the progeny trial further seed has been collected from the parent ortets, and has been used in a glasshouse inoculation trial (burdon et al., 1976). progeny of the select parents showed better resistance overall in the glasshouse than control lots, although the select families differed markedly among themselves. these results suggested (a) that the field selection had been reasonably effective, and (b) that glasshouse inoculation would be a valid and effective screening technique. nevertheless, it was still clearly desirable to be able to confirm the glasshouse result in the progeny test in the field. in a field assessment of dieback there are two major problems: (i) obtaining a satisfactory quantitative measure of the observable occurrences of dieback; (ii) expressing the incidence on a scale of variation that has satisfactory statistical properties for analysis of variance. in obtaining a quantitative measure one must weigh up several considerations: that dieback on the leader is of greater practical importance than dieback on laterals. that laterals, because they represent many more potential infection sites than the leader, offer the prospect of a more precise expression of inherent susceptibility. that there might or might not be a good genetic correlation in susceptibility to dieback between the leader and the laterals. that whereas cumulative incidence of dieback would provide more information, past occurrences are often difficult to identify with certainty. this report covers the first assessment of dieback in the field progeny trial, in which a number of alternative measures of dieback incidence were tried. material and methods the select parents and the progeny trial details of the selection of parents and the establishment of the progeny trial are given in gti work plan 96, so only a brief account is given here. twenty-six trees were originally selected, at eight years after planting, in fenton'smillfla~tarawera forest. they were selected for tree form and dominance as well as for virtual freedom from dieback. most of the trees provided sufficient wind-pollinated seed for a progeny trial, although in some cases the number of available cones was very small. progenies were raised in the nursery, and planted out during the winter of 1971, on two sites: (i) tarawera forest (plot r904) on a flat river terrace. (ii) kaingaroa forest (cpt 1350, plot r944/13) on undulating terrain typical of the northern boundary area of the forest. 3 the layout conformed basically to randomised complete blocks, with 12 replicates of 8-tree plots at each site. however, some progenies were not represented at tarawera, while at kaingaroa some were missing in certain block replicates and their place taken with additional plots of controls. two p. radiata controls were used: (i) seed collection from the ra 1 seed orchard (al 1) (ii) kaingaroa unselcted bulk seed collection, seedlot r69/854. in addition, one lot of "blue" p. muricata was included. on neither site was there any severe outbreak of dieback, but it was decided that it was necessary to assess for whatever dieback was present, in march 1978. the trees were 6~ years old from planting, when dieback incidence was expected to peak, and were approaching the stage when inspection of the crowns could become very difficult. assessment stem diameter and tree form characters were assessed in addition to dieback. because of the relatively low incidence of dieback, individual occurrences were counted instead of each tree being rated for general prevalence of the disease. the following data were recorded on each tree: 1. d.b.h. o.b. (mm) 2o stem straightness (1-9 scale); 1 = v. crooked, 9 = v. straight 3. branch habit quality (1-9 scale); 1 heavy, rough, irregular; 9 = light, even, strongly multinodal type 4. stem malformation score (1-6); 1 multiple forks 2 = two forks or one multifork 3 single fork 4 large ramicorn (s) 5 = small ramicorn ( s) 6 no forks or ramicorns 5. number of definite occurrences of dieback on leader 6. number of doubtful occurrences of dieback on leader 7. number of definite occurrences of dieback on laterals 8. number of doubtful occurrences of dieback on laterals. assessment was done by crews of two, with one person measuring diameter and booking, and one scoring tree form and counting occurrences of dieback. each replicate within a site was scored and counted entirely by one individual. 4 derivation of variables for analysis preliminary analysis (fri statspack program flq4) was made of overall frequency distributions, site by site, for individual scores and combinations of dieback counts on each tree in order to decide what transformations of variables were worth using. measurements of d..b~h.o.b. and stem straightness and branch habit quality scores were used in the original form for analysis of variance. malformation scores were subjected to a normalising transformation as follows: the scores were transformed to give class intervals corresponding to the intervals (arbitrary units) in a normal distribution between the means of the percentile classes cover both sites pooled) represented by the respective scores. this was achieved by using tne formula x' where x' b (a + x) the transformed variable, x = the original score, and a and b were constants chosen empirically to give roughly the desired intervals. dieback counts were used to derive alternative variables (measures of dieback) as shown in table 1. the general idea was to adopt the transformations where they materially reduced statistical interactions and improved the resolution of family differences. the use of a 0-1 scale was tried because, despite the sacrifice of information, there are corrections available with this scale to give heritability estimates that relate to an underlying continuous scale of variation (dempster and lerner, 1950; van vleck, 1972). for the later stages of statistical analysis certain variables were dropped on the basis of early analyses. statistical analysis analysis of variance was complicated by several ty.pes of imbalance in the classification: (i) some progenies (families) being represented only at kaingaroa. (ii) unequal numbers of surviving trees per plot. (iii) not all progenies being represented in all block replicates. (iv) controls being represented by more than one plot in some replicates at kaingaroa. accordingly, analyses of variance were carried out as follows: 1. involving those p. radiata lots (14 progenies plus two controls that were represented in all reps on both sites (see table 2a)) the method of unweighted means was used, linking analyses of subclass means (fri stats pack programs flpl and flql) with estimates of withinsubclass variance (program flpl on basic data). 5 2. involving tarawera data only (see table 2b) again, the method of unweighted means was used (fri stats pack programs flpl and flp7 on subclass means). 3. involving kaingaroa data only (i) hendersons method i (see table 2c) the unadjusted mean squares were obtained using fri stats pack programs flpl and flq6. expectations of mean squares were calculated from subclass numbers using program kmat, and variance components estimated using program flfx. as can be seen from the expectations of mean squares all f tests are only approximate. (ii) least squares analysis (fri stats pack program fltg) this gives an exact test for interaction, but biassed tests for main effects in the presence of interaction. it also gives estimates of lot means that are adjusted for rep effects, and rep means that are adjusted for lot effects, but without taking account of interaction. p. muricata was omitted from these analyses. lots were provisionally treated as a random effect; and sites were considered both as a fixed effect and as a random effect, since the appropriate approach was debatable. where a lot was represented by more than one plot in a block replicate the plots were pooled. this approximation, which was made to simplify the analysis and to bring it within current computer capacity, presumably gives a slight underestimate of the statistical lots x replicates interaction. individual tree heritabilities (h2) were estimated as follows: a 4 02 "'2 (within sites) f 0~ ... a h + 02 + 02 f rf w ... a 4 0 2 h2 (over both sites) f a ... ... ... 02 + 02 + 02 ;+ 02 f fs rf:s w 0~ being included in the denominator only if sites are regarded as conforming tosa random effect. the use of the coefficient of 4 in the numerator involves assuming that the families represent a random group of half-sib families. genetic correlations between two traits (ra at a site were estimated as xy where covf xy j is the between-families covariance between the two traits, estimated from mean cross-products in a manner analogous to the estimation of variance components, and 02 and 0 2 are the between-families (lots) components of variance fx fy for the respective traits. 6 genetic correlations between traits at different sites (rg ) were calculated as k~ (cf. burdon, 1977a) where rk~ is the phenotypic correlation between family means at sites k and ~ 2 2 are heritabilities (repeatabilities) of family and h -fk and h _ f~ means at the respective sites. in correlating family performances between the field and in the glasshouse, on one hand, and performances in the field and parent clone monoterpene composition, on the other hand, there was a complication. the clone labelled 870-387 in the archive had essentially the monoterpene of the ortet that was nominally clone 870-385 (burdon et al., 1977a). in this case, therefore, it was not quite certain which progenies in the inoculation trial corresponded to the progenies of clones 870-385 and 870-387 in the field trial. accordingly, the correlations were calculated making the two alternative assumptions as to identity. case a crossreferenced progeny 870-387 in the field with clone 870-387 in the archive and lot 70 in the inoculation trial (burdon et al., 1976). case b crossreferenced progeny 870-387 in the field with clone 870-385 in the archive (lot 72 being absent from the inoculation trial). results general at tarawera the growth was appreciably faster and the incidence of dieback much higher than at kaingaroa (tables 3 and 4). branch habit quality scores and malformation scores, however, were slightly poorer at kaingaroa, but these latter comparisons are not rigorous. the distributions of the dieback counts were strongly non-normal (table 3), even after transformation. in fact the use of transformations did not materially affect the results of analyses of variance (tables 5, 6, 7, 8, 10, 11). hence some reservation must attach to most of the analyses of variance and resulting estimates of parameterso the significant site x lot interactions (table 5) are particularly suspect. lot differences clear differences between lots were evident for all variables at tarawera (table 7) with good repeatabilities of lot means (table 8). at kaingaroa there were clear differences between lots in respect of d.b.h. o.b. and the tree form traits, but not in respect of dieback variables (tables 10 and 11). lot x site interactions were unimportant. although analyses of variance suggested interactions for dieback variables (table 5), the use of genetic correlation analysis (table 13) makes it clear that such interactions were essentially an artifact of the non-normality of the data. comparing the controls with the progenies, neither control differed significantly from the progenies as a group in respect of dieback, either at tarawera or kaingaroa (tables 9 and 12). at tarawera al 1 was slightly, 7 but not significantly (p > oo05) better than r69/854 for all traits. at kaingaroa al 1 was significantly (p > 0.05) better than r69/854 in both d.b.h.o.b._and stem straightness. there it was significantly superior to the progenies as a group in d.b.h.o.b., while r69/854 was significantly (p < 0.01) worse than the progenies overall in stem straightness. with the general lack of clear differences between the controls and the progenies it was deemed unnecessary to segregate the controls for obtaining heritability estimates. in fact, none of the estimated heritabilities (tables 6, 8 and 11) were very high, the highest values (ca. 0.25) being for stem straightness and branch habit qualityo inclusion of dieback counts on the laterals, in addition to leader dieback, gave a modest improvement in resolution of lot differences at tarawera, but taking account of uncertain cases of dieback did not improve resolution. the more elaborate counts tended to show greater effects of replicates (which were confounded with observers) and more lot x replicate interaction. interrelationships between traits in field trial estimates of intercorrelations between traits (table 14) suggest that there were no material differences in lot rankings for dieback between the leaders and the laterals. (the kaingaroa results are too imprecise to be very informative on this point). the expected pattern of strong phenotypic and genetic (between-lot) correlations was observed between malformation at tarawera and the incidence of dieback at either site (tables 13 and 14). the negative signs in the listed correlations reflect the fact that malformation was recorded on an inverse scale. variancesandbetween-trait covariances for lots and lot means are shown in table 14, in case it proves worthwhile to rank the families using a multitrait selection index. relationship between field performance and response in glasshouse field performance of progenies and their inoculation responses did not correlate at all satisfactorily (table 15, figs 1 & 2), irrespective of assumptions concerning the identity of progenies (viz. case a vs .case b). in fact, the correlations, which in general were non-significant (p > 0.05), tended to be of the opposite sign to what could be expected. the only significant correlations, between inoculation responses and malformation score, were in the'wrong'direction and were presumably fortuitous. in this situation no useful purpose was seen in pursuing estimates of genetic correlations. looking at figs 1 and 2 (in which the expected association would be negative (owing to the nature of the scales used), it can be seen that the performance of the control r69/854, relative to the progenies as a group, was not actually inconsistent between the two studies. considering the performance of individual progenies, however, even allowing for approximations and some uncertainties as to appropriate estimates of errors of progeny means, there are clearly some important discrepancies between the studies in progeny rankings. this is irrespective of assumptions as to identity of progeny [s5c[l387. 8 relationship between field performance and parental monoterpenes the correlations in tables 16 and 17 were calculated between progeny means observed in the field and mean levels of individual monoterpenes in parent clone material kept at fri headquarters (for details see burdon et al., 1977a). there was no convincing evidence of meaningful correlations, the occasional statistically significant correlations being readily attributable to chance in view of the large number of correlations being calculated. comparison between !?._. radiata and p. muri.cata predictably, the p. radiata grew considerably faster than the p. muricata (table 18). also it showed less dieback, although the differences were only significant (p < 0.05) at tarawera. however, the dieback in the p. muricata could have been accentuated by deer damage (which was concentrated in this species) at tarawera. in respect of stem diameter and the tree form traits the p. muricata performed much better relative to p. radiata at kaingaroa than at tarawera, presumably because it was not appreciably affected by dieback and not damaged by deer at kaingaroa. in fact the p. muricata was significantly straighter at kaingaroa. on both sites the p. muricata showed dramatically less needle cast than the p. radiata. at tarawera dothistroma pini was strongly implicated in the needle cast, but at kaingaroa naemacyclus niveus appeared to be the prime culprit. discussion the scoring procedure for dieback represented the basis for the study, and the ideal approach was by no means clear. although it was not explored exhaustively, several lessons seem clear enougho unless the incidence of dieback is high, it seems inevitable that dieback records will have some undesirable statistical properties which demand caution in the use of analysis of variance. these statistical properties will not readily be overcome by transformation of data. nevertheless, there appeared to be satisfactory resolution of family differences at tarawera, although it must be remembered numbers of trees per lot were fairly large and the number of block replicates higher than in most gti progeny trials. refinements of the scoring system appeared to add relatively little to the information obtained in this study. recording dieback on branches as well as the leader slightly improved resolution of family differences, and could give more satisfactory estimates of between-trait covariances. recording uncertain cases of dieback achieved virtually nothing, and seemed to introduce an important element of observer bias. with a large number of trees per family, adequate block replication, and the sort of dieback incidence that was observed at tarawera, there would seem to beno great advantage in recording more than whether or not each tree had definite leader dieback; with fewer trees per family say, in the region of 25 it might be worth incorporating counts of dieback on laterals in the measure of dieback occurrence. however, the ideal situation for genetic studies of dieback resistance would probably be where most trees have multiple occurrences of dieback, so that one could visually rate individuals for the general amount of dieback. 9 pattern of dieback incidence in relation to other studies the major and most disturbing result is the conflict between inoculation responses and the dieback figures for lots in the field. this conflict is sharp, since reasonably good resolution of lot differences was obtained in both studies. it is clear-cut in the rankings among the progenies themselves rather than in the comparisons between the controls and progenies. several possible explanations must be considered in some detail, although none appears altogether satisfactory: (i) that inappropriate controls were used in the respective studies. (ii) that the progeny samples from individual parents differed between the two studies. (iii) that rankings of genotypes for dieback resistance differ according to environment. (iv) that rankings for resistance change with age of trees. (v) that different fungal strains were involved in the two studies, with tree genotypes having resistance that is specific to pathogen strains. (vii) that certain lots were incorrectly identified at some stage. the controls were not ideal in that they were of different origin from the stand in which the selection was done, and so neither was necessarily representative of the effective base population. nevertheless, one control (r69/854) was common to the two studies, and its performance relative to that of the progenies as a group was not actually inconsistent. the progeny samples used in the field had certain deficiencies, which would mean that they by no means conformed to half-sib progenies of the respective parents. in some cases very few cones were available on the parents, and these cones would not have included consecutive pollination years or consecutive clusters of cones within a pollination season, but this would seem unlikely to have caused radical discrepancies. the seed collections used for the inoculation trial, made five years later, would have come from a more select sample of pollen parents. this could account for a slightly better performance of select material relative to controls, but it cannot account for very different progeny rankings. it would not be surprising if rankings of progenies for resistance did differ between the field progeny test and the inoculation trial, because it is well recognised that short-cut screening procedures can prove inapplicable to field conditions. what is noteworthy is that the inoculation trial results accord slightly better than the field results with the original circumstance of selection in the field, insofar as the selectparent progenies clearly excelled the controls only in the glasshouse. logically, this suggests that the glasshouse inoculation conditions might have corresponded better to the conditions at fentons mill flat prior to selection than did conditions in the field progeny trial. this, of course, would mean that any genetic gains in resistance would presumably be very specific to particular sites. but even though dieback was not very prevalent in the progeny trial the tarawera trial site and fentons mill flat seem very similar. moreover, dieback incidence differed sharply between the two trial sites without material differences in lot rankings so, all told, the possible explanation seems implausible. 10 diagnosis of diplodia-associated dieback is always a problem, since it must be made inductively on the basis of gross visual symptoms combined with proper examination of a very small sample of cases. this problem would have applied alike both in the parent stand and in the progeny trial, while there is no reason to suspect that incorrect diagnosis was an important factor. variations in progeny rankings with age of trees, although likely enough in itself, would hardly account for the observed resultsc the seedlings in the glasshouse were of course much younger than the progeny trial material when it was assessed. nevertheless, the parents, when selected, were older than the progeny trial at assessment, and the selection was endorsed rather more by the inoculation trial than by the field progeny trial. specificity of response of tree genotypes to strain of pathogen, with the presence of several different pathogen strains, is always a possibility, but again it does not provide a convincing ~xplagation for the results. the inoculation trial used fungus spores from a single isolate. this isolate was from "death valley" in tarawera forest, where the dieback was similar to that at fentons mill flat, although more extreme. the available evidence (chou, 1977) does not suggest that isolates of diplodia vary much in pathogenicity, although the comparisons were not precise and reflected only the average pathogenicity of an isolate to a sample population of seedling genotypes. the possibility of identification errors always haunts the experimenter. in this study there was one case where identification was in serious doubt, but it clearly had no bearing on the general picture. the two control lots performed roughly as might be expected in relation to each other and to the progenies, which would argue against any general misidentification, but it is difficult to be entirely confident. it is clear from figs 1 and 2 that, even though reasonable repeatabilities of lot means were obtained, much better resolution of lot differences would have been desirable in order to give a precise picture of the extent of the discrepancies between the two studies. other aspects of results the pattern of estimated heritabilities is consistent with other results obtained with p. radiata, in that stem straightness and quality of branch habit appeared to be more heritable than stem diameter or malformation rating. dieback, as a trait which shows a threshold effect and has obvious elements of chance in its expression, could not be expected to show a high individual-tree heritability, unless the overall incidence was very high indeed. the initial field selection, since it could be expected to cause greater truncation of between-family variance in high-heritability traits, has probably damped down inherent differences between traits in apparent heritability. the general lack of lot x site interaction was reassuring, even with two fairly similar trial sites. however, it should be noted that in the case of dieback, which showed undesirable data characteristics, it was necessary to use genetic correlation analysis in place of conventional analysis of variance in order to obtain the correct picture. the genetic correlation analysis could have been pursued in further detail, but this seemed unnecessary. the lack of correlations back variables is consistent although the confused picture situation is still not clear. 11 between parental monoterpene levels and diewith the finding of burdon et al. (1977b), of progeny resistance means that the general the seed orchard lot was generally superior to the bulk seed collection, except in the case of the obviously inconclusive figures for dieback at kaingaroa. although many of the differences were not statistically significant individually, this gives further confirmation of the efficacy of the main breeding programme. comparison of ~· muricata and p. radiata the general growth and form of p. muricata was as expected. the deer damage to p. muricata at tarawera may not mean much, since the occasional plots of p. muricata would have had the novelty value that tends to attract animals. it does seem that p. muricata is the more susceptible to diplodiaassociated shoot dieback, at least on these warm sites. however, it was much more resistant to needle casts. indications for future work the sharp conflicts in results make it very difficult to decide what to do next, if anything. the progeny trial plantings certainly need thinning within the next year, and despite likely problems of visibility it is recommended that another but less elaborate assessment of dieback be made during this autumn. few decisions should probably be made until the results of such an assessment are known. in any case field and glasshouse studies should probably be made of resistance in juvenile clones to infection and dieback, as already prescribed in pathology work plan no. 117, experiments 4 and 5. (see also addenda to gti work plan no. 96). however, as an adjunct to any such work the possible specificity of clonal responses to different fungal isolates could be studied. if conflicts in results cannot be resolved it might be appropriate to check the identity of progenies in the field using, say, monoterpene analysis. even so this may not be very quick or easy, and further groundwork on the technique needs to be done. despite reservations concerning the nature of the progeny trial, it is not recommended that any immediate attempt be made to repeat the trial with more recent seed collections and additional control lots. a more promising approach might be control-crossing between the selectionso acknowledgements thanks are due to messrs j. riley, j. tombleson, d. briscoe and j. geany and mrs s. gallagher for carrying out the field assessments. 12 references burdon, r.d. 1977: genetic correlation as a concept for studying genotypeenvironment interaction in forest tree breeding. silvae genetica 26: 168-75. burdon, r.d., chou, c.kcs• and currie, d. 1976: response to inoculating with diplodia pinea in progenies of apparently reistant trees of pinus radiata. n.z. for. serv., for. res. inst., genetics and tree impr. int. rep. 122 (unpublished). burdon, r.d., _gaskin, r.e., low, c.b. and zabkiewicz, j.a. 1977a: clonal repeatability of monoterpene composition of cortical oleoresin of pinus radiata. ibid, 136. ------~ 1977b: monoterpene composition of cortical oleoresin from p. radiata trees selected for resistance to diplodia shoot dieback. ibid, 140. chou, c.k.s. 1977: a shoot dieback in pinus radiata caused by diplodia pinea. ii. inoculation studies. n,z. j. for. sci., 6: 409-20. dempster, e.re and lerner, i.m. 1950: heritability of threshold characters. genetics, 35: 212-34. van vleck, l.d. 1972: estimates of heritability of threshold characters. journal of dairy science, 55: 218-25. 13 table 1: derivation of dieback variables for analysis weighting (w) given to each occurrence variable dieback: used in formula for variable --------------------------------------analysis leader: lateral: definite doubful definite doubtful dldrdef 1 0 0 0 if i:w = 0, x= 0; i:w >0, x dldrexp 1 0 0 0 (i:w)0.6 dldrgen 2 1 0 0 i:w edldrgen 2 1 0 0 (2:w)0.65 dlatgen 0 0 2 1 i:w edlatgen 0 0 2 1 (i:w)0.65 alldbkge 10 5 2 1 i:w elldbkge 10 5 2 1 (i:w)0.65 defdbkge 5 0 1 0 (i:w)0.65 dldrexu represents an empirical normalising transformation of dldrdef dldrgen .is a composite score covering both definite and doubtful occurrences of leader dieback ~~~~is a composite score covering both definite and doubtful occurrences of dieback on laterals edlatgen represents an empirical normalising transformation of dlatgen alldbkge is a composite score covering definite and doubtful occurrences of dieback on both leader and laterals an normalising transformation of defdbk~e (or dfdbkge) is a score covering defl~ite occurrences of dieback on both leader and laterals, subjected to an empirical normalising transformation 1 15 table 2: analysis of variance models m.s. source d. f. expectation of mean squares a. both sites combined 1. t sites (s) 1 ( 1 /7.167 0'2 + 0'2 ) + 16 cr2 + 12 0'2 + 192 cr 2 w fr:s rs sf s 2. lots (f) 15 ( 11 7.167 0'2 + 0'2 ) w fr:s + [12 0'2 -,+ sf 24 0'2 f 15 ( 11 7.167 0'2 w 3. s x f + 0'2 ) fr:s + 12 0'2 sf 22 ( 1 /7.167 0'2 + 0'2 ) 16 0'2 w fr:s rs 4. reps:sites (r:s) 5. f x r:s(syn. plots) 330 ( 11 7.167 0'2 w + 0'2 ) fr:s 2582 11 7.167 2* 0' w 6. within plots b. tarawera lots (f) 16 ( 1 /7.078 0'2 w + 0'2 ) fr + 12 0'2 f reps (r) 11 ( 1 /7.078 0'2 w + 0'2 ) fr + 17 0'2 r f x r (syn. plots) 176 ( 1 /7.078 0'2 + 0'2 ) w fr within plots 1283 11 7.078 0'2 w -------------------------------------------------------------------------------c. kaingaroa lots (f) 21 0'2 + 8.3609 cr2 + 1.0613 cr 2 + 2 w fr r 87.9558 cj'f reps (r) 11 0'2 + 8.8038 cr2 + 161.7227 cr 2 + 1.0612 0'~ w fr r f x r (syn. plots) 208 0'2 + 7. 9799 0'2 0.1072 0'2 0.0561 0'2 w fr r f within plots 1700 0'2 w * obtained by dividing within plots m.s. by harmonic mean of numbers of trees in plots taf = 1.+ (3. + 4. 5.) where cr 2 within-plots variance w cr2 = lots x (reps within sites) (syn. plots) variance fr:s = reps within sites variance = lots variance sites variance = lots x reps variance term in square brackets omitted if sites are treated as a fixed effect. 17 table 3: frequency distributions for the different variables, site by site bounds of variable distribution class interval -----------------------~------------------------kaingaroa malftran d1drdef dldrexp dldrgen ed1drgen dlatgen edlatgen alldbkge ~lldbkge dfdbkge tarawera lower upper -5.03 2.83 0 1 0 2.30 0 8 0 3.48 0 40 0 11.00 0 70 0 15.82 0 10.08 malftran -5.03 2.83 dldrdef 0 1 dldrexp 0 2.30 dldrgen 0 8 ed1drgen 0 3.48 dlatgen 0 40 edlatgen 0 11.00 alldbkge 0 70 elldbkge 0 15.82 dfdbkge 0 10.08 overall malftran d1drdef dldrexp d1drgen edldrgen dlatgen edlatgen alldbkge elldbkge dfdbkge -5.03 0 0 0 0 0 0 0 0 0 2.83 1 2.30 8 3.48 40 11.00 70 15.82 10.08 1 2 3 21 0 124 1749 0 0 1749 0 0 1690 57 171 1690 0 57 1824 83 16 1548 194 125 1684 200 35 1528 120 187 1597 71 179 4 5 6 7 8 0 0 372 277 399 0 0 0 0 0 0 175 0 15 0 5 13 0 3 1 0 171 5 13 3 8 4 4 2 0 41 15 9 3 6 17 4 0 1 0 65 27 12 1 1 54 35 3 1 1 55 0 209 0 0 363 184 266 948 0 0 0 0 0 0 0 948 0 0 0 436 0 85 0 836 112 411 24 77 0 9 15 836 0 112 0 411 24 77 9 1054 249 93 47 26 10 6 0 595 258 342 111 94 57 15 13 809 362 154 91 43 20 3 4 556 170 248 268 142 51 43 4 645 123 252 232 145 44 39 3 76 0 333 0 0 735 461 665 2697 0 0 0 0 0 0 2697 0 0 0 611 0 100 2526 169 582 29 90 0 12 2526 0 169 0 582 29 90 2878 332 109 55 30 14 8 2143 452 467 152 109 66 18 2493 562 189 108 47 20 4 2084 290 435 333 169 63 44 2242 194 431 286 180 47 40 0 0 16 12 0 19 4 5 4 9 10 0 748 0 192 1 1 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 410 0 539 17 1 2 1 15 3 0 2 0 2 0 1 4 1 3 1 0 1158 0 731 18 2 2 2 16 4 0 2 0 2 0 1 4 1 3 1 19 table 4: overall means for p. radiata at kaingaroa and tarawera (based on lots that were fully represented at both sites) variable kaingaroa tarawera tarawera kaingaroa d.b.h.o.b. 151.02 162.02 11.0 *** straightness 5.57 5. 77 [o.20:1 n.s. branching 4.69 5.07 [0.38] ** malf(tran) 0,.88 0.17 -0.71 ** dldrdef++ 0.095 0.365 -0.250 *** dldrexp++ 0.099 0.406 -0.307 *** dldrgen ++ 0.242 0.995 -0.753 *** edldrgen++ 0.184 0.702 -0.518 *** dlatgen ++ 1.000 3.831 -2.831 *** edlatgen ++ o. 570 1.906 -1.336 *** alldbkge ++ 2.211 8.798 -6.587 *** elldbkge ++ 1.052 3.462 -2.410 *** dfdbkge ++ 0.579 2.038 -1.459 *** ++ denotes high score undesirable n.s. denotes not significant (p >0.05) ** denotes highly significant (p <0.01) *** denotes very highly significant (p <0.001) table 5: f ratio and significance levels in anova. inyqlying both sites variable sites t blocks:sites lots lots x sites lots x reps:sites --------------------------l,k d.f. 22,330 d.f. 15,330 d.f. 15,15 d.f. 15,330 d.f. 330,2582 d.f. d.b.h.o.b. 23.16 *** 4.55 *** 6.27 *** 4.29 ** 1..34 n.s. 1.11 n.s. straightness [1. 77 n.s~l 8.40 *** 10.87 *** 10.87 *** <1 n.s. 1.25 ** branching [10.69 **] 3.59 *** 10.97 '*** 10.97 *** <1 n.s. ~1 n.s. malftran 15.32 ** 3.27 *** 4o76 *** 3.98 ** 1.20 n.s. 1.21 * dldrdef 82.39 *** 3.03 *** 3.21 *** 1.62 n.so 1.98 * 1.01 n.s. dldrexp 92.39 *** 2.81 *** 2.92 *** 1.59 n.s. 1.83 * 1.70 *** dldrgen 116.10 *** 2.41 *** 2.62 *** l 70 n.s. 1. 54 n.s. lo23 ** n 1-' edldrgen 110 0 73 *** 2.82 *** 3 0 21 *** 1.71 n.s. 1.88 * 1.01 n.s. dlatgen 27.66 *** 8.42 *** 3.91 *** 2.05 n.s. 1.91 * 1.70 *** edlatgen 32.26 *** 10.61 *** 3.69 *** 2.03 n.s. 1.82 * 1.68 *** alldbkge 76.71 *** 5.30 *** 3.75 *** 1.72 n.s. 2.18 ** 1.23 ** elldbkge 77.80 *** 6.74 *** 3.89 *** 1.74 n.s. 2.29 ** 1.25 ** defdbkge 73.01 *** 6.17 *** 3.95 *** 1.61 n.s. 2.46 ** 1.25 ** n.s. denotes not significant (p >0.05) * denotes significant (p <0.05) ** denotes highly significant (p <0.01) *** denotes very highly significant (p <0.001) tk is variable, but exact values are clearly immaterial table 6: estimates of variance components and heritabilities from anova involving both sites variables 02 s ,. ,. ... 02 2 2 r:s 0 f (a) (} f (b) ... ... ... ,.. " a a (}2 02 02 h2h2h~a) h~b) fs fr:s w fa fb d.b.h.o.b. 57.80 22.73 22.51 21.06 2.89 10.19 661.06 0.84 0.79 0.13 0.17 straightness 0.0151 0.2067 0.1840 0.1840 0 0.0887 2.57124 0.91 0.,91 0.26 0.26 branphing 0.0655 0.0586 0.1501 0.1501 0 0 2.7175 0.91 0.91 0.21 0.,21 malftran 0.2380 0.0893 0.0988 0.0937 0.1028 0.1097 3. 7291 o. 79 0.75 0.10 0.09 dldrdef 0.03583 0.00268 0.00194 0.00108 0.00172 0.00025 0.14924 0.69 0.38 0.051 0.028 dlatgen 3.8617 1.3826 0.3614 0.2840 0.2261 1.2294 12.,5524 0.74 0.42 0.10 0.08 edlatgen 0.8647 0.2792 0.0520 0.0362 0.0362 0.1881 1.9827 0.73 0.51 0.09 0.06 alldbkge 21.4109 2. 2 519 0.9596 0.5505 0.8238 1.5481 48.9916 0.73 0.42 0.07 0.04 n elldbkge 2.,8659 0.3225 0.1083 0.0622 0.09211 0.1821 5.1421 0.74 0.43 0.08 0.05 w defdbkge 1.0489 0.1185 0.0450 0.0228 0.04557 0.07321 2.1092 0.75 0.38 0.08 0.04 ,. assumes sites are a random effect of(b) .. 2 assumes sites are a fixed effect 0 f (a) ,. a 02 2 f 1 f h f (a) = = ---,. " "' f cr2 + cr2 + cr2 f fr:s/24 w/7.167 .. "' 02 2 f 1 f h f(b) = ~ ~ a 2 = -(}2 + 02 + 02 + (} f f fs/2 fr:s/24 w/7.167 " 4 d2 2 f h (a) = a a1\. (}2 + 02 + (}2 f fr:s w a a 4 o 2 2 f h (b) = a a a a 02 + 02 + 02 + 02 f fs fr:s w 25 table 7: f ratios and significance levels at tarawera variable reps lots reps x lots 11,176 d.f. 15,176 d.f. 176,1283 d.f. d.b.h.o.b. 3.81 *** 4.40 *** <1 n.s. straightness 4.88 *** 4.64 *** 1.37 ** branching 1.77 * 4 .. 46 *** <1 noso malftran 1.81 * 3.06 *** 1.17 n.s. dldrdef 3.69 *** 2. 79 ** <1 n.s. dldrexp 3.42 *** 2 .. 48 ** <1 n.s .. dldrgen 2.56 ** 2.56 ** <1 n .. s. edldrgen 2.87 *** 2 .. 87 ** <1 n.so dlatgen 8.25 *** 3.00 ** 2.56 *** edlatgen 9.74 *** 2.93 *** 1.61 ** a11dbkge 4.66 *** 3.18 *** 1.13 n.s. e11dbkge 5.14 *** 3.51 *** 1.15 n.s" dfdbkge 5.78 *** 3.68 *** 1.11 n.s. table 8: estimates of variance components and heritabilities variable d.b.h.o.b. straightness branching malftran dldrdef dlatgen alldbkge elldbkge dfdbkge at tarawera statistic ---~--------~--------a-------a-------~-----~------(j2 (j2 a2 f r fr 30.3 17.7 0 0.149 0.118 0.141 0.116 0.018 0 0.124 0.035 0.107 0.00459 0.00487 0 0 0 785 2.009 1.814 2.412 2.412 1.485 0.263 0.306 0.1609 0.115 0.145 0.0531 " + (j2 w/12x7.078 a 2 4 (j-<= . j.. (j2 h2h2 w f 782.0 o. 77 0.15 2.665 0.78 0.20 2.986 0.78 0.15 4.351 0.67 o.ll 0.22167 0.64 0.08(0.30)t 20.506 0.67 0.14 83.643 0.69 0.13 7.770 o. 72 0.13 3.257 0.73 0.13 tadjusted to continuous underlying scale of variation table 9: lot means at tarawera variable lot -----------------------------------------------------------------------------------------------------------------------d.b.h. straight branch malftran dldrdef dldre:xp dldrgen edldrgen dlatgen edlatgen a11dbkge elldbkge dfdbkge o.b. -ness -ing 378 151 6.14 5.00 -0.46 0.49 0.54 1.25 0.90 4.,69 2.20 10.98 4.20 2.52 380 160 5.99 5.,22 0.06 0.32 0.36 0.92 0.64 3.69 1.92 8.30 3.39 1.,94 381 154 5.24 4.50 0.59 0.42 0.46 lc08 0.75 3.94 2.04 9.35 3. 72 2.25 383 166 6.42 5.54 -0.00 0.27 0.30 0.80 0.58 2.47 1 .. 36 6.,51 2.69 1.44 386 159 5.43 5.46 0.63 0.34 0.38 0.89 0.63 3.68 1.85 8.16 3.,20 1.92 387 156 6.41 5.67 0.80 0.28 0.31 0.71 0.50 5.10 2.37 8.67 3.37 2.05 388 156 5.80 4.93 0.11 0.39 0.43 1.06 0.72 4.07 1.92 9.40 3.56 2.13 391 161 5.87 5.03 0.77 0.,28 0.30 0.73 0.53 2.78 1.57 6.44 2.71 1.58 392 160 5.10 4.58 0.06 0.32 0.39 0.99 0.65 3.25 l68 8.23 3.23 1.91 393 163 5.83 5.13 -o.oo 0.44 0.48 1.19 0.87 3.98 1.95 9.95 3.91 2.28 394 161 5.63 4.88 0.02 0.28 0.33 0.86 0.58 3.23 1.72 7.56 3.12 1.82 395 172 6.26 5.73 0.82 0.19 0.22 0.61 0.43 2.45 1.33 5.51 2.,39 1.32 5.91 -0.15 0.44 1.08 0.77 2.98 1.68 8.39 3.48 2.03 n 399 164 4.37 0.40 -...] 400 171 6.28 5.28 -0.25 0.38 0.44 1.07 0.75 4.34 2.11 9.71 3.69 2.21 401 173 5.14 4.90 -0.50 0.50 0.55 1.32 0.92 6.76 3.02 13.40 4.92 2.96 all 166 5.58 5.09 0.38 0.34 0.39 1.00 0.70 2.96 1.51 7.99 3.19 . 1.80 r69/ 160 5.28 4.95 -0.07 0.43 0.47 1.17 0 .. 87 4.53 2.10 10.42 3.94 2.29 854 site 161.9 5.79 means 5.08 0.166 0.363 0.404 0.989 0.698 3.82 1.91 8. 77 3.46 2.03 lsd 8.29 0.577 0.510 0.682 0.141 0.159 0.338 0.247 1.74 0.664 2.93 0.90 0.58 there are no significant differences between the two controls between either control and the selections as a whole 29 table 10: f ratios and approximate significance levels at kaingaroa effect variable ------------------------------------------------------~---------reps lots reps x lots (a) reps x lots (b) ( 11' 2 08 d. f. ) ( 21' 2 08 d. f.) (208,1700 d.f.) (208,1700 d.f.) d.b.h.o.b. 6.81 *** 2.88 *** 1.48 straightness 17.61 *** 6.38 *** 1.15 branching 7.10 *** 6.08 *** 1.08 malftran 3.47 ** 2.68 *** 1.21 dldrdef 1.87 * 1.54cp;:;0.05) 1.02 dldrexp 1.86(p;:oe05) 1.57 (pr;0.05) lo03 dldrgen 3.24 *** 1.57(p:;0.05) 1.09 edldrgen 3.80 *** 1.47 n.s. 1.07 dlatgen 13.84 *** 1.40 n.s. 2.12 edlatgen 17.40 *** 1.34 n.s. 2.11 a11dbkge 9.20 *** 1.33 n.s. 1. 70 elldbkge 11.78 *** 1.,32 n.s. 1.64 dfdbkge 8.18 *** 1.42 n.s. 1.63 (a) denotes f ratio obtained from unadjusted mean squares (b) exact test for interaction in least squares anova ' 1.56 *** 1.15 n.s. 1.09 n.s .. 1.24 * 1.03 n.s. 1.04 n.s. 1.09 n.s. 1.07 nose 2.12 *** 2.,11 *** 1.71 *** 1.64 *** 1.62 note: where interaction was negligible the tests for main effects in least squares anova gave essentially the same results as presented here. table 11: estimates of variance components and heritabilities at kaingaroa variable d.b.h.o.b. straightness branching malftran dldrdef dlatgen e11dbkge dfdbkge statistic ---a----------a----------a----------a-------a----------------02 02 02 02 h2h2 f r fr w f 17.81 30.54 35.71 579.34 0.63 0.11 0.1695 0.2903 0.0526 2.4629 0.82 0.25 0.1520 0.0990 0.0277 2.4554 0.83 0.23 0.0703 0.1715 0.0872 3.1470 0.60 0.09 t 0.00055 0.00048 0.00027 o. 08795 0.35 0.025(0.09) 0.0363 0.8293 0.7612 5.3180 0.22 0.024 0.0119 0.3174 0.2379 2.9213 0.17 0.015 0.00740 0.08152 oa0903 1.1367 0.25 0.024 a 4 0 2 f ,., a ..,. 02 + 02 + 02 f fr w a 02 f (applicable to unadjusted lot means) (m.s. f) + 87.96 tadjusted to continuous underlying scale of variation table 12: lot means at kaingaroa, adjusted for rep effects variable lot (progeny/ seedlot) d.b.h.o.b. (mm) straightness branching malftran dldrdef dldrexp dldrgen edldrgen dlatgen edlatgen alldbkge elldbkge dfdbkge 378 379 380 381 383 384 385 386 387 388 390 391 392 393 394 395 396 397 400 401 all r69/854 ~ all r69/854 141 146 146 144 150 158 155 152 154 150 148 148 149 153 152 159 145 148 157 152 158 148 7.73 -2.56 lsd {approx) 8.69 5.73 5.27 5.91 5.38 5.84 5.10 5.30 5.22 6.29 5.41 5.46 5.53 4.51 5.85 5.68 6.20 5.72 5.39 5.75 4.75 5.62 4.84 0.13 -0.64 0.500 4.27 5.54 4.73 4.27 4.89 4.27 4.92 5.14 5.48 4.79 4.69 4.07 4.33 4.88 4.26 5.18 4.33 3.90 5.15 4.81 4.88 4.69 0.17 -0.02 0.483 significance of comparisons involving controls * * * n.s. n.s. ** 'v· n.s. n.s. n.s. 0.34 0.93 1.02 1.06 0.75 0.64 0.79 1.30 1.37 0.70 0.76 0.81 1.12 0.97 0.60 1.21 0.40 . 0.70 0.16 0.16 1.19 0.60 0.30 -0.29 0.089 0.175 0.118 0.070 0.094 0.151 0.082 0.172 0.062 0.065 0.128 0.056 0.071 0.121 0.119 0.022 0.085 0.111 0.065 0.066 0.085 0.127 0.090 0.186 0.142 0.070 0.099 0.167 0.132 0.178 0.062 0.065 0.128 0.057 0.071 0.121 0.125 0.022 0.090 0.121 0.071 0.072 0.085 0.127 0.098 0.030 0.023 -0.007 -0.006 0.43 0.40 0.16 0.24 0.40 0.40 0.41 0.17 0.14 0.32 0.15 0.18 0.33 0.27 0.08 0.22 0.26 0.17 0.20 0.20 0.28 0.26 0.02 0.00 n.s. between the controls n.s. n.s. between al 1 and progenies ) between r69/854 and progenies~ 0.32 0.26 0.13 0.18 0.27 0.22 0.30 0.14 0.11 0.25 0.13 0.14 0.27 0.20 0.07 0.17 0.19 0.13 0.15 0.16 0.22 0.19 0.03 0.00 1.35 1.13 0.75 0.55 1.05 0.52 1.45 0.83 0.72 0.57 0.16 1.47 0.64 1.03 0.79 0.93 1.11 1.11 1.08 2.26 0.79 o. 77 -0.16 -0.19 tp=0.05 0.76 0.63 0.44 0.34 0.62 0.35 0.72 0.53 0.45 0.36 0.17 0.81 0.44 0.58 0.46 0.56 0.56 0.58 0.60 1.06 0.47 0.45 -0.08 -0.09 3.49 3.11 1.56 1. 75 3.04 2.51 3.49 1.67 1.42 2.18 0.93 2.35 2.30 2.39 1.17 2.03 2.43 1.97 2.08 3.27 2.21 2.06 -0.04 -0.17 m.s.rf 0'~ ---+-x 1.96 88 20 1.60 1.28 0.76 0.85 1.36 0.99 1.48 0.90 o. 72 1.02 0.56 1.15 1.13 1.10 0.64 1.00 1.04 0.89 1.01 1.45 1.07 0.93 0.03 -0.11 0.95 0.69 0.45 0.43 0.81 0.59 0.85 0.47 0.40 0.52 0.20 0.59 0.54 0.67 0.32 0.55 0.62 0.51 0.54 0.83 0.62 0.51 0.04 -0.07 missing subclasses 0 6 2 0 0 7 0 0 0 0 4 0 0 0 0 0 4 0 0 0 0 0 w .... table 13: estimated correlations between mean levels of dieback in lots at tarawera and kaingaroa respectively a. phenotypic correlations between lot means tarawera kaingaroa variables ----------------------------------------------------------------------------1 variables d.b.h.o.b. malftran dldrdef dlatgen elldbkge dfdbkge d.b.h.o.b. 0.68 ** 0.05 -0.22 0.47 0.15 0.18 malftran 0.30 0.61 ** -0.27 -0.47 -0.51 * -0.53 * dldrdef -0.46 -0.29 0.31 0.37 0.42 0.43 dlatgen -0.11 0.01 -0.01 0.51 * 0.27 0.28 elldbkge -0.30 -0.19 0.16 0.49 0.38 0.40 dfdbkge -0.30 -0.17 0.12 0.47 0.34 0.35 -... h2f 0.63 0.60 0.35 0.36 0.17 0.25 b. genetic correlations d.b.h.o.b. 0.98 0.07 -0.42 0.87 0.41 0.41 malftran 0.46 0.96 -0.56 -0.96 -1.51 -1.29 dldrdef -0.72 -0.47 0.65 0.77 1.27 1.08 dlatgen -0.17 0.02 -0.02 1.04 0.80 0.68 elldbkge -0.45 -0.29 0.32 0.96 1.09 0.94 dfdbkge -0.45 -0.26 0.24 0.92 0.97 0.82 h2f 0.77 0.61 0.64 0.67 0.72 0.73 w w 35 table 14: estimates of genetic and phenotypic variances and correlations among lots at tarawera. variances are shown on the diagonals of matrices, covariances above diagonals and correlations below d.b.h.o.b. straight. br qual malftran dldrdef d1atgen genetic d.b.h.o.b. 30.3 0.109 0.386 -0.423 -0.086 straight. 0.05 0.149 0.094 o. 0112 -0.0142 br qual 0.21 0.71 0.116 0.0497 -0.0143 malftran -0.22 0.08 0.41 0.124 -0.0179 dldrdef -0.23 -0.54 -0.62 -0.75 0.0046 0.054 dlatgen 0.90 0.785 alldbkge 1.01 0.97 phenotypic d.b.h.o.b. 39.2 0.244 0.587 -0.401 -0.103 straight. 0.09 0.192 0.106 0.0219 -0.015 br qual 0.24 0.63** 0.150 0.065 -0.0171 malftran -0.15 0.12 0.41 0.184 0.0245 dldrdef -0.19 -0.39 -0.52* -0.68** 0.00715 0.063 dlatgen 0.69 1.177 alldbkge 0.91 0.91 * denotes significant (p <0.05) for phenotypic correlations ** denotes highly significant (p <0.01) phenotypic variances (o 2 ) are estimated as p a11dbkge 0.106 1.341 2.412 0.145 1.848 3.520 where families mean square calculated from subclass means. glasshouse response variable case b infection dbk. dbk ratio score d # case a infection dbk dbk ratio scored# h2f table 15: field vs glasshouse correlations, involving progeny means dieback variable d.b.h.o.b. malftran dldrdef dlatgen elldbkge .pfdbkge kaingaroa -0.24 0.50 -0.40 -0.03 -0.23 -0.22 -0.32 0.55 -0.30 0.18 -0.10 -0.10 -0.38 0.50 -0.15 0.27 -0.01 -0.01 0.26 -0.50 0.33 -0.12 0.13 0.13 -0.13 0.32 -0.28 -0.24 -0.21 -0.25 -0.21 0.32 -0.12 -0.03 -0.08 -0.10 -0.30 0.28 0.05 0.11 0.04 0.01 -0.15 -0.28 0.19 0.08 0.10 0.13 0.77 0.67 0.64 0.67 0.72 0.73 h2f(a)t h2f(brt 0.66 0.58 o. 76 0.60 0.&30 0.60 0.66 0.58 0.76 0.60 0.80 0.67 ------------------------------------------------------------------------------------------------------------case b infection dbk dbk ratio scored# case a infection dbk dbk ratio score d # h2f 0.00 -0.09 -0.24 0.07 o.oo -0.09 -0.25 0.07 0.63 0.45 -0.41 0.53* -0.21 0.53* 0.05 -0.49 0.31 0.25 -0.41 0.27 -0.21 0.31 0.06 -0.26 0.31 0.60 0.35 th 2f(a) assumes fixed effects in inoculation trial tth2 f assumes random effects in inoculation trial -rr reverse scale to infection/dieback incidence. tarawera -0.16 -0.41 -0.33 0.66 0.58 -0.08 -0.26 -0.19 0.76 0.60 0.02 -0.03 0.02 0.07 0.30 0.23 0.80 0.67 -0.16 -0.41 -0.33 0.66 0.58 -0.08 -0.26 -0.19 0.76 0.60 0.03 0.02 0.03 0.07 0.30 0.23 0.80 0.67 0.36 0.17 0.25 w -..] table 16: correlations between parental monoterpene levels (% total monoterpenes in cortical oleoresin) and mean incidence of dieback in progenies in field planting at tarawera (13 d.f.) variable case a d.b.h.o.b. malftran dldrdef dlatgen elldbkge dfdbkge case b d.b.h.o.b. malftran dldrdef dlatgen elldbkge dfdbkge monoterpene (see burdon et al., 1977a) h2~--~------------------------------------------------------------------------------------------------------f ~-pinene camphene s-pinene sabinene ~ 3 -carene myrcene limonene s-phell~n~rene x-terpinene terpinolene -o.ll 0.31 -0.07 0.10 -0.02 0.00 0.02 0.10 0.05 -0.06 -0.01 -0.01 0.02 0.25 ().02 0.54* 0.25 oo28 0.29 -0.13 0.27 0.29 0.32 0.30 -0.02 -0.11 0.06 0.09 0.09 oalo -0.04 -0.08 0.04 0.12 0.09 0.10 0.21 -0.17 0.35 ooll 0.25 0.22 0.20 -0.15 0.34 0.13 0.25 0.23 0.36 0.02 -0.26 -0.34 -0.31 -0.34 0.14 0.31 -0.40 -0.08 -0.30 -0.28 0.12 -0.07 -0.47 -0.44 -0.41 -0.42 0.22 -0.26 -0.25 -0.51 -0.30 -0.33 -0.36 -0.16 0.02 -0.00 0.06 0.05 -0.37 -0.14 0.02 0.01 0.06 0.05 -0~32 0.27 -0.15 0.21 -0.05 -0.01 -0.,18 -0.02 -0.02 -0.03 -0.03 -0.03 0.67*** -0.,25 0.17 -0.19 oooo -0.03 0.63* 0.02 -0.05 0.03 -0.02 -0.03 0.30 -0.17 0.31 0.03 0.18 0.16 0.27 -0.10 0.28 0.09 0.19 0.17 0.77 0.67 0.64 0.68 0.72 0.73 0.77 0.67 0.64 0.68 0.72 0.73 ------------------------------------------------------------------------------------------------------------------------------h2c 0.98 0.,98 0.96 h2= repeatability of clonal means c 0.98 0.95 0.91 0.98 0.99 0.89 0.98 w 1.0 table 17: correlations between parental monoterpene levels (% total monoterpenes in cortical oleoresin) and mean incidence of dieback in progenies in field planting at kaingaroa (19 d.f.) monoterpene variable ~-pinene camphene s-pinene sabinene ~ 3 -carene myrcene limonene s-phellenqrene y-terpinene terpinolene case a d.b.h.o.b. malftran dldrdef dlatgen elldbkge dfdbkge case b d.b.h.o.b. malftran dldrdef dlatgen elldbkge dfdbkge h2c -0.15 -0.17 0.07 0.05 0.11 0.06 -0.06 -0.02 -0.01 -0.02 -0.01 -0.04 0.98 -0.17 -0.08 0.25 0.36 0.39 0.30 -0.03 0.15 0.12 0.24 0.19 0.14 0.98 -0.28 -0.06 -0.10 -0.10 -0.10 -0.16 -0.29 -0.09 -0.09 -0.09 -0.08 -0.15 0.96 h 2 = repeatability of clonal means c -0.17 0.24 -0.07 0.15 0.05 0.01 -0.18 0.22 -0.06 0.16 0.06 0.02 0.98 0.29 0.06 -0.15 -0.01 '-:"0.15 -0.07 0.18 -0.16 -0.05 0.09 o.o1 0.06 0.95 0.19 -0.34 0.15 -0.04 0.25 0.21 0.28 -0 0 22 0.09 -0.11 0.16 0.14 0.,91 -0.03 0.07 -0.12 -0.18 -0.21 -0.18 -0.04 0.05 -doll -0.17 -0.20 -0.17 0.98 0.21 -0.24 0.59* 0.09 0.45* 0.46* 0.31 o.oo 0.41 -0.02 0.23 0.26 0.99 0.23 0.08 -0.32 0.20 -0.07 -0.,04 0.08 -0.14 -0.18 0.28 0.10 0.10 0.,89 -0.11 0.27 -0.11 0.15 0.01 -0.01 -0.14 0.19 -0.08 0.18 0.06 0.04 0.98 h2f 0.63 0.60 0.35 0.22 0.17 0.25 0.63 0.60 0.35 0.22 0.17 0.25 .!::> 1-' 43 table 18: comparisons between p. radiata and p. muricata variable d.b.h.o.b. straightness branching malftran dldrdef++ dlatgen++ elldbkge++ dfdbkge++ p.rad. 5.07 0.17 0.36 3.83 8.80 2.04 tarawerat p.mur. diff. 97.8 64.2 5.85 -0.08 4.67 0.40 -0.89 l 0 qq o. 71 0.35 7.41 3.58 18.90 9.10 3.85 1.81 tsome p. radiata familes missing site kaingaroa p ;~~~~~i--;~~~~:--;~;;~----;-*** 153.7 117.6 36.1 *** n.s. 5.61 6.56 -1.05 * n.s. 4.80 4.55 0.25 ** 0.84 1.05 -0.21 n.s. *** 0.08 0.12 0.04 n.s. * 0.66 0.92 0.26 n.s. *** 1.76 2.56 0.80 n.s. *** 0.46 0.66 0.20 n.s. tbased on means of block means (adjusted for missing subclasses) for those blocks in which p. muricata was represented ++ high score undesirable ·~ 1 1 j .,....,..,.," p---;,~ ."'i "f?'o ~ "..,)0 ' ·~ 1 ·~ '~ ~ j~ ')-. . s . ~'!"~i vi -... ~ b ..j j t4 1 0 lj , f c-:f -2 ll \. t.~ '-. j + ~ ~ ~~ 0 j -c ~ a j i ~ ~ vi j j j ~ '-..,./ ~ \... -j j '-lj 0 /'\ "'l ~ ;;:a • j "> ,) :ic<j j ~ j t.~ <:~<.... .) . ::j -t~ j' 1). ...... 0 ~ .::t "+ f:t 411 0 1 .j ~ i) ia ........ ·s ~ 0 m • () •• i>"'1 ~i ~ i • v growth and survival of eucalyptus viminalis in a frost-prone site in southern brazil, and implications for genetic management evandro v. tambarussi1,2*, eder d.b. silva3, rodolfo m.l. da costa4, jéssica f.f. santos5, vitor a. jatzek1 and rafael t. resende5,6 1 departamento de produção vegetal, faculdade de ciências agronômicas, universidade estadual paulista (fca/unesp), av. universitária, 3780, botucatu-sp, 18610-034, brazil. 2 programa de pós-graduação em recursos genéticos, escola superior de agricultura “luiz de queiroz”, universidade de são paulo (esalq/usp), av. pádua dias, 11, piracicaba-sp 13418-900, brazil 3 programa de pós-graduação em agronomia, universidade estadual do centro-oeste (unicentro), rua padre, r. salvatore renna, 875 santa cruz, guarapuava pr, 85015-430, brazil 4 suzano papel e celulose, centro de pesquisa de três lagoas. rodovia br 158, km 292, acesso a direta no km 04, s/n, fazenda buriti, três lagoas-ms, 79601-970, brazil 5 programa de pós-graduação em genética e melhoramento de plantas (ppggmp), escola de agronomia, universidade federal de goiás (ufg), av. esperança, s/n chácaras de recreio samambaia, goiânia go, 74690-900, brazil 6 programa de pós-graduação em ciências florestais, universidade federal de brasília (unb), faculdade de tecnologia, brasília – df, 70910-900, brazil *corresponding author: evandro.tambarussi@unesp.br (received for publication 11 may 2022; accepted in revised form 6 february 2023) abstract background: as the climate in southern brazil is cold with frequent frosts, eucalyptus species that can resist these climatic conditions are needed for commercial plantations. this study aims to evaluate provenances and families of eucalyptus viminalis labill. and compare them to 11 other eucalyptus spp. to select superior genotypes with high production potential and resistance to frost. methods: a total of 58 open-pollinated families from 16 provenances in australia were planted in a frost-prone site in irati, paraná state, brazil, using a randomised block design, with three replicates, linear plots of three plants, and a 2 × 2 m spacing, for a total experimental area of one hectare. another eleven eucalyptus species were planted in linear plots ranging from four to 59 plants alongside the experiment. results: survival for e. viminalis ranged from 0 to 68%, indicating genetic variability for frost resistance in this species. the variation among provenances was high (56%) for total genetic variation, indicating relatively high additive genetic differentiation among them. the other half of the total additive genetic variation was within (24%) and among families (20%), showing good variability among genotypes of each provenance. conclusions: in relation to the other 11 species, e. viminalis shows promise for frost-prone sites as the results are comparable to other species used in this region and clones from breeding programmes developed for these conditions. thus, e. viminalis offers the potential for selecting superior genotypes to be cloned for immediate genetic gains, as well as for the next generation of breeding. new zealand journal of forestry science tambarussi et al. new zealand journal of forestry science (2023) 53:5 https://doi.org/10.33494/nzjfs532023x236x e-issn: 1179-5395 published on-line: 22/03/2023 © the author(s). 2023 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access keywords: bayesian analysis; multi-species conservation; provenance and progeny trial; tree breeding; selection gain; spatial modeling. http://creativecommons.org/licenses/by/4.0/), tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 2 introduction in the subtropical regions of brazil, growth in the establishment of eucalyptus plantations has been limited due to the occurrence of severe and frequent frosts and the minimal number of eucalypt species that are adapted to this climatic condition (müller et al. 2017; frigotto et al. 2020). predicting the occurrence and intensity of frosts to address this problem becomes impractical, given the fiveto seven-year cycle of eucalyptus cultivation (diniz et al. 2021), and the fact that several frosts can occur throughout a single production cycle (alvares et al. 2018). thus, it is vital to produce genetic materials that are resistant to the adverse conditions found in these regions and can meet economic and productive demands (wendling et al. 2021). some eucalypt species and their genotypes are known to present the ability to limit the damage caused by frost (resistant) or grow despite frequent frosts and withstand the damage they cause (tolerant). the species with good productivity that have stood out for use in the southern states of brazil are eucalyptus benthammi maiden et cambage, eucalyptus dunnii maiden, and eucalyptus saligna sm. (lengowski et al. 2020); in the southeastern us the species used is e. benthammi (hall et al. 2019); and in southern china the species used are eucalyptus amplifolia naudin, e. benthamii, e. dunnii, and eucalyptus dorrigoensis (blakely) l.a.s.johnson & k.d.hill (arnold et al. 2015). in europe, previous studies recommend eucalyptus nitens (h.deane & maiden) maiden and eucalyptus gunni hook. f. in countries such as portugal, spain, and france (cerasoli et al. 2016). eucalyptus viminalis labill. is a species that can reach 90 m in height, with smooth, powdery bark that is white to pale brown and sheds in long ribbons. due to the presence of lignotubers (ladiges et al. 1974), the species regrows easily. like most eucalypt species, e. viminalis occurs in australia, where it extends from tasmania (tas) to the border between new south wales (nsw) and queensland. the best quality stands are found in victoria (vic) and nsw, where the terrain is mountainous with rigorous winters and frequent frosts, a fact that favors the planting of the species in southern brazil (flores et al. 2016). the wood is light yellow or pink in color, with high basic density (up to 0.748 g.cm-3) compared to other eucalypts, and it is widely used for the production of cellulose, fiber, and particle board, in sawmills, and as poles. eucalyptus viminalis shows several favorable traits, such as stem form, high productivity, and frost tolerance, and has been planted in several countries (gonzaga et al. 1983; louppe et al. 2009). due to its wood quality traits, growth, and resistance to water stress, disease, and frost, in addition to effective regeneration through resprouting (gonçalves et al. 2013), the species is indicated for interspecific hybridisation based on reciprocal recurrent selection (santos et al. 2013). the species has also performed well in tests with different eucalyptus species in argentina, where e. viminalis was ranked among the most productive species, with good wood quality, survival, and stem form (cappa et al. 2010). the introduction of new provenances and species with an adequate genetic base may help to meet the demand for more productive materials in eucalyptus spp. cultivation in the ecological conditions of subtropical regions (pinto junior & silveira 2021). therefore, it is important to assess the behavior of these different provenances, as they may present diverse adaptive potential (kageyama 1983; cruz et al. 2020; chaves et al. 2021). genetic and phenotypic parameters used to predict expected gains with the selection of superior genotypes and the best provenances in forest species, have been estimated with the restricted maximum likelihood/ best linear unbiased predictor (reml/blup) methods, and/or bayesian models via markov chain monte carlo (mcmc). this is mainly due to the precision that these methods offer in estimates obtained from unbalanced classifications (resende 2002; araújo et al. 2014; konzen et al. 2017; munhoz et al. 2021; oliveira et al. 2021; flôres júnior et al. 2021). in this context, the comparison between species from different trials may be biased, as the effect of the environment has a significant impact on the manifestation of phenotypic expression. ideally, the assessment of multiple species should be done in a combined experiment (santos et al. 2013); however, experiments commonly contain only one species and its progeny. in these cases, models with spatial effects within sites can be adjusted to circumvent the lack of perfect randomisation (rodriguez-alvarez et al. 2018), mapping the residual environmental effects of each plot by variograms, and enabling a more accurate comparison between diverse but spatially proximal experiments. this comparison of genotypic performance is valuable, as it enables the validation of a recommended genotype of a new species by comparing it to other species that are known to be suitable, in the present case, for frost resistance in other regions. the institute for forestry research and studies (ipef) has developed a robust, ongoing multi-species assessment program in different regions of brazil, with several species and strategic provenances of eucalyptus and corymbia (baroni et al. 2020; araujo et al. 2021). the present research is part of this ongoing programme and aims to evaluate provenances and progenies of e. viminalis and compare them to other eucalyptus spp. to inform the selection of superior materials with high productive potential and resistance to typical climatic conditions in the subtropical region of brazil. in addition, we outline future strategies for the genetic improvement of this potential eucalypt species for such region. methods study area and experimental design study area and trial site the studied population was established in december 2017 in irati, paraná (pr), brazil, at the state university of the central-west (coordinates 25º32’25”s and 50°39’51”w; elevation of 812 m; figures 1 & 2). the soil at the study site is classified as dystrophic tb haplic cambisol (cxbd) (embrapa 2018). the climate in the region is humid subtropical without a dry season (cfb), with severe and frequent frosts and evenly distributed rainfall; the average temperature in the hottest month does not reach 22 °c and in the coldest month it is greater than 10 °c, with precipitation from 1100 to 2000 mm (köppen & geiger 1928). this region of paraná is the coldest region in brazil, with frosts occurring frequently throughout the winter. the minimum and maximum temperature data (see additonal file) were obtained from the national meteorological institute (2021) for the period from december 2017 to october 2021. during this period, 75 days (2.68% of days) reached a minimum temperature below 5 °c and plant survival after eight months of age (after three periods of frost – table 1) was 84.7%. data for frost occurrence (per day) were taken from inmet (2022) (table 1). genetic material and field layout a total of 58 open-pollinated families in 16 provenances from australia were planted (table 2). the experiment was designed in randomised blocks with three replicates, linear plots of three plants, and a 2 × 2 m spacing, with double border rows of trees surrounding the plots. the total experimental area was 1 ha. simultaneously, 11 other eucalyptus species (table 2) were planted adjacent to the e. viminalis experiment in linear plots ranging from four to 59 plants (see supplementary material s1 for a detailed map of the experiment in the field). these species were planted to act as a control (table 2). all plants of each of the plots of the different species consisted of a mix of seeds from trees of the same population. trait assessment diameter at breast height (dbh, cm) was estimated by measuring the circumference at breast height (cbh, cm) collected with a measuring tape and the result divided by π. total height of the tree (ht, m) was measured with a vertex iv® hypsometer (långsele, sweden). dbh and ht were measured when plants were 46 months old (3.83 years), which is an age close to the length of eucalyptus rotation cycles used in brazil, and considered a reliable length of time for early selection practices tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 3 table 1: description of the study sites figure 1: (a) experimental area of the provenance and progeny test of eucalyptus viminalis; (b) eucalyptus viminalis during frost; (c) eucalyptus benthamii; and (d) eucalyptus longirostrata during frost. all images are from the experimental trial in irati, pr. (lima et al. 2011; castro et al. 2016). the stem volume (vol, m3 tree−1) was estimated as: v = 0.0000018 × dbh1.77298 × ht1.37336 as used previously for e. dunnii (figueiredo filho & amaral 2014). statistical model and genetic parameters models were fitted in the hierarchical bayesian framework in the r software environment (r core team 2020). to avoid using monte carlo markov chains, parameters were fitted through integrated nested tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 4 laplace approximation (inla) in the inla package in r (rue et al. 2009). the statistical model consisted of: y = µ + zp + wf + ma + su + ε where y is a vector containing the response variable, in this case wood volume; µ is the estimated intercept; p is the vector of estimated genetic effects of provenances; f is the vector of estimated family effects nested within provenances; a is the vector of estimated individual additive genetic effects nested within families and nested within provenances; u is the vector of estimated spatial effects based on the position of each plant; and ϵ is the vector of residual effects. z, w, m and s are incidence matrices for the respective effects. the genetic effects were parameterised respectively in the linear mixed model described above, with the following probability distributions: p ~ n(0, iσ2p) f ~ n(0, iσ2f) a ~ n(0, aσ2a) ϵ ~ n(0, iσ2ϵ) considering estimated variance among provenances (σ2p), variance among families within provenances (σ2f), variance among individuals within families within provenances (σ2a), environmental variance (σ 2 ϵ) and identity matrix (i) as fitted in a bayesian approach with a prior loggamma(1,1). the relationship among individuals is parameterised based on single-generation pedigree information (a) (wright 1922), thus capturing the relationship at the half-sibling level. such parametrisation is well known in animal breeding and is computationally efficient since the sparse relationship matrix (a-1) can be calculated directly (henderson 1976). the spatial term was set using the following parameterisations: stochastic partial differential equations (spde) in which the spatial term is defined as: u ~n(0,q σ2u), where q is a spatial covariance structure built via spde (lindgren et al. 2011; lindgren & rue 2015) and defined by the function: w(s) = (κ2 – δ)0.5α 𝑥(s), where w(s) is gaussian spatial white noise process, κ is scale parameter, δ is the laplacian ( ∂2/∂ki 2 ) ), and α is a smoothing parameter (α = ν + 0.5γ) with dimension domain γ. the heritability was estimated at all levels for genetic effects: narrow-sense heritability as: h2(a) = σ 2 a /(σ 2 p + σ 2 f + σ 2 a + σ 2 ϵ) day/month/year minimum temperature of 1200 utc (°c) frost intensity 20/05/2018 2.8 weak 21/05/2018 0.6 strong 26/08/2018 -0.2 strong 03/08/2019 0.1 strong 26/05/2020 1.8 weak 27/05/2020 1.6 weak 28/05/2020 3.0 weak 30/05/2020 3.6 weak 03/07/2020 0.4 strong 22/08/2020 -2.3 strong 23/08/2020 2.2 moderate 25/05/2021 0.5 strong 29/06/2021 -1.8 strong 30/06/2021 -2.0 strong 01/07/2021 -0.8 strong 19/07/2021 -3.3 strong 20/07/2021 -2.1 strong 21/07/2021 0.0 strong 22/07/2021 3.1 weak 28/07/2022 -0.7 strong 29/07/2022 -3.4 strong 30/07/2022 -3.9 strong 31/07/2’022 -1.0 strong 20/05/2022 0.6 strong 21/05/2022 1.2 moderate 22/05/2022 2.0 moderate 30/06/2022 1.4 moderate 01/07/2022 2.9 moderate 30/07/2022 1.2 moderate table 1: frost occurrence dates, temperature, and intensity in the eucalyptus viminalis experimental area in irati, paraná state (pr), brazil. 2 i = 1 ̭ tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 5 sp ec ie s o ri gi n so ur ce la ti tu de lo n gi tu de el ev at io n ( m ) n o. fa m il ie s n o. li vi n g pl an ts su rv iv al * ( % ) a ve ra ge ( m 3 t re e− 1 ) ( sd ) e. v im in al is g le nb og s f cs ir o 36 °3 8’ 14 9° 26 ’ 10 00 2 6 29 1. 41 ( 1. 03 ) w in ge llo s f cs ir o 34 °4 4’ 15 0° 10 ’ 60 0 1 1 17 6. 24 ( ---) m ou nt s un da y v ic cs ir o 37 °2 0’ 14 6° 26 ’ 96 0 2 5 25 2. 88 ( 2. 45 ) pr oc to rs r oa d ta s cs ir o 42 °5 7’ 14 7° 19 ’ 40 1 2 33 0. 29 ( 0. 09 ) ta lla ga nd a sf cs ir o 35 °2 6’ 14 9° 34 ’ 90 0 2 2 13 3. 60 ( 4. 30 ) ta lla ga nd a sf cs ir o 35 °5 8’ 14 9° 35 ’ 90 0 2 2 13 3. 79 ( 2. 37 ) n im m it ab el n sw ky lis a (b ul k) 36 °2 9’ 14 9° 16 ’ 10 75 1 1 11 0. 17 ( ---) n oo je e v ic ky lis a (b ul k) 37 °5 2’ 14 6° 00 ’ 27 5 13 36 27 3. 96 ( 3. 09 ) g eo rg et ow n ta s ky lis a (b ul k) 41 °0 6’ 14 6° 82 ’ 10 1 0 0 u ri ar ra a ct ky lis a 35 °1 5’ 14 8° 55 ’ 62 5 3 4 22 1. 72 ( 1. 35 ) w ar re n n sw ky lis a 31 °4 2’ 14 7° 48 ’ 20 0 1 2 22 2. 92 ( 2. 96 ) m ar ti ns c re ek v ic ky lis a 12 66 48 3. 91 ( 3. 62 ) ti m ba rr a v ic ky lis a 37 °0 7’ 14 7° 59 ’ 19 8 4 11 28 1. 99 ( 2. 12 ) er ri nu nd ra v ic ky lis a 37 °1 7’ 14 8° 53 ’ 44 0 8 40 40 4. 31 ( 4. 30 ) b al d h ill s v ic ky lis a 37 °2 6’ 14 3° 51 ’ 47 7 4 10 30 1. 12 ( 1. 13 ) fe de ra ti on ’ r an ge ky lis a 37 °2 8’ 14 5° 52 ’ 14 83 6 43 68 3. 90 ( 2. 23 ) e. b en th am ii ss o k ow en -p ur e se ed co m pa ny 35 °1 8’ 14 9° 18 ’ 65 0 1 34 58 5. 04 ( 6. 02 ) e. b ad je ns is d eu a n p ip ef ( bu lk ) 1 6 17 2. 39 ( 1. 59 ) e. b ot ry oi de s o rb os t ip ef ( bu lk ) 37 °4 3’ 14 8° 15 ’ 80 1 3 75 0. 72 ( 0. 86 ) e. c la do ca ly x ss o h am ilt on v ic ip ef ( bu lk ) 37 °4 9’ 14 2° 04 ’ 20 3 1 15 52 1. 62 ( 1. 23 ) e. d ea ne i k ed um ba v al le y ip ef ( bu lk ) 33 °4 9’ 15 0° 23 ’ 14 0 1 7 78 2. 93 ( 2. 89 ) e. lo ng ir os tr at a ip ef ( bu lk ) 1 21 10 0 4. 77 ( 2. 31 ) e. m ac ar th ur ii pa dd ys r iv er , n sw ip ef ( bu lk ) 1 23 46 1. 82 ( 2. 42 ) e. n ob ili s st yx r iv er s f ip ef ( bu lk ) 1 24 50 3. 27 ( 2. 93 ) e. o bl iq ua d ay le sf or d ip ef ( bu lk ) 1 4 15 2. 25 ( 0. 95 ) e. o cc id en ta lis o ld n ew ga te r oa d ip ef ( bu lk ) 1 5 13 1. 52 ( 1. 60 ) e. s m it hi n ar oo m a ip ef ( bu lk ) 36 °1 0’ 15 0° 04 ’ 15 0 1 7 21 1. 85 ( 1. 60 ) e. b en th am ii_ cl on e co m pa ny 1 10 83 9. 26 ( 8. 51 ) ta b le 2 : d et ai ls o f e uc al yp tu s v im in al is a nd o th er e uc al yp tu s s pp . t ri al s fo r w oo d vo lu m e (v o l) a t 4 6 m on th s of a ge , i nc lu di ng p ro ve na nc e, lo ca ti on , a nd d es cr ip tiv e at tr ib ut es of th e po pu la ti on s. sd : s ta nd ar d de vi at io n; * ra te o f s ur vi va l f or fa m ili es w it hi n th e or ig in s it e; ““ d at a is u na va ila bl e. family level heritability as: h2(fa) = σ 2 a + σ 2 f/(σ 2 p + σ 2 f + σ 2 a + σ 2 ϵ) and provenance level heritability as: h2(pfa) = σ 2 p + σ 2 f + σ 2 a/(σ 2 p + σ 2 f + σ 2 a + σ 2 ϵ) in addition, a measure of effective population size (ne) for selected individuals was calculated according to the following equation (resende & bertolucci 1995): ne = 4nf kf/(kf + 3 + (σ 2 kf / kf)) where nf is the number of families; kf is the average number of selected individuals per progeny; σ2kf is the variance of the number of selected individuals per family. the experiment used in the present study was a preliminary analysis to support two other subsequent aims: (1) to select high-performance families to be used in breeding programmes; and (2) to conserve genetic variability in this e. viminalis population pool in brazil. results descriptive results survival at 46 months varied widely ranging from 0 to 68% for e. viminalis, and for the other eucalyptus species from 13% for e. occidentalis to 100% for e. longirostrata (table 2). high survival rates were also found for e. botryoides (75%), e. deanei (78%), and e. benthamii (83%), which indicates the presence of genetic variability for cold resistance in these species. of the survival rates for e. viminalis from the 16 tested provenances, mt sunday and georgetown tas showed 0%, while the other provenances ranged from 11 to 68%. federation range showed the best survival among the e. viminalis provenances. the locations (figure 2) of the provenances occur naturally within coordinates that range between latitudes 31°42’ to 42°57’; and longitudes 143°51’ to 150°10’and with elevation ranging from 10 to 1483 m. this wide environmental variation can explain the diverse results obtained for survival and wood volume. compared to the other species planted in the same location, e. viminalis showed a higher mortality. when we evaluated the average individual volume there was also considerable variation, which was expected since the test includes a diversity of e. viminalis provenances. the three provenances that presented the best performance were federation range, uriarra act, and bald hills vic, in that order. however, some of the provenances presented individuals with good performance when compared to e. benthamii, which is a species with good growth that is adapted to conditions of intense cold. this demonstrates that there is a possibility of obtaining gains with the selection of the best individuals and the establishment of a breeding programme for the species. we found some e. viminalis families with individuals that had the same growth in wood volume as other species (additional file). estimated parameters most of the observed genetic variance was among provenances (σ2p= 0.185) (table 3). the variance observed among provenances was high, corresponding to 56% of the total genetic variation (figure 3). the other half of the total genetic variation was retained within families (24%) and among progenies (20%). breeding selection the results of the two selection strategies (i.e., selection aimed at breeding improvement or conservation) are presented in table 4. the selection of 10, 20, 30, 40, 50, and 60 individuals were defined and applied to both strategies. these values are based on the operational capacity of deploying new experiments in the next phase of the breeding programme. to obtain enhanced productivity, selection was made across families, with the only criterion being the highest predicted values. for conservation, selection was within and between families, identifying the best families and then the best individuals within these families. the best individual of the best family was selected, followed by the best individual of the second-best family, and so on. after selecting the best of each family, the second best was selected, and so on. in selection for production, which was more restrictive, provenances ranging from three to six individuals were included for selection of the 10 to 60 best across all provenances. in these provenances, errinundra vic, martins creek vic, and noojer vic, appear to offer the best performance with four to five tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 6 figure 2: provenance in australia of eucalyptus viminalis seeds collected for use in this study. the experimental station is located in irati, paraná state, brazil. ̭ ̭ ̵ ̵ ̵ ̵ families showing the highest results. the ne results for these three provenances were 12.88, 16.64, and 24.25, respectively, with the low values being the result of the selection of only a few provenances and families within the production-focused strategy. selection aimed at production obtained predicted gains of 104% in relation to the general average. this very high value is due to the average being relatively low (4.01 m³ in 1000 plants). when compared to the commercial clone adapted to the study conditions, we found negative gains from -23% to -37% for the selected e. viminalis individuals. in selection focused on conservation, the goal was to select the best individuals within the families to help maximise the ne. thus, it is possible to obtain gains relative to the average volume of 24% to 50%. relative to the e. benthamii clone, negative gains of -37% to -52% were found, suggesting much lower values in terms of volume yield. however, the ne was well above the average, and in the range of 20.00 to 59.01 individuals (table 4). the individuals selected for conservation will be recombined with each other to generate more variability that can be tested in further progeny trials for local environmental conditions. in the selection-for-production scenario, the behavior of e. viminalis is very similar compared to the other studied species (figure 4), where the selection response for e. viminalis (green line) and all other species (blue line) showed a minimal difference for individual tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 7 ‘ estimated parameter mean hpd (95%) genetic σ2p 0.19 0.08-0.32 σ2f 0.08 0.05-0.11 σ2a 0.07 0.05-0.09 non-genetic σ2u 0.05 0.03-0.08 spatial range 2.84 1.52-5.61 σ2ϵ 0.04 0.02-0.06 heritability h2(pfa) 0.89 0.82-0.95 h2(fa) 0.40 0.25-0.55 h2(a) 0.18 0.11-0.27 table 3: estimated genetic variance and heritability for wood volume (vol) in eucalyptus viminalis population at 46 months of age. hpd = highest probability density interval; σ2p = variance within provenance; σ2f = variance among provenances and families; σ 2 a = variance among provenances and individuals within families; σ2u = spatial variance; σ2ϵ = residual variance; h 2 (pfa) = heritability at pfa; h2(fa) = heritability fa; and h 2 (a) = heritability at the individual level (a). figure 3: percentage of genetic variance for a eucalyptus viminalis population for wood volume (vol) at 46 months age: provenance in green; families (half-sibs) in blue; and individuals in yellow. selection target n np nf kf σ 2 kf ne blup (m3 tree−1 × 1000) gains (%) comparative gains (%) e. viminalis e. longirostrata1 e. benthamii2 clone3 production* 20 3 9 2.00 2.22 11.78 8.77 82 28 32 -23 30 4 13 2.14 3.36 16.60 8.17 70 19 23 -29 40 5 17 2.35 3.24 23.77 7.75 61 13 16 -32 50 6 20 2.50 3.00 29.85 7.43 54 8 12 -35 60 6 25 2.73 3.16 39.61 7.17 49 4 8 -37 conservation** 20 12 20 1.00 0.00 20.00 7.23 50 5 9 -37 30 14 30 1.00 0.00 30.00 6.46 34 -6 -3 -44 40 15 40 1.00 0.00 40.00 5.99 24 -13 -10 -48 50 15 50 1.00 0.00 50.00 5.65 17 -18 -15 -51 60 16 58 1.03 0.03 59.01 5.56 16 -19 -17 -52 * for production, the best-ranked individuals are selected across families. ** for conservation, the best individual is selected within families. n: number of individuals selected; np = number of provenances in the selection; nf = number of families; kf = average number of selected individuals per family; σ2kf = variance of the number of selected individuals per family. 1 selection compared with total average of e. longirostrata (n=21, blup =6.87); 2 selection compared with total average of e. benthamii (n=34, blup =6.66); 3 selection compared with total average of e. benthamii commercial clone adapted to these conditions (n=1, blup =11.46). table 4: two selection strategies, one for genetic conservation and the other for recombination for production, and effective population size (ne). ̭ ̭ ̭ selection. these results indicate that the variability present in e. viminalis can produce individuals with similar performance to other species currently being used in commercial plantations. discussion suitability of e. viminalis for frost-prone conditions compared to other eucalyptus spp. some of the e. viminalis provenances showed individuals with high performance compared to other eucalypt species, particularly in comparison to e. benthamii, which is a species commonly used in southern brazil with good growth that is adapted to conditions of severe frost (graça et al. 1999). as such, the results suggest the possibility of obtaining genetic gains with the selection of the best individuals and the establishment of a breeding programme for these prospective individuals. in southern brazil, commercial plantations of eucalyptus spp. have been established using frosttolerant germplasm since the severe frosts that occurred in the 1970s (oliveira & pinto junior 2021). for all tested species (except e. longirostrata) planting in this region has been recommended due to the similar climatic conditions between the regions where the species occurs in australia and subtropical regions in brazil. however, e. viminalis is currently in high demand, as it can be crossed with other species to achieve high-quality wood for pulp and fiber board (cappa et al. 2010). in irati, paraná, several frosts can occur every year (table 1 and figure s1) which cause significant damage to the tropical eucalyptus species that generally have higher levels of productivity. plantations of e. viminalis have been recommended since the 1970s (sturion et al. 1988; higa et al. 1997), but to date their implementation has been minimal. higa et al. (1997) tested 20 subtropical eucalyptus species in paraná and found that e. viminalis from forest lands sf, nsw, was among the provenances that produced the most wood at eight years of age. more provenance and progeny tests have been carried out in paraná for the species (sturion et al. 1988); however, no large plantation has been established to examine its productivity at an industrial scale. beyond its tolerance to frost, there are several other reasons why e. viminalis should be planted in southern brazil, the most important being its high basic wood density (sturion et al. 1988) and the coppice shoot quality and quantity after thinning (higa & sturion 2000). however, as new studies on e. viminalis are carried out, important information about their resistance to frost will become available along with other environmental considerations, such as different soil management treatments and/or tree spacing (resende et al. 2018). it is also important to highlight the possible interactions between wood quality and tree growth in cold conditions, as this will certainly have an impact on wood growth patterns, and consequently stem form (cao et al., 2020; legowski et al., 2020). is there sufficient genetic variability in the e. viminalis population? the genetic variation identified in this study suggests wide variability among families from each e. viminalis provenance. the high levels of variability found between provenances in this trial is expected since its base population has not yet been genetically improved. further, materials were collected from sites across a wide geographical distribution, which although similar (below the tropic of capricorn – see figure 1) have different climates and conditions to those in which the provenances were exposed (kageyama & jacob 1980). our results corroborate the work by konzen et al. (2017), who found a marked genetic variance within eucalyptus spp. and corymbia spp. provenances, ranging from 63.6% to 73.1%. the variation observed within e. viminalis tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 8 figure 4: selection trend for wood volume (vol). (a) depicts selection gains considering the best individual, independent of provenance; (b) depicts selection for conservation, i.e., selecting the best individuals within families. the blue line is the selection differential for other species and the green line is for e. viminalis. the dotted green line is ne for e. viminalis at different selection levels. provenances indicates its potential for improved frost resistance, as well as higher productivity and improved wood quality. genetic variation between provenances of the same species can be quite useful, since these differences can result in distinct behaviors in a given environment when diverse geographic populations are used as a seed source (brune 1978). furthermore, the values obtained for heritability were considered intermediate h2(a) = 0.18 and h 2 (fa) = 0.40 and high h2(pfa) = 0.89. ziegler and tambarussi (2022) indicate that values ranging from 0.028< h2(a) ≤ 0.36 are moderate, and h2(a) > 0.53 are very high. individual heritability is consistent with values generally found in the literature for eucalyptus populations (resende 2002; henriques et al. 2017; paludeto et al. 2020) and are classified as intermediate. a high estimate indicates that there are high levels of genetic control at the provenance, family, and individual levels. significant heritability values will also favor greater genetic selection gains, as discussed further in the following subsection. the existence and magnitude of genetic diversity is fundamental at the beginning of breeding programmes, and throughout their development, since it is of paramount importance for the definition of strategies for conservation, management, and genetic improvement of a species (duarte et al. 2015; brandão et al. 2015). existing genetic variation indicates that the selection will be effective for conservation when applied between and within families of each provenance, and for the selection of superior genotypes when applied between and within families of the best provenance. selection practices for the next e. viminalis breeding cycle several selection methods can be used in a tree breeding programme, and it is up to the breeder to verify which methods meet their short-, medium-, and long-term demands. however, it is always important to consider not only the expected genetic gains, but also the maintenance of genetic variability that will enable long-term gains (vencovsky 1987; costa et al., 2016; araujo et al. 2021). for example, in the practice of forest tree breeding, the recurrent selection process is highly dependent on the existing genetic variability (as described in the previous subsection). therefore, the continuation of a new selection cycle must contain individuals selected for the trait of interest, but also those that contain enough genetic variability so that subsequent cycles continue to be successful (yamashita et al. 2018). the appropriate size of a selected population should be based on two considerations: (1) what are the best individuals, i.e., those that provide the greatest genetic gains; and (2) its effective population size (ne). to conduct a more parsimonious selection in relation to these two parameters (selection gain vs. ne), we can choose to perform such selection at the meeting point between the two. for example, in the individual selection (figure 4a), this occurred when we selected the best 50 individuals. furthermore, the individual selection demonstrated that it is more robust since it reached a higher ne even without including the provenances in the selection criteria. when we conduct selection considering individuals from the best families (figure 4b), the ne curve plateaus only when selecting at least 190 individuals, which is impractical for a recurrent forest selection programme. the gains predicted from selection in relation to the species average varied from 49 to 82% in the strategy aimed at obtaining new, improved cultivars, and 16 to 50% in the strategy aimed at genetic conservation. this was expected as these studied e. viminalis populations are considered wild, without any breeding or improvement for brazilian conditions. in comparison, costa et al. (2016) obtained gains for volume ranging from 15.5 to 20.5% for e. benthamii. it is important to note that e. benthamii shows better performance in terms of productivity as it has been subjected to improvement programmes. the site where our trial was implemented is located in a region that typically produces timber from e. benthamii; thus, the clone used in this experiment is a traditional regional material with superior performance. nevertheless, this does not undermine the potential shown here for e. viminalis. without much effort put into genetic improvement, we found that e. viminalis can provide materials that are competitive with current e. benthamii clones used in the region after only a few cycles of recurrent selection. in other studies, silva et al. (2018) obtained gains ranging from 25.7% to 26.6% for an e. urophylla progeny, which has been the focus of most breeding programmes in brazil. meanwhile, araújo et al. (2021) designed several improvement and conservation strategies for e. dunnii, in which the selection method and number of selected individuals varied, and obtained estimated gains ranging from 8 to 11%. in order to define the best strategy for the species, their study sought a balance between genetic gains and effective size, which is similar to the aims of the present study. in addition to the improvement and conservation of the species, as proposed herein, another strategy that can be used to accelerate the development of new, more productive cultivars is hybridisation, which can be done with species that show better performance in productivity. subsequently, the best individuals obtained in these crosses are cloned. as such, complementarity of the species traits used in crosses can be achieved (assis 2014). associated with the strategies discussed herein, is the integration of ideal silvicultural practices, such as tree spacing, fertilisation, and weed competition control (binkley et al. 2017). for e. viminalis, this is particularly important, as the species is still poorly studied, and this type of information is lacking. with this, the ideal is to construct a species improvement programme combined with the development of management practices to achieve success with the species or its hybrids in the region. it is important to highlight that, other studies with larger experimental scopes, including different environment testing, should be conducted for better comparison between species and the examination of species × environments interaction. furthermore, the number of individuals per family was limited due to the tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 9 ̭ ̭ ̭ ̭ ̭ tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 10 poor availability of seminal propagules. however, the research results presented here can be very helpful for conducting initial e. viminalis breeding programmes, both for regions with frequent frosts, and to compare the performance of e. viminalis materials with other eucalypts traditionally used in southern brazil. conclusions based on the preliminary tests presented in this study, there is potential for the commercial use of e. viminalis in cold and frost-prone regions in brazil. the results for the best individuals are comparable to other species used in this region and to clones from breeding programmes grown in similar conditions. competing interests the authors declare that there are no conflicts of interests. author contributions rmlc writing original draft, review & editing; edbs conceptualisation, formal analysis, writing original draft, review & editing; jffs writing original draft, review & editing; vaj data curation, investigation, rtr methodology, supervision, writing original draft, review & editing; evt conceptualisation, project administration, resources, writing original draft, review & editing. acknowledgements we would like to thank the “instituto de pesquisa e estudos florestais (pcmf/ipef)” for donating the seeds to establish this trial. evandro v. tambarussi is supported by research fellowship (grant number 304899/2019-4) and post-doctoral fellowship (grant number 200727/2020-6) from “conselho nacional de desenvolvimento científico e tecnológico” (cnpq). supplemental information data archiving statement phenotype data (https:// f i g s h a r e . c o m / a r t i c l e s / d a t a s e t / s e l e c t i o n _ o f _ eucalyptus_viminalis_for_frost_compared_to_species_ o f _ eu c a ly p t u s _ s p p _ to _ t h e _ s u b t ro p i c a l _ re g i o n _ o f _ brazil/17284874) references alvares, c.a., sentelhas, p.c., & stape, j.l. (2018). modeling monthly meteorological and agronomic frost days, based on minimum air temperature, in center-southern brazil. theoretical and applied climatology, 134, 177-191. https://doi. org/10.1007/s00704-017-2267-6 araújo, d., sebbenn, a.m., zanatto, a.c.s, zanata, m., morais, e., moraes, m.l.t., & freitas, m.l.m.(2014). variação genética para caracteres silviculturais em progênies de polinização aberta de astronium graveolens jacq. (anacardiaceae). cerne, 20, 61-68. https://doi.org/10.1590/s010477602014000100008 de araujo, m.j., lee, d.j., tambarussi, e.v., paula, r.c., & silva, p.h.m. (2021). initial productivity and genetic parameters of three corymbia species in brazil: designing a breeding strategy. canadian journal of forest research, 51, 25-30. https://doi. org/10.1139/cjfr-2019-0438 arnold r., li b., luo j., bai f., & baker, t. (2015). selection of cold-tolerant eucalyptus species and provenances for inland frost-susceptible, humid subtropical regions of southern china. australian forestry, 78, 180-193. https://doi.org/10.1080/00 049158.2015.1063471 assis, t.f. (2014). melhoramento genético de eucalyptus: desafios e perspectivas. in: 3º encontro brasileiro de silvicultura, campinas. 520 p. baroni, g. de r., de pieri, c., furtado, e.l., & da silva, p.h.m. (2020). genetic parameters of eucalyptus pilularis resistance to rust in controlled and field conditions. australasian plant pathology, 49, 65-68. https://doi.org/10.1007/s13313-019-00671-x binkley, d., campoe, o.c., alvares, c., carneiro, r.l., cegatta i., & stape, j.l. (2017). the interactions of climate, spacing and genetics on clonal eucalyptus plantation across brazil and uruguay. forest ecology and management 405, 271-283. https:// doi.org/10.1016/j.foreco.2017.09.050 brandão, m.m., vieira, f. de a., nazareno, a.g., & carvalho, d. (2015). genetic diversity of neotropical tree myrcia splendens (myrtaceae) in a fragmentcorridor system in the atlantic rainforest. flora morphology, distribution, functional ecology of plants, 216, 35-41. https://doi.org/10.1016/j. flora.2015.07.006 brune, a. (1978). genética e melhoramento florestal. universidade federal de viçosa, viçosa. cao, p.b., ployet, r., nguyen, c., dupas, a., ladouce, n., martinez, y., & teulières, c. (2020). wood architecture and composition are deeply remodeled in frost sensitive eucalyptus overexpressing cbf/ dreb1 transcription factors. international journal of molecular sciences, 21(8): 3019. https://doi. org/10.3390/ijms21083019 cappa, e.p., pathauer, p.s., & lopez, g.a. (2010). provenance variation and genetic parameters of eucalyptus viminalis in argentina. tree genetics & genomes, 6, 981-994. https://doi.org/10.1007/ s11295-010-0307-9 castro, c.a.d.o., resende, r.t., bhering, l.l., & cruz, c.d. (2016). brief history of eucalyptus breeding in brazil under perspective of biometric advances. ciência rural, 46, 1585-1593. https:// doi.org/10.1590/0103-8478cr20150645 about:blank about:blank about:blank about:blank about:blank https://doi.org/10.1007/s00704-017-2267-6 https://doi.org/10.1007/s00704-017-2267-6 https://doi.org/10.1590/s0104-77602014000100008 https://doi.org/10.1590/s0104-77602014000100008 https://doi.org/10.1139/cjfr-2019-0438 https://doi.org/10.1139/cjfr-2019-0438 https://doi.org/10.1080/00049158.2015.1063471 https://doi.org/10.1080/00049158.2015.1063471 https://doi.org/10.1007/s13313-019-00671-x https://doi.org/10.1016/j.foreco.2017.09.050 https://doi.org/10.1016/j.foreco.2017.09.050 https://doi.org/10.1016/j.flora.2015.07.006 https://doi.org/10.1016/j.flora.2015.07.006 https://doi.org/10.3390/ijms21083019 https://doi.org/10.3390/ijms21083019 https://doi.org/10.1007/s11295-010-0307-9 https://doi.org/10.1007/s11295-010-0307-9 https://doi.org/10.1590/0103-8478cr20150645 https://doi.org/10.1590/0103-8478cr20150645 tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 11 cerasoli, s., caldeira, m., pereira, j., caudullo, d., & rigo, d. (2016). eucalyptus globulus and other eucalypts in europe: distribution, habitat, usage and threats. in: san-miguel-ayanz, j., de rigo, d., caudullo, g., houston durrant, t., mauri, a. (eds.). european atlas of forest tree species. pp. 90-91. chaves, m.d.g., resquín, f., & balmelli, g. (2021). performance of eucalyptus tereticornis provenances in subtropical climate. agrociencia uruguay, 25: e322. https://doi.org/10.31285/agro.25.322 costa, r.m.l. da, estopa, r.a., biernaski, f.a., & mori, e.s. (2016) predição de ganhos genéticos em progênies de eucalyptus benthamii maiden & cambage por diferentes métodos de seleção. scientia forestalis, 44: 109. https://doi.org/10.18671/scifor. v44n109.10 cruz, s.l., pedrozo, c.â., oliveira, v.x.a, silva, a.m., resende, m.d.v., & gonçalves d.a. (2020). parâmetros genéticos e seleção inicial de procedências e progênies de taxibranco (tachigali vulgaris) em roraima. ciência florestal, 30: 258. https://doi.org/10.5902/1980509831631 de araujo, m.j., de paula, r.c., de moraes, c.b., pieroni, g., & silva, p.h.m. (2021). thinning strategies for eucalyptus dunnii population: balance between breeding and conservation using spatial variation and competition model. tree genetics & genomes, 17: 42. https://doi.org/10.1007/s11295-02101523-w diniz, é.s., lorenzon, a.s., de castro, n.l.m., marcatti, g.e., santos, o.p., júnior, j.c.d., cavalcante, r.b.l, fernandes-filho, e.i., & amaral, c.h. (2021). forecasting frost risk in forest plantations by the combination of spatial data and machine learning algorithms. agricultural and forest meteorology 306: 108450. https://doi.org/10.1016/j. agrformet.2021.108450 duarte, j.f., carvalho, d. de, & vieira, f. de a. (2015). genetic conservation of ficus bonijesulapensis r.m. castro in a dry forest on limestone outcrops. biochemical systematics and ecology, 59, 54-62. https://doi.org/10.1016/j.bse.2015.01.008 embrapa. (2018). sistema brasileiro de classificação de solos, 5a̲ edição revista e ampliada. https://www. embrapa.br/solos/sibcs; accessed apr. 21st 2022., brasília, df. figueiredo filho a., amaral s. do. (2014). compêndio de equações de volume e de afilamento de espécies florestais, 1st ed. flores, t.b., alvares, c.a., souza, v.c., & stape, j.l. (2016). eucalyptus  no brasil: zoneamento climático e guia para identificação. piracicaba: ipef, 447 p. flôres júnior, p.c., ishibashi, v., monteiro de matos, j.l., martinez, d.t., & higa, a.r. (2021). feasibility of early pinus taeda l. selection to assess growth variables in progeny test. revista floresta, 51, 937945. https://doi.org/10.5380/rf.v51i4.74441 frigotto, t., navroski, m.c., de aguiar, n.s. d., felippe, d., borsoi, g.a., de oliveira pereira, m., & lovatel, q.c. (2020). desempenho de espécies e procedências de eucalyptus no planalto norte catarinense, brasil. scientia forestalis, 48: e3273. https://doi. org/10.18671/scifor.v48n127.14 gonçalves, j.l. d. m., alvares, c.a., higa, a.r., silva, l.d., alfenas, a.c., stahl, j., ferraz, s.f.d.b., lima, w.d.p., brancalion, p.h.s., hubner, a., bouillet, j.-p.d., laclau, j.-p., nouvellon, y., & epron, d. (2013). integrating genetic and silvicultural strategies to minimize abiotic and biotic constraints in brazilian eucalypt plantations. forest ecology and management, 301, 6-27. https://doi. org/10.1016/j.foreco.2012.12.030 gonzaga, j.v., busnardo, c.a., & dias, c. (1983). caracterização da qualidade da madeira de eucalyptus viminalis introduzido na região de guaíba-rs. in: iii congresso latino-americano de celulose e papel. são paulo, 1053-1071. graça, m.e.c., shimizu, j.y., & tavares, f.r. (1999). capacidade de rebrota e de enraizamento de eucalyptus benthamii. boletim de pesquisa florestal, 39, 135-138. hall, k.b., stape j., bullock, b.p., frederick, d., wright, j., scolforo, h.f., & cook, r. (2019). a growth and yield model for eucalyptus benthamii in the southeastern united states. forest science, 66, 2537 https://doi.org/10.1093/forsci/fxz061 henderson, c.r. (1976). a simple method for computing the inverse of a numerator relationship matrix used in prediction of breeding values. biometrics, 32, 69-83. https://doi.org/10.2307/2529339 henriques, e.p., moraes, c.b. de, sebbenn, a.m., filho, m. t., de moraes, m.l.t., & mori, e.s. (2017). estimativa de parâmetros genéticos para caracteres silviculturais e densidade do lenho em teste de progênies de eucalyptus urophylla. scientia forestalis, 45, 119128. https://doi.org/10.18671/scifor.v45n113.11 higa, r.c.v., higa, a.r., trevisan ,r., & souza, m.v.r. de. (1997). comportamento de vinte espécies de eucalyptus em área de ocorrência de geadas na região sul do brasil. conferência iufro sobre silvicultura e melhoramento de eucaliptos, salvador. pp. 106-110. higa, r.c.v., & sturion, j.a. (2000). efeito do espaçamento na capacidade de brotação de eucalyptus viminalis. boletim de pesquisa florestal, 40, 77-83. inmet. instituto nacional de meteorologia. https:// portal.inmet.gov.br. accessed aug. 12, 2022. kageyama, p.y. (1983). melhoramento genético de pinheiros tropicais no brasil. silvicultura, 8, 17-21 kageyama, p.y., & jacob w.s. (1980). variação genética https://doi.org/10.31285/agro.25.322 https://doi.org/10.18671/scifor.v44n109.10 https://doi.org/10.18671/scifor.v44n109.10 https://doi.org/10.5902/1980509831631 https://doi.org/10.1007/s11295-021-01523-w https://doi.org/10.1007/s11295-021-01523-w https://doi.org/10.1016/j.agrformet.2021.108450 https://doi.org/10.1016/j.agrformet.2021.108450 https://doi.org/10.1016/j.bse.2015.01.008 https://www.embrapa.br/solos/sibcs https://www.embrapa.br/solos/sibcs https://doi.org/10.5380/rf.v51i4.74441 https://doi.org/10.18671/scifor.v48n127.14 https://doi.org/10.18671/scifor.v48n127.14 https://doi.org/10.1016/j.foreco.2012.12.030 https://doi.org/10.1016/j.foreco.2012.12.030 https://doi.org/10.1093/forsci/fxz061 https://doi.org/10.2307/2529339 https://doi.org/10.18671/scifor.v45n113.11 https://portal.inmet.gov.br https://portal.inmet.gov.br entre e dentro de populações de araucária angustifolia (bert) o. ktze. circular técnica 115. 7p. konzen, e.r., navroski, m.c., pereira, m. de o., nascimento, b., meneguzzi, a., & souza, p.f. (2017) genetic variation for growth variables of eucalyptus benthamii maiden & cambage and e. smithii r. t. baker provenances in southern brazil. cerne, 23, 359-366. https://doi.org/10.1590/01047760201 723032357 ladiges, p.y., & ashton, d.h. (1974) variation in some central victorian populations of eucalyptus viminalis labill. australian journal of botany, 22, 81-102. https://doi.org/10.1071/bt9740081 lengowski, e.c., júnior, e.a.b., vatras, b., neto, b.b.m., barros, j.m.r., & nisgoski, s. (2020). properties of wood from frost-tolerant eucalyptus planted in brazil. wood and fiber science, 52, 431-435. https://doi.org/10.22382/wfs-2020-041 lima, j.l., souza, j.c.d., ramalho, m.a.p., andrade, h.b., & sousa, l.c.d. (2011). early selection of parents and trees in eucalyptus full-sib progeny tests. crop breeding and applied biotechnology, 11, 10-16. https://doi.org/10.1590/s198470332011000100002 lindgren, f., & rue, h. (2015). bayesian spatial modelling with r inla. journal of statistical software, 63, 1-25. https://doi.org/10.18637/jss.v063.i19 lindgren, f., rue h., & lindström, j. (2011). an explicit link between gaussian fields and gaussian markov random fields: the stochastic partial differential equation approach: link between gaussian fields and gaussian markov random fields. journal of the royal statistical society: series b . statistical methodology, 73, 423-498. https://doi. org/10.1111/j.1467-9868.2011.00777.x louppe, d., oteng-amoako, a., & brinks, m. (2009). plant resources of tropical africa .7.1 timbers 1. afoc 22. pp. 105-106. https://doi. org/10.1163/2031356x-02202016 müller, b.v., rocha, m.p. da, klitzke, r.j., silva, j.r.m., & cunha, a.b. (2017). produção de madeira serrada com cinco espécies de eucalipto resistentes à geada. advances in forestry science, 4, 195-201. munhoz, l.v., santos, o.p. dos, valente, b.m. dos r.t., & tambarussi, e.v. (2021). genetic control of productivity and genotypes by environments interaction for eucalyptus dorrigoensis in southern brazil. cerne 27: e-102594. https://doi.org/10.159 0/01047760202127012594 oliveira, e.b. de, & pinto junior, j.e. (2021). o eucalipto e a embrapa: quatro décadas de pesquisa e desenvolvimento. brasília, df. 1163p. oliveira, l., marques, a, lopes, e.d, gonçalves, j., martins, n.s., pena, c.a.a., arbex, d.c., & laia, m.l.(2021). produtividade, adaptabilidade e estabilidade genotípica de clones de eucalyptus spp. e corymbia spp. em diferentes espaçamentos de plantio. scientia forestalis, 49: e3664. https:// doi.org/10.18671/scifor.v49n131.21 paludeto, j.g.z., perek, m., munhoz, l.v., santos, j.r.m., & pesck, v.a., tambarussi, e.v. (2020). variabilidade genética em população base de eucalyptus viminalis em idade juvenil. scientia forestalis, 48: 1-9. https://doi.org/10.18671/scifor.v48n126.07 pinto junior, j.e., & silveira, r.a. (2021). a introdução do eucalipto no brasil pela embrapa: bases institucionais e sua estruturação para a pesquisa com eucaliptos e corímbias. in: oliveira, e.b. de; pinto junior, j.e. (eds.). o eucalipto e a embrapa: quatro décadas de pesquisa e desenvolvimento p. 36. r core team. (2020). r: a language and environment for statistical computing. r foundation for statistical computing, vienna, austria resende, m.d.v. de. (2002). genética biométrica e estatística no melhoramento de plantas perenes. embrapa informação tecnológica, brasilia, df. 975p. resende, m.d.v., & bertolucci, f.l.g. (1995). maximization of genetic gain with restriction on effective population size and inbreeding in eucalyptus grandis. in: proceedings. crc for temperate hardwood forestry, hobart, pp 167-170. resende, r.t., soares, a.a., forrester, d.i., marcatti g.e., dos santos, a.r., takahashi, e.k., silva, f.f., grattapaglia, d., resende, m.d.v., & leite, h.g. (2018). environmental uniformity, site quality and tree competition interact to determine stand productivity of clonal eucalyptus. forest ecology and management , 410, 76-83. https://doi. org/10.1016/j.foreco.2017.12.038 rodriguez-alvarez, m.x., boer, m., eilers, p., & van eeuwijk, f. (2018). spats: spatial analysis of field trials with splines. version 1.0-15. https:// cran.r-project.org/package=spats. accessed jun. 15, 2019. rue, h., martino, s., & chopin, n. (2009). approximate bayesian inference for latent gaussian models by using integrated nested laplace approximations. journal of the royal statistical society: series b (statistical methodology), 71, 319-392. https://doi. org/10.1111/j.1467-9868.2008.00700.x santos, g.a., resende, m.d.v., silva, l.d., higa, h., & assis, t.f. (2013). adaptabilidade de híbridos multiespécies de eucalyptus ao estado do rio grande do sul. revista árvore, 37, 759-769. https:// doi.org/10.1590/s0100-67622013000400019 silva, p.h.m., brune, a., pupin, s., moraes, m.l.t., sebbenn, a.m., & paula, r.c. (2018). maintenance of genetic diversity in eucayptus urophylla s.t. blake populations with restriction of the number of trees per family. silvae genetica, 67, 34-40. https://doi. tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 12 https://doi.org/10.1590/01047760201723032357 https://doi.org/10.1590/01047760201723032357 https://doi.org/10.1071/bt9740081 https://doi.org/10.22382/wfs-2020-041 https://doi.org/10.1590/s1984-70332011000100002 https://doi.org/10.1590/s1984-70332011000100002 https://doi.org/10.18637/jss.v063.i19 https://doi.org/10.1111/j.1467-9868.2011.00777.x https://doi.org/10.1111/j.1467-9868.2011.00777.x https://doi.org/10.1163/2031356x-02202016 https://doi.org/10.1163/2031356x-02202016 https://doi.org/10.1590/01047760202127012594 https://doi.org/10.1590/01047760202127012594 https://doi.org/10.18671/scifor.v49n131.21 https://doi.org/10.18671/scifor.v49n131.21 https://doi.org/10.18671/scifor.v48n126.07 https://doi.org/10.1016/j.foreco.2017.12.038 https://doi.org/10.1016/j.foreco.2017.12.038 https://cran.r-project.org/package=spats https://cran.r-project.org/package=spats https://doi.org/10.1111/j.1467-9868.2008.00700.x https://doi.org/10.1111/j.1467-9868.2008.00700.x https://doi.org/10.1590/s0100-67622013000400019 https://doi.org/10.1590/s0100-67622013000400019 https://doi.org/10.2478/sg-2018-0005 org/10.2478/sg-2018-0005 sturion, j.a., pereira, j.c.d., & chemin, m.s. (1988). qualidade da madeira de eucalyptus vimanalis para fins energéticos em função do espaçamento e idade de corte. boletim de pesquisa florestal, 16, 55-59. vencovsky, r. (1987). tamanho efetivo populacional na coleta e preservação de germoplasma de espécies alógamas. ipef 35, 79-84. wendling, i., dutra, l.f., gabira, m.m., gabira, m.m., vieira, l.m., & degenhardt, j.(2021). produção de mudas de eucalipto. in: e.b de oliveira & j.e. pinto junior (eds.). o eucalipto e a embrapa: quatro décadas de pesquisa e desenvolvimento, embrapa, brasília, df. p. 36. wright, s. (1922). coefficients of inbreeding and relationship. the american naturalist, 56: 330-338. https://doi.org/10.1086/279872 yamashita, m., mullin, t.j., & safarina, s. (2018). an efficient second-order cone programming approach for optimal selection in tree breeding. optimization letters, 12, 1683-1697. https://doi.org/10.1007/ s11590-018-1229-y ziegler, a.c.f., & tambarussi, e.v. (2022). classifying coefficients of genetic variation and heritability for eucalyptus spp. crop breeding and applied biotechnology, 22: e40372222. https://doi. org/10.1590/1984-70332022v22n2a12 tambarussi et al. new zealand journal of forestry science (2023) 53:5 page 13 https://doi.org/10.2478/sg-2018-0005 https://doi.org/10.1086/279872 https://doi.org/10.1007/s11590-018-1229-y https://doi.org/10.1007/s11590-018-1229-y https://doi.org/10.1590/1984-70332022v22n2a12 https://doi.org/10.1590/1984-70332022v22n2a12 incorporating competition factors in a mixed-effect model with random effects of site quality for individual tree above-ground biomass growth of pinus kesiya var. langbianensis mingchuan nong 1,#, yan leng 1,2#, hui xu1,#, chao li1, guanglong ou1* 1 key laboratory of state forestry administration on biodiversity conservation in southwest china, southwest forestry university, kunming, china; 2 school of biotechnology and engineering, dianxi science and technology normal university, lincang, china *corresponding author: olg2007621@126.com # these authors contributed equally to this work (received for publication 7 august 2018; accepted in revised form 21 november 2019) abstract background: accurate biomass estimation has critical effects on quantifying carbon stocks and sequestration rates, and aboveground biomass (agb) growth models are a key component of tree biomass estimation. the study objective was to develop a growth model for agb of an individual tree by combining competition factors and site quality using a mixed-effect model. methods: the agb of 128 sampling trees was investigated for simao pine (pinus kesiya var. langbianensis) at three typical sites near pu’er city of yunnan province, china. richards’ equation was used for the basic growth model (bm) of the agb, and a mixed-effect model with random effect of site quality (mem) based on bm and a mixed-effect model with fixed effect of competition factors (memc) based on mem were built using s-plus. results: both mixed-effect models are significantly better than the basic model in fitting and predicting the individual tree agb growth for simao pine, but the mem is better than the memc. moreover, the mixed-effect model with competition factors and site quality is the optimal estimation model due to its highest prediction precision (p=86.08%) as well as the lowest absolute average relative error (rma=54.34%) and average relative error (ee =6.45%). conclusion: a model including site quality and competition factors can be used to improve the tree agb growth estimation for the individual tree agb growth of simao pine. new zealand journal of forestry science nong et al. new zealand journal of forestry science (2019) 49:11 https://doi.org/10.33494/nzjfs492019x27x e-issn: 1179-5395 published on-line: 11/12/2019 © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access as the potential climate mitigation benefits of forest management and for assessing climate change impacts on forest ecosystems (temesgen et al. 2015; cosmo et al. 2016). direct measurements of biomass are time consuming and expensive, so the usual procedure is to destructively sample a subset of trees for the development of allometric models that predict biomass from commonly measured tree characteristics like diameter (dbh) and height (h) (houghton 2005; saintintroduction forests are important components of terrestrial ecosystems, and they play vital roles in the global carbon balance (woodwell et al. 1978; talhelm et al. 2014; sleeter et al. 2018). trees are an important element of forest ecosystems’ carbon storage (goodale et al. 2002; houghton 2005; houghton et al. 2009; luo et al. 2014). accurate estimates of tree biomass are critical for quantifying carbon stocks and fluxes as well keywords: above-ground biomass, competition factors, growth model, mixed-effect model, pinus kesiya var. langbianensis, richards’ equation, site quality nong et al. new zealand journal of forestry science (2019) 49:11 page 2 andre et al. 2005; somogyi et al. 2007; sileshi 2014; chave et al. 2014; temesgen et al. 2015). equations for tree biomass estimation can be categorized into regional biomass conversion factors, stand-level and tree-level biomass equations (di cosmo et al. 2016; jagodziński et al. 2018). tree-level biomass models are used to estimate the total and the components of individual trees with easily measurable tree inventory attributes (e.g. dbh, h). the tree-level biomass model became a research priority because of its better performance and higher predicting precision compared with regional and stand-level biomass models (temesgen et al. 2015). more than 2600 biomass models related to more than 100 tree species have been constructed all over the world (e.g. ter-mikaelian & korzukhin 1997; chojnacky 2003; jenkins et al., 2003, 2004; zianis 2005; muukkonen 2007). although many studies on tree-level biomass models have already been developed, the existing equations are generally not good enough and need improvement (temesgen et al. 2015). moreover, almost all of these studies were static models and focused on the relation among the individual tree biomass and predictors, whereas few studies have considered the growth of individual tree biomass using growth equations. furthermore, spatial and temporal autocorrelation between individuals and groups may be neglected in modelling tree biomass growth (li & zhang 2010). it is necessary to develop comprehensive biomass estimation models with better prediction performance that consider differences in stand density and among sites (temesgen et al. 2015). forest biomass estimation variation can be observed between different ecological zones and sites (henry et al. 2011, de-miguel et al. 2014). trees have different growth rates due to differences between sites, and models perform better when site quality is considered in the model (vanninen et al. 1996; rohner et al. 2013). furthermore, site quality also has an important effect on tree growth, and it can be quantified as a site index (si) which represents the potential productivity of the forestland by using the dominant height of stands at a particular standard age (running & mcleod 1988; carmean et al. 1989; sturtevant & seagle 2004; waring et al. 2006). stand density can be used to reflect the effects of crowding and competition among trees on tree growth (curtis 1985; zeide 2005). competition results from interactions between many biotic and abiotic factors, and this affects the forest structure (sahney et al. 2010). with increasing stand density, individual trees can be restricted in their growth and even die (bragg 2001). many studies have shown that tree size is strongly affected by competition (coomes & allen 2007; coates et al. 2004, 2009). by incorporating the size and distance of neighboring trees into the model, predictions of growth and mortality can be improved (mctague & weiskittel 2016). biomass growth equations are generally constructed by using forestry investigation data and contain multiple repeated observations of individual trees. the characteristics of these data led to spatial and temporal correlations among observations in the same sampling unit (lhotka & loewenstein 2011; timilsina & staudhammer 2013). a previous parameter estimation of the individual tree growth equation was estimated by ordinary least squares (ols). this would violate the regression assumptions of homoscedasticity of variances and independence of residuals and lead to inaccurate estimates if the growth model was developed based on non-independent data by ols (budhathoki et al. 2010; njana et al. 2016). to address these problems, many researchers have tried to develop new growth models by incorporating mixedeffect modelling techniques (budhathoki et al. 2010; lhotka & loewenstein 2011). a mixed-effect model, which consists of fixed effects and random effects, provides a flexible and powerful tool for the analysis of grouped data (pinheiro & bates 2000). fixed effects can indicate the average relationships of a dependent variable with independent variable(s), and random effects can reflect the difference among groups (razali et al. 2015). the advantage of mixed-effect models is that they can fit growth and yield data in forestry fairly well via multilevel random effects (gregoire et al. 1995), and the prediction accuracy of such models can be improved through modifications of random effects (calama & montero 2004). in recent years, mixed-effect models have been widely applied in forestry due to their better fitting performance and prediction precision (ramon et al. 2006), and they have been used to estimate the dominant height growth (fang & bailey 2001), diameter increment (lhotka & loewenstein 2011), tree stand basal area (gregoire et al. 1995), basal area growth of individual trees (budhathoki et al. 2008), wood density (li & jiang 2013), stand volume (li 2012), and biomass (zhang & borders 2004; fehrmann et al. 2008; fu et al. 2012; njana et al. 2016) by incorporating different random effects (e.g. tree-level, plot-level). therefore, it is crucial to accurately predict biomass growth by constructing a comprehensive model with good prediction performance using a mixed-effect model that considers competition factors and site quality. simao pine (pinus kesiya var. langbianensis), a geographic variation of p. kesiya, is distributed in mountain areas from the northern tropical zone to the southern subtropical zone of yunnan province, china. its distribution area and stocking volume account for 11% of the forestland of yunnan province (compilation committee of yunnan forest 1986). it has been an important tree species for afforestation in yunnan due to its rapid growth (southwest forestry college & forestry department of yunnan province 1988). moreover, simao pine forests provide a range of goods and services and have important economic and ecological values (wu & dang 1992; wen et al. 2010; yue & yang 2011; li 2011). therefore, it is important to be able to estimate biomass growth comprehensively and with high prediction accuracy by incorporating site index and competition factors to assess the potential value of these forests and to guide forest management. in this study, natural forests of simao pine were studied by sampling the above-ground biomass (agb) of 128 trees at three sites. mixed-effect models incorporating site quality and a competition factor were used to construct the agb growth model based on a transformation of richards’ equation. this study aimed to: (1) explore a comprehensive biomass growth model with high prediction accuracy for estimating the agb growth of simao pine; and (2) explain the impacts on improving agb estimation from site quality and competition factors. methods study region the study region is located in mojiang county, simao district and lancang county which belong to pu’er city, southwest of yunnan province, p. r. china. in this city, mountain areas comprise 98.3% of the overall region, and the study region is located between 22°02′n to 24°50′n and 99°09′e to 102°19′e. three typical geographic areas with different climates, tongguan town of mojiang county (site i), yunxian town of simao district (site ii) and nuofu town of lancang county (site iii), were selected as study sites (fig. 1), and the elevation of the sites are from 1400 m to 1600 m above sea level. the mean annual temperatures and annual precipitation both decrease from south to north (fig. 2). lancang county has the highest annual mean temperature (19.7 ℃) and annual precipitation (1586.5 mm), and mojiang county has the lowest values (the annual mean temperature is 18.4 ℃ and the annual precipitation is 1306 mm) (ou et al. 2016). data collection and processing data investigation a total sample of 128 pines in 45 plots with an area of 900 m2 were selected and investigated in the study areas (fig. 1). plot information including latitude, longitude, degree of slope, and aspect of slope was recorded. tree age (a), diameter at breast height (dbh) and tree height (h) of the 128 sampled trees were recorded (fig. 3). tree age was determined by counting the number of annual growth rings of the stump of each felled sample trees. then, we also recorded the basic characteristics of the surrounding trees within 5 metres of the sample trees, including tree species, dbh, h and the distance to calculate the competition index (ci). we calculated the average height of dominant trees (ht) and the average age of stands (a) to calculate the site index (si). the average height of the dominant trees for each plot is the mean of the three highest trees, and the average age of the stand of each plot is the mean age of the standard trees with a dbh similar to the average dbh of the plot. biomass measurement according to the sample collection method for tree biomass modelling in china, the biomass of each tree component was measured one at a time (zeng et al. 2011). the biomass of the stem was measured by the method of volume density. firstly, felled trees were segmented and weighed, and the fresh weight was measured. secondly, the segment length and diameter were measured, and the volume of the trunk was calculated. branches and leaves were measured by the method of the graded sample branch. dead branches and fruits were measured by the method of total weight. thirdly, the samples from the nong et al. new zealand journal of forestry science (2019) 49:11 page 3 figure 1: location of the study sites. figure 2: monthly mean temperatures and monthly gross precipitation of three typical sites. the data are average values from 1980 to 2010 measured by the respective county weather stations. the lines are the monthly mean temperatures, and the bars are the monthly gross precipitation. different components were dried to a constant weight at 105 ℃ using an oven, and the sample density of the wood and bark was also measured using the drainage method. finally, the biomass of wood and bark of the sampled trees was calculated using the volume and the corresponding sample density, and the branch biomass and needle biomass of sample trees were calculated nong et al. new zealand journal of forestry science (2019) 49:11 page 4 using fresh weight and the corresponding dry matter ratio. the sample trees were divided into two sets by random selection; one set with 96 sample trees was used to fit the models, and the other one was used for the test. the basic characteristics are listed in table 1. figure 3: scatter of the diameters at breast height (dbh) and tree height (h) vs tree age. a: dbh vs age, b: h vs age. data set si class (m) n age (years) dbh (cm) height (m) agb (kg) mean standard deviation mean standard deviation mean standard deviation mean standard deviation fitting 12 7 28.43 2.59 21.24 2.83 16.84 1.82 197.50 50.37 14 34 38.12 2.65 22.49 1.81 16.86 0.79 297.96 63.59 16 19 38.58 4.44 20.71 3.00 14.86 1.42 294.57 111.64 18 26 38.73 2.20 34.00 2.24 21.45 1.17 622.36 83.73 20 10 36.40 4.54 31.59 4.71 23.26 3.17 684.21 197.22 total 96 37.49 1.50 26.11 1.30 18.37 0.67 418.06 46.68 testing 12 2 32.00 2.00 28.75 3.55 19.30 0.10 345.37 71.94 14 5 39.00 5.67 19.66 3.20 16.78 1.52 185.89 60.91 16 13 50.00 5.59 27.68 3.55 18.71 1.57 395.02 92.28 18 10 44.70 3.68 37.88 2.57 23.80 2.26 704.72 96.04 20 2 48.00 2.00 40.50 5.60 32.25 2.75 830.54 283.31 total 32 45.38 2.77 30.48 2.08 20.88 1.19 483.24 61.21 all 12 9 29.22 2.07 22.91 2.50 17.38 1.44 230.36 45.79 14 39 38.23 2.40 22.12 1.62 16.85 0.71 283.59 56.11 16 32 43.22 3.57 23.54 2.34 16.42 1.09 335.38 75.59 18 36 40.39 1.92 35.08 1.77 22.11 1.05 645.24 65.68 20 12 38.33 3.98 33.08 4.08 24.76 2.83 708.60 167.35 total 128 39.46 1.35 27.20 1.11 19.00 0.59 434.35 38.18 table 1. basic characteristics of the sampled trees competition index calculation to calculate the competition index (ci), we used the formula of hegyi (1974), equation 1. (1) where cii is a competition index, di is the diameter of sample tree i, dj is the diameter of competition tree j around the sample tree, and distij is the distance from sample tree i to competition tree j. site index calculation to derive the site index (si), the dominant height and age of each plot were measured. si was calculated for simao pine using the equation by wang (2003), equation 2. (2) where si is the site index, ht is the average height of dominant trees of each plot, a is the average age of each plot, and the base age is 20 years according to wang (2003). the site index of the plots includes five types from 12 m to 20 m according to an interval of 2 m in this study (table 1). model fitting basic model (bm) richards’ growth equation developed from bertalanffy’s growth theory is used to describe biological growth changes over time. growth equations such as monomolecular, gompertz, and logistic equations are richards’ growth equations of the special form. richards’ equation is widespread in forestry because of its flexibility and excellent fitting performance (richards 1959, liu & li 2003, rohner et al. 2013). in the present study, richards’ growth equation is used as a basic biomass growth model. the general expression of richards’ growth equation is listed in equation 3. (3) where parameters a, b and c are given in equations 4–6: (4) (5) (6) where is the response variable describing the change in biomass with tree age (t), a is the asymptote of the maximum parameter for tree growth, b is the growth rate parameter that indicates the rate a tree approaches its asymptotic biomass, c is the parameter related to m, m is the power exponent of anabolism, η is the anabolism constant, and β is the catabolism constant. parameter a in equation 3 is the most unstable parameter; thus, a transformation of the equation is constructed to solve this problem by using the parameter a, which is an expected-value parameter when t = t0 to replace parameter a (fang & bailey 2001). therefore, we finally selected the transformation of richards’ equation to construct a basic biomass growth model. the equation is shown in equation 7. (7) where y is the organ biomass of an individual tree, a is the progressive parameter of organ biomass growth, b is the rate of growth, c is the curve shape parameter, t is the tree age, and t0 is a fixed reference age that may be fixed at any positive value according to the specific research situation (fang & bailey 2001). its value was set at 20 years in this study. mixed-effect model without fixed effects from competition factors (mem) forestry growth and yield data are affected by the sampling region (e.g. different regions may have different site conditions) and the correlation among the trees in the same sampling position. the data from within a sampling unit is dependent; thus, autocorrelation and heterogeneity are common in these data (gregorie 1987). the mixed-effect modelling technique can partly remove the negative impact of heterogeneity and autocorrelation within-plots by using variance (e.g. power, exponential function) and covariance (e.g. timeautocorrelation function) structure. this technique can also explain the plot parameter variability by selecting appropriate covariates (fang & bailey 2001). in our study, site quality was taken as a random effect to select the mixed parameters, and power and exponential functions were used for describing the variance structures. three time-autocorrelation functions, including the autoregressive time correlation structure with order 1 (ar(1)), continuous time ar(1) structure (car(1)) and autoregressive moving-average structure with both order 1 (arma(1,1)) function, were used to describe covariance structures. the mixed parameter selection to determinate the model forms, variance and covariance structure were all given by the research of pinheiro & bates (2000). mixed-effect model with fixed effects from competition factors (memc) based on the mem, competition factors were considered fixed effects to input the equation parameters of the mem, and both ci and the quadratic effect (ci2) are incorporated into the models. model evaluation in this study, log likelihood (loglik), akaike information criterion (aic) and bayesian information criterion (bic) were used to evaluate the model fitting results (equations 8–10). nong et al. new zealand journal of forestry science (2019) 49:11 page 5 (8) (9) (10) where is the maximum likelihood estimation of θ for the likelihood function of model , x is the random sample, q is the number of unknown parameters, and n is the sample capacity. moreover, the sum relative error (rs), mean relative error (ee), absolute mean relative error (rma) and prediction precision (p) were used to test the model prediction performance (equations 11–14). (11) (12) (13) (14) where yi is the measured value, is the estimated value, is the mean value of the estimated value, ta is the distribution value of t (when confidence level is a = 0.05), n is the sample capacity, and t is the parameter number of the regression curve equation. results mixed-effect model without fixed effects from competition factors (mem) mixed parameter selection the fitting results of different parameter combinations are listed in table 2. models with mixed parameters have lower aic and bic and higher loglik values than the basic model, referred to as the transformation of richards’ equation (bm). the optimal fitting result emerges when the parameter a is regarded as the mixed parameter (aic=1305.367, bic=1318.189, loglik= -647.684). the mixed-effect model is listed in equation 15. (15) where y is the above-ground biomass, a is the progressive parameter of organ biomass growth, ua is the parameter for the random effect from the site index, b is the growth rate of agb, t is the tree age, t0 is the standard age with a value of 20 years, and c is the shape parameter. variance and covariance structure selection a comparison of the variance and covariance structure fitting results of the mem are listed in table 3. the optimal fitting result emerged when the power function is regarded as a variance structure but none are regarded as covariance structures (aic=1224.4, bic=1239.8, loglik=-606.2). mixed-effect model with fixed effects from competition factors (memc) basic model construction based on the mem, when parameter a is regarded as a random effect and the basic individual tree competition index is regarded as a fixed effect, the optimal model form is listed in equation 16. comparisons of different effects in the agb growth mixed-effect model are listed in table 4. the optimal fitting results emerge in the mixed-effect model with competition factors and site index (memc) (aic=1298.1, loglik=-640.1). the fitting results of parameters are listed in table 5. parameter b is extremely significant at the p=0.01 level. (16) where y is the above-ground biomass, ci is the competition index, ci2 is the quadratic effect of ci, a is the progressive parameter of organ biomass growth, ua is the parameter for random effects from the site index, b is the growth rate of agb, b1 and b2 are the estimated parameters of the fixed effect from and to parameter b, respectively, t is the tree age, t0 is the standard age with a value of 20 years, c is the shape parameter, and c1 and c2 are the estimated parameters of the fixed effect from ci and ci2 to parameter c, respectively. variance and covariance structure selection the fitting results of the variance and covariance structures are listed in table 6. the optimal fitting result emerged when the power function is regarded as a variance structure but none are regarded as covariance structures (aic=1230.2, bic=1250.7, loglik=-607.1). the optimal fitting results of the mixed-effect model taking the competition factor as a fixed effect are listed in table 7, and the optimal model form is shown in equation 17. (17) nong et al. new zealand journal of forestry science (2019) 49:11 page 6 mixed parameter loglik aic bic lrt p-value no -667.0 1342.0 1352.3 a -647.7 1305.4 1318.2 38.7 <0.0001 b -652.6 1315.1 1327.9 28.9 <0.0001 c no convergence a, b -647.7 1307.4 1322.8 38.7 <0.0001 a, c -647.7 1307.4 1322.8 38.7 <0.0001 b, c no convergence a, b, c -647.7 1309.4 1327.3 38.7 <0.0001 table 2. mixed parameter selection of the mixed-effects model for agb growth where y is the above-ground biomass, a is the progressive parameter of organ biomass growth, ua is the parameter for random effects, is the growth rate of agb, t is the tree age, t0 is the standard age with a value of 20 years, c is the shape parameter, c1 is a mixed parameter of parameter c, c2 is a mixed parameter of parameter c, ci is the basic competition index, and ci2 is the quadratic effect of ci. nong et al. new zealand journal of forestry science (2019) 49:11 page 7 model evaluation the final fitting results of bm, mem and memc are listed in table 8. the parameter a did not differ significantly between the three models due to overlapping intervals of the estimated value, although bm has the lowest value with 61.799, and the parameters b and c have significant differences among the three models. mem has the no. variance structure covariance structure loglik aic bic lrt p-value 1 no no -647.7 1305.4 1318.2 2 power no -606.3 1224.4 1239.8 92.839 <0.0001 3 exponential no -617.8 1247.5 1262.9 59.889 <0.0001 4 no ar(1) -647.7 1307.3 1322.7 0.099 0.7532 5 no car(1) -647.7 1307.7 1322.8 0.015 0.9015 6 no arma(1,1) -647.7 1305.7 1322.7 0.095 0.9777 table 3. mixed-effects models considering variance and covariance structures for agb growth table 4. comparison of different effects on the agb growth mixed-effects model model loglik aic bic lrt p-value mixed-effects model with competition factors and regional effect -640.1 1298.1 1321.2 mixed-effects model with regional effect -647.7 1305.4 1318.2 25.0 0.0001 table 5. estimated parameters of the agb growth mixed-effects model incorporating competition factors as fixed effects parameter estimated value standard deviation df t-value p-value a 62.4101 34.2398 87 1.823 0.0718 b 0.0930 0.0225 87 4.128 0.0001 b1 -0.0015 0.0008 87 -1.812 0.0734 b2 0.00003 0.00001 87 1.956 0.0537 c 17.2559 10.4085 87 1.658 0.1009 c1 -0.5845 0.3855 87 -1.516 0.1331 c2 0.0088 0.0061 87 1.443 0.1526 table 6. mixed-effects models incorporating competition factors as fixed effects considering variance and covariance structures for agb growth (* the model with no significant parameters b1 and b2.) no. variance structure covariance structure loglik aic bic lrt p-value 1 no no -640.1 1298.1 1321.2 2 power no -608.1 1236.2 1261.9 63.9 <0.0001 3 exponential no -616.2 1252.5 1278.1 47.7 <0.0001 4 no ar(1) -639.4 1298.8 1324.4 1.4 0.2394 5 no car(1) -639.4 1298.8 1324.4 1.4 0.2394 6 no arma(1,1) -639.6 1299.2 1324.8 1.0 0.3209 7* power no -607.1 1230.2 1250.7 66.0 <0.0001 optimal fitting performance because of the lower values of aic and bic and the highest value of loglik (table 8), but the difference between the mem and memc is not significant because of the higher p value (p value = 0.113) according to the likelihood ratio test (lrt). while memc has the minimum value of ee (6.45%) and rma (54.34%), and the highest value of prediction precision table 2: confusion matrix p (86.08%), bm has the minimum value of rs (7.71%) (table 8). moreover, the heteroscedasticity of the residual is not found in either mixed-effect models, but it is obvious in the bm, and the memc has the narrowest interval of standardised residual (fig. 4). therefore, the memc is the optimal model for the individual tree agb growth of simao pine. nong et al. new zealand journal of forestry science (2019) 49:11 page 8 table 7. fitting results of the three agb growth models. estimation parameter bm mem memc estimated value p-value estimated value p-value estimated value p-value a 61.7990±14.5400 <0.0001 76.3382±19.2527 0.0001 76.9496±11.6512 <0.0001 b 0.0516±0.0200 0.0346 0.0111±0.0115 0.3389 0.0656±0.0152 <0.0001 c 6.3479±1.0760 0.0013 2.9552±0.6778 <0.0001 9.1919±2.9250 0.0023 c1 -0.1017±0.0341 0.0037 c2 0.0003±0.0001 0.0033 loglik -667.0 -606.2 -607.1 aic 1342.0 1224.4 1230.2 bic 1352.3 1239.8 1250.7 d-matrix d=[28.1220] d=[2.9887×10-9] heteroscedastic function value residual error 1.4507 1.7662 2.3904 table 8. comparison of the three agb growth models. model fitting index testing index loglik aic bic lrt p-value rs(%) ee (%) rma (%) p (%) bm -667.0 1342.0 1352.3 7.71 9.82 61.93 85.81 mem -606.2 1224.4 1239.8 121.6 <0.0001 -62.08 -37.46 55.16 79.85 memc -607.1 1230.2 1250.7 119.9 <0.0001 -34.26 6.45 54.34 86.08 figure 4: scatter of standardised residual vs predicted agb for three models. a: basic model (bm); b: mixed effect model without fixed effects from competition factors (mem); and c: mixed effect model with fixed effects from competition factors (memc). discussion the different combinations of mixed parameters were selected, and the mixed model with mixed parameter a was considered as the optimal basic model because of the low aic (table 2). parameter a represents the estimated asymptotic biomass at a standard age; however, trees have different growth potentials at the same standard age depending on the site quality. many factors would influence tree maximum diameter at an equal basal age, such as water-holding capacity, elevation, and slope (rohner et al. 2013). thus, selecting parameter a as a mixed parameter in the model can improve the estimation and indicates that the site quality has a significant effect on the age-biomass relationship. site quality had important effects on tree growth (robichaud & methven 1993). past studies generally used site characteristics (a plot-level variable) as variables for biomass growth models and showed that site characteristics had a significant effect on tree growth (lee et al. 2004). westfall (2016) reported that mixed models, including plot random effects, can reduce prediction bias and variance for populations to a great extent compared to fixed effect models; lhotka & loewenstein (2011) reported analogous results. similarly, our study found that mixed-effect models, including the random effects of site quality, are better than basic models (i.e. higher loglik and lower aic and bic), and incorporating random effects can improve the biomass growth model for simao pine. our results are consistent with previous studies for both static models (subedi & sharma 2011; rohner et al. 2013; ou et al. 2016; chen et al. 2017; huff et al. 2018), and growth models (budhathoki et al. 2008; timilsina & staudhammer 2013; westfall 2016). moreover, the competition factor is an important predictor variable for the individual tree model because it intensively affects tree growth (lhotka & loewenstein 2011). in our study, memc had a better estimation due to incorporating the competition factor as a fixed effect compared with mem, and the predictive ability was significantly improved (i.e. largest p and smallest ee and rma). specifically, the predictive accuracy increased to 86%. thus, it can obviously improve the mixed model predictive accuracy if competition is regarded as an independent variable, and it also indicates that tree competition is the critical element for predicting individual tree biomass growth. therefore, mixed-effect models have been used in forestry because of their superior fitting and prediction accuracy compared with traditional models (huff et al. 2018). the mixed-effect model, only including the random effects of site quality, can greatly improve the fit performance of the model, and the mixed-effect model incorporating competition factors can further improve the prediction ability. in addition, autocorrelation among measurement data may result in biased estimates of the model parameter for biological data (budhathoki et al. 2010). the mixed-effect model was then introduced to address this challenge by defining the variance and covariance structures of random effects in parameter estimations (west et al. 2007; smith et al. 2014; demiguel et al. 2014; njana et al. 2016). we also found that our prediction model had issues with heteroscedasticity which was corrected by using the power variance function (the optimal variance function) in the estimation process (budhathoki et al. 2008). therefore, the mixed-effect model had a good fitting performance which is consistent with previous studies (subedi & sharma 2011; de-miguel et al. 2014). in contrast, all of the time-autocorrelation covariance structures in our study did not improve the model performance which indicates that the estimation models considering time autocorrelation cannot improve fitting. this might be attributable to the biomass sample data without a time series (eisfelder et al. 2017). however, it is unrealistic and impossible to obtain the time series data of the biomass because destructive sampling was performed to collect biomass data (temesgen et al. 2015). therefore, we investigated sample trees with different ages (from 8 years to 80 years) in different locations to replace the time series data. the agb change rule along with the tree ages can be explained by using the mixed-effect model considering site quality and competition factors. thus, the investigated methods for estimating the agb growth would be reasonably practical. additionally, we did not consider spatial autocorrelation among trees in the same plot because simao pine is an intolerant tree species and because sampling trees of different ages occurs at distant locations. furthermore, the data characteristics of the independent variables are important for selecting the form of mixed effects. a random effect is applicable if the variable is a grouped one, but a fixed effect is appropriate if the data are a continuous variable in a mixed-effect model (pinheiro & bates 2000). in this study, the site index and competition index were incorporated into the mixed-effect models as random and fixed effects, respectively. the site index (si) is often used as grouped data in forestry. si tables of the dominant tree species are established to predict potential productivity of forestland; si is based on the dominant height classes with a 2-m interval at a standard stand age (meng 2006; duan et al. 2009 ). thus, it was considered as a random effect into the mixed-effect model in this study. the competition index is a continuous variable, and it was calculated using the distances between each sample tree and its neighbouring competing trees and their dbh in this study (hegyi 1974). thus it was incorporated into the mixed-effect model as a fixed effect. in addition, the national continuous forest inventory of china (ncfi) is being conducted using permanent plots and carried out every five years to reflect the dynamic change of forest resources at the national scale since 1987 (kang 2011). fang et al. (2001, 2002) calculated forest biomass carbon using the ncfi database by the variable biomass expansion factor method, with estimation error of less than 3% at the regional and national scale. at the stand level, the location of each tree in each permanent plot and the height of the stands had been recorded (kang 2011). therefore, the ci of each tree can be calculated and the si of the plots can be found by looking up the si tables of the different dominant tree species. so the comprehensive estimation method has the potential to improve biomass growth estimation and nong et al. new zealand journal of forestry science (2019) 49:11 page 9 reflect the dynamical change of stand biomass. conclusions to improve estimation of individual tree agb growth, a nonlinear mixed-effect model was developed by incorporating random effects of site quality and fixed effects of competition factors based on a transformation of richards’ function. we found that the mixed-effect model was significantly better than the bm because of its better fitting and prediction performance, and the memc is the optimal estimation model due to its highest prediction precision. therefore, comprehensive biomass growth estimation considering the site index and competition index can be used to predict the agb growth of simao pine trees, and it is a potential method for agb growth estimation of other tree species. list of abbreviations agb: above-ground biomass; bm: basic model; mem: mixed-effect model without fixed effect from competition factors; memc: mixed-effect model with fixed effect from competition factors; dbh: diameter at breast height; h: tree height; ci: competition index; si: site index. competing interests the authors declare that they have no competing interests. authors’ contributions go conceived the study, analyzed the data, and wrote the manuscript. mn and ly participated in field work and analyzed the data, cl participated in field work, and hx reviewed the manuscript. all authors read and approved the final manuscript. acknowledgements we acknowledge the support from the forestry bureau of mojiang county, simao district and lancang county for assistance with field work, and the key laboratory of state forestry administration on biodiversity conservation in southwest china (southwest forestry university) for assistance with measurements of plant samples at the laboratory. we thank enliang li, zhigang liang and junfeng wang for assistance with the field work. wenjun wu and mingquan huang edited the manuscript, and guangxing wang (usa) reviewed and provided valuable comments. additionally, thanks to the anonymous external reviewers for their valuable comments. funding this study was supported by funding from the national natural science foundation of china (31560209, 31760206, 31660202), and the scientific research foundation of southwest forestry university (111416), and the ten-thousand talents program of yunnan province, china (ynwr-qnbj-2018-184). ethics approval and consent to participate not applicable consent for publication not applicable availability of data and materials please contact author for data requests references bragg, d.c. (2001). a local basal area adjustment for crown width prediction. northern journal of applied forestry, 18(1), 22-28. budhathoki, c.b., lynch, t.b., & guldin, j.m. (2008). nonlinear mixed modeling of basal area growth for shortleaf pine. forest ecology and management, 255(8-9), 3440-3446. https://doi.org/10.1016/j. foreco.2008.02.035 budhathoki, c.b., lynch, t.b., & guldin, j.m. (2010). development of a shortleaf pine individual-tree growth equation using non-linear mixed modeling techniques. in: stanturf, john a. (ed.), proceedings of the 14th biennial southern silviculture research conference; 2007 february 26-march 1; athens, ga. [general technical report srs-121]. asheville, nc, usa: southern research station, usda forest service, pp. 519-520. calama, r., & montero, g. (2004). interregional nonlinear height–diameter model with random coefficients for stone pine in spain. canadian journal of forest research, 34(1), 150-163. https://doi. org/10.1139/x03-199 carmean, w.h., hahn, j.t., & jacobs, r.d. (1989). site index curves for forest tree species in the eastern united states. [general technical report nc-128]. st. paul, mi, usa: north central forest experiment station, usda forest service, 142 p. chave, j., réjou-méchain, m., búrquez, a., chidumayo, a., colgan, m.s., delitti, w.b.c., duque, a., eid, t., fearnside, p.m., goodman, r.c., henry, m., martinez-yrizar, a., mugasha, w.a., muller-landau, h.c., mencuccini, m., nelson, b.w., ngomanda, a., nogueira, e.m., ortiz-malavassi, e., pelissier, r., ploton, p., ryan, c.m., saldarriaga, j.g., & vieilledent, g. (2014). improved allometric models to estimate the aboveground biomass of tropical trees. global change biology, 20(10), 3177-3190. https://doi. org/10.1111/gcb.12629 chen, d., huang, x., zhang, s., & sun, x. (2017). biomass modeling of larch (larix spp.) plantations in china based on the mixed model, dummy variable model, and bayesian hierarchical model. forests, 8(8), 268. https://doi.org/10.3390/f8080268 chojnacky, d.c. (2003). allometric scaling theory applied to fia biomass estimation. in: mcroberts, ronald e.; reams, gregory a.; van deusen, paul c.; moser, john w. (eds.), proceedings of the third annual forest inventory and analysis symposium; 2001 october 17-19; traverse city, mi. [general technical nong et al. new zealand journal of forestry science (2019) 49:11 page 10 report nc-230]. st. paul, mn: usda forest service, north central research station. 208 p. https://doi. org/10.2737/nc-gtr-230 coates, k.d., canham, c.d., & lepage, p.t. (2004). a neighborhood analysis of canopy tree competition: effects of shading versus crowding. canadian journal of forest research, 34(4), 778-787. https:// doi.org/10.1139/x03-232 coates, k.d., canham, c.d., & lepage, p.t. (2009). above versus below-ground competitive effects and responses of a guild of temperate tree species. journal of ecology, 97(1), 118-130. https://doi. org/10.1111/j.1365-2745.2008.01458.x compilation committee of yunnan forest. (1986). yunnan forest. beijing: china forestry publishing house. coomes, d.a., & allen, r.b. (2007). effects of size, competition and altitude on tree growth. journal of ecology, 95(5), 1084-1097. https://doi. org/10.1111/j.1365-2745.2007.01280.x cosmo, l.d., gasparini, p., & tabacchi, g. (2016). a national-scale, stand-level model to predict total above-ground tree biomass from growing stock volume. forest ecology and management, 361, 269-276. https://doi.org/10.1016/j. foreco.2015.11.008 curtis, r.o. (1985). stand density measures: an interpretation. forest science, 16, 403-414. de-miguel, s., pukkala, t., assaf, n., & shater, z. (2014). intra-specific differences in allometric equations for aboveground biomass of eastern mediterranean pinus brutia. annals of forest science, 71(1), 101112. https://doi.org/10.1007/s13595-013-03344 di cosmo, l., gasparini, p., & tabacchi, g. (2016). a national-scale, stand-level model to predict total above-ground tree biomass from growing stock volume. forest ecology and management, 361, 269-276. https://doi.org/10.1016/j. foreco.2015.11.008 duan, j., ma, l., jia, l., hou, y., & gong, n. (2009). establishment and application of site index table for pinus tabulaeformis plantation in the low elevation area of beijing. scientia silvae sinicae, 45(3), 7-12. eisfelder, c., klein, i., bekkuliyeva, a., kuenzer, c., buchroithner, m.f., & dech, s. (2017). aboveground biomass estimation based on npp timeseries? a novel approach for biomass estimation in semi-arid kazakhstan. ecological indicators, 72, 13-22. fang, j., chen, a., peng, c., zhao, s., & ci, l. (2001). changes in forest biomass carbon storage in china between 1949 and 1998. science, 292, 2320-2322. fang, j., chen, a., zhao, s., & ci, l. (2002). estimating biomass carbon of china’s forests: supplementary notes on report published in science (291: 23202322) by fang et al. (2001). acta phytoecologica sinica, 26(2), 243-249. fang, z., & bailey, r. l. (2001). nonlinear mixed effects modeling for slash pine dominant height growth following intensive silvicultural treatments. forest science, 47(3), 287-300. fehrmann, l., lehtonen, a., kleinn, c., & tomppo, e. (2008). comparison of linear and mixed-effect regression models and a k-nearest neighbour approach for estimation of single-tree biomass. canadian journal of forest research, 38(1), 1-9. https://doi.org/10.1139/x07-119 fu, l.y., zeng, w.s., tang, s.z., sharma, r.p., & li, h.k. (2012). using linear mixed model and dummy variable model approaches to construct compatible single-tree biomass equations at different scales a case study for masson pine in southern china. journal of forest science, 58(3), 101-115. goodale, c.l., apps, m.j., birdsey, r.a., field, c.b., heath, l.s., houghton, r.a., jenkins, j.c., kohlmaier, g.h., kurz, w., liu, s., nabuurs, g.j., nilsson, s., & shvidenko, a.z. (2002). forest carbon sinks in the northern hemisphere. ecological applications, 12(3), 891–899. gregorie, t.g. (1987). generalized error structure for forestry yield models. forest science, 33(2), 423444. gregoire, t.g., schabenberger, o., & barrett, j.p. (1995). linear modelling of irregularly spaced, unbalanced, longitudinal data from permanentplot measurements. canadian journal of forest research, 25(1), 137-156. hegyi, f. (1974). a simulation model for managing jackpine stands. in: fries, j. (ed.), growth models for tree and stand simulation. stockholm, sweden: royal college of forestry, pp. 74-85. henry, m., picard, n., trotta, c., raphaël j.m., valentini, r., bernoux, m., & saint-andré, l. (2011). estimating tree biomass of sub-saharan african forests: a review of available allometric equations. silva fennica, 45(3b), 477-569. houghton, r.a. (2005). aboveground forest biomass and the global carbon balance. global change biology, 11(6): 945–958. https://doi.org/10.1111/j.13652486.2005.00955.x houghton, r.a., hall, f., & goetz, s.j. (2009). importance of biomass in the global carbon cycle. journal of geophysical research, 116(g2). https://doi. org/10.1029/2009jg000935 huff, s., poudel, k.p., ritchie, m., & temesgen, h. (2018). quantifying aboveground biomass for common shrubs in northeastern california using nonlinear mixed effect models. forest ecology and management, 424, 154-163. https://doi. org/10.1016/j.foreco.2018.04.043 jagodziński, a.m., dyderski, m.k., gęsikiewicz, k., horodecki, p., cysewska, a., wierczyńska, s., & maciejzyk, k. (2018). how do tree stand parameters affect young scots pine biomass? – allometric equations and biomass conversion and expansion factors. forest ecology and management, 409, 7483. https://doi.org/10.1016/j.foreco.2017.11.001 jenkins, j.c., chojnacky, d.c., heath, l.s., & birdsey, r.a. (2003). national-scale biomass estimators for united states tree species. forest science, 49(1): 12-35. nong et al. new zealand journal of forestry science (2019) 49:11 page 11 jenkins, j.c., chojnacky, d.c., heath, l.s., & birdsey, r.a. (2004). comprehensive database of diameter-base biomass regressions for north american tree species [general technical report ne-319]. newtown square, pa, usa: usda forest service, northeastern research station, 45 p. kang, x.g. (2011). forest management, 4th edition. beijing: china forestry publishing house. lee, w.k., gadow, k.v., chung, d.j., lee, j.l., & shin, m.y. (2004). dbh growth model for pinus densiflora and quercus variabilis mixed forests in central korea. ecological modelling, 176(1-2),187-200. https:// doi.org/10.1016/j.ecolmodel.2003.11.012 lhotka, j.m., & loewenstein, e.f. (2011). an individualtree diameter growth model for managed uneven-aged oak-shortleaf pine stands in the ozark highlands of missouri, usa. forest ecology and management, 261(3),770-778. https://doi. org/10.1016/j.foreco.2010.12.008 li, c.m.,& zhang, h.r. (2010). modeling dominant height for chinese fir plantation using a nonlinear mixedeffects modeling approach. scientia silvae sinicae, 46(3), 89-95. li, c.m. (2012). the simultaneous equation system of total volume in fir plantation. scientia silvae sinicae, 48(6), 80-88. li, j. (2011). dynamics of biomass and carbon stock for young and middle aged plantation of simao pine (pinus kesiya var. langbianensis). beijing: beijing forestry university. li, y.x., & jiang, l.c. (2013). modeling wood density with two-level linear mixed effects models for dahurian larch. scientia silvae sinicae, 49(7), 91-98. https:// doi.org/10.11707/j.1001-7488.20130713 liu, z.g., & li, f.r. (2003). the generalized chapmanrichards function and applications to tree and stand growth. journal of forestry research, 14(1) 19-26. https://doi.org/10.1007/bf02856757 luo, x., he, h.s., liang, y., wang, w.j., wu, z., & fraser, j.s. (2014). spatial simulation of the effect of fire and harvest on aboveground tree biomass in boreal forests of northeast china. landscape ecology, 29(7),1187-1200. https://doi.org/10.1007/ s10980-014-0051-x mctague, j.p., & weiskittel, a.r. (2016). individual-tree competition indices and improved compatibility with stand-level estimates of stem density and long-term production. forests, 7(10), 238. https:// doi.org/10.3390/f 7100238 meng, x.y. (2006). forest mensuration (3rd ed). beijing: china forestry publishing house. muukkonen, p. (2007). forest inventory-based largescale forest biomass and carbon budget assessment: new enhanced methods and use of remote sensing for verification. helsinki, finland: department of geography, university of helsinki. njana, m.a., bollandsås, o.m., eid, t., zahabu, e., & malimbwi, r.e. (2016). aboveand belowground tree biomass models for three mangrove species in tanzania: a nonlinear mixed effects modelling approach. annals of forest science, 73(2),353-369. https://doi.org/10.1007/s13595-015-0524-3 ou, g.l., wang, j.f., xu, h., chen, k.y., zheng, h.m., zhang, b., sun, x.l., xu, t.t., & xiao, y.f. (2016). incorporating topographic factors in nonlinear mixed-effects models for aboveground biomass of natural simao pine in yunnan, china. journal of forestry research, 27(1), 119-131. https://doi.org/10.1007/s11676015-0143-8 pinheiro, j.c., & bates, d.m. (2000). mixed effects models in s and s-plus. new york: springer verlag. ramon, c.l., george, a.m., walter, w.s., russell, d.w., & oliver, s. (2006). sas for mixed models, 2nd ed. cary, nc, usa: sas institute inc. razali, w. w., razak, t. a., azani, a. m., & kamziah, a.k. (2015). mixed-effects models for predicting early height growth of forest trees planted in sarawak, malaysia. journal of tropical forest science, 27(2), 267-276. https://doi.org/10.2307/43582392 richards, f.j. (1959). a flexible growth function for empirical use. journal of experimental botany, 10(29), 290-300. https://doi.org/10.1093/ jxb/10.2.290 robichaud, e., & methven, i.r. (1993). the effect of site quality on the timing of stand breakup, tree longevity, and the maximum attainable height of black spruce. canadian journal of forest research, 23(8), 1514-1519. https://doi.org/10.1139/x93191 rohner, b., bugmann, h., & bigler, c. (2013). estimating the age-diameter relationship of oak species in switzerland using nonlinear mixed-effects models. european journal of forest research, 132(5-6),751764. https://doi.org/10.1007/s10342-013-07105 running, s.w.,& mcleod, s.d. (1988). comparing site quality indices and productivity in ponderosa pine stands of western montana. canadian journal of forest research, 18(3), 346-352. https://doi. org/10.1139/x88-052 sahney, s., benton, m.j., & ferry, p.a. (2010). links between global taxonomic diversity, ecological diversity and the expansion of vertebrates on land. biology letters, 6(4), 544-547. https://doi. org/10.1098/rsbl.2009.1024 saint-andre, l., m’bou, a.t., mabiala, a., mouvondy, w., jourdan, c., roupsard, o., deleporte, p., hamel, o., & nouvellon, y. (2005). age-related equations for aboveand below-ground biomass of a eucalyptus hybrid in congo. forest ecology and management, 205(1-3), 199-214. https://doi.org/10.1016/j. foreco.2004.10.006 sleeter, b.m., liu, j., daniel, c., rayfield, b., & loveland, t.r. (2018). effects of contemporary land-use and land-cover change on the carbon balance of terrestrial ecosystems in the united states. environmental research letters, 13(4). https://doi. org/10.1088/1748-9326/aab540 sileshi, g.w. (2014). a critical review of forest biomass estimation models, common mistakes and corrective measures. forest ecology and management, 329, 237-254. https://doi. nong et al. new zealand journal of forestry science (2019) 49:11 page 12 org/10.1016/j.foreco.2014.06.026 smith, a., granhus, a., astrup, r., bollandsås, o.m., & petersson, h. (2014). functions for estimating aboveground biomass of birch in norway. scandinavian journal of forest research, 29(6), 565-578. https://doi.org/10.1080/02827581.201 4.951389 somogyi, z., cienciala, e., mäkipää, r., muukkonen, p., lehtonen, a., & weiss, p. (2007). indirect methods of large-scale forest biomass estimation. european journal of forest research, 126(2), 197-207. https://doi.org/10.1007/s10342-006-0125-7 southwest forestry college, & forestry department of yunnan province. (1988). iconographia arbororum yunnanicorum. kunming, china: yunnan science and technology press. sturtevant, b.r., & seagle, s.w. (2004). comparing estimates of forest site quality in old second-growth oak forests. forest ecology and management, 191(1-3), 0-328. subedi, n., & sharma, m. (2011). individual-tree diameter growth models for black spruce and jack pine plantations in northern ontario. forest ecology and management, 261(11), 2140-2148. https:// doi.org/10.1016/j.foreco.2011.03.010 talhelm, a.f., pregitzer, k.s., kubiske, m.e., zak, d.r., campany, c.e., burton, a.j., dickson, r.e., hendrey, g.r., isebrands, j.g., lewin, k.f., nagy, j, & karnosky, d.f. (2014). elevated carbon dioxide and ozone alter productivity and ecosystem carbon content in northern temperate forests. global change biology, 20(8), 2492-2504. temesgen, h., affleck, d., poudel, k., gray, a., & sessions, j. (2015). a review of the challenges and opportunities in estimating above ground forest biomass using tree-level models. scandinavian journal of forest research, 30(4), 326-335. https:// doi.org/10.1080/02827581.2015.1012114 ter-mikaelian, m.t., & korzukhin, m.d. (1997). biomass equations for sixty-five north american tree species. forest ecology and management, 97(1), 1–24. https://doi.org/10.1016/s03781127(97)00019-4 timilsina, n., & staudhammer, c.l. (2013). individual tree-based diameter growth model of slash pine in florida using nonlinear mixed modeling. forest science, 59(1), 27-37. https://doi.org/10.5849/ forsci.10-028 vanninen, p., ylitalo, h., sievänen, r., & mäkelä, a. (1996). effects of age and site quality on the distribution of biomass in scots pine (pinus sylvestris l.). trees, 10(4): 231-238. https://doi.org/10.1007/ bf02185674 wang, h.l. (2003). study on stand growth models for pinus kesiya var. langbianensis natural secondary woodland. kunming, china:southwest forestry college. waring, r.h., milner, k.s., jolly, w.m., phillips, l., & mcwethy, d. (2006). assessment of site index and forest growth capacity across the pacific and inland northwest usa with a modis satellite-derived vegetation index. forest ecology and management, 228(1-3), 1-291. wen, q.z., zhao, y.f., chen, x.m., yang, z.x., ai, j.l., & yang, x.s. (2010). dynamic study on the values for ecological service function of pinus kesiya forest in china. forest research, 23(5), 671-677. west, b., welch, k.b., & galecki, a.t. (2007). linear mixed models: a practical guide using statistical software. boca raton fl, usa: chapman and hall/crc. westfall, j.a. (2016). strategies for the use of mixedeffects models in continuous forest inventories. environmental monitoring and assessment, 188(4), 1-11. https://doi.org/10.1007/s10661-016-52520 woodwell, g.m., whittaker, r.h., reiners, w.a., likens, g.e., delwiche, c.c., & botkin, d.b. (1978). the biota and the world carbon budget. science, 199(4325), 141-146. https://doi.org/10.1126/ science.199.4325.141 wu, z.l.,& dang, c.l. (1992). the biomass of pinus kesiya var. langbianensis stands in pu-er district, yunnan. journal of yunnan university: natural science edition, 14(2), 161-167. yue, f.,& yang, b. (2011). study on carbon sink of pinus kesiya forests. jiangsu agricultural sciences, 39(5), 467-469. zeide, b. (2005). how to measure stand density. trees,19(1), 1-14. zeng, w.s., zhang, h.r., & tang, s.z. (2011). modeling method on the standing trees. beijing: china forestry publishing house. zhang, y., & borders, b.e. (2004). using a system mixed-effects modeling method to estimate tree compartment biomass for intensively managed loblolly pines an allometric approach. forest ecology and management, 194, 145-157. https:// doi.org/10.1016/j.foreco.2004.02.012 zianis, d., muukkonen, p., mäkipää, r., & mencuccini, m. (2005). biomass and stem volume equations for tree species in europe. silva fennica monographs, 4(4). nong et al. new zealand journal of forestry science (2019) 49:11 page 13 optimisation of ultrasound assisted extraction of antiacetylcholinesterase and antioxidant compounds from manuka (leptospermum scoparium) for use as a phytomedicine against alzheimer’s disease hafiz majid1,2 and filipa v.m. silva1,3* 1chemical and materials engineering department, university of auckland, private bag 92019, auckland 1142, new zealand 2faculty of science and natural resources, universiti malaysia sabah, 88400 kota kinabalu, sabah, malaysia 3leaf, instituto superior de agronomia, universidade de lisboa, tapada da ajuda, 1349-017 lisboa, portugal *corresponding author: filipavinagresilva@gmail.com (received for publication 28 march 2020; accepted in revised form 5 november 2020) abstract background: alzheimer’s disease is a progressive mental deterioration related to ageing and senility. approved drugs that inhibit acetylcholinesterase (ache) enzyme activity in the human brain are one of the ways to control the natural progression of this disease. the present study reports on the optimisation of ultrasound-assisted extraction of antiacetylcholinesterase and antioxidant compounds from manuka leaves using response surface methodology. methods: a box-behnken design was used to investigate the effect of extraction temperature (40–60°c), time (1–20 min), and ethanol concentration (30–70%) on ache inhibition, antioxidant activity, and extraction yield. results: the values of ache, radical scavenging activity (rsa) and yield predicted by the models generated were similar to the experimental values. extraction time, ethanol concentration and temperature were significant in all the responses. optimum extraction conditions for maximum ache inhibition (74%), rsa (79%) and yield (50%) were successfully validated experimentally and the ic50 of the optimised extracts were reduced to 28.5 (from 66.0) and 2.37 (from 32.4) μg/ ml for ache and antioxidant activity, respectively. the optimisation enabled an increase in the extraction yield from 21% to 49%. conclusions: in view of the significant bioactive properties determined, with possible beneficial effects on memory deficit, we would encourage the use of the manuka leaf extract for the development of new phytopharmaceuticals to improve brain function and control dementias such as alzheimer’s disease. one other application could be as a beverage for the preparation of tea infusions. new zealand journal of forestry science majid & silva. new zealand journal of forestry science (2020) 50:12 https://doi.org/10.33494/nzjfs502020x103x e-issn: 1179-5395 published on-line: 25/11/2020 © the author(s). 2020 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access aged 65 to 74 years, and 47.2% for elders above 85 years old (taylor et al. 2002; adewusi & steenkamp 2011; huang et al. 2010). the most developed pharmacological approach for the symptomatic treatment of ad is the inhibition of acetylcholinesterase (ache) (mukherjee et al. 2007). based on the cholinergic hypothesis, the aim of this treatment is to increase the level of acetylcholine introduction with an ageing population mental disability or dementia are increasing problems among the elderly affecting more than 20 million people worldwide. alzheimer disease (ad) is the most prevalent dementia affecting more than 20 million people worldwide. the disease is highly associated with age, with 3% of affected people keywords: leptospermum scoparium; plant extracts; yield; acetylcholinesterase; enzyme inhibition majid & silva new zealand journal of forestry science (2020) 50:12 page 2 (ach) in synaptic regions in order to restore cognitive functions (nordberg & svensson 1998). tacrine was the first approved drug, followed by various ache inhibitors including donepezil, galantamine and rivastigmine (schelterns & feldman 2003). in recent years, there has been significant interest in natural products in health relevant areas, particularly in developing anti ad agents from medicinal plants. because of this, a large number of plants and their crude extracts have been studied for their ache inhibition activity (syad & devi 2014). leptospermum scoparium, also known as manuka or ‘’tea tree”, grows as a shrub or small tree across new zealand (forster & forster 1776). it is about 2 m tall but occasionally reaches 4 m or more. there is already a great commercial interest and several plantations in new zealand for the production of the manuka honey, known worldwide not only for its flavour, but also for its medicinal properties. early reports show that māori used the bark, leaves, seeds, and sap of manuka for food, medicine and timber (porter & wilkins 1999). in medicinal use, different preparations of the leaves have been taken orally, directly chewed, applied as salve, or inhaled to treat cold, dysentery, as vapor baths, and to ease internal and external pains (brooker et al. 1987; stephens et al. 2005). in new zealand, there is growing commercial interest on manuka products, especially in the essential oils distilled from the leaves part. beneficial properties of the oils such as potent antimicrobial activity, including antifungal, antioxidant, insecticidal, and anthelminthic activities were demonstrated (lis-balchin et al. 2000; porter & wilkins 1999). interestingly, previous studies in this area revealed that the crude extracts from leptospermum scoparium leaf inhibit ache activity (majid & mathew 2015, majid & silva 2020a). the chemical composition study from dichloromethane extract of manuka leaf showed traces of flavonoids and triterpenoids compounds, which may be the factor attributed to the inhibition activity (häberlein & tschiersch 1994; mayer 1990). it is known that some class of triterpenoids and flavonoids have significant anticholinesterase activity, and are regarded as promising candidates to be used as cholinesterase inhibitors in clinical practice (ahmed et al. 2013). ultrasound or ultrasonic assisted extraction (uae) has been known to increase the extraction efficiency of bioactives from a wide range of plant materials (hossain et al. 2012). this extraction technique has been used to extract substances such as terpenoids, polysaccharides, polyphenols, flavonoids, and other class of compounds from different parts of the plant (pan et al. 2012; wang et al. 2008; cacace & mazza 2003). the advantages of uae are reduced extraction time and solvent use, and higher extraction yield (majid & silva, 2020b). ultrasound is more efficient than a conventional extraction method as it allows more effective mixing and faster mass/energy transfer (silva et al. 2007). through ultrasound technique, the production of plant extracts is enhanced by the acoustic cavitation that breaks the cells’ walls mechanically, greatly facilitating the mass transfer between the solid and liquid phase (wang et al. 2008). despite the simplicity of this technique, many factors such as solvent type, solvent concentration, extraction time, extraction temperature, and solvent to solid ratio need to be optimised to achieve maximum efficacy of the extraction (liyana-pathirana & shahidi 2005). response surface methodology (rsm) is used for developing, improving and optimising processes involving several variables. compared to the one factor at a time optimisation approach, which is more laborious and time consuming, rsm is able to reduce the number of experiment trials needed to study the interactions of multiple variables (majid & silva, 2020b). the rationale for the use of rsm is to determine the optimum conditions for the extraction and to generate a mathematical model, which relates ache inhibition with the extraction parameters. the main objectives of this study were to optimise the uae parameters namely the solvent concentration, the extraction time, and the temperature using rsm, aiming to maximise the extraction of antiache related compounds from manuka leaves and the yield of plant extract produced. in addition, the antioxidant activity of the extracts produced was also quantified for the same experimental design. models to predict ache inhibition, yield of extract production and radical scavenging activity (rsa) as a function of extraction parameters were generated and validated. the specific objectives were: 1. to select the best solvent and appropriate range of extraction parameters to extract antiacetylcholinesterase (ache) compounds from manuka leaves; 2. to study the effect of temperature, extraction time, and ethanol concentration on ache inhibition using response surface methodology; 3. to experimentally validate the optimum extraction conditions; 4. to determine the ic50 concentrations for ache inhibition and dpph rsa of the optimised extracts; and 5. to compare ic50 values of optimised extracts with non-optimised extracts. methods plant materials dried leptospermum scoparium leaves were purchased from a local herbal supplier “kiwiherbs ltd”, wellington, new zealand. the material was ground into fine particles using a grinder and sifted through a 0.25 mm sieve to obtain a uniform particle size. the material was then vacuum-sealed and stored in a dark condition at room temperature until use in the extraction experiments. ultrasound assisted extraction for all the extractions carried out, one gram of dried and ground manuka leaves were extracted using 50 ml of solvent in a beaker (fixed ratio of 20 mg plant/ ml solvent). the process of ultrasonic extraction was performed by using a 200w and 24 khz frequency ultrasonic processor equipped with a 3 mm diameter micro tip (up200s, hielscher ultrasonics gmbh, teltow, germany). this sonotrode was dipped about half way into the mixture contained in the beaker (plant leaves + solvent). the maximum acoustic energy intensity for this sonotrode is 460 w/cm2 according to the specifications of the manufacturer manual (hielscher 2007). as the area of 3 mm tip is 0.0707 cm2, a value of 32.5 w is obtained for the acoustic power at the maximum amplitude of 210 µm. the power/energy input was configured by setting the amplitude of the sonicator probe, as higher amplitude is proportional to higher energy/power. as 50 ml of mixture was sonicated, the maximum acoustic power density was 0.65 w/ml (which corresponds to 210 mm amplitude). prior to each experiment, the temperature of the solvent inside the beaker was stabilised at the experiment’s set temperature, then the dry manuka leaves were added and sonication time started. during the sonication, the beaker with the mixture was submerged in an ice water bath. the extraction mixture temperature was continuously monitored and kept approximately constant (±2°c) by adding ice water to the bath surrounding the beaker. preparation of plant extract and determination of extraction yield following the extraction, the samples of solvent containing the crude extract were filtered (whatman #1 filter paper) to obtain a clear extract solution and remove solid residues of the plant. then the clear solution was dried at 50°c under vacuum using a rotary evaporator (rotavapor r-215, büchi labortechnik ag, flawil, switzerland) to remove the solvent and obtain the crude extract. the extracts were then further air-dried in fume hood at room temperature until constant weight is obtained. the final weight of the dried crude extract was recorded for the determination of extraction yield, and then stored at -20°c for further analysis. the yield of extraction in percentage of extraction was calculated as follows: (1) overview of experiments the optimisation of ultrasound assisted extraction (uae) from manuka leaves was performed in two three stages. the first stage involved the selection of appropriate extraction solvent and the identification of variables with a significant effect on the extraction yield, ache inhibition and rsa. for this set of experiments, the effects were analysed by changing one factor at a time while keeping the other variables constant: 0.52 w/ml of acoustic power density, 50°c temperature, 10 min extraction time, 50% ethanol-water solvent. in the second stage of experiments, further optimisation of extraction conditions (section experimental design and response surface methodology) was carried out through response surface methodology (rsm) using a box-behnken experimental design to investigate the simultaneous effect of extraction temperature (40–60°c), extraction time (1–20 min) and ethanol concentration (30–70%) on plant extraction yield (%), plant extract ache inhibition (%) and antioxidant activity assessed by dpph radical scavenging activity (%). the ultrasound amplitude was set to 168 μm which corresponds to 0.52 w/ml of acoustic power density in all the experiments. lastly in the third stage of experiments optimised conditions predicted by the models for maximum inhibition of ache and rsa were tested experimentally and the experimental ic50 concentrations were determined for those conditions (section validation of model). the effect of different ultrasound extraction parameters on ache inhibition and rsa of manuka extracts solvent for extraction five different solvents (acetone, chloroform, ethyl acetate, methanol, and ethanol) were examined at 100% concentration. one gram of dried and ground manuka leaves were extracted in a beaker containing 50 ml of each solvent. the extraction was performed for 10 min at 50°c, using acoustic power density of 0.52 w/ml. the samples were filtered and freed of solvent by rotary evaporation. the crude extracts were stored then stored at -20°c prior to analysis. for this set of experiments two ic50 values of ache inhibitions and dpph rsa were determined for each solvent by carrying out triplicate tests for four concentrations of the dried extract and performing a non-linear regression (details shown in section determination of ic50 extract concentrations). ethanol concentration mixtures of ethanol-water were selected as extraction solvents. the concentration of ethanol-water was set at 0%, 30%, 50%, 80% and 100%. dried and ground manuka leaves (1 g) were extracted with the different concentrations of ethanol (50 ml). the extraction performed at fixed extraction conditions for 10 min at 50°c, with acoustic power density of 0.52 w/ml. the samples were filtered and freed of solvent by rotary evaporation. the crude extracts were stored then stored at -20°c prior to analysis. the inhibition of ache and rsa for an extract concentration of 50 μg/ml were determined in triplicates and average ± standard deviation presented. extraction temperature manuka leaves were extracted at different temperature of 30, 40, 50, 60, and 70°c. one gram of dried manuka leaves were sonicated with 50 ml 50% ethanol-water solvent. the extraction was achieved at 10 min using acoustic power density of 0.52 w/ml. the samples were filtered and evaporated to dryness by using a rotary evaporation. the crude extracts were stored at -20°c prior to analysis. the inhibition of ache and rsa for an extract concentration of 50 μg/ml were determined in triplicate and average ± standard deviation presented. extraction time extraction of manuka leaves were executed for a time ranging from 2 to 60 minutes. one gram of dried manuka leaves were extracted with 50 ml of ethanolwater (50%). the extraction was done at 50°c using majid & silva. new zealand journal of forestry science (2020) 50:12 page 3 acoustic power density of 0.52 w/ml. the samples were filtered and dried by rotary evaporation to obtain crude extracts. the crude extracts were stored at -20 °c prior to analysis. the inhibition of ache and rsa for an extract concentration of 50 μg/ml were determined in triplicates and average ± standard deviation presented. acoustic power density extraction sonication was performed at a setting of 0.13, 026, 0.39, 0.52, and 0.65 w/ml. dried manuka leaves (1 g) were extracted with 50% concentration of ethanol at different acoustic power density. the extraction was performed for 10 min at 50 °c. the samples were filtered, dried using rotary evaporation and crude extracts were then stored at -20°c prior to consequent analysis. the inhibition of ache and rsa for an extract concentration of 50 μg/ml were determined in triplicates and average ± standard deviation presented. response surface methodology to further investigate the combined effect of extraction temperature, time and ethanol concentration on ache inhibition, extraction yield and rsa: experimental design, optimisation and experimental validation of model predictions the extraction parameters were further optimised using rsm based on a three level, three variables box behnken design (bbd). temperature (t, °c), time (t, min), majid & silva. new zealand journal of forestry science (2020) 50:12 page 4 and ethanol concentration in water (e, %) were the independent variables optimised for the extraction of manuka leaves, while the dependent variables were the yield of extraction, and the ache inhibition (%) and dpph radical scavenging activity rsa (%) for a dried extract concentration of 50 μg/ml. the extraction temperature between 40 and 60°c, extraction time between 1 to 20 min, and ethanol concentration in water from 30 to 70% were investigated. the variables and their levels are coded at three levels, -1 (the lowest value), 0 (midpoint value) and +1 (highest value). the complete design with actual experimental parameters is presented in table 1. the design experiment includes 17 experimental points, including five replicates that were used for estimation of pure error sum of squares. each experimental condition was performed in triplicates and average values (± sd) were taken as response for the dependent variables. once the experiments were performed, all the response data were fitted a quadratic polynomial equation. the experimental results of the response surface were analysed using design-expert version 12 software (statease inc., minneapolis, mn, usa). analysis of variance (anova) was then carried out for each response to determine the statistical significance and suitability of the model. the significances of all terms were analysed by calculating the f-value and p-value (p<0.05), while the quality of the equation models was expressed by the adjusted coefficient of determination (adjusted r2). the run t, temp (°c) t, time (min) e, ethanol concentration (%) ache inhibition (%) dpph rsa (%) yield (%) 1 50 1 30 29.87 ± 7.00 50.99 ± 2.57 28.87 ± 0.89 2 60 10.5 30 42.10 ± 3.62 71.46 ± 3.71 32.72 ± 0.78 3 60 1 50 51.53 ± 3.48 61.67 ± 3.79 28.79 ± 0.75 4 60 20 50 61.36 ± 4.97 74.48 ± 3.18 44.28 ± 1.36 5 50 20 30 52.79 ± 3.17 58.30 ± 4.44 31.68 ± 1.81 6 40 20 50 71.22 ± 2.39 60.73 ± 4.58 38.30 ± 0.61 7 50 1 70 54.63 ± 2.97 46.45 ± 4.72 28.69 ± 1.07 8 60 10.5 70 55.76 ± 3.56 76.16 ± 2.43 44.67 ± 1.11 9 50 10.5 50 72.59 ± 4.66 68.73 ± 3.38 41.26 ± 0.74 10 40 10.5 30 34.95 ± 2.75 59.65 ± 3.35 32.16 ± 1.66 11 50 10.5 50 73.25 ± 2.09 67.02 ± 5.14 40.58 ± 0.61 12 50 10.5 50 67.20 ± 3.42 68.71 ± 2.95 40.52 ± 1.24 13 50 10.5 50 69.17 ± 2.92 70.81 ± 7.64 41.20 ± 1.05 14 40 10.5 70 61.14 ± 3.85 57.54 ± 2.62 39.58 ± 0.70 15 50 20 70 65.18 ± 2.55 64.58 ± 2.77 46.20 ± 2.18 16 40 1 50 38.46 ± 2.60 55.89 ± 2.79 34.80 ± 1.55 17 50 10.5 50 71.25 ± 4.42 69.29 ± 3.44 40.40 ± 0.91 table 1: experimental results of box-behnken design used to investigate the effect of ethanol concentration, ultrasound extraction temperature and time on the ache inhibition, radical scavenging activity (rsa) and yield of manuka leaf extracts (50 μg/ml dried extract concentration, 0.52 w/ml acoustic power density).* * ache is the acetylcholinesterase alzheimer’s disease enzyme, and rsa is the dpph radical scavenging activity; ache, rsa and yield data are mean ± standard deviation of triplicate experiments for the same processing conditions. relationship between the dependent and independent variables was presented using a response surface plot. to obtain optimum extraction (factor) conditions, each response (ache inhibition, dpph rsa, yield) was set to “maximum” to achieve the highest value, while the factors (temperature, time, ethanol concentration) were set within the ranges investigated in the rsm study, by using the point prediction post analysis tool of design expert statistical software. validation of model the optimum conditions of extraction for maximum ache inhibition, rsa, and yield were predicted from the polynomial models generated by rsm as explained in previous section. then, the inhibition of ache, rsa at 50 μg/ml, and yields at optimum conditions were determined experimentally. finally, the ic50 values of the optimised crude dried extracts and reference compounds with respect to their ache inhibition and rsa were also determined. enzyme inhibition and dpph radical scavenging activity chemicals chromatography-grade methanol and ethyl acetate and analytical-grade ethanol and acetone were obtained from ecp laboratory research and chemicals, new zealand. analytical-grade chloroform was obtained from sigma-aldrich (nz). phosphate buffer, potassium hyroxide, 5,5’-dithiobis-(2-nitrobenzoic acid) (dtnb), sodium hydrogen carbonate, acetylthiocholine iodide substrate (atchi) acetylcholinesterase from human erythrocytes (ache), dimethyl sulfoxide, 2,2-diphenyl1-picrylhydrazyl (dpph), donepezil and ascorbic acid standards were purchased from sigma-aldrich (nz). deionised water was used for all experiments. acetylcholinesterase activity and inhibition the assay for measuring ache activity was measured by the microplate assay using ellman’s colorimetric method and modified by rauter et al. (2007) (rauter et al. 2007). manuka extract mixtures were prepared in concentrations of 2.2 mg/ml in 12.5% dmso-water which gives a final test concentration of 50 μg of extract/ ml. reagents were prepared as follows: 0.1 m phosphate buffer was freshly prepared before each analysis (136.1 mg of kh2po4 in 10 ml water, adjusted at ph 8.0 with koh); 0.01 m dtnb solution (3.96 mg dtnb in 1 ml water containing 1.5 mg sodium hydrogen carbonate). 0.022 m atchi solution (6.4 mg atchi in 1 ml water); 1.32 unit/ml ache solution (4.4 mg of ache enzyme (10 ml, 1.02041 u) in 1 ml buffer at ph 8.0. the assay was achieved by adding 5 ml of manuka extract, 200 ml phosphate buffer, 5 ml of ache enzyme, and 5 ml dtnb reagent in a 96-well microplate, which was kept for 15 min at 30°c. then, 5 ml of atchi substrate solution was added to the mixture to start the enzymatic reaction. absorbances were determined using a microplate reader (enspire multimode plate reader, perkinelmer, turku, finland) at 405 nm for every 45 s, 6 times consecutively at a controlled temperature of 30°c. the experiments were conducted in triplicate. the rate of enzyme inhibition was calculated using the equation: (2) where vextract is the rate of colour change of the extract (δabs/δtime) and vmax is the maximum rate of colour change of the blank (δabs/δtime) not containing any inhibition compound. for the ic50 study, the dried crude extract of manuka and donepezil (control) were initially dissolved in 12.5% dmso and diluted in distilled water to obtain final test concentrations between 1 to 1000 mg/ml. no inhibition was detected at the highest concentration of dmso used (12.5%). donepezil is a standard alzheimer drug and was used as reference for comparison purposes. radical scavenging antioxidant activity the antioxidant activity of plant extracts was determined using free-radical scavenging effect on the 2,2-diphenyl1-picrylhydrazyl (dpph) radical with slight modification (rauter et al. 2012). methanolic dpph solution (100 mm) was prepared at least a day before analysis, this is to ensure a fully dissolved solution with a stable wavelength measurement. manuka extract mixtures were prepared in concentrations of 1 mg/ml in 12.5% dmso-water giving a final test concentration of 50 μg/ml. similar to ache, for the ic50, four different concentrations of the same extract were tested in triplicates. an aliquot of 10 ml manuka extract was mixed with 190 ml of dpph solution in a clear 96-well microplate. the mixture was shaken vigorously before being kept in dark at room temperature for 40 min. all the test solutions were measured at 517 nm using a microplate reader. the results were shown as percentage of dpph inhibition; these values represent the radical scavenging capacity of the extracts. the percentage of dpph radical scavenging activity (rsa, %) is calculated as follows: (3) where absblank is the absorbance of dpph solution without extract after 40 min while absextract is the absorbance after 40 min of the dpph solution containing the extract mixtures. determination of ic50 extract concentrations the ic50 value of plant extract is the concentration of dried extract that inhibits 50% of enzyme activity or causes 50% rsa. it was estimated from the inhibition/ antioxidant results for different concentrations (3 replicates for each of the four extract concentrations tested). by plotting the % inhibitions/rsa against the extract concentrations (μg/ml), the ic50 value was determined through a nonlinear regression analysis using graphpad prism 6 (graphpad software inc, la jolla, majid & silva. new zealand journal of forestry science (2020) 50:12 page 5 ca, usa). the results for ic50 were reported with 95% confidence interval limits. donepezil and ascorbic acid ic50 values were also determined, as those are reference pure compounds for alzheimer ache inhibition and antioxidant, respectively. the ic50 of references are expected to be much lower than those found for manuka crude extracts, as they are pure compounds. results and discussion the first section presents single factor experiments ache inhibition and rsa results for a dried extract concentration of 50 μg/ml. based on those results, the effects of ethanol concentration (30–70% v/v), extraction temperature (40–60°c) and time (1–20 min) on ache inhibition and antioxidant activity of manuka extracts were considered for the response surface methodology study and results are presented in second section. last section shows the experimental validation of optimal conditions predicted by the rsm models. single factor experiments: effects on ache inhibition and rsa activity effect of the extraction solvent on ic50 of manuka extracts selection of solvent can play an important role in extraction of targeted compounds from complex samples. in the first step of this study, the efficiency of five commonly used solvent, acetone, chloroform, ethyl acetate, methanol, and ethanol on the extraction of antiache and antioxidant compounds from manuka leaves was compared. the comparison was made based on ic50 values, which is the extract concentration needed to inhibit 50% of the enzymatic activity and 50% rsa (figure 1). a low ic50 value is preferable as it represents good activity of a tested extract. methanol and ethanol were the most efficient solvents in the recovery of antiache compounds when compared to ethyl acetate, chloroform, and acetone. although methanol (ic50 = 31.6 μg/ml) was slightly better than ethanol (ic50 = 66.01 μg/ml), as methanol is toxic, ethanol was chosen as the extraction solvent due to its major advantages from environmental and human consumption safety aspects, in addition to recommendation by the us food and drug administration for extraction purposes. the manuka extracts prepared with ethyl acetate, acetone and chloroform were not so good in extracting the bioactive compounds for ache inhibition, presenting much higher ic50 concentrations (209 to 538 μg/ml). the different extraction efficiencies of these solvents may be explained by their differences in polarities, which showed strong preference to solvents with high polarity (tian et al. 2013). in terms of antioxidant activity, methanol and ethanol produced extracts comparable to that of the standard antioxidant ascorbic acid (ic50 12.01 µg/ml). the significant differences between the range of antioxidant and antiache results demonstrated that the latter may be contributed by a much narrower class of compound(s), which were extracted specifically by certain type of solvents. regarding rsa results, the ic50 for all solvents tested was very similar, with ic50 results ranging from 15.8 to 69.7 μg/ml. effect of ethanol concentration regarding the extraction solvent, different proportions of ethanol-water were investigated. figure 2 shows the effect of ethanol concentration at 0, 30, 50, 80, and 100% on ache inhibition (%) and dpph rsa (%) of 50 μg/ml manuka extracts produced after 10 min, 0.52 w/ml, and 50°c extraction. the results show that the extractions of antiache and antioxidant compounds are highly dependent on the ethanol concentration in water. the highest activity was observed for 50% ethanol concentration (70.3%), followed by 80% ethanol (54.1%), while the lowest activity was registered for pure water (12.7%). similar trend was registered for majid & silva. new zealand journal of forestry science (2020) 50:12 page 6 figure 1: effect of solvent type on ache inhibition and dpph rsa of manuka extracts produced with 0.52 w/ml – 50°c ultrasound for 10 min. results are ic50 values (µg dried extract/ ml causing 50% enzyme inhibition or rsa). the error bar represents the values of 95% confidence interval. figure 2: effect of ethanol concentration in water on ache inhibition and dpph radical scavenging activity of 50 μg/ml manuka extracts produced using 50°c and 0.52 w/ml acoustic power density for 10 minutes (the error bars are standard deviations; results with different letters for ache or roman numbers for rsa are significantly different). rsa results. the combination of water with other organic solvents is able to produce moderately polar solvents that have more universal capabilities, ensuring the extraction of many type of compounds (chirinos et al. 2007). in addition, the existence of water also allowed an effective swelling of the plant, which further increase the surface area for solute-solvent contact (yang & zhang 2008). a similar effect was found in the extraction of antioxidant compounds from wheat bran and peanut skins (nepote et al. 2005; wang et al. 2008). effect of extraction temperature the selection of an appropriate range of extraction temperatures was also studied. the experiments were carried out at temperatures between 30 to 70°c under fixed extraction conditions. the effects of extraction temperature on ache inhibition and dpph rsa are shown in figure 3. the activities for ache inhibition and dpph rsa increased when the temperature was increased from 30 to 40–50°c (ache 70%, rsa 56%), and then declined at higher extraction temperature (≥60°c). this is possibly due to the increase of molecular movement at higher temperature, which also increase the solubility that leads to higher extraction rate (yang et al. 2010). however, extraction at temperature higher than 60°c reduced both responses. it may be explained by oxidation and degradation of compounds responsible for ache inhibition and antioxidant properties in manuka extract solutions. effect of extraction time figure 4 shows the influence of extraction time on ache inhibition and dpph rsa. the figure compares the ultrasound extractions using 50% ethanol-water concentration, at 50°c and 0.52 w/ml for different treatment times (2, 5, 10, 20, 30, 40, 50, 60 min). the majid & silva. new zealand journal of forestry science (2020) 50:12 page 7 results show that for antiache compounds, the extraction produced the maximum inhibition just after 5 minutes of ultrasound extraction (68.9%). the extraction reached equilibrium from 10–30 min, then dropped after this point. for dpph rsa, the response increased significantly in the initial 10 min (78.7%) with a maximum at 40 min extraction time (84.7%), then rsa decreased from 40 to 60 min. in both cases, the extraction process responses occur in three stages, rapid increase, slow/equilibrium, decrease stage. during the rapid increase stage, also known as “washing” phenomena, the cell wall of manuka leaves powder cracked within a certain period of time as the result of the acoustic cavitation effect, allowing better penetration of the solvent into the cells and enhancing the release of dissolved compounds out of the solid structure of the leaves (tian et al. 2013). the “slow extraction” may be explained by the decrease in solvent’s permeability into manuka cell structure, caused by the release of various impurities into the solvent during washing stage. besides that, increasing solute content also lower the diffusion rate and mass transfer between the plant matrix into the solvent (şahin and şamlı 2013). the decreasing stage can be observed in ache inhibition (after 30 min) and antioxidant activity (after 40 min). this decrease may have occurred due to the heating effect at 50°c of overexposure to ultrasound treatment that caused degradation of the active compounds present in the extracts (şahin and şamlı 2013). therefore, with a negative response at longer extraction time, it is unnecessary to study beyond this range. effect of acoustic power density the effect of acoustic power density (0.13, 0.26, 0.39, 0.52, and 0.65 w/ml) on ache inhibition and antioxidant activity of the manuka extracts was studied while setting the other parameters fixed as followed: figure 3: effect of extraction temperature on ache inhibition and dpph radical scavenging activity by 50 μg/ml of manuka extracts produced using 50% ethanol concentration for 10 min at 0.52 w/ml acoustic power density (the error bars are standard deviations; results with different letters for ache or roman numbers for rsa are significantly different). figure 4: effect of extraction time on ache inhibition and dpph radical scavenging activity by 50 μg/ml manuka extracts produced using 50% ethanol concentration at 50°c and 0.52 w/ml acoustic power density (the error bars are standard deviations; results with different letters for ache or roman numbers for rsa are significantly different). 50% ethanol-water concentration, extraction time 10 min at temperature 50°c. as shown in figure 5, higher ultrasound acoustic power produced better results for both the responses in general, reaching a maximum for 0.52 w/ml (69.6% ache inhibition, 80.8% rsa). the ultrasound energy has great influence in the extraction of bioactive compounds from the leaves. it is known that the extraction of various substances plant material by sonication was achieved due to the presence of cavitation microscopic bubbles generated by ultrasonic waves travelling through the solvent. the type and amount of bubbles created are proportional to the amplitude, power and intensity of ultrasonic waves. the higher the ultrasound energy (amplitude) the greater the production of cavitation bubbles; the collapse of these bubbles are believed to produce high-shear gradients which disrupts the plant cell walls, this enhance the penetration of solvent into solid matrix and accelerate the release of active components into the extraction solvent, leading to a higher extraction efficiency (tian et al. 2013). from the result, 168 µm which is 0.52 w/ml was chosen as the optimum amplitude/power density. experimental results for box-behnken design, polynomial models and response surface plots the responses of ache inhibition, dpph rsa and extraction yield for the 17 experimental conditions are reported in table 1. the values of variables for each experimental design condition are also presented. the response of ache inhibition varied from 29.9±7.0 to 73.3±2.1%, dpph rsa that was used to evaluate the antioxidant activity ranged between 46.45±4.72 to 76.16±2.43%, while extraction yield values ranged from 28.7±1.1 to 46.20±2.2%. multiple regression analysis was performed based on the results in table 1. the best mathematical model for each response was fitted by the statistical software. table 2: confusion matrix equations 4–6 show the quadratic models for ache inhibition, dpph rsa and yield in terms of their real values (t, t, e): the quality of the adjustments for each model was given by adj r2 and coefficient of variation (c.v.). for ache inhibition the fit statistical indicators were 0.973 and 4.0%, for dpph rsa adj r2 was 0.906 and c.v. 3.98% and yield adj r2 was 0.955 and c.v. 3.35%, demonstrating acceptable models and reliability of the experimental results. a high f-value and p-value lower than 0.05 are also indicators of good models (and parameters significance). the anova results show p-values <0.0001 for ache inhibition, dpph rsa and yield (additional files a–c). three-dimensional response surface plots are graphical representations of the regression equation, which provide a method to visualise the relationship between any two factors on selected responses. based on the rsm polynomials, the 3d response surfaces were generated using design-expert software and 3 examples are show in figure 6. in all response surface graphs the response (z-axis) was plotted against any two variables while the other factor was kept at its ‘0’ level, the midpoint value of the range investigated (50°c, 50% ethanol). in terms of ache inhibition, time and ethanol concentration demonstrated quadratic effect on ache inhibition (figure 6a). at this temperature, the range of response varied between 28 to 75%. at any time and ethanol concentration, ache produced the highest inhibition at around 50% ethanol water concentration and after 10 minutes of extraction. figure 6b shows the effect of temperature and time on dpph rsa of manuka extracts. the dpph rsa increased with increase of extraction temperature. the response curves demonstrated higher dpph rsa at around medium extraction time (10 min). the 3d response of ethanol concentration and time on extraction yield was shown in figure 6c. from the graph, it can be observed that lower concentration of ethanol concentration (30%) produced no obvious increase in extraction yield (3% increment) within 20 minutes of extraction time. however, at higher ethanol, as in case of 70% concentration, the yield of extraction increased from 29% to 46% when extraction time increased from 1 to 20 min. experimental validation of optimal conditions estimated by the rsm models using the models generated by the rsm analysis (equations 4–6), the predicted t, t and ethanol concentration which maximise the ache inhibition, majid & silva. new zealand journal of forestry science (2020) 50:12 page 8 (4) (5) (6) figure 5: effect of ultrasound acoustic power on ache inhibition and dpph radical scavenging activity by 50 μg/ml manuka leaves extracts produced using 50% ethanol in water concentration for 10 min at 50˚c (the error bars are standard deviations; results with different letters for ache or roman numbers for rsa are significantly different). rsa activity and yield of extraction were estimated. three different optimum extraction conditions for each response are shown in table 2 together with the predicted and real experimental values. optimum conditions were 50°c–15 min–50% ethanol for ache inhibition, 60°c– 14 min–56% for dpph radical scavenging activity and 60°c–20 min–70% ethanol for maximum extraction yield. then, the three responses were experimentally determined for those optimum extraction conditions, to compare the predictions with real experimental results. the close values of predicted and experimental data (absolute errors <2%) show that the models are adequate tools to predict ache inhibition, rsa and yield of manuka extracts. the ic50 concentrations of optimised manuka extracts, and reference compounds were also determined experimentally and can be compared with those from non-optimised extracts (table 3). regarding ache inhibition, the ic50 of ultrasound optimised manuka extract (28.48 μg/ml) was lower than the value determined for the ultrasound non-optimised extracts (66.01 μg/ml obtained with 100% ethanol and 157.5 μg/ml obtained with 100% water) and the extract obtained by maceration with 100% water (148 μg/ml) (mathew 2015). although the ic50 concentration of ache inhibition for optimised manuka extract (28.48 µg/ml) was higher than the reference compound donepezil (0.7551 µg/ml), it still shows a comparable value as majid & silva. new zealand journal of forestry science (2020) 50:12 page 9 figure 6: response surface plots showing the effects of temperature (°c), extraction time (min), and ethanol concentration (%) on a) ache inhibition (%) by 50 μg/ml manuka extract, b) dpph radical scavenging activity (%) by 50 μg/ml manuka extract, and c) yield (%). the rest of the variables were set at midpoint (50°c, 10.5 min, 50% ethanol). in a crude extract the active compound(s) was not purified. regarding dpph rsa, optimised manuka extract had lower value of ic50 (2.4 μg/ml) than nonoptimised extract (32.4 μg/ml) and ascorbic acid (12.0 μg/ml), indicating potent antioxidant activity. the yield increased from 14% (water maceration) and 21% (ultrasound with water) to 49% (optimised ultrasound conditions with ethanol 70%). majid & silva. new zealand journal of forestry science (2020) 50:12 page 10 extraction method acoustic power density (w/ml) temp. (°c) time (min) solvent and concent. (%) ic50 (µg/ml) a yield (%) ache dpph rsa optimised ultrasound ache 0.52 50 15 ethanol 50% 28.48 (25.43 to 31.89) optimised ultrasound rsa 0.52 60 14 ethanol 56% 2.367 (2.190 to 2.560) optimised ultrasound yield 0.52 60 20 ethanol 70% 49.19 non-optimised ultrasound 0.52 50 10 ethanol 100% 66.01 (64.59 to 67.45) 32.43 (29.03 to 36.23) nd non-optimised ultrasoundc 0.422 70 2.3 water 100% 157.5 (142.2 to 174.4) nd 21.10 macerationc 70 40 water 100% 148.0 (138.3 to 158.5) nd 14.05 reference compounds donepezilb 0.7551 (0.6129 to 0.9302) ascorbic acidb 12.01 (10.23 to 14.11) table 3: yields and ic50 concentrations for acetylcholinesterase (ache) inhibition and radical scavenging activity (rsa) of manuka leaf extracts: comparison of ultrasound optimised extract with ultrasound non-optimised extracts, extract produced by maceration, and reference compounds for alzheimer enzyme inhibition and antioxidant activity. nd – not determined a ic50 is the plant extract concentration presenting 50% activity. the values inside parenthesis are the 95% confidence interval b donepezil and ascorbic acid are known ache inhibitor and antioxidant compounds, respectively, both used as references cmathew mb. 2015. optimisation of plant extraction conditions for the inhibition of alzheimer’s disease enzyme acetylcholinesterase. final year project 2015, chemical and materials engineering department, university of auckland, new zealand. responses optimum conditions maximum values temperature t (°c) time t (min) ethanol concentration e (%) rsm predicted experimental result a ache (%) 50 15 50 74 72.69 ± 3.84 rsa (%) 60 14 56 79 80.14 ±3.88 yield (%) 60 20 70 50 49.19 ± 2.08 table 2: optimum ultrasound (0.52 w/ml) extraction conditions from manuka leaves for maximum ache inhibition, radical scavenging activity (rsa) and extraction yield: comparison of values predicted by the model and experimentally determined for a dried extract concentration of 50 μg/ml. a mean ± standard deviation of triplicate experiments for the same processing conditions conclusions ultrasound assisted extraction offers an advantage for the extraction of antiache and antioxidant compounds from the leaf of manuka. response surface methodology was successfully employed to optimise the extraction conditions. the quadratic models generated could predict accurately the effect of manuka leaf extraction on ache inhibition, rsa and yield of extraction. ethanol concentration, temperature and time affected all the responses individually and combined with some of the other variables. studies involving the chemical analysis of extracts produced under optimum conditions, namely the identification of compounds responsible for biological activities are an important area of future research. in view of the significant bioactive properties of manuka extracts found in this study, with possible beneficial effects on memory deficit, we would encourage the use of the manuka leaf extract for the development of new phytopharmaceuticals to improve brain function and control dementias such as alzheimer disease. one other application of manuka could be the preparation of tea infusions or incorporation as an ingredient in beverages or solid foods. competing interests the authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. authors’ contributions hm planned and carried out the experiments and wrote the first draft of this manuscript. fs planned the experiments, revised the manuscript, supervised the overall research and journal submission/revision. acknowledgements we thank the ministry of higher education malaysia, government of malaysia for a phd scholarship to hm. the authors acknowledge “the biocide tool box for new zealand manufacturing exporters”, funded by the new zealand ministry of business, innovation and employment. additional files additional file a: ache inhibition by manuka leaf extracts. additional file b: antioxidant dpph radical scavenging activity (rsa) by manuka leaf extracts. additional file c: manuka leaves extraction yield. references adewusi, e.a., steenkamp, v. (2011). in vitro screening for acetylcholinesterase inhibition and antioxidant activity of medicinal plants from southern africa. asian pacific journal and tropical medicine, 4(10), 829–835. 835. https://doi.org/10.1016/s19957645(11)60203-4 ahmed, f., ghalib, r.m., sasikala, p., ahmed, k.k.m. (2013). cholinesterase inhibitors from botanicals. pharmacognosy reviews, 7(14): 121. https://doi. org/10.4103/0973-7847.120511 brooker, s.g., cambie, r.c., cooper, r.c. (1987). new zealand medicinal plants. new zealand: heinemann. cacace, j.e., mazza, g. (2003). optimization of extraction of anthocyanins from black currants with aqueous ethanol. journal of food science, 68(1), 240–248. h t t p s : / / d o i . o rg / 1 0 . 1 1 1 1 / j . 1 3 6 5 2 6 2 1 . 2 0 0 3 . tb14146.x chirinos, r., rogez, h., campos, d., pedreschi, r., larondelle, y. (2007). optimization of extraction conditions of antioxidant phenolic compounds from mashua (tropaeolum tuberosum ruíz & pavón) tubers. separation and purification technology, 55(2), 217–225. https://doi.org/10.1016/j. seppur.2006.12.005 forster, j.r., forster, g. (1776). characteres generum plantarum quas in itinere ad insulas maris australis collegerunt, descripserunt, delinearunt, annis. london, white, cadell & elmsley. 150 p. https:// doi.org/10.5962/bhl.title.4448 häberlein, h., tschiersch, k.p. (1994). triterpenoids and flavonoids from leptospermum scoparium. phytochemistry, 35(3), 765–768. https://doi. org/10.1016/s0031-9422(00)90601-0 hielscher. (2007). up200s/up400s: instruction manual. hielscher ultrasonics gmbh: teltow, germany. hossain, m.b., brunton, n.p., patras, a., tiwari, b., o’donnell, c.p., martin-diana, a.b., barryryan, c. (2012). optimization of ultrasound assisted extraction of antioxidant compounds from marjoram (origanum majorana l.) using response surface methodology. ultrasonics and sonochemistry, 19(3), 582–590. https://doi. org/10.1016/j.ultsonch.2011.11.001 huang, l., shi, a., he, f., li, x. (2010). synthesis, biological evaluation, and molecular modeling of berberine derivatives as potent acetylcholinesterase inhibitors. bioorganic and medicinal chemistry, 18(3), 1244–1251. https://doi.org/10.1016/j. bmc.2009.12.035 lis-balchin, m., hart, s.l, deans, s.g. (2000). pharmacological and antimicrobial studies on different tea-tree oils (melaleuca alternifolia, leptospermum scoparium or manuka and kunzea ericoides or kanuka), originating in australia and new zealand. phytotherapy research, 14(8), 623–629. https://doi.org/10.1002/10991573(200012)14:8<623::aid-ptr763>3.0.co;2-z majid & silva. new zealand journal of forestry science (2020) 50:12 page 11 liyana-pathirana, c., shahidi, f. (2005). optimization of extraction of phenolic compounds from wheat using response surface methodology. food chemistry, 93(1), 47–56. https://doi. org/10.1016/j.foodchem.2004.08.050 majid, h., mathew, m. (2015). optimized ultrasoundassisted water extraction from leptospermum scoparium leaves containing acetylcholinesterase inhibitory activity. presented in 9th cigr section vi international technical symposium, creating value from bioresources. majid, h, silva, f.v.m. (2020a). inhibition of enzymes important for alzheimer’s disease by antioxidant extracts prepared from 15 new zealand medicinal trees and bushes. journal of the royal society of new zealand, 50(4), 538-551. https://doi.org/10. 1080/03036758.2020.1741403 majid, h, silva, f.v.m. (2020b). improvement of butyrylcholinesterase enzyme inhibition and medicinal properties of extracts of aristotelia serrata leaves by ultrasound extraction. food and bioproducts processing, 124, 445-454. https://doi. org/10.1016/j.fbp.2020.10.004 mayer, r. (1990). flavonoids from leptospermum scoparium. phytochemistry, 29(4), 1340–1342. https://doi.org/10.1016/0031-9422(90)85462-o mukherjee, p.k., kumar, v., houghton, p.j. (2007). screening of indian medicinal plants for acetylcholinesterase inhibitory activity. phytotherapy research, 21(12), 1142–1145. https://doi.org/10.1002/ptr.2224 nepote, v., grosso, n.r., guzmán, c.a. (2005). optimization of extraction of phenolic antioxidants from peanut skins. journal of the science of food and agriculture, 85(1), 33–38. https://doi.org/10.1002/jsfa.1933 nordberg, a., svensson, a.l. (1998). cholinesterase inhibitors in the treatment of alzheimer’s disease. drug safety, 19(6), 465–480. https://doi. org/10.2165/00002018-199819060-00004 pan, g., yu, g., zhu, c., qiao, j. (2012). optimization of ultrasound-assisted extraction (uae) of flavonoids compounds (fc) from hawthorn seed (hs). ultrasonics sonochemistry, 19(3), 486–490. https://doi.org/10.1016/j.ultsonch.2011.11.006 porter, n.g., wilkins, a.l. (1999). chemical, physical and antimicrobial properties of essential oils of leptospermum scoparium and kunzea ericoides. phytochemistry, 50(3), 407–415. https://doi. org/10.1016/s0031-9422(98)00548-2 rauter, a.p., branco, i., lopes, r.g., justino, j., silva, f.v.m., noronha, j.p., cabrita, e.j., brouard, i., bermejo, j. (2007). a new lupene triterpenetriol and anticholinesterase activity of salvia sclareoides. fitoterapia, 78(7–8), 474–481. . https://doi. org/10.1016/j.fitote.2007.02.013 rauter, a.p., dias, c., martins, a., branco, i., neng, n.r., nogueira, j.m., goulart, m., silva, f.v.m., justino, j., trevitt, c. (2012). non-toxic salvia sclareoides brot. extracts as a source of functional food ingredients: phenolic profile, antioxidant activity and prion binding properties. food chemistry, 132(4), 1930–1935. https://doi.org/10.1016/j. foodchem.2011.12.028 şahin, s., şamlı, r. (2013). optimization of olive leaf extract obtained by ultrasound-assisted extraction with response surface methodology. ultrasonics sonochemistry, 20(1), 595–602. https://doi. org/10.1016/j.ultsonch.2012.07.029 schelterns, p., feldman, h. (2003). treatment of alzheimer’s disease: current status and new perspectives. the lancet neurology, 2(9), 539–547. https://doi.org/10.1016/s1474-4422(03)005027 silva, e.m., rogez, h., larondelle, y. (2007). optimization of extraction of phenolics from inga edulis leaves using response surface methodology. separation and purification technology, 55(3), 381–387. https://doi.org/10.1016/j.seppur.2007.01.008 stephens, j.m.c., molan, p.c., clarkson, b.d. (2005). a review of leptospermum scoparium (myrtaceae) in new zealand. new zealand journal of botany, 43(2), 431–449. https://doi.org/10.1080/002882 5x.2005.9512966 syad, a.n., devi, k.p. (2014). botanics: a potential source of new therapies for alzheimer’s disease? botanics, 4: 11. https://doi.org/10.2147/btat.s33554 taylor, j.p., hardy, j., fischbeck, k.h. (2002). toxic proteins in neurodegenerative disease. science, 296(5575), 1991–1995. https://doi.org/10.1126/ science.1067122 tian, y., xu, z., zheng, b., lo, y.m. (2013). optimization of ultrasonic-assisted extraction of pomegranate (punica granatum l.) seed oil. ultrasonics and sonochemistry, 20(1), 202–208. https://doi. org/10.1016/j.ultsonch.2012.07.010 wang, j., sun, b., cao, y., tian, y., li, x. (2008). optimisation of ultrasound-assisted extraction of phenolic compounds from wheat bran. food chemistry, 106(2), 804–810. https://doi.org/10.1016/j. foodchem.2007.06.062 yang, y., zhang, f. (2008). ultrasound-assisted extraction of rutin and quercetin from euonymus alatus (thunb.) sieb. ultrasonics and sonochemistry, 15(4), 308–313. https://doi.org/10.1016/j. ultsonch.2007.05.001 yang, l., cao, y.l., jiang, j.g., lin, q.s., chen, j., zhu, l. (2010). response surface optimization of ultrasound-assisted flavonoids extraction from the flower of citrus aurantium l. var. amara engl. journal of separation science, 33(9), 1349–1355. https://doi.org/10.1002/jssc.200900776 majid & silva. new zealand journal of forestry science (2020) 50:12 page 12 diameter distribution model development of tropical hybrid eucalyptus clonal plantations in sumatera, indonesia: a comparison of estimation methods joni waldy1*, john a. kershaw jr2, aaron weiskittel3 and mark j. ducey4 1 faculty of forestry and environmental management, university of new brunswick, 28 dineen dr. fredericton, nb e3b 5a3, canada/ forest planning division, asia pacific resources international holdings limited (april), indonesia 2 faculty of forestry and environmental management, university of new brunswick, 28 dineen dr. fredericton, nb e3b 5a3, canada 3 center for research on sustainable forests, university of maine, orono, me 04469, usa 4department of natural resources and the environment, university of new hampshire, 114 james hall, 56 college road, durham, nh 03824, usa *corresponding author: jwaldy@unb.ca (received for publication 2 march 2021; accepted in revised form 22 november 2021) abstract background: effective forest management and planning often requires information about the distribution of volume by size and product classes. size-class models describe the diameter distribution and provide information by diameter class, such as the number of trees, basal area, and volume per unit of area. a successful diameterdistribution model requires high flexibility yet robust prediction of its parameters. to our knowledge, there are no studies regarding diameter distribution models for eucalyptus hybrids in indonesia. therefore, the aim of this study was to compare different recovery methods for predicting parameters of the 3-parameter weibull distribution for characterising diameter distributions of eucalyptus hybrid clone plantations, on sumatera island of indonesia. methods: the parameter recovery approach was proposed to be compatible with stand-average growth and yield models developed based on the same data. three approaches where compared: moment-based recovery, percentilebased prediction and hybrid methods. the ultimate goal was to recover weibull parameters from future stand attributes, which were predicted from current stand attributes using regression models. results: in this study, the moment method was found to give the overall lowest mean error-index and kolmogorov– smirnov (ks) statistic, followed by the hybrid and percentile methods. the moment-based method better fit long tails on both sides of the distribution and exhibited slightly greater flexibility in describing plots with larger variance than the other methods. conclusions: the weibull approach appeared relatively robust in determining diameter distributions of eucalyptus hybrid clone plantation in indonesia, yet some refinements may be necessary to characterize more complex distributions. new zealand journal of forestry science waldy et al. new zealand journal of forestry science (2022) 52:1 https://doi.org/10.33494/nzjfs522022x151x e-issn: 1179-5395 published on-line: 11/01/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access et al. 2011), since they provide detailed yield estimates by size (de miguel et al. 2010). size-class models predict diameter distributions and provide estimates of forest attributes (such as the number of trees, basal area, and volume per unit area) by diameter class (poudel & cao 2013). introduction effective forest management and planning often requires information about the distribution of volume by size and product classes (burkhart & tomé 2012). size-class and individual-tree models are required for predicting distributions of multiple products (weiskittel keywords: 3-parameter weibull; parameter recovery method; moment-based estimates; percentile estimates; hybrid parameter estimates; indonesia, eucalyptus hybrids mailto:jwaldy@unb.ca http://creativecommons.org/licenses/by/4.0/), waldy et al. new zealand journal of forestry science (2022) 52:1 page 2 the concept of diameter distributions is not new to forestry literature (de liocourt 1898). several probability density functions are used to describe the structure of forest stands, including weibull distributions (bailey & dell 1973). the weibull distribution is one of the most widely used diameter distribution models (burkhart & tomé 2012) and one of the best-performing models among other distributions (eisfeld et al. 2005). the weibull distribution is a flexible distribution function capable of fitting a variety of distribution shapes (poudel & cao 2013). it has a lower bound and calculations of proportions of trees across diameter classes is straightforward and does not require numeric integration (cao 2012). in addition, the parameters of the weibull distribution are generally well correlated with several stand-level attributes (bailey & dell 1973), such as dominant height, quadratic mean diameter, and mean diameter (sghaier et al. 2016). because of its flexibility, the weibull distribution is applicable to many different species and forest structures from pure, single cohort, even-aged stands to multispecies, multicohort, uneven-aged stands (mattney & sullivan 1982; hyink & moser 1983; knowe et al. 1994; knowe & stein 1995; siipilehto 1999; cao 2004; newton et al. 2005). the three-parameter weibull distribution function is commonly used to quantify tree diameter distributions because of: its flexibility in fitting a variety of shapes and degrees of skewness; the relative simplicity of estimating its parameters; and the cumulative distribution has a closed-form solution (bailey & dell 1973; bowling et al. 1989; little 1983; matney & sullivan 1982; rennolls et al. 1985; schreuder & swank 1974; zarnoch et al. 1991). the probability density function (pdf) for the threeparameter weibull distribution is specified by: (1) = 0, otherwise integration yields a closed-form cumulative distribution function (cdf): (2) the parameter α is referred to as the location parameter and defines the minimum value of the distribution, β is the scale parameter, γ is the shape parameter and 𝑥 is diameter at breast height (clutter et al. 1983). the β and γ parameters must be positive, while α mathematically can be positive, zero, or negative (provided 𝑥 α ≥ 0). for diameter distribution applications ≥ 0. for shape (γ) parameters less than 1, the weibull distribution assumes a classic inverse j-shape distribution typically found in uneven-aged stands, while when γ equals 1, the negative exponential distribution results. mound shape curves typical of even-aged stands are produced for γ greater than 1 (burkhart & tomé 2012). when γ is equal 3.6, the weibull distribution is symmetrical, similar to a normal distribution shape. right-skewed curves are defined for γ less than 3.6 and left-skewed curves for γ greater than 3.6. as γ approaches infinity, the distribution approaches a spike over a single point (burkhart & tomé 2012). the location parameter often is assumed known in many cases, so it is logical to set this parameter to the smallest value or the lower limit of diameter measurement (kershaw et al. 2016). parameter estimates based on maximum likelihood methods for the weibull distribution requires individual tree data (bolker 2008). maximum likelihood estimation has several desirable statistical properties, such as consistency and asymptotic normality (bury 1999; royle & dozario 2008), and provides better estimates compared to other methods (zhou & mctague 1996). however, it requires more computational resources (cao & mccarty 2006) and precise individual tree measurements. in most forestry applications, diameter distributions are generally predicted from characteristics measured in a stand of interest. hyink and moser (1983) presented a generalized framework for estimating diameter distributions using parameter prediction methods (ppm) and parameter recovery methods (prm). in the ppm approach, the future parameters of the distribution model are directly predicted from the current parameters and other information about the stand such as density, basal area and volume. (kershaw et al. 2016). the ppm uses the location (α), scale (β), and shape (γ) parameters as the dependent variables, which have been previously estimated using maximum likelihood methods (cao 2004). in the prm approach, future values of stand parameters are directly predicted, and the parameters of the diameter distribution are derived from the stand parameters (hyink & moser 1983; kershaw et al. 2016). the parameter prediction approach using the weibull distribution has been applied to many different species and forest types: pine (pinus taeda l. and pinus echinate mill.) plantations (smalley & bailey1974 a,b); slash pine (pinus elliottii) plantations (dell et al. 1979); longleaf pine (pinus palustris mill.) plantations (leduc et al. 2001); loblolly pine (pinus taeda) plantations (clutter et al. 1984; feduccia et al. 1979; smalley & bailey 1974a); natural slash pine (pinus elliotti) stands (schreuder et al. 1979); mixed stands of western hemlock and douglasfir (pseudotsuga menziesii and tsuga heterophylla) (little 1983); sitka spruce (picea sitchensis) plantations (rennolls et al. 1985); and norway spruce (picea abies) plantations (kilkki & päivinen 1986; kilkki et al. 1989; siipilehto 1999) and many others (vanclay 1994; weiskittel et al. 2011). while parameter predictions can be easily derived from regression equations fit to precisely estimated diameter distributions, these models often produce poorer parameter predictions, low r2 values for α and γ parameters; and are not responsive to silvicultural treatments (hyink & moser 1983). because ppm does not work very well in many cases, this approach was not considered in this study. according to siipilehto and mehtätalo (2013), there are two main options for prm approaches: moment-based and percentile-based estimation. the prm moment-based approach solves for the weibull parameters typically using the moments of the diameter distribution that are estimated from regression equations using a variety of stand characteristics (bowling et al. 1989; hyink & moser 1983; lindsay et al. 1996; matney & sullivan 1982; newton et al. 2004; strub & burkhart 1975) or from stand-level models predicting future stand conditions (clutter et al. 1983; waldy et al. 2021). the percentile-based method predicts the weibull parameters using percentiles of the diameter distribution that also can be estimated from stand characteristics (bailey et al. 1989; brooks et al. 1992; lohrey & bailey 1976; knowe 1992; magnussen 1986; mctague & bailey 1987). other methods, such as a hybrid approach (mctague & bailey 1987), cumulative distribution function regressions (cdfr; cao 2004), and modified cdfrs (poudel & cao 2013) also have been proposed. a successful diameter-distribution model requires robust predictions of its parameters. the prm approach was proposed because its estimates were compatible with stand-average growth and yield models developed from the underlying diameter distribution data (hyink & moser 1983). other methods were proposed because of the nature of the underlying data and/or the objectives of the study. in this study, the overall objective of this analysis was to evaluate different parameter recovery methods for predicting parameters of the weibull pdf for characterising diameter distributions of eucalyptus hybrid clone plantations in sumatera, indonesia. the specific objectives were to: (1) compare moment-based, percentile-based and hybrid methods; (2) determine the best approach for robust estimation of diameter distributions across a full range of current and future predicted stand conditions; and (3) based on observed performance, predict moments and/or percentiles from stand and site characteristics using nonlinear regression analyses. methods study site this study was conducted in sector teso east of pt. riau andalan pulp and paper, a member of asia pacific resources international holding limited (april) group and used inventory plot data from eucalyptus hybrid plantations. teso east is 19,600 ha in size and is located in the central region of sumatera island, riau province, indonesia in the kampar and kuantan singingi regencies between 101° 18′ e and 101° 32′ e, and 00° 09′ n and 00° 03′ n (figure 1). the region is characterised by a wet tropical climate with average rainfall ranging from 2000– 3000 mm per year, and the average rainy days is around 160 days per year. the annual average temperature is 27.6oc with an average minimum of 21.8oc and an average maximum of 35oc. until now, estate teso east has had its fifth rotation. the first rotation was planted with acacia mangium in 1995 and then eucalyptus sp. was planted on a large scale starting in 2010. based on the soil characteristics, this study location was dominated by soil horizon b (topsoil) and c (parent material). this plantation area is relatively flat with slopes ranging from 0–15% and low elevation ranging from 30–90 metres above sea level. data collection in april plantations, the plot layout used systematic sampling with random starting points. initial sampling intensity was 1% of total stand area (one plot represents 4 ha area) with an additional 2–5% sampling intensity for a pre-harvest inventory (one year before harvest). plots were circular with a radius of 11.28 m (0.04 ha). first measurements were made at six months after planting and regularly continued at twelve-month intervals until harvesting. dbh was measured beginning at 18 months. all live trees with dbh of 1.0 cm and greater on each waldy et al. new zealand journal of forestry science (2022) 52:1 page 3 table 1: description of the study sites figure 1: location of the study area in the sector teso east, central region of sumatera plot were measured using a diameter tape at 1.3 metres above the ground as measured from the uphill side of the stem. each tree was assigned a status (live or dead), and assessed for wind damage, pests, and diseases. this study only used the most recently established eucalyptus clone stands that were planted 2013 to 2016 with an initial spacing of 3 x 2 m (1667 trees per ha) and had at least three consecutive measurements. there are 2808 measurements in this study, where clone a and b accounted 2,476 and 332 plots, respectively. table 1 summarises stand attributes and their associated increments by clone. in comparison to clone a, clone b is recently developed by the april group concession. the maximum age of existing inventory data for clone b is 42 months. because of its promising growth (table 1), it was important to include this clone in the modeling conducted in this study. maximum likelihood estimation (mle) weibull parameters were estimated using the individual tree dbh measurements from each plot at each measurement period using maximum likelihood estimation (mle) methods (johnson et al. 1995; casella & berger 2001). the location parameter (α) was set to 1 cm in this study (minimum dbh measured), and the mle estimates of the scale (β) and shape (γ) parameters were used as reference distributions to compare with the recovered parameter estimates. the likelihood function of the weibull pdf (eq. 1): (3) mles where obtained by minimising the negative of the logarithm of the likelihood function. moment-based parameter recovery in moment-based parameter recovery, regression equations, as a function of mean top height, stand density, age, or relative density, are used to predict the arithmetic mean diameter (d̄, the first moment) and waldy et al. new zealand journal of forestry science (2022) 52:1 page 4 the quadratic mean diameter (dq, the square root of the second moment). as with the mle estimates, we set α =1, our minimum measured dbh. weibull parameters were then recovered from the arithmetic mean diameter (d̄) and quadratic mean diameter (dq) using (burkhart & tomé 2012): (4) (5) where was predicted dq, d̄ was predicted d̄, and α, β, and γ were recovered weibull parameters. eqs. 4 and 5 were numerically solved to recover the parameters using the algorithms developed by kershaw and maguire (1995). percentile-based parameter recovery percentile estimation of the weibull parameters are relatively easy to obtain given the closed form solution of the cumulative distribution and the geometric interpretation of the distribution parameters (zarnoch & dell 1985). similar to the moment estimates, the percentile estimates are functions of the distribution parameters and are often highly correlated with forest stand characteristics (borders et al. 1987; knowe 1992; knowe et al. 1992). if three sample percentiles are known, each can be equated to its corresponding weibull cumulative distribution function value and the three equations can be solved iteratively for estimates of α, β, and γ (burkhart & tomé 2012). if the location parameter is assumed known, then only two percentiles are required to estimate β and γ. given the weibull cumulative distribution function (equation 2), and letting dp represent the estimated p th percentile value of diameter in the sample, then: ̭ stand attributes clone a (n = 2476) clone b (n = 332) mean min max stdev mean min max stdev a (month) 34.3 12 63.6 14.4 29.1 18 42 9.6 ht (m) 17.3 5.8 30 5.7 18 10.6 27.7 4.4 ba (m²ha-1) 12.4 1.8 26.9 5.2 14.2 3.9 22.9 3.8 sd (trees∙ha-1) 1,536 175 2,101 209 1,569 450 1,851 203 dq (cm) 9.7 3.4 18.5 2.5 10.6 7.8 14.5 1.4 vol (m³ha-1) 73 0.05 234.4 53.37 85.71 14.69 189.8 42.63 table 1: summarised stand attributes and their increment for the inventory plots (n = 2808) used for modeling (a = age; ht = top height; ba = basal area; sd = stand density; dq = quadratic mean diameter; and vol = stand volume). (6) solving eq.6 for dp yields: (7) the scale parameter, β, is given by: (8) given two percentiles p1 and p2 where p1 < p2, γ is estimated using: (9) theoretically, any two percentiles can be used; however, bailey et al. (1989) found best performance resulted when percentiles represented a broader proportion of the distribution. in this study, we used the 25th and 99th diameter percentiles to recover parameters β and γ of the weibull distribution. hybrid parameter recovery methods moments and percentiles are combined in the hybrid approach (bailey et al. 1989; brooks et al. 1992; knowe et al. 2005; lee & coble 2006; coble & lee 2008; and jiang & brooks 2009). as in the other approaches, the location parameter was expected to be known (i.e., α = 1), and the β and γ parameters were recovered from a moment estimate and two percentile estimates using (bailey et al. 1989): (10) (11) here we used estimates of dq, 25th and 99th diameter percentiles to estimate β and γ. moment and percentile estimation the quadratic mean diameter, arithmetic mean diameter, and the percentiles were calculated for each plot and measurement period using the individual tree dbh data. several regression equation forms were used to predict the diameter moments or percentiles (e.g., matney & farrar 1992; baldwin & feduccia 1987; cao 2004) based on stand characteristic. in this study, the logistic equation was used to predict arithmetic mean diameter (d̄) and a modified general form of the regression equation from cao (2004) to predict diameter percentiles with additional dq variables to ensure compatibility with stand-level models (waldy et al. 2021). dq was estimated using the relationship of basal area and stand density. in waldy et al. (2021), the stand density model from clutter et al. (1983) and basal area model derived from the schumacher polymorphic equation were the best fit models based on several models evaluated, for projecting future attributes to any point in time. these time-based models gave dq estimates with rmse = 0.72 cm and explained 85% of the variability (waldy et al. 2021). arithmetic mean diameter (d̄) was estimated using a logistic equation and the estimated dq: (12) finally, percentiles were estimated using: (13) where d̄ was arithmetic mean diameter (cm); di was the ith diameter percentile (cm); dq was the quadratic mean diameter (cm); rd was a relative density measure defined as the ratio of actual density to the maximum density attainable in a stand with the same mean tree size; ht was the top height (m); age was stand age (months); bi were regression parameters and ϵ was a random error term. equation (12) assured that dq > d̄ because the logistic component, , and exp(x) > 0. fitted eqs. 12 and 13 were used to predict each plot × measurement period moments. nonlinear regression was used to fit eqs 12 and 13 and appropriate goodness of fit criteria were used to evaluate the moment and percentiles estimates. we then included clone and site class as random effects into the regression equations and nonlinear mixed effects methods (pinheiro & bates 2000) were used estimate fixed and random effects for each model. a likelihood ratio test used to assess significance of random effects (weiskittel et al. 2011). model evaluation the kolmogorov-smirnov (ks) statistic (massey 1951) and an error-index (ei; reynolds et al. 1988) were computed for each method to evaluate the three prediction methods. using a significance level of 5%, the ks test was used to compare the estimated cumulative frequency and the observed frequency. the method producing the lowest average ks statistics and errorindex values was considered the best method. all estimation and analyses were carried out using the r statistical language (r core team 2020). waldy et al. new zealand journal of forestry science (2022) 52:1 page 5 ̭ results moment and percentile prediction models using nonlinear least square analysis, all coefficients for equation 12 and 13 to predict arithmetic mean diameter and diameter percentiles were significant (p < 0.05). consequently, all parameters associated with stand characteristics (dq, rd, ht, and age) were included for mixed-effect analysis that involved clone and site class within clone as random effects (table 2). based on the likelihood ratio test, the best random-effects models included all coefficients as random effects of clone and site class within the clone (table 2). parameter estimates and their associated standard errors (in parentheses), random effects standard deviations, and goodness-of-fit statistics for the arithmetic mean diameter, 25th and 99th diameter percentiles are shown in table 2. the full model with fixed and random effects accounted for 99%, 77% and 93% of the variation for d̄, d25 and d99 prediction models, respectively (table 2). the d25 percentile model had the lowest performance compared with the d̄ and d99 models. differences between clones were greater than differences across sites within clones for almost all modelling parameters, except b1 and b2 associated with the d99 model (table 2). coefficient estimates (fixed + random effects) by clone and site classes for all prediction models are shown in table 3. characteristics of maximum likelihood estimates the estimated weibull scale parameters, β, based on the mle method ranged from 3.95 to 15.11 and the shape parameters, γ, ranged from 1.68 to 11.50 (table 4). based on the k-s test, estimated diameter distributions were not significantly (p > 0.05) different from the observed diameter distributions for 68% (ci = 95%) and 80% (ci = 99%) of the observed eucalyptus hybrid clone diameter distributions (table 5). the scale parameter increased, and the shape parameter decreased, with increasing age and site class for both clones (figure 2). for a similar stand age, clone a had lower scale and shape parameters than clone b. in addition, clone b had smaller variation in scale parameter but higher variation in shape parameter than clone a (figure 2). diameter distribution model comparisons table 4 summarises the parameter estimates for fit data using maximum likelihood (mle) and the three prms (mom, pct, and hyb). the results of the three prms indicated that all the three prms provided relatively similar mean parameter estimates for the weibull distribution function, with the average scale parameter in the range of 9.53–10.37, and the average shape parameter in the range of 2.80–3.46. the mom and hyb methods were more similar to the mle based on the waldy et al. new zealand journal of forestry science (2022) 52:1 page 6 factor parameter d̄ d25 d99 fixed effects b0 3.7564 (1.1495) 1.2951 (0.1785) 1.9872 (0.1591) b1 1.5021 (0.4841) 0.0269 (0.0088) -0.0155 (0.0029) b2 -0.7535 (0.2975) 0.0127 (0.0036) 0.0288 (0.0012) b3 15.3432 (6.7380) 0.6449 (0.1969) -0.3663 (0.3397) random effects (clone) s(b0) 2.8489 0.2334 0.2135 s(b1) 0.7263 0.0096 <0.0001 s(b2) 0.6312 0.0040 <0.0001 s(b3) 13.2230 0.2348 0.4693 (site class | clone) s(b0) 2.8490 0.1454 0.1070 s(b1) 1.1114 0.0129 0.0064 s(b2) 0.7182 0.0050 0.0027 s(b3) 16.0999 0.2332 0.1466 goodness-of-fit fixed rmse 0.2874 1.1311 1.1510 r² 0.9821 0.5451 0.8631 bias -0.0829 -0.5099 0.7320 fixed + random rmse 0.2156 0.7953 1.0592 r² 0.9899 0.7738 0.9327 bias -0.0038 -0.0046 -0.0117 table 2: parameter estimates and their associated standard errors (in parentheses), random effects standard deviations and goodness-of-fit statistics for arithmetic mean diameter and percentiles prediction models. average scale parameter, the pct and hyb methods were relatively closer to the mle based on the average shape parameter (table 4; figure 3). shape parameters for hyb method were similar with the pct that derived from the same variables and formulation (table 4; figure 3). like most other applications, the shape parameter was more difficult to model than the scale parameter (figure 3). based on the statistical model evaluation of three prms, the mom had the best fit based on the ks statistic at 0.1757, followed by the hyb and pct with ks statistic at 0.1988 and 0.2125, respectively (table 5). the ks statistic also indicated the coverage (# of observed distributions fitting within the 95% ci) of estimated distributions were 49% (mom), 26% (pct), and 37% (hyb); and 66% (mom), 44% (pct), and 54% (hyb) at 99% confidence (table 5). the mom also had the lowest mean error-index at 26.5218, followed by hyb (27.7307) and pct (31.4708) (table 5). in terms of the differences in precision for predicting the weibull parameter, the mom has the lowest variability with the standard deviation of error index at 6.7118 and the pct had the highest variability at 9.7699. although all approaches allow for a direct mathematical link between the predicted overall stand characteristics and a diameter distribution that is waldy et al. new zealand journal of forestry science (2022) 52:1 page 7 equation parameter estimate coefficient by clone and site class a:22 a:24 a:26 a:28 b:24 b:26 b:28 d̄ b0 5.0580 6.8023 6.4207 5.9698 1.4566 -0.6481 1.2354 b1 0.1972 0.6093 0.3989 1.7608 2.7261 3.2856 2.0468 b2 -0.8457 -1.5070 -1.3428 -1.6237 -0.2855 0.2443 -0.0292 b3 2.6035 0.2958 -0.3092 7.9372 31.4552 41.3175 27.9144 d25 b0 0.9454 1.1477 1.2233 0.9345 1.4989 1.4012 1.6833 b1 0.0070 0.0174 0.0041 0.0415 0.0351 0.0320 0.0421 b2 0.0237 0.0155 0.0178 0.0096 0.0101 0.0125 0.0039 b3 0.9622 0.8341 0.5222 1.1938 0.4514 0.5397 0.2436 d99 b0 2.1217 2.1818 2.3095 2.3301 1.7646 1.7967 1.6186 b1 -0.0063 -0.0224 -0.0156 -0.0165 -0.0126 -0.0115 -0.0235 b2 0.0289 0.0302 0.0269 0.0257 0.0285 0.0278 0.0333 b3 -0.7517 -0.7901 -1.0401 -0.9254 0.1114 0.0389 0.3257 table 3: coefficient estimates (fixed + random effects) of nonlinear mixed effect model by clone and site classes for arithmetic mean diameter and percentiles prediction models. r α is referred to as the location parameter and defines the minimum value of the distribution, β is the scale parameter, γ is the shape parameter and 𝑥 is diameter at breast height d̄ ϵ method scale parameter (β) shape parameter (γ) mean stdev min max mean stdev min max mle 10.23 1.85 3.95 15.11 3.39 1.20 1.68 11.50 mom 10.24 1.71 4.29 15.08 2.80 0.78 1.89 14.02 pct 9.53 1.59 5.10 14.58 3.46 0.97 2.35 7.98 hyb 10.37 1.79 4.24 15.85 3.46 0.97 2.35 7.98 table 4: average of the parameter estimates of the weibulll distribution using fit data and three prediction methods. method ks ei mean stdev coverage (%)a mean stdev mle 0.1499 0.0699 68, 80 23.6921 6.4569 mom 0.1757 0.0724 49, 66 26.5218 6.7118 pct 0.2125 0.0759 26, 44 31.4708 9.7966 hyb 0.1988 0.0825 37, 54 27.7307 8.1979 a percentage of distributions that fall within 95% or 99% confidence intervals of the ks dmax table 5: means and standard deviations of the goodness-of-fit statistics produced by three diameter distribution prediction methods. consistent with those characteristics, the moment-based method (mom) indicated the best fit to the observed data when compared to the other methods (pct and hyb). for evaluation, some graphical examination of the performance of the three prms were conducted. figure 4 illustrates the three methods for typical plots of the diameter distributions observed in the eucalyptus hybrid clone plantations. the plots represent a range of clones, stand ages, and the variation of distributions typically observed in the region. figure 4a shows a unimodal distribution for clone a at 30 months and illustrates that the three methods perform equally well for modeling the distribution of that plot. figure 4b shows a multimodal distribution. in this case, none of the three methods fitted the plot well and all missed the valley (8–9 cm) and the peak (13–14 cm). however, the mom and pct were better fits for the peak (6–7 cm) than the hyb. in figure 4c for clone b at 30 months with a distribution that was close to normal, the mom was the better fit than the others. the pct and hyb show a similar pattern for this plot. while in figure 4d with an irregular distribution, the mom tends to deal with tails of both sides and mom exhibited slightly greater flexibility in describing the larger variance than the two other methods. waldy et al. new zealand journal of forestry science (2022) 52:1 page 8 β (c m ) γ figure 2: boxplots of: weibull scale β (a); and shape γ (b) parameter by clone, age and site class estimated from full sample of the trees in the plot using maximum likelihood estimation method. (a = clone "a"; b = clone "b"; sc = site class; the black dots indicate extreme values). (a) (b) waldy et al. new zealand journal of forestry science (2022) 52:1 page 9 figure 3: parameter estimation comparison between mles and three parameter recovery estimation methods for scale (a; above) and shape (b; below). the dotted line represents the 1:1 line, while the red line is the observed trend using a smoothing line. figure 4: model evaluation for the four example plots that represent: (a) clone a, 30 months; (b) clone a, 42 months; (c) clone b, 30 months; and (d) clone b, 42 months. the histogram represents the observed diameter distribution, and three curves represent diameter distribution models prediction. waldy et al. new zealand journal of forestry science (2022) 52:1 page 10 discussion to be compatible with a recently developed standlevel growth and yield model (waldy et al. 2021), the predicted quadratic mean diameter played an important role in recovering parameters of the weibull distribution that characterised the future diameter distributions. in this study, the prediction models were quite good at predicting arithmetic mean diameter and 99th percentiles but relatively poor at predicting the 25th percentile. this means that stand attributes, especially dq and age, have a stronger correlation with the arithmetic mean diameter and higher percentiles than lower percentiles, which has been found in similar analyses. for example, cao (2012) found similar trends in predicting future diameter distributions of loblolly pine (pinus taeda l.). he found that stand attributes explained more variability in estimating higher percentiles (50th percentile and above) than lower percentiles. the method of moments is one of the most accurate methods for estimating the weibull distribution parameters (al-fawzan 2000; nanang 1998; ueno & osawa 1987; shifley & lentz 1985). moments are the preferred method in growth and yield models because they ensure numeric compatibility and generally require fewer equations (weiskittel et al. 2011). our study indicated the mom had a higher percentage of coverage for estimated distributions that fit the observed diameter distributions than pct and hyb. this higher percentage for mom in this study was likely because of sufficient sample sizes to model moment based recovery with an average of 60 trees per plot (plot size 0.04 ha). bankston (2019) reported that larger plot sizes resulted in more accurately predicted diameter distributions. however, he suggested a plot size of 0.08 ha might be sufficient for model building for data from unthinned stands. a substantial decrease in error was no longer evident when plot size increased from 0.08 to 0.10 hectares. shiver (1988) suggested a sample size of not less than 50 trees is required to obtain satisfactorily accurate estimates of the weibull parameters, while simulation studies from saborowski (1994) indicated that a sample with n = 80 could generally be expected to produce satisfactory results. to get a better diameter distribution that represented plot-level measurements, modifying the plot size is not a logical decision for the well-established forest inventory system applied by the april group. instead of using weibull distributions, other probability density functions (e.g., beta, johnson’s sb, gamma or lognormal) are potentially worth examining to describe the structure of a eucalyptus hybrid clone in this region. goodwin (2020) suggested several variants of the weibull distribution. while some of those models produced better estimates of diameter distributions, goodwin’s recommendation for plantations was to use the 3-p weibull distribution, as done in this study. using simulated development of the mom model predictions, the peakedness of the distribution was reduced for both clones with increasing stand age. the diameter distribution of clone a tended to become more positively skewed and the variation increased with increasing stand age (figure 5), partly because of mortality in the lower tree strata, and thinnings from below, which remove the suppressed trees or crown thinning (thinning from above), which remove dominant and co-dominant trees (van laar & akca 2007). the number of trees decreased in lower diameter classes and increased in upper classes, shifting curves to the right and increasing the flattening degree with increasing age. in clone b, the shape curves for all ages were almost similar with no significant increases in variation. these findings were relatively logical because the mortality for this clone was very low, so the diameter growth tends to be relatively uniform. conclusions in this study, the moment method was found to give the lowest mean error-index and ks statistic, followed by the hybrid and percentile methods. although all three methods had difficulty in describing multimodal diameter distributions, the moment method tended to be more robust with tails on both sides of the distribution and exhibited slightly greater flexibility in describing the larger variance than the two other methods. overall, the weibull approach appeared relatively robust in estimating diameter distributions of eucalyptus hybrid clone plantation in indonesia yet some refinements may be necessary to characterise more complex distributions and longer rotations. figure 5: simulated development of diameter distribution over ages using mom method. simulations are based on prediction standlevel variables (sd, ba, ht, rd and dq) using site class = 26 and initial stand density 1667 trees ha-1. list of abbreviations cdf cumulative distribution function dbh diameter at breast height ei error index hyb hybrid ks kolmogorov-smirnov mle maximum likelihood estimation mom moment pct percentile pdf probability density function ppm parameter prediction method prm parameter recovery method competing interests the authors declare that they have no competing interests. authors’ contributions jw was the primary author, completed the statistical analyses and writing the manuscript draft. jak contributed to the idea, review and editing the manuscript. aw and md contributed to review and editing the manuscript. all authors read and approved the final manuscript. acknowledgements the authors are grateful to pt. riau andalan pulp and paper, a member of april group that provided forest inventory data for this study. references al-fawzan, m.a. (2000). methods for estimating the parameters of the weibull distribution. interstat, statistics on the internet. bailey, r.l., & dell, t.r. (1973). quantifying diameter distributions with the weibull function. forest sciences, 19(2), 97-104. bailey, r.l., burgan, t.m., & jokela, e.j. (1989). fertilized mid-rotation-aged slash pine plantations-stand structure and yield prediction models. southern journal of applied forestry, 13(2), 76-80. https:// doi.org/10.1093/sjaf/13.2.76 baldwin, v.c., jr., & feduccia, d.p. (1987). loblolly pine growth and yield prediction for managed west gulf plantations. [res. pap. so-236] (27 p.). new orleans, la: u.s. department of agriculture, forest service, southern forest experiment station. https://doi.org/10.2737/so-rp-236 bankston, s. (2019). sample-plot size and diameter moments/percentiles prediction model effects on stand diameter distribution recovery accuracy. thesis. department of forestry, mississippi state university. bolker, b.m. (2008). ecological models and data in r. 1st ed. princeton, nj, usa: princeton university press. 396 p. borders, b.e., souter, r.a., bailey, r.l., & ware, k.d. (1987). percentile-based distributions characterize forest stand tables. forest science, 33(2), 570-576. bowling, e.h., burkhart, h.e., burk, t.e., & beck, d.e. (1989). a stand-level multispecies growth model for appalachian hardwoods. canadian journal of forest research, 19(4), 405-412. https://doi. org/10.1139/x89-064 brooks, j.r., borders, b.e., & bailey, r.l. (1992). predicting diameter distributions for site-prepared loblolly and slash pine plantations. southern journal of applied forestry, 6(3), 130-133. https://doi. org/10.1093/sjaf/16.3.130 burkhart, h.e., & tomé, m. (2012). modelling forest trees and stands. dordrecht, new york: springer. https://doi.org/10.1007/978-90-481-3170-9 bury, k. (1999). statistical distributions in engineering. cambridge: cambridge university press. https:// doi.org/10.1017/cbo9781139175081 cao, q.v. (2004). predicting parameters of a weibull function for modelling diameter distribution. forest science, 50(5), 682-685. cao, q.v. (2012). use of the weibull function to predict future diameter distributions from current plot data. in j.r. butnor (ed.), proceedings of the 16th biennial southern silvicultural research conference. [e-general technical report srs-156]. (pp. 53-58). asheville, nc, usa: us department of agriculture forest service, southern research station. cao, q.v., & mccarty, s.m. (2006). new methods for estimating parameters of weibull functions to characterize future diameter distributions in forest stands. in k.f. connor (ed.), proceedings of the 13th biennial southern silvicultural research conference. [e-gen. tech. rep. srs-92]. (pp. 338-340). asheville, nc: u.s. department of agriculture, forest service, southern research station. casella, g., & berger, r.l. (2001). statistical inference, 2nd ed. duxbury-thomson learning, 660 p. clutter, j.l., fortson, j.c., pienaar, l.v., brister, g.h., & bailey, r.l. (1983). timber management: a quantitative approach. john wiley & sons, inc. 333 p. clutter, j.l., harms, r.w., brister, g.h., & rheney, j.w. (1984). stand structure and yields of site-prepared loblolly pine plantations in the lower coastal plain of the carolinas, georgia, and north florida. [e-gen. tech. rep. se-27]. (173 p.). asheville, nc: u.s. department of agriculture, forest service, southeastern forest experiment station. https:// doi.org/10.2737/se-gtr-27 waldy et al. new zealand journal of forestry science (2022) 52:1 page 11 https://doi.org/10.1093/sjaf/13.2.76 https://doi.org/10.1093/sjaf/13.2.76 https://doi.org/10.2737/so-rp-236 https://doi.org/10.1139/x89-064 https://doi.org/10.1139/x89-064 https://doi.org/10.1093/sjaf/16.3.130 https://doi.org/10.1093/sjaf/16.3.130 https://doi.org/10.1007/978-90-481-3170-9 https://doi.org/10.1017/cbo9781139175081 https://doi.org/10.1017/cbo9781139175081 https://doi.org/10.2737/se-gtr-27 https://doi.org/10.2737/se-gtr-27 coble, d.w., & lee, y.j. (2008). a new diameter distribution model for unmanaged slash pine plantations in east texas. southern journal of applied forestry, 32(2), 89-94. https://doi.org/10.1093/sjaf/32.2.89 de liocourt, f. (1898). de l’amenagement des sapinieres. [bulletin trimestriel vol. 4]. (pp. 396-409). societe forestiere de franche-comte et belfort. de miguel, e.p., machado, s.a., figueiredo filho, a., & arce, j.e. (2010). using the weibull function for prognosis of yield by diameter class in eucalyptus europhylla stands. cerne, 16(1), 94-104. https:// doi.org/10.1590/s0104-77602010000100011 dell, t.r., fedducia, d.p., campbell, t.e., mann, w.f. jr., & polmer, b.h. (1979). yields of un-thinned slash pine plantations on cutover sites in the west gulf region. [res. pap. so-148]. (94 p.). new orleans, la: u.s. department of agriculture, forest service, southern forest experiment station. eisfeld, r.l., sanquetta, c.r., arce, j.e., maestri, r., & weber, k.s. (2005). growth and yield modelling of pinus taeda l. using probability function. floresta, 35(2), 317-328. https://doi.org/10.5380/ rf.v35i2.4619 eslami, a.r., karimi, b., payam, h., & derakhshan, o.k. (2011). investigation of the structure and distribution diameter classes models in beech forests of northern iran. african journal of agricultural research, 6(10), 2157-2165. feduccia, d.p., dell, t.r., mann, w.f. jr., campbell, t.e., & polmer, b.h. (1979). yields of un-thinned loblolly pine plantations on cutover sites in the west gulf region. [res. pap. so-148] (94 p.). new orleans, la: u.s. department of agriculture, forest service, southern forest experiment station. goodwin, a.n. (2020). a blind spot in the use of the weibull function for modeling diameter distributions. forest science, 67(2), 125-134. https://doi.org/10.1093/forsci/fxaa042 hyink, d.m., & moser, j.w. (1983). a generalized framework for projecting forest yield and stand structure using diameter distributions. forest science, 29(1), 85-95. jiang, l., & brooks, j. r. (2009). predicting diameter distributions for young longleaf pine plantations in southwest georgia. southern journal of applied forestry, 33(1), 25-28 https://doi.org/10.1093/ sjaf/33.1.25 johnson, n.l., kotz, s., & balakrishnan, n. (1995). continuous univariate distributions, 2nd ed. wiley. 756 p. kershaw, j.a., jr., ducey, m.j., beers, t.w., & husch, b. (2016). forest mensuration, 5th ed. wiley/blackwell, hobokin, nj. https://doi. org/10.1002/9781118902028 kershaw, j.a., jr., & maguire, d.a. (1995). crown structure in western hemlock, douglas-fir, and grand fir in western washington: trends in branch-level mass and leaf area. canadian journal of forest research, 25(12), 1897-1912. https://doi.org/10.1139/x95206 kilkki, p., & paivinen, r. (1986). weibull function in the estimation of the basal area dbh-distribution. silva fennica, 20(2), 149-156. https://doi. org/10.14214/sf.a15449 kilkki, p., maltamo, m., mykkanen, r., & paivinen, r. (1989). use of the weibull function in estimating the basal area dbh-distribution. silva fennica, 23(4), 311-318. https://doi.org/10.14214/ sf.a15550 knowe, s.a. (1992). basal area and diameter distribution models for loblolly pine plantations with hardwood competition in the piedmont and upper coastal plain. southern journal of applied forestry, 16(2), 93-98. https://doi.org/10.1093/sjaf/16.2.93 knowe, s.a., foster, g.s., roussau, r.j., & nance, w.l. (1994). eastern cottonwood clonal mixing study: predicted diameter distributions. canadian journal of forest research, 24(2), 405-414. https://doi. org/10.1139/x94-054 knowe, s.a., harrington, t.b., & shula, r.g. (1992). incorporating the effects of interspecific competition and vegetation management treatments in diameter distribution models for douglas-fir saplings. canadian journal of forest research, 22(9), 1255-1262. https://doi. org/10.1139/x92-167 knowe, s.a., radosevich, s.r., & shula, r.g. (2005). basal area and diameter distribution prediction equations for young douglas-fir plantations with hardwood competition: coast ranges. western journal of applied forestry, 20(2), 73-93. https:// doi.org/10.1093/wjaf/20.2.77 knowe, s.a., & stein, w.i. (1995). predicting the effects of site preparation and protection on development of young douglas-fir plantations. canadian journal of forest research, 25(9), 1538-1547. https://doi. org/10.1139/x95-167 leduc, d.j., matney, t.g., belli, k.l., & baldwin, v.c., jr. (2001). predicting diameter distributions of longleaf pine plantations: a comparison between artificial neural networks and other accepted methodologies. [res. pap. srs-25] (24 p.). asheville, nc: u.s. department of agriculture, forest service, southern research station. https:// doi.org/10.2737/srs-rp-25 lee, y.j., & coble, d.w. (2006). a new diameter distribution model for unmanaged loblolly pine plantations in east texas. southern journal of applied forestry, 30(1), 13-20. https://doi. org/10.1093/sjaf/30.1.13 waldy et al. new zealand journal of forestry science (2022) 52:1 page 12 https://doi.org/10.1093/sjaf/32.2.89 https://doi.org/10.1590/s0104-77602010000100011 https://doi.org/10.1590/s0104-77602010000100011 https://doi.org/10.5380/rf.v35i2.4619 https://doi.org/10.5380/rf.v35i2.4619 https://doi.org/10.1093/forsci/fxaa042 https://doi.org/10.1093/sjaf/33.1.25 https://doi.org/10.1093/sjaf/33.1.25 https://doi.org/10.1002/9781118902028 https://doi.org/10.1002/9781118902028 https://doi.org/10.1139/x95-206 https://doi.org/10.1139/x95-206 https://doi.org/10.14214/sf.a15449 https://doi.org/10.14214/sf.a15449 https://doi.org/10.14214/sf.a15550 https://doi.org/10.14214/sf.a15550 https://doi.org/10.1093/sjaf/16.2.93 https://doi.org/10.1139/x94-054 https://doi.org/10.1139/x94-054 https://doi.org/10.1139/x92-167 https://doi.org/10.1139/x92-167 https://doi.org/10.1093/wjaf/20.2.77 https://doi.org/10.1093/wjaf/20.2.77 https://doi.org/10.1139/x95-167 https://doi.org/10.1139/x95-167 https://doi.org/10.2737/srs-rp-25 https://doi.org/10.2737/srs-rp-25 https://doi.org/10.1093/sjaf/30.1.13 https://doi.org/10.1093/sjaf/30.1.13 lindsay, s.r., wood, g.r, & woollons, r.c. (1996). stand table modelling through the weibull distribution and usage of skewness information. forest ecology and management, 81(1-3), 19-23. https://doi. org/10.1016/0378-1127(95)03669-5 little, s.n. (1983). weibull diameter distributions for mixed stands of western conifers. canadian journal of forest research, 13(1), 85-88. https:// doi.org/10.1139/x83-012 lohrey, r.e., & bailey, r.l. (1976). yield tables and stand structure for un-thinned longleaf pine plantations in louisiana and texas. [res. pap. so-133] (53 p). new orleans, la: u.s. department of agriculture, forest service, southern forest experiment station. magnussen, s. (1986). diameter distributions in picea abies described by the weibull model. scandinavian journal of forest research, 1(1-4), 493-502. https:// doi.org/10.1080/02827588609382440 massey, f.g.j.r. (1951). the kolmogorov-smirnov test for goodness of fit. journal of the american statistical association, 46, 68-78. https://doi.org/10.1080/0 1621459.1951.10500769 matney, t.g., & farrar, r.m. (1992). a thinned/un-thinned loblolly pine growth and yield simulator for planted cutover site-prepared land in the mid-gulf south. southern journal of applied forestry, 16(2), 70-75. https://doi.org/10.1093/sjaf/16.2.70 matney, t.g., & sullivan, a.d. (1982). compatible stand and stock tables for thinned and unthinned loblolly pine stands. forest science, 28(1), 161-171. mctague, j.p., & bailey, r.l. (1987). compatible basal area and diameter distribution models for thinned loblolly pine plantations in santa catarina, brazil. forest science, 33(1), 43-51. nanang, d.m. (1998). suitability of the normal, lognormal and weibull distributions for fitting diameter distributions of neem plantations in northern ghana. forest ecology and management, 103(1), 1-7 https://doi.org/10.1016/s03781127(97)00155-2 newton, p.f., lei, y., & zhang, s.y. (2004). a parameter recovery model for estimating black spruce diameter distribution within the context of a stand density management diagram. the forestry chronicle, 80(3), 349-358. https://doi. org/10.5558/tfc80349-3 newton, p.f., lei, y., & zhang, s.y. (2005). stand-level diameter distribution yield model for black spruce plantations. forest ecology and management, 209, 181-192. https://doi.org/10.1016/j. foreco.2005.01.020 pinheiro, j.c., & bates, d.m. (2000). mixed effects models in s and s-plus. new york: springer. 528 p. https:// doi.org/10.1007/978-1-4419-0318-1 poudel, k.p., & cao, q.v. (2013). evaluation of methods to predict weibull parameters for characterizing diameter distributions. forest science, 59(2), 243252. https://doi.org/10.5849/forsci.12-001 r core team. (2020). r: a language and environment for statistical computing. r foundation for statistical computing. https://www.r-project.org/ (accessed on 14 sept 2020). rennolls, k., geary, d.n., & rollinson, t.j.d. (1985). characterizing diameter distributions by the use of weibull distribution. forestry: an international journal of forest research, 58(1), 57-66 https:// doi.org/10.1093/forestry/58.1.57 reynolds, m.r., burk, t.e., & huang, w.c. (1988). goodness of-fit tests and model selection procedures for diameter distribution models. forest science, 34(2), 373-399. royle, j.a., & dorazio, r.m. (2008). hierarchical modelling and inference in ecology: the analysis of data from populations, metapopulations and communities. san diego, ca: academic press. 444 p saborowski, j. (1994). minimum sample size for estimating the weibull parameters of diameter distributions. proceedings of iufro symposium. stellenbosch, pp 1-7 schreuder, h.t., hafley, w.l., & bennet, f.a. (1979). yield prediction for un-thinned natural slash pine stands. forest science, 25(1), 25-30. schreuder, h.t., & swank, w.t. (1974). coniferous stands characterized with the weibull distribution. canadian journal of forest research, 4(4), 518-523. https://doi.org/10.1139/x74-075 sghaier, t., cañellas, i., calama, r., & sánchez-gonzález, m. (2016). modelling diameter distribution of tetraclinis articulata in tunisia using normal and weibull distributions with parameters depending on stand variables. iforest biogeosciences and forestry, 9, 702 -709. https://doi.org/10.3832/ ifor1688-008 shifley, s.r., & lentz e. (1985). quick estimation of the three-parameter weibull to describe tree size distributions. forest ecology and management, 12, 195-203. https://doi.org/10.1016/03781127(85)90034-9 shiver, r.d. (1988). sample size and estimation methods for the weibull distribution in un-thinned and thinned pine plantation diameter distributions. forest science, 24(3), 809-814. siipilehto, j. (1999). improving the accuracy of predicted basal-area diameter distribution in advanced stands by determining stem number. silva fennica, 33(4), 281-301. https://doi.org/10.14214/sf.650 siipilehto, j., & mehtätalo, l. (2013). parameter recovery vs. parameter prediction for the weibull distribution validated for scots pine stands in waldy et al. new zealand journal of forestry science (2022) 52:1 page 13 https://doi.org/10.1016/0378-1127(95)03669-5 https://doi.org/10.1016/0378-1127(95)03669-5 https://doi.org/10.1139/x83-012 https://doi.org/10.1139/x83-012 https://doi.org/10.1080/02827588609382440 https://doi.org/10.1080/02827588609382440 https://doi.org/10.1080/01621459.1951.10500769 https://doi.org/10.1080/01621459.1951.10500769 https://doi.org/10.1093/sjaf/16.2.70 https://doi.org/10.1016/s0378-1127(97)00155-2 https://doi.org/10.1016/s0378-1127(97)00155-2 https://doi.org/10.5558/tfc80349-3 https://doi.org/10.5558/tfc80349-3 https://doi.org/10.1016/j.foreco.2005.01.020 https://doi.org/10.1016/j.foreco.2005.01.020 https://doi.org/10.1007/978-1-4419-0318-1 https://doi.org/10.1007/978-1-4419-0318-1 https://doi.org/10.5849/forsci.12-001 https://doi.org/10.1093/forestry/58.1.57 https://doi.org/10.1093/forestry/58.1.57 https://doi.org/10.1139/x74-075 https://doi.org/10.3832/ifor1688-008 https://doi.org/10.3832/ifor1688-008 https://doi.org/10.1016/0378-1127(85)90034-9 https://doi.org/10.1016/0378-1127(85)90034-9 https://doi.org/10.14214/sf.650 finland. silva fennica. 47(4), 1-22. smalley, g.w., & bailey, r.l. (1974a). yield tables and stand structure for loblolly pine plantations in tennessee, alabama, and georgia highlands. [res. pap. so-96]. (81 p.). new orleans, la: u.s. department of agriculture, forest service, southern forest experiment station. smalley, g.w., & bailey, r.l. (1974b). yield tables and stand structure for shortleaf pine plantations in tennessee, alabama, and georgia highlands. [res. pap. so-97]. (57 p.). new orleans, la: u.s. department of agriculture, forest service, southern forest experiment station. strub, m.r., & burkhart, h.e. (1975). a class-intervalfree method for obtaining expected yields from diameter distributions. forest science, 21(1), 6769. ueno, y., & osawa, y. (1987). the applicability of the weibull and the expanded weibull distributions. journal of the japanese forest society, 69(1), 24-28. van laar, a., & akça, a. (2007). forest mensuration. 2nd edition. new york: springer https://doi. org/10.1007/978-1-4020-5991-9 vanclay, j.k. (1994). modelling forest growth and yield: applications to mixed tropical forests. cab international. waldy, j., kershaw, j.a., jr., weiskittel, a., & ducey, m.j. 2021. comparison of time-based versus statespace stand growth models for tropical hybrid eucalyptus clonal plantations in sumatera, indonesia. canadian journal of forest research, 51(8), 1178-1187 https://doi.org/10.1139/cjfr2020-0499 weiskittel, a.r., hann, d.w., kershaw, j.a., & vanclay, j.k. (2011). forest growth and yield modeling. hoboken, nj: wiley. https://doi. org/10.1002/9781119998518 zarnoch, s.j., & dell, t.r. (1985). an evaluation of percentile and maximum likelihood estimators of weibull parameters. forest science, 31(1), 260268. zarnoch, s.j., feduccia, d.p., baldwin, v.c., jr., & dell, t.r. (1991). growth and yield predictions for thinned and un-thinned slash pine plantations on cutover sites in the west gulf region. [res. pap. so-264]. (32 p.). new orleans, la, usa: us department of agriculture, forest service, southern forest experiment station. https://doi.org/10.2737/sorp-264 waldy et al. new zealand journal of forestry science (2022) 52:1 page 14 https://doi.org/10.1007/978-1-4020-5991-9 https://doi.org/10.1007/978-1-4020-5991-9 https://doi.org/10.1139/cjfr-2020-0499 https://doi.org/10.1139/cjfr-2020-0499 https://doi.org/10.1002/9781119998518 https://doi.org/10.1002/9781119998518 https://doi.org/10.2737/so-rp-264 https://doi.org/10.2737/so-rp-264 new zealand journal of forestry science panagiotidis et al. new zealand journal of forestry science (2019) 49:2 https://doi.org/10.33494/nzjfs492019x26x e-issn: 1179-5395 published on-line: 04/03/2019 detection of fallen logs from high-resolution uav images d. panagiotidis*, azadeh abdollahnejad, peter surový, karel kuželka department of forest management, faculty of forestry and wood sciences, czech university of life sciences (culs), kamýcká 129, prague 165 21, czech republic *corresponding author: panagiotidis@fld.czu.cz (received for publication 1 october 2018; accepted in revised form 15 february 2019) abstract background: high-resolution images from unmanned aerial vehicles (uavs) can be used to describe the state of forests at regular time periods in a cost-effective manner. the purpose of this study was to assess the performance of a line template matching algorithm, the hough transformation, for detecting fallen logs from uav-based high-resolution rgb images. the suggested methodology does not aim to replace any known aerial method for log detection, rather it is more oriented to the detection of fallen logs in open forest stands with a high percentage of log visibility and straightness. methods: this study describes a line template matching algorithm that can be used for the detection of fallen logs in an automated process. the detection technique was based on object-based image analysis, using both pixel-based and shape descriptors. to determine the actual number of fallen logs, and to compare with the ones predicted by the algorithm, manual visual assessment was used based on six high-resolution orthorectified images. to evaluate if a line matched, we used a voting scheme. the total number of detected fallen logs compared with the actual number of fallen logs based on several accuracy metrics. to evaluate predictive models, we tested the cross-validation mean error. finally, to test how close our results were to chance, we used the cohen`s kappa coefficient. results: the detection algorithm found 136 linear objects, of which 92 of them were detected as fallen logs. from the 92 detected fallen logs, 86 were correctly predicted by the algorithm and 24 were falsely detected as non-fallen* logs. the calculated amount of observed agreement (po) was equal to 0.78, whereas the expected agreement by chance (pe) was 0.61. finally, the kappa statistic was 0.44. conclusions: our methodology had high reliability for detecting fallen logs based on total user‘s accuracy (94.9%), whereas a kappa of 0.44 indicated there was good agreement between the observed and predicted values. also, the crossvalidation analysis denoted the efficiency of the proposed method with an average error of 16%. keywords: unmanned aerial vehicle; forest; windthrow; computer vision; pattern recognition; hough transformation algorithm © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access introduction wind damage is a serious threat to managed forests because it can reduce landscape quality, timber yields, and wildlife habitat. windthrow is one of the most significant disturbances in norway spruce (picea abies (l.) h.karst.) forests in central europe, mainly because the root system is sensitive to any dislocation of its primary taproot (puche 2003). damage can range from small-scale gaps in forests to catastrophic landscape disturbances (kuuluvainen 1994), and it varies both spatially and temporally. according to the czech national forest inventory *author correction 20/09/2019 panagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 2 (nfi) from 2001 to 2004, norway spruce comprised 47.7% of forest cover area (úhúl 2007); it is the most widespread tree species within the czech republic. norway spruce has fast growth and is relatively easy to manage, thus many forest owners have favored spruce monocultures, and this has greatly altered the natural forest ecosystems of central europe. over the past century, the rapid change in the ratio of autochthonous to non-autochthonous species composition (böhm 1981) and the widespread establishment of unstable spruce monocultures has gradually led to the loss of silvicultural flexibility, and many areas have become more susceptible to biotic and abiotic damages (klimo et al. 2000). in addition, wind damage to forest stands is expected to increase in the future because of the risks from extreme and unpredictable weather conditions related to climate change (seidl et al. 2017). damage assessment after windthrow disturbances is an important component of forest management and ecological monitoring. fallen logs are typically assessed by field surveys (ståhl et al. 2001), but they present some unique challenges. field surveys for log detection are usually time-consuming and labor intensive (bütler & schlaepfer 2004); for example, it is expensive to assess the spatial distribution of fallen logs over a wide area. also, because fallen logs are typically on or near the ground, they can be obscured by understory vegetation, moss, or litter, which can adversely affect data integrity (rondeux & sanchez 2010). inaccessible forests and areas with restrictions, such as national parks, also pose additional challenges. one method to reduce costs and provide practical solutions for forest applications is to use remote sensing technologies, such as airborne and satellite imagery (e.g. bütler & schlaepfer 2004; pasher & king 2009). currently, there are three extraction methods from remote sensing techniques: (a) visual interpretation; (b) individual windthrown tree extraction; and (c) area extraction. however, all three methods require: (i) interpretation consistency; and (ii) trained and experienced staff. remote visual interpretation is a relatively simple method that has high accuracy. fransson et al. (2007) proposed a method based on visual interpretation over simulated windthrown forests, at both stand and individual tree levels, using synthetic aperture radar (sar) images from the swedish airborne carabasii and lora systems after a severe storm. wang et al. (2010) studied the detection of fallen logs by comparing before and after images of a hurricane disaster using standard aerial images (altitude of 4–5 km) and satellite images. in recent years though, many authors have proposed automated approaches for the extraction of windthrown trees. szantoi et al. (2012) used a sobel method for edge detection combined with spectral information based on color filtering. for detecting the fallen logs, they considered 15 statistical combinations of spectral bands using high-resolution digital photos obtained with a leica airborne digital sensor (ads40); they used manned aircraft at two different flying heights (1,500 and 3,000 m). although they could identify areas of high tree densities, it was difficult to achieve extraction of individual windthrown trees from standard aerial photographs and satellite images due to obstruction and resolution issues. unmanned aerial vehicles (uavs) deployed with light detection and ranging (lidar) sensors is a modern automated approach to collect data for the construction of 3d forest maps. the ability of the laser to penetrate the canopy allows for the creation of a digital terrain model (dtm), even in the case of dense forests (maguya et al. 2014), and it can define geometric objects (i.e. linear and circular) located near the ground based on an area extraction approach (bailly et al. 2008). unlike other aerial surveys, lidar was an effective tool for individual windthrown tree extraction because it is not subject to the issue of canopy obstruction (tran et al. 2015; niemi & vauhkonen 2016). blanchard et al. (2011) tried to extract individual fallen logs using an objectoriented image analysis approach with airborne laser scanner (als) data. their study showed a completeness of 73%; however, because of the over-division, they were not able to achieve complete delineation of fallen trees (blanchard et al. 2011). even though laser scanner-based systems produce high-quality data, they are still quite expensive. lindberg et al. (2013) and hu & yuan (2016) used als data to perform binary classifications, based on height characteristics to eliminate the interference of foreign objects under a closed canopy. however, similar to the object height model (ohm), this approach still suffered from interference problems caused by objects with shapes similar to fallen trees. both techniques were based on the template matching method, with a reported correctness of 32% and 38% at the individual tree level. mücke et al. (2013) provided an area extraction method using full-waveform als data for the extraction of fallen logs based on area-perimeter ratios and ohm; they achieved a log completeness of 75.6% with an accuracy of 89.9%. in another study, mokroš et al. (2017) tried to combine unmanned aerial system (uas) imagery with als, to determine the size of windthrown areas within forest stands. their results showed that the windthrown areas were successfully identified using both the uas and uas/als techniques, and they performed substantially better than landsat. although, the results from uas overestimated the volume provided by the field measurements, the difference between them was only 4.93%. consumer-grade uavs equipped with imaging sensors represent a more efficient tool to acquire highly detailed and spatially continuous 3d data, and they provide a low-cost alternative, which is a core challenge of forest attribute data acquisition at the individual tree level (panagiotidis et al. 2017; abdollahnejad et al. 2018; surový et al. 2018). uavs, in combination with photogrammetric methods, have also been used for pre-harvest and disturbance assessments. because uav images are typically of high-resolution, they also represent new potential to recognize fallen logs (inoue et al. 2014), and detect coniferous seedlings (feduck et al. 2018) and tree stumps (puliti et al. 2018). however, only one study has exclusively focused on the prediction of fallen logs at the individual tree level based on uav imagery (duan et al. 2017). although they achieved acceptable extraction accuracy, the flying height was at 500 m above the ground. in general, increasing the altitude will increase the image overlap, but the spatial resolution (as a function of flying height) will be lower. nevertheless, whether using laser or camera sensors, the analyses and processing for fallen log detection in forested areas rely on similar approaches, which may be conducted using different methods, such as rasterization of point clouds using a random sample consensus (ransac) algorithm, image processing (tittmann et al. 2011), and object-based image analysis (blanchard et al. 2011). alternatively, we propose the hough transformation algorithm (hough 1962), which is a robust feature extraction technique used in image analysis and computer vision; it can be directly applied to a binary image to detect multiple objects of interest after “pixel filtering” (duan et al. 2017). the principle of the algorithm lies in the image transformation and the use of non-maxima suppression to locate and distinguish peaks in hough images. such post-processing might require tuning of extra parameters, especially in the case of higher image complexity when objects of interest are difficult to distinguish, either because they are close to each other or because they are obscured by noisy pixels. the hough transformation has several advantages: (a) it is robust to partial or slightly deformed shapes; (b) it is tolerant of noise; (c) it can find multiple occurrences of a shape during the same processing pass; and (d) it is robust to the presence of additional structures in the image (hough 1962). its greatest strength lies in specialised vision, such as manufacturing quality control and analysis of aerial photographs (antolovic 2008). unfortunately, the computational load increases rapidly with the greater number of parameters that define the detected shape; for example, lines have two parameters, circles three, and so on. consequently, the hough algorithm can be used to detect any linear, circular, and ellipsoid objects, such as fallen logs, tree rings (aschoff & spiecker 2004), and stem cross-sections (olofsson et al. 2014). the aim of this study was to examine the feasibility and performance of a line template matching algorithm (hough transformation) for fallen log detection from uav-based high-resolution rgb images under open canopy conditions. to evaluate the accuracy of the method, several accuracy metrics were used, including the total number of detected fallen logs compared with the actual number of fallen logs, and we tested the crossvalidation mean error. methods the study area (fig. 1) is located in west bohemia, and it is largely surrounded by dense forests. the altitude is approximately 600 m a.s.l. and the topography is generally flat. we selected six randomly distributed experimental plots (50 × 50 m) with open canopy cover as shown in additional file: figure s1. the most common tree species were norway spruce and scots pine (pinus sylvestris l.), with scattered individuals of deciduous species, such as norway maple (acer platanoeides l.) and birch (betula spp.); norway spruce was significantly more abundant than the other species. the forested area extends geographically from 50°1104200n, 13°1601200e to 50°0903100n, 13°1303000e, and we used wgs84 as the coordinate reference system (fig. 1). image acquisition a commercial uav quadcopter dji mavic pro (dá-jiāng innovations science and technology co. ltd., shenzhen, china) with 12 mp resolution was used for the imagery data collection; it was powered by the compact onboard gimbal, which ensured crisper and cleaner image acquisition. the copter was guided with the help of dji go version v. 4.1.9 software (dá-jiāng innovations science and technology co. ltd., shenzhen, china) with a planned route at a height of 60 m at a speed of 2 m/s. the camera was set to automatic mode with a time lapse of 3 s. the route was uploaded to the driving unit of the copter using an ipad tablet (apple inc., california, united states), though for security reasons, the takeoff and landing of the copter were guided by manual radio control. the flight trajectory followed a double zig-zag pattern, which consists of two zig-zag flights in perpendicular directions. overlap of photos at the front was 80% and to the sides it was 70%. the overlaps were calculated using two applications, including the pix4dcapture v. 4.2.0 software (pix4d s.a, lausanne, switzerland), which allows automatic picture taking at a predefined distance, in our case it was 10 metres, and the dji go with a builtpanagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 3 figure 1: study area in the czech republic. source: http:// download.geofabrik.de/europe/czech-republic. html for a more detailed map (arcmap v.10.1). in intervalometer, which takes a photo every 5 s, and based on the selected travel speed (2 m/s), the distance would be 10 or 15 metres. given the approximate size of the photo at ground level, this would correspond to 70 to 80% overlap. the quadcopter needed approximately 6 to 7 min to complete a flight based on the predefined parameters (i.e. number of waypoints and flight speed). we performed six flights in total, with one flight per plot using the same flight parameters. the total number of images acquired per flight was approximately 250 to 300. the flights were conducted on november 11, 2017 at 11:00 – 12:00 local time, while the weather conditions were favorable for photographing. data processing collected imagery data were processed in photoscan professional v. 1.4.2 software (agisoft lcc, st. petersburg, russia). to increase the accuracy of the alignment process and to generate high-resolution orthomosaics (2.68 cm/pixel) from the 3d point clouds, accuracy was set to high quality. to geo-referencing the images, they were orthorectified based on their gps (dji mavic pro) coordinates using a built-in rectification algorithm in photoscan. no ground control points (gcps) were used to improve the final positional accuracy of the models (tomaštík et al. 2017; rangel et al. 2018) due to the inability to access the plots. manual visual assessment to determine the actual number of fallen logs used for the reference data, we performed a manual visual assessment to identify individual fallen logs from the digital orthophotos; we marked the beginning and end of each fallen log in each plot using adobe photoshop v. 19.1.1 software (adobe systems incorporated, california, united states). the total number of fallen logs was 110 across all six plots. pre-processing of raw image data usually, fallen logs display elongate geometry and a bright spectrum due to lateral optical scattering, which makes them observably different from other objects. the technique used for the interpretation of uav images to identify fallen logs was based on object-based image analysis using both spectral and shape characteristics (jones & purves 2008). because of the ambiguity problem of the range measurement and the degree of forest structural complexity in some of the plots, preliminary work was done to improve and prepare the images for further processing. the ‘imread’ function was initially used to read the image, then the rgb image was converted to binary using the ‘imbinarize’ function. to better identify line patterns in each plot, an image thresholding was applied based on the reflectance of defined spectral log values; however, we did encounter a few problems in this phase. the edges of the objects were often comprised of noisy pixels due to extreme changes in pixel intensity compared to the neighboring pixels, and also, in some cases, slender logs had partial interference from living trees, such as branches and leaves. to overcome these issues and to produce smoother line patterns, we used a step filtering process that included different types of mathematical morphological operations. we used the ‘lagmatrix’ for smoothing the line patterns and the ‘bwareaopen’ tool was used to remove small objects with an area smaller than a particular number of pixels; we also used ‘bwareafilt’ to retain only those pixels contained within the largest areas, which was based on the number of pixels. to address the problem of slender logs, we used additional functions such as ‘imdilate’ to dilate the lines and ‘imclose’ to perform a close operation based on a line structuring element of the number of desired neighboring pixels. the parameters used for each of the above morphological tools are described in additional file: table s1. as a final step, edge detection (ziou & tabbone 1998) using the canny approach (canny 1986) was applied to determine the boundaries of line objects (fig. 2). the canny approach was preferred because it ensures optimal detection performance of linear objects and it dramatically reduces the amount of data to be processed. the canny edge detection process has the following attributes: (i) application of a gaussian filter for smoothing the images (noise removal); (ii) finding the intensity gradients in each image; (iii) application of non-maximum suppression to get rid of spurious responses to edge detection; (iv) application of a double threshold to determine potential edges; and (v) it finalises the detection of edges by suppressing all the other weak edges that are not connected to strong edges. the whole pre-processing phase was conducted in matlabr2017b professional edition (mathworks©, inc., massachusetts, united states). an overview of the workflow can be seen in fig. 3. panagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 4 figure 2: illustration of log segmentation using the edge detection technique (plot 1) principle of the hough algorithm fallen logs have evident linear characteristics, therefore hough transformation was adopted to extract the fallen logs from the binary images (hough 1962). for this purpose, we used the following mathematical model (equation 1), as suggested by duda & hart (1972): ρ = s1∗ cos(θ) + s2 ∗ sin(θ) (1) where the parameter ρ is the distance from the origin to the closest point on the straight line and θ is the angle between the x-axis and the line connecting the origin with the closest point (fig. 4b). the hough transformation algorithm (ye et al. 2015; mukhopadhyay & chaudhuri 2015) was used to detect all the line patterns in each image based on the decision from the edge detector, and it represented them in the form of a 2d-array, named parameter space. the parameter space is a graphical representation of an image, as evident in the example of fig. 4, that was used to better illustrate the entire concept of the hough algorithm. the use of the hough algorithm for log detection is quite simple and it is based on all the points that the edge detector indicated as an edge in the binary images. note that every point that the edge detector found, voted for the possibilities of having a line. for example, from point p1 (fig. 4a), a defined number of lines (theoretically infinite) will pass through it and each line will have a particular distance (ρ) and a particular angle (θ; fig. 4a). the process will continue for another line that will pass from the same point and which will also have its own (ρ, θ). consequently, any line that will pass through that point, will eventually get one vote (principle of voting scheme). following the exact same process for another point, p2, (fig. 4a, b), we will get one more vote, and so on. however, there will be a line (fig. 4a, b) where both points p1 and p2 are located and which will get two votes, as a common line between the two points. the process of voting continues for the rest of the points until it identifies the rest of the lines of interest in the binary images. log detection the rule for log detection is that when there are multiple points on the same line, it will receive multiple votes. to calculate the total number of these multiple votes, we created the accumulator, which is a new transformed coordinate system array (ρ, θ) (fig. 4c, d). for every point (e.g. p1, p2 etc.) we used their coordinates and for different (θ) values in equation 1, and we were able to define the position of (ρ) inside the accumulator. consequently, each point formed its own sinusoidal curve in the parameterised space. as long as the process continued, sinusoidal curves started to accumulate; when the process was finished, we looked for the maxima (highest number of votes), which indicated the line we were looking for (single fallen log). finally, to find and extract the entire number of fallen logs in the image, the ‘houglines’ function was applied using a loop method. every time the loop was running, panagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 5 figure 3: flow diagram of proposed automated fallen log detection process figure 3: flow diagram of proposed automated fallen log detection process figure 4: example of points mapping p1 and p2 from cartesian space (a); to the slope-intercept parameter space (c); mapping p1 and p2 from cartesian space (b); and to the (ρ, θ) parameter space (d). source: liu et al. (2017). it identified and plotted a single line. the houghline property contains the coordinates of the points (p1 (s1,s2), p2 (s1,s2)) in the form of point (1) and point (2), which represent the beginning and end of each line or log. the entire process repeated until it reached the desired number of points based on the ‘houghpeaks’ function, and the output of this process was then stored in a 2 × 2 matrix. for the entire analysis (pre-processing and processing) we have created two separate .mat files using the econometrics toolbox in matlabr2017b. evaluation of accuracy and error assessment the accuracy of the detected and actual fallen logs was evaluated for each plot using the following accuracy metrics: overall accuracy (%) = cd/mva * 100 (2) where cd stands for correct detected and mva represents the manual visual assessment of the number of fallen logs. user's accuracy (%) = cd/dfl * 100 (3) where dfl stands for detected as fallen log. because all six plots had relatively low structural complexity (i.e., open-canopy, few piled logs, and low percentage of ground cover), it was possible to visually investigate the user‘s accuracy based on the digital orthophotos. commission error (%) = 100 − user's accuracy (%) (4) due to the small number of sample plots, it was not possible to divide them into train and test groups. also, the division of classes based on individual trees and random selection of them as train and test groups by the algorithm cannot guarantee the independence of train and test samples because the algorithm can randomly select train and test samples from the same plot. therefore, we used crossvalidation technique to evaluate predictive models based on the leave-one-out cross-validation type as follows: cvn= 1/n ((yi− ŷi)/(1 − hi )) 2 (5) where cv stands for cross-validation, n is the number of samples, yi is the observed quantitative output, ŷi is the predicted values, and hi is leverage. additionally, to test how far or close to chance the result was, we used the cohen's kappa coefficient as follows: k=(po − pe )/(1 − pe ) (6) where po is the observed agreement and pe is the expected agreement by chance. results to test the detection performance and demonstrate some of the strengths and weaknesses of the hough algorithm, we sampled plots with different spatial patterns of fallen log distributions. to visualise the results in each plot, the binary images were used as background. binary images have the practical advantage of contrast, especially in images with increased complexity (e.g. forest stands) in small scale figures. in plot 1 the algorithm detected 15 of the 22 fallen logs (only 16 of the fallen logs could be displayed in the binary image); however, the correct number of detected fallen logs was 15. the overall accuracy was 68.2%, user‘s accuracy was 100%, and the commission error was 0% (fig. 5a; table 1). for plot 2, all 10 of the 10 fallen logs were automatically detected, thus resulting in an overall accuracy of 100%, user‘s accuracy of 100%, and a commission error of 0% (fig. 5b; table 1). plot 3 had the greatest number of fallen logs. we recorded a total of 41 fallen logs, of which 30 were automatically detected. however, the correct number of detected fallen logs was 28; the algorithm produced additional lines for two different fallen logs. the estimated overall accuracy for this plot was 68.3%, user‘s accuracy was 93.3%, and the commission error was 6.7% (fig. 6; table 1). panagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 6 ∑ 1 n=1 figure 5: white lines represent the fallen logs and green lines represent the detected fallen logs in (a) plot 1; and (b) in plot 2 plot 4 also had 100% accuracy, as the algorithm detected all five of the five (actual fallen logs) fallen logs, thus resulting in an overall accuracy of 100%, user's accuracy of 100%, and a commission error of 0% (table 1; fig. 7a). in plot 6, the actual number of fallen logs was equal to 22, however, the algorithm detected 23 fallen logs. nevertheless, the correct number of detected fallen logs was 20 because the algorithm produced additional lines for three fallen logs, thus resulting in an overall accuracy of 90.9%, user‘s accuracy of 87%, and a commission error of 13% respectively (fig. 7b; table 1). finally, for plot 5, the total number of actual fallen logs was 10, of which nine fallen logs were automatically detected. similar to plots 3 and 6, the algorithm produced two additional lines from a single fallen log. as a result, the correct number of detected fallen logs was eight, thus resulting in an overall accuracy of 80%, user‘s accuracy of 88.9%, and a commission error of 11.1% (fig. 8; table 1). the cross-validation technique indicated that the applied methodology could detect fallen logs with a mean error of 16%. during the process, the algorithm found 136 linear objects, of which 92 of them were detected as fallen logs. from the 92 detected fallen logs, 86 were panagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 7 correctly predicted by the algorithm and 24 were falsely detected as non-fallen* logs (table 2). the calculated amount of observed agreement (po) was equal to 0.78, whereas the expected agreement by chance (pe) was 0.61. finally, the kappa statistic was 0.44. figure 6: white lines represent the fallen logs and green lines represent the detected fallen logs for plot 3 figure 7: white lines represent the fallen logs and green lines represent the detected fallen logs in: (a) plot 4; and (b) in plot 6 plot number mvaa detected as fallen loga correct detecteda overall accuracy (%) user’s accuracy (%) commission error (%) 1 22 15 15 68.2 100 0 2 10 10 10 100 100 0 3 41 30 28 68.3 93.3 6.7 4 5 5 5 100 100 0 5 10 9 8 80 88.9 11.1 6 22 23 20 90.9 87 13 total 110 92 86 84.6 94.9 5.1 table 1: statistical summary table for all six research plots including their total accuracies a manual visual assessment (mva), detected as fallen log, and correct detected represent the number of individual fallen logs. *author correction 20/09/2019 discussion different compositions of norway spruce and other species presented in this study, allowed us to using a lowcost uav, we tested the efficiency of the hough algorithm for the detection of fallen logs using high-resolution orthorectified images of forest sites dominated by norway spruces in the czech republic. the proposed algorithm exhibited good detection rates based on both spectral and shape descriptors; we were able to support the statistical results with visual image interpretation from the digital orthophotos (fig. 5-8). although we were unable to use gcps in the research plots, which might have improved the positional accuracy of the final models (orthomosaics), the geo-referencing process is generally not fully correlated with the accuracy of local coordinates (elatawneh et al. 2014). it might help improve the accuracy, but from a practical perspective, the gcps installation would be senseless for forest practitioners, especially in fallen log areas that are usually inaccessible for humans and the importance of quickly evaluating the damage is crucial. because fallen logs are generally defined as being > 2.5 cm in diameter (harmon et al. 1986), we assumed that the high spatial resolution of the aerial photos would allow us to detect even the smaller fallen logs. satisfactory detection was observed in the case of partially occluded fallen logs (plots 2, 6; fig. 5b, 7b), where the use of shape filtering functions made it possible to unify particular log sections. however, the detection was difficult in cases where: (a) the fallen logs were either too close to each other; either the perpendicular distance between two or more parallel fallen logs was too small or the lines were about figure 8: white lines represent the fallen logs and green lines represent the detected fallen logs for plot 5 table 2: confusion matrix to intersect at some point, thus forming acute angles, which produced duplicated lines (plots 3, 6, 5; fig. 6, 7b, 8). (b) the fallen logs were not clearly distinguishable, either as a result of occlusion (i.e., other logs, branches, leaves, etc.), or because of the shadowing effect from the neighboring standing trees, which attributed darker tones to some of the fallen logs, as in the case of plot 1 (fig. 5a) where six actual fallen logs did not appear. in other cases (plot 5; fig. 8), although we tried to give weight to the log values, we were unable to optimise the detection without having adverse effects on the detection of other linear objects. in fact, it was difficult to define an appropriate threshold value that would ensure that the occluded parts could be “filled” to form a uniform log without any intervention (i.e., merging) with the neighboring pixels as a part of another line of interest. (c) the fallen logs were slender; in general, a high percentage of log completeness is mainly associated with larger diameter trees (blanchard et al. 2011). in our case, some of the fallen logs appeared discreetly in the binary image due to their size (usually slender), which resulted in insufficient intensity of pixel values such as in the case of plot 3 (fig. 6), and we assumed that these fallen logs should be smaller than 30 cm in diameter at breast height (dbh). because of the difficulty in distinguishing them as complete linear objects, some of them remained undetectable by the algorithm. that is consistent with the findings of nyström et al. (2014), who reported only 43% completeness for fallen logs with dbh < 30 cm, and inoue et al. 2014, where they were able to identify approximately 80% to 90% of fallen logs that were > 30 cm in dbh, but they failed to distinguish many of the fallen logs that were narrower or shorter. another limitation was related to the percentage of log curvature; whenever there was a visible abrupt change in curvature along the log, as in the case of plots 2 and 4 (figs. 5b, 7a, respectively), where the algorithm was unable to continue its process and detect the rest of the fallen log. due to the high percentage of user‘s accuracy in all research plots (table 1), we concluded that the proposed methodology produced an output similar to the reference data, and it reliably detected individual fallen logs. in terms of the evaluation of accuracy of detecting fallen logs in different densities, the statistical analysis of accuracy metrics (table 1), and particularly the results from the commission error, revealed that the amount of error more likely depends on the tree positions/formation (e.g., piled logs, logs which are very close to each other, etc.), rather than the actual number of fallen logs within each plot, which is consistent with the results of lindberg et al. (2013). also, if we consider fig. 7a, we can clearly see that plot 4, with five fallen logs, had the same amount of error as plot 1 (fig. 5a), which had 22 fallen logs. similarly, plot 3 (fig. 6) had 41 fallen logs and plot 5 (fig. 8) had 10 fallen logs. the proposed technique can be a feasible extraction solution because it is capable of producing low commission errors (< 13%) across a relatively broad range of tree heights and sizes, similar to the findings of duan et al. 2017. the kappa statistic (0.44) suggested panagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 8 n = 136 predicted: yes predicted: no total actual: yes 86 24 110 actual: no 6 20 26 total 92 44 136 moderate agreement between the predicted and the actual values. finally, there was no need to proceed in any radiometric correction of the images because (a) of the low uav flight path altitude, (b) all sample plots were located on flat areas, and (c) there was almost negligible shadowing effects. however, special caution should be taken in similar studies that use optical cameras deployed on uavs due to the angular variation of reflection from objects. conclusion the main contribution of our study is the development and demonstration of the hough transformation algorithm for individual windthrown tree detection. based on our findings, we propose that this detection technique is more feasible in cases where: (a) the percentage of canopy cover is small (open canopy stands); preferably clear-cut areas and silvopastoral systems, where the percentage of log visibility is high; and (b) log form is straight (i.e. pure cylindrical or conical forms). although there were a few limitations, for example difficulties to obtain permission for access into the plots in order to collect field measurements (e.g., log diameters), which could help us better identify diameter limits that produced log incompleteness, we showed that the proposed methodology has high reliability for the detection of fallen logs based on total user‘s accuracy (94.9%), and the kappa statistic suggested that there was good agreement between the observed and predicted values. moreover, the cross-validation analysis denoted the efficiency of the proposed method with an average error of 16%. although the results showed good relation between the two approaches (logs detected by the algorithm versus manual approach), further research is needed to refine the accuracy of the proposed method. the largest noticeable problem in using the hough algorithm in our study was the influence of isolated pixels, which were solvable using particular shape filtering functions in some cases. additional files figure s1: orthophotos of the study site. table s1: all the parameters used for the detection of linear patterns during image pre-processing. ethics approval not applicable. consent for publication not applicable. availability of data please contact the primary author for further information. competing interests the authors declare that they have no competing interests. funding we acknowledge that this work was financially supported by (a) the project eva4.0 [grant no. cz.02.1. 01/0.0/0.0/16_019/0000803] of the faculty of forestry and wood sciences (ffws) from the czech university of life sciences (culs) in prague; (b) the ministry of agriculture of czech republic [grant no. qj1520187]. author’s contributions dp; aa; kk are currently researchers at the department of forest management in the czech university of life sciences (culs) in prague, czech republic. graduate of the forest sciences, ph.d. program, department of forest management, faculty of forestry and wood sciences, culs. ps is an associate professor in forest management with specification in remote sensing and head of the department of forest management of the faculty of forestry and wood sciences in culs prague, czech republic. acknowledgements we gratefully acknowledge project eva4.0 [grant no. cz.02.1.01/0.0/0.0/16_019/0000803] of the faculty of forestry and wood sciences (ffws) from the czech university of life sciences (culs) in prague and the ministry of agriculture of czech republic [grant no. qj1520187] for financial support to this study. references abdollahnejad, a., panagiotidis, d., surový, p., & ulbrichová, i. (2018). uav capability to detect and interpret solar radiation as a potential replacement method to hemispherical photography. remote sensing, 10(3), 423. antolovic, d. (2008). review of the hough transform method, with an implementation of the fast hough variant for line detection. indiana university. bloomington, usa: department of computer science. aschoff, t., & spiecker, h. (2004). algorithms for the automatic detection of trees in laser scanner data. in: proceedings of the isprs working group viii/2 “laser-scanners for forest and landscape assessment”, freiburg, germany, (pp. 71-75). bailly, j.s., lagacherie, p., millier, c., puech, c., & kosuth, p. (2008). agrarian landscapes linear features detection from lidar: application to artificial drainage networks. international journal of remote sensing, 29, 3489–3508. blanchard, s.d., jakubowski, m.k., & kelly, m. (2011). object-based image analysis of downed logs in disturbed forested landscapes using lidar. remote sensing, 3(11), 2420–2439. böhm, p. (1981). sturmschäden in schwaben von 1950 bis 1980. allgemeine forst und jagdzeitung, 36, 1380. bütler, r., & schlaepfer, r. (2004). spruce snag quantification by coupling colour infrared aerial photos and a gis. forest ecology and management, 195, 325–339. panagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 9 canny, j.f. (1986). a computational approach to edge detection. ieee transactions on pattern analysis and machine intelligence, 8(6), 679-698. duan, f., wan, y., & deng, l. (2017). a novel approach for coarse-to-fine windthrown tree extraction based on unmanned aerial vehicle images. remote sensing, 9(4), 306. duda, r.o., & hart, p.e. (1972). use of the hough transformation to detect lines and curves in pictures. communications of the acm, 15(1), 11-15. elatawneh, a., wallner, a., manakos, i., schneider, t., & knoke, t. (2014). forest cover database updates using multi-seasonal rapideye data—storm event assessment in the bavarian forest national park. forests, 5, 1284-1303. feduck, c., mcdermid, g.j., & castilla, g. (2018). detection of coniferous seedlings in uav imagery. forests 9(7): 432. fransson, j. e. s., magnusson, m., folkesson, k., & hallberg, b. (2007). mapping of wind-thrown forests using vhf/uhf sar images. in: proceedings of the 2007 ieee international geoscience and remote sensing symposium, barcelona, spain, (pp. 2350–2353). harmon, m.e., franklin, j.f., swanson, f.j., sollins, p., & gregory, s.v. (1986). ecology of coarse woody debris in temperate ecosystems. advances in ecological research, 15, 133–302. hough, p.v.c. (1962). method and means for recognizing complex patterns. u.s. patent 3,069,654. hu, x., & yuan, y. (2016). deep-learning-based classification for dtm extraction from als point cloud. remote sensing, 8(9): 730. inoue, t., nagai, s., yamashita, s., fadaei, h., ishii, r., okabe, k., taki, h., honda, y., kajiwara, k., & suzuki, r. (2014). unmanned aerial survey of fallen trees in a deciduous broadleaved forest in eastern japan. plos one, 9(10), e109881. jones, c. b., & purves, r. s. (2008). geographical information retrieval. international journal of geographical information science, 22(3), 219–228. klimo, e., hager, h., & kulhavý, j. (eds.) (2000). spruce monocultures in central europe–problems and prospects. european forest institute proceedings, 33, 1–208. kuuluvainen, t. (1994). gap disturbance, ground microtopography, and the regeneration dynamics of boreal coniferous forests in finland: a review. annales zoologici fennici, 31(1), 35–51. lindberg, e., hollaus, m., mücke, w., fransson, j.e.s., & pfeifer, n. (2013). detection of lying tree stems from airborne laser scanning data using a line template matching algorithm. isprs annals of the photogrammetry remote sensing and spatial information sciences, ii-5/w2, 169–174. liu, w., zhang, z., li, s., & tao, d. (2017). "road detection by using a generalized hough transform", remote sensing, 9(6): 590. maguya, a s., junttila, v., & kauranne, t. (2014). algorithm for extracting digital terrain models under forest canopy from airborne lidar data. remote sensing, 6(7), 6524–6548. mokroš, m., výbošťok, j., merganič, j., hollaus, m., barton, i., koreň, m., tomaštík, j., & čerňava, j. (2017). early stage forest windthrow estimation based on unmanned aircraft system imagery. forests, 8(9): 306. mücke, w., deak, b., schroi, a., hollaus, m., & pfeifer, n. (2013). detection of fallen trees in forested areas using small footprint airborne laser scanning data. canadian journal of remote sensing, 39, s32–s40. mukhopadhyay, p., & chaudhuri, b.b. (2015). a survey of hough transform. pattern recognition, 48(3), 993–1010. niemi, m.t., & vauhkonen, j. (2016). extracting canopy surface texture from airborne laser scanning data for the supervised and unsupervised prediction of area-based forest characteristics. remote sensing, 8(7), 582. nyström, m., holmgren, j., fransson, j.e., & olsson, h. (2014). detection of windthrown trees using airborne laser scanning. international journal of applied earth observation, 30, 21–29. olofsson, k., holmgren, j., & olsson, h. (2014). tree stem and height measurements using terrestrial laser scanning and the ransac algorithm. remote sensing, 6(5), 4323–4344. panagiotidis, d., abdollahnejad, a., surový, p., & chiteculo, v. (2017). determining tree height and crown diameter from high-resolution uav imagery. international journal of remote sensing, 38(8-10), 2392–2410. pasher, j., & king, d.j. (2009). mapping dead wood distribution in a temperate hardwood forest using high resolution airborne imagery. forest ecology and management, 258, 1536–1548. puche, j. (2003). growth and development of the root system of norway spruce (picea abies) in forest stands – a review. forest ecology and management, 175, 253–273. puliti, s., talbot, b., & astrup, r. (2018). tree-stump detection, segmentation, classification, and measurement using unmanned aerial vehicle (uav) imagery. forests, 9(3): 102. rangel, j.m.g., gonçalves, g.r., & pérez, j.a. (2018). the impact of number and spatial distribution of gcps on the positional accuracy of geospatial products derived from low-cost uass. international journal of remote sensing, 39, 7154-7171. rondeux, j., & sanchez, c. (2010). review of indicators and field methods for monitoring biodiversity within national forest inventories. core variable: deadwood. environmental monitoring and assessment, 164, 617–630. seidl, r., thom, d., kautz, m., martin-benito, d., peltoniemi, m., vacchiano, g., wild, j., ascoli, d., petr, m., & honkaniemi, j. (2017). forest disturbances under climate change. nature climate change, 7, 395–402. ståhl, g., ringvall, a., & fridman, j. (2001). assessment of coarse woody debris: a methodological overview. ecological bulletins, 49, 57–70. surový, p., ribeiro, n. a., & panagiotidis, d. (2018). panagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 10 estimation of positions and heights from uav-sensed imagery in tree plantations in agrosilvopastoral systems. international journal of remote sensing, 39(14), 4786–4800. szantoi, z., malone, s., escobedo, f., misas, o., smith, s., & dewitt, b. (2012). a tool for rapid post-hurricane urban tree debris estimates using high resolution aerial imagery. international journal of applied earth observation and geoinformation, 18, 548– 556. tittmann, p., shafii, s., hartsough, b., & hamann, b. (2011). tree detection and delineation from lidar point clouds using ransac. in: proceedings of silvilaser 2011 – 11th international conference on lidar applications for assessing forest ecosystems, hobart, australia, (pp. 583595). tomaštík, j.; mokroš, m.; saloň, š.; chudý, f.; tunák, d. (2017). accuracy of photogrammetric uav-based point clouds under conditions of partially-open forest canopy. forests, 8(5): 151. tran, t.h.g., hollaus, m., nguyen, b.d., & pfeifer, n. (2015). assessment of wooded area reduction by airborne laser scanning. forests, 6(5), 1613–1627. úhúl (2007). national forest inventory in the czech republic 2001–2004: introduction, methods, results. brandýs nad labem, czech republic 224. wang, w., qu, j.j., hao, x., liu, y., & stanturf, j.a. (2010). post-hurricane forest damage assessment using satellite remote sensing. agriculture and forest meteorology, 150, 122–132. ye, h., shang, g., wang, l., zheng, m. (2015). a new method based on hough transform for quick line and circle detection. in: proceedings of the international conference on biomedical engineering and informatics, shenyang, china, (pp. 52–56). ziou, d., & tabbone, s. (1998). edge detection techniques—an overview. international journal of pattern recognition and image analysis, 8, 537– 559. list of abbreviations uavs: unmanned aerial vehicles; nfi: czech national forest inventory; sar: synthetic aperture radar; lidar: light detection and ranging; dtm: digital terrain model; als: airborne laser scanner; ohm: object height model; ransac: random sample consensus; sfm: structure from motion; cd: correct detected; mva: manual visual assessment; dfl: detected as fallen log; cv: crossvalidation; dbh: diameter at breast height. panagiotidis et al. new zealand journal of forestry science (2019) 49:2 page 11 new zealand journal of forestry science (2019) 49:2 https://doi.org/10.33494/nzjfs492019x26x additional file 1 detection of fallen logs from high-resolution uav images d. panagiotidis*, azadeh abdollahnejad, peter surový, karel kuželka figure s1: orthophotos of the study site. new zealand journal of forestry science (2019) 49:2 https://doi.org/10.33494/nzjfs492019x26x 1 additional file 2 detection of fallen logs from high-resolution uav images d. panagiotidis*, azadeh abdollahnejad, peter surový, karel kuželka table s1: all the parameters used for the detection of linear patterns during image pre-processing. morphological operator parameters alagmatrix (n) bbwareaopen (n) cbwareafilt (n) dstrel r imdilate imclose plot 1 3 120 30 plot 2 3 120 30 disk 12 disk 4 plot 3 3 130 55 plot 4 3 120 10 plot 5 7 120 28 disk 10 disk 3 plot 6 3 140 50 disk 1 a value n stands for the number of lags b remove objects containing fewer than n pixels c retaining only the n objects with the largest areas d strel is an essential part of morphological dilation and erosion operations, it refers to the creation of a diskshaped structuring element, whereas r specifies the distance from the structuring element origin to the points of the disk sawn timber and structural products from ‘kawa’ poplar (populus deltoides marshall x p. yunnanensis dode) grown in northland, new zealand dean satchell1* and john moore2 1 northland regional council, private bag 9021, whangarei 0148, new zealand 2 timberlands limited, p.o. box 1284, rotorua 3046, new zealand *corresponding author: deans@nrc.govt.nz (received for publication 23 may 2022; accepted in revised form 26 october 2022) abstract background: while poplar (populus spp.) is an important source of raw materials for the wood processing sector in many regions of the world, in new zealand it has a reputation for producing poor grade recoveries of sawn timber that is unsuitable for structural applications. however, the ‘kawa’ poplar clone (populus deltoides marshall x p. yunnanensis dode), which has relatively high wood density, could yield structural timber. this, along with evidence demonstrating building code compliance, would improve utilisation options for this species in new zealand. methods: sawn timber conversion and grade recovery were quantified for a 28-year-old pruned stand of ‘kawa’ poplar grown in northland, new zealand. a sample of 90 mm x 45 mm structural boards were tested to determine their mechanical properties and the resulting strength class. boron preservative retention and penetration were measured to determine whether timber could be treated to the level required under new zealand’s building standards. density, modulus of elasticity and modulus of rupture were assessed on small defect-free specimens taken from different radial and vertical positions within trees to determine intra-stem and inter-stem variation in these properties. results: the overall conversion of logs to sawn timber was 53%, with approximately 94% of this recovery consisting of graded timber. the most common sources of downgrade were knots, pruning wounds, and end-splits. approximately 70% of the sawn boards were graded as clears, with smaller recoveries of cladding and structural boards. the average length of clear section was approximately 2.5 m. mechanical testing of structural boards demonstrated that they have characteristic values sufficient to meet the requirements for the sg10 strength class. preservative treatment achieved the h1.2 specification. density, modulus of elasticity and modulus of rupture were all higher in specimens cut from the outside of the log compared with those taken from near the pith at all heights up the stem. conclusions: mechanical properties and boron treatment results indicate suitability for structural applications in accordance with new zealand’s building code. ‘kawa’ poplar also produced high grade recoveries suggesting potential for commercial sawn timber production, especially for structural appearance products. new zealand journal of forestry science satchell & moore new zealand journal of forestry science (2023) 53:1 https://doi.org/10.33494/nzjfs532023x238x e-issn: 1179-5395 published on-line: 16/11/2022 © the author(s). 2023 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access paper sector (stanton et al. 2002). despite its relatively low density and mechanical properties, hybrid poplar timber has been used in structural applications in north america, although generally at a disadvantage compared with softwood species (balatinecz & kretschmann 2001). poplar is a light-coloured timber with an attractive appearance and lustre, so can also be used for introduction globally, poplar (populus spp.) is increasingly being used to provide a fast-growing source of raw material to different parts of the forest products sector (ball et al. 2005; truax et al. 2014). there has been considerable focus on hybrids of fast growing poplar species to provide feedstocks for biomass energy and the pulp and keywords: hybrid poplar; structural timber; timber preservation; mechanical properties; building code; end-splitting; timber grading; pruning wounds; knots; stiffness; strength http://creativecommons.org/licenses/by/4.0/), satchell & moore new zealand journal of forestry science (2023) 53:1 page 2 other solid wood applications such as furniture and a range of engineered wood products including plywood, laminated veneer lumber and cross-laminated timber (hematabadi et al. 2021). ‘kawa’ (populus deltoides marshall x p. yunnanensis dode) is the first hybrid poplar bred in new zealand and was released in 1986 (wilkinson 2000). ‘kawa’ has been the primary poplar cultivar supplied by northland’s only commercial poplar nursery, stix, from 1990 to 2010 for pastoral soil conservation, (murray hunter, pers. comm) and from the northland regional council nursery since 2016. despite this information on nursery sales, unfortunately there is no detailed information on the overall extent of the resource. the national exotic forest description groups poplars together with other hardwood species, excluding eucalyptus species. the total area of “other hardwoods” in new zealand is roughly 12,500 ha with the northland resource being 2300 ha in size (ministry for primary industries 2020). while these statistics are only approximate, they do indicate that most of the northland resource is 16-30 years old, and therefore potentially able to be harvested and processed. wilkinson (2000) recommended ‘kawa’ poplar as suitable for sawn timber production because it produces a relatively high-density wood. however, the same author also reported that poplar has a lower sawn timber recovery than radiata pine (pinus radiata d.don), which is the main commercial timber species grown in new zealand, and knots constitute a considerable defect in the sawn timber. distortion resulting from the presence of tension wood also occurs in drying of poplar, earning the genus a reputation for a high reject rate after drying (williams et al. 1986). no data on sawn timber and grade recovery data from new zealand-grown ‘kawa’ poplar have been available from which to determine its suitability as a source of sawn timber, however ‘kawa’ has good form and monopodial architecture typical of most commercial conifer species, suggesting opportunity for sawlog production. because much of the northland ‘kawa’ resource comprises straight trees that are maturing with appropriate diameters for sawing, it is of interest to both existing and prospective growers to determine whether the timber may have value and what applications it is most suitable for. furthermore, the relatively small size of the resource and its distributed nature means that it is likely to be processed using portable sawmilling equipment. therefore, there is interest in determining the timber grade recovery when processing ‘kawa’ logs using this type of equipment. low wood density is associated with low strength properties in poplar (wilkinson 2000). wood density also varies significantly among poplar clones, with basic density varying between 300-400 kg m-3 for the most common hybrid clones, with little density variation among individual trees of a specific clone (wilkinson 2000). for example, ‘kawa’ had an individual tree density range of only 9 kg m-3, compared to an individual tree density range of over 100 kg m-3 for radiata pine (wilkinson, 2000). an earlier study using material from 8-year-old ‘kawa’ trees reported a mean basic density of 365 kg m-3 with a mean stiffness of 5.4 gpa at 12% moisture content (wilkinson, 2000). ‘kawa’ poplar also showed no marked radial density change. while these values are higher than many of the other new zealand-grown poplar species tested, they are still significantly lower than those reported for 25-year-old radiata pine which had a mean basic density of 415 kg m-3 and a mean stiffness of 8.2 gpa at 12% moisture content, suggesting that stiffness of ‘kawa’ may not be adequate for structural applications. this result agrees with the conclusions of williams et al. (1986) that poplar generally has “moderate to low strength properties”. a more recent study by jones (2016) based on a single ramet of the ‘kawa’ poplar clone grown in the north island found that while density at breast height was over 400 kg m-3, dynamic modulus of elasticity was less than 6 gpa. despite these results, anecdotes from end users of ‘kawa’ timber are that it is sufficiently stiff and strong to be used in structural applications (peter davies-colley, pers. comm). in addition to having the required mechanical properties, structural timber products are required to hold sufficient durability for a minimum 50-year building life under new zealand’s building code (clause b2 durability, 2nd edition amendment 12). two new zealand standards, nzs 3602:2003 (standards new zealand 2003a) and nzs 3640:2003 (standards new zealand 2003b) provide the acceptable solutions for achieving building code compliance for durability performance. nzs 3640:2003 provides definitions of the hazard classes and their preservative treatment requirements and nzs 3602:2003 lists species and level of treatment (as hazard class per nzs 3640:2003) required for specific structural applications. the hazard class for interior structural applications is h1.2 for “service conditions protected from the weather but with a risk of moisture content conducive to decay” (nzs 3640:2003). the preservative treatment for radiata pine is prescribed in nzs 3640:2003 for h1.2 hazard class service conditions. like radiata pine, poplar wood is non-durable (williams et al. 1986). sawn softwood timber requires complete sapwood penetration along with a retention level in the sapwood cross section of 0.4% boric acid equivalent (bae). because poplar is a hardwood and is not listed in nzs 3602:2003 for h1.2 service conditions, the route to compliance is via a similar materials test as per building code verification method b2/vm1 (1.0 durability evaluation, building code clause b2). this test compares the relative durability performance of the test material (e.g., borontreated poplar) with a reference softwood (e.g., radiata pine boron-treated to h1.2). there is little published information on preservative treatment of poplar timber. one study in the united states using 100 mm x 50 mm sawn boards from yellow poplar (liriodendron tulipifera l) found that wrapped diffusion storage for at least eight weeks after dipping was required to achieve adequate borate penetration (chen et al. 1997). in new zealand, pressure treatment of poplar with boron has yielded unsatisfactory penetration results (tripti singh pers. comm), but williams et al. (1986) tested the boron diffusion method, demonstrating that 50-mm-thick poplar boards could be effectively treated to above the 0.2% bae required at that time by the timber preservation authority (1980) for preservation of wood against borer attack. however, hazard class h1.2 now requires 0.4% bae (nzs 3640:2003) to protect against decay and no research data are available demonstrating that the boron diffusion method consistently achieves this. for ‘kawa’ poplar to be used in structural applications, more information is needed on its physical and mechanical properties, and the ability of preservative treatment to improve its durability so that it can meet the requirements of the new zealand building code for structural applications. therefore, the objective of this study was to assess the sawn timber recoveries and performance of structural products cut from a plantation of 28-year-old pruned ‘kawa’ poplar grown in northland. the specific questions that this research project sought to answer were: (1) can ‘kawa’ poplar be processed with minimal degrade using best practice portable sawmilling followed by air-drying, to enable on-farm production of structural timber? (2) to what degree does log diameter and log position within the stem from which it was cut affect sawn timber and grade recoveries? (3) what levels of defect and knot defect can be expected in northland plantation ‘kawa’ poplar at 28 years of age? (4) to what degree does radial position and vertical position in tree affect mechanical properties and density? (5) what relationships exist between density, stiffness, and strength, among trees and within trees? (6) can the boron diffusion method be used on ‘kawa’ poplar to meet the preservative penetration and retention requirements under nzs 3640:2003 for the h1.2 hazard class? methods tree harvesting and log preparation seventeen pruned ‘kawa’ poplar trees were harvested from a small 28-year-old stand growing near titoki, northland, new zealand (latitude 35°44’08’’s, longitude 174°01’37’’e). the trees were growing on a reasonably sheltered pastoral hill country site, on a moderately fertile clay loam soil. trees were harvested mechanically, and 71 logs were cut to length on site, with the bark and limbs removed mechanically. diameter under bark and the diameter of visible heartwood were measured on both the large end and the small end of all logs. sawlog lengths were measured, and logs were marked to identify tree and log position in each sawn board. only those logs with a small-end diameter (sed) greater than 18 cm were milled into boards. timber processing and sawn timber preparation a total of 43 logs from 15 trees were milled two weeks after harvest. this period was deemed as commercially appropriate for minimising end-splitting of logs, which increases as the interval between cross-cutting logs and sawmilling increases. forty sawlogs were 3.7 m long, one was 4.0 m long, one was 3.6 m and one was 3.5 m long, and their small end diameters ranged from 18 cm up to 44.5 cm, with a mean small end diameter of 30.6 cm. heartwood diameters ranged from 7.0 cm up to 40 cm, with a mean of 20.2 cm. corresponding values of heartwood percentage (on an area basis) ranged from 9.7% up to 76.3%, with a mean of 32.8%. there was no evidence of discoloured heartwood or “biological blackheart” which can occur in poplar (johansson & hjelm 2013). sawlogs were sawn on a wood-mizer 3 mm kerf horizontal bandsaw into 25 mm and 50 mm slabs; slabs were then edged using a wood-mizer twin-blade edger (wood-mizer, indianapolis, usa). wider central slabs were ripped through the centre of their width into two slabs, before edging into straight boards, avoiding inclusion of pith. edging included visual judgment calls that focused on grade recoveries in preference to volume recoveries. resulting nominal board widths were 75 mm, 100 mm, 150 mm, 200 mm, 250 mm and 300 mm. nominal board thicknesses were 25 mm and 50 mm. boards were oversized during sawing to allow for shrinkage during drying. sawn recoveries were calculated from nominal board dimensions but excluded those sections containing taper and wane. after air drying outdoors under cover for eight months, boards were visually graded as either “clears”, “cladding” or “structural”. two grading methods were used. the first targeted clearwood lengths (“best grade”) and the second targeted longest graded lengths in preference to shorter lengths of a higher grade (“long length grade”). criteria for the three grades are described in table 1 (criteria for higher grades always exceeds those for lower grades). the lengths of defect were marked as if docked and allocated to the following categories: knots; pruning wounds; end-splits; pith; excessive crook and excessive bow. lengths of end splits were measured from the end of the board to where the visible crack ended. knots were further categorised as follows: (1) knot (only classified as defect according to the grade allocated); (2) large knot defect; (3) spike knot (only classified as defect according to the grade allocated); (4) large spike knot defect; (5) pruning wound (only classified as defect according to the grade allocated); and (6) large pruning wound defect. the presence of pith was recorded as a defect. crook and bow were also recorded as defects where these exceeded the limits given in nzs 3631:1988 (standards new zealand 1988). bending strength and stiffness testing of structural grade boards air dry 100 mm x 50 mm boards were machined to 90 mm x 45mm and a sample of 31 structural-grade boards was randomly selected for bending tests to satchell & moore new zealand journal of forestry science (2023) 53:1 page 3 table 1: description of the study sites determine modulus of elasticity and modulus of rupture. testing was undertaken at the new zealand forest research institute limited (scion) in rotorua using a baldwin universal testing machine. specimens were tested as a joist (on edge) in accordance with as/nzs4063.1:2010 (standards new zealand 2010a) and as/nzs4063.2:2010 (standards new zealand 2010b) over a span to depth ratio of 18:1 (i.e. 1620 mm). a short cross section was then cut from an undamaged clearwood section close to the failure point of each test specimen for determining density and moisture content. moisture content was measured using the oven drying method. nominal density was calculated for each section from the oven dry weight and the volume at the time of testing. density at the time of testing was calculated for each section from the test weight and volume at time of testing. mechanical testing of small defect-free specimens small defect-free specimens were prepared from sections with 25 mm x 25 mm cross-sectional dimensions that were ripped from two radial positions (inner/outer) at up to five height positions from seven trees. inner specimens were cut as close as possible to the pith, but no closer than 1 cm from the pith line. outer specimens were cut no closer than 2 cm from the outside of the log to avoid physical damage caused by machine debarking. the first vertical height position was at 1.85 m +/1 m and then at 3.7 m increments +/1 m up the stem. sections containing only clearwood were selected. wood with incipient decay or discolouration was excluded. growth ring orientation was marked on sample ends and only specimens with growth rings parallel to a face were used for testing. specimens were machined to their final cross-sectional dimensions of 20 mm x 20 mm and crosscut into 500 mm lengths, resulting in 140 small clear specimens (two or three replicates for each radial position and height position within each tree). specimens were conditioned at 20°c and 65% rh until they achieved constant mass (approximate equilibrium moisture content of 12%). bending tests were conducted using an instron universal testing machine at scion, rotorua. the growth rings were aligned as much as possible with the direction of loading, in accordance with astm d143-94 (american society for testing and materials 2000). the test span was 280 mm with a centre point load. from the load-deflection data the fibre stress at the proportional limit (fspl), fibre stress at maximum load (modulus of rupture) and satchell & moore new zealand journal of forestry science (2023) 53:1 page 4 modulus of elasticity were calculated. following testing, each entire specimen was used for determination of density and moisture content. moisture content was measured using the oven drying method. oven-dry density (a proxy for basic density) was calculated for each test specimen from the oven dry weight and volume measured at the time of testing. density at the time of testing was calculated for each test specimen from the weight and volume at the time of testing. boron treatment retention and penetration a boron solution was prepared by adding 2.3 kg of granular boric acid (inkabor orthoboric acid 99.99% h3bo3 min, cas no:10043-35-3 ec no:233-1239-2) and 3 kg of borax pentahydrate (disodium tetraborate pentahydrate na2b4o7.5h2o etimaden etibor-48 cas no:12179-04-3 ec no:215-540-4) per 10 litres of water used. the solution was gently heated and stirred until the salts dissolved and no visible solids remained. the solution was used within 12 hours of dissolving the boron salts and before any crystallisation occurred in the solution. rough-sawn freshly milled boards with crosssectional dimensions of 100 mm x 50 mm and 150 mm x 50 mm were immersed in the solution for five seconds, then block stacked and fully wrapped with a black polythene cover and stored outdoors for 41 days to allow the boron to diffuse into the wood. boards were then filleted and allowed to air dry for eight months. sixteen samples representative of the batch of treated timber were selected and prepared for analysis by machining to 90 mm x 45 mm and then cross-cutting a single 20 cm length that contained defect-free wood, no closer than 50 cm from board ends. boron retention and penetration tests were conducted by independent verification services ltd in hamilton. variamine blue rt (vbrt) salt solution with ammonia buffer was used as a staining test to determine the heartwood sapwood boundary following as/nzs 1605.1:2018 (standards new zealand 2018a). however, this test was unsuccessful for determining the boundary between heartwood and sapwood, so two other staining tests (methyl orange and ferric chloride) were trialled for heartwood/sapwood differentiation. these tests also proved unsuccessful, so it was decided that analysis would be based on total cross section retention and penetration. boron penetration was analysed using the turmeric acid test as per as/nzs 1605.2:2018 grade criteria clears clearwood on the front face and edges, with the board also meeting the requirements of cladding grade. cladding dressing grade as per nzs 3631:1988 (standards new zealand 1988) with two additional requirements for cladding from nzs 3602:2003 (standards new zealand 2003a): 1. all holes, resin and bark pockets shall be excluded; 2. knot size shall not exceed 50 mm, or 25 mm width for spike knots. structural no. 1 framing grade as per nzs 3631:1988. table 1: criteria used to allocate boards to specific grades. (standards new zealand 2018b). boron retention samples were prepared using nitric acid micro digestion at 100°c for 60 minutes as per awpa a7-12 (american wood protection association 2012). boron content was measured using inductively coupled plasma optical emission spectrometry (icp-oes), with correction for moisture as per awpa a21-08 (american wood protection association 2008). boric acid equivalent (bae) was calculated as elemental boron x 5.717. data analysis individual log volumes were calculated from data on their large and small-end diameters and length using smalian’s formula (avery & burkhart 2015). the nominal volumes of the individual boards cut from each log were summed to give the total sawn timber recovery per log. these values were divided by log volume to give the proportional recovery. because these data are bounded by zero and one, beta regression was used to model sawn timber recovery as a function of log characteristics (e.g., log diameter and taper). the recovery of different grades was calculated from the grade assigned to a board (based on the two different grading methods) and its overall nominal dimensions. the volume downgraded due to defects such as knots and end splits was calculated from the nominal cross-sectional dimensions of the board and the length affected by the particular defect. beta regression was used to model the relative proportion of volume with these defects and log characteristics. data from the bending tests on structural boards were used to calculate the characteristic strength, stiffness and density values using the procedures set out in as/nzs4063.2:2010 (standards new zealand 2010b). these, in turn, were used to assign the timber to a strength class according to nzs3603:1993 (standards new zealand 1993). ordinary least squares regression was used to explore inter-relationships among density, stiffness and bending strength. a linear mixed effects model was fitted to examine the effects of radial and longitudinal position on density, bending strength and stiffness measured on the small defect-free specimens. fixed effects were included for radial position (inner and outer) and log height class. because the data had a hierarchical structure with multiple specimens collected from the same log within a tree, a term was included in the model to account for the random effect of the different trees. the model had the following form: yijk = µ + αposi + βlogj + γposi × logj +ak + εijk (1) where yijk is the density, bending stiffness or bending strength of a small defect-free specimen cut from the ith radial position in the jth log cut from the kth tree and εijk represents the within-group error. the random effect of the kth tree is assumed to be normally distributed (ak ~ n(0, ψ)) where ψ is the variance-covariance matrix. results sawn timber the 43 logs that were processed had a total volume of 14.2 m3. these logs were processed into 7.6 m3 of nominal sized (i.e., dry) sawn boards, giving an overall conversion of 53%. the most common section size produced by piece count was 150 mm x 50 mm (26% of boards), while the most common section size by volume was 200 mm x 50 mm (34% of volume; table 2). sawn timber recovery for individual logs ranged from 32% up to 62% and increased with increased log diameter (p=0.01; figure 1). it also decreased with increasing log taper (p=0.009). log taper was highest in the butt logs (on average around 2 cm/m), and lowest in the second and third logs cut from the stem (approximately 0.7 cm/m). overall, the beta regression model that included terms for log diameter and log taper explained 47% of the variation in sawn timber recovery. the fitted model had the following form: conv = − 1.598 + 0.08845diam − 0.0009543diam2 − 0.118taper (2) where conv is the proportion of log volume converted into timber, diam is the average log diameter (cm), and taper is the change in log diameter per unit length (cm/m). satchell & moore new zealand journal of forestry science (2023) 53:1 page 5 nominal width (mm) nominal thickness (mm) number of boards total volume of boards (m3) 75 25 6 0.04 100 25 55 0.49 100 50 50 0.90 150 25 50 0.64 150 50 92 2.48 200 25 23 0.42 200 50 71 2.56 250 25 1 0.02 300 50 1 0.06 total 349 7.59 table 2: number and volume of boards produced with different nominal dimensions sawn timber grade recovery of the total volume of timber produced, there was a 94% recovery of graded timber based on the “best grade” method (table 3), with an average graded board length of 2.6 m. most of the downgraded volume was due to end splits, knots and pruning wounds. end splits are of similar importance as knot/pruning defects as being the most significant cause of degrade. the volume docked due to knots and pruning wounds was broken down into different categories, which showed that the most common downgrade was due to large spike knots, followed by large knots and then pruning wounds (table 4). recovery of graded timber was slightly higher (circa 95%) under the “long length” grading method due to a reduction in degrade due to knots and pruning wounds (table 3). the average length of “long length” graded boards was slightly longer at 2.8 m. most of the recovered timber volume was assigned to the “clears” grade (table 5). under the “best grade” method, 74% of the recovered volume was assigned to the “clears” grade with the average length of clear sections being 2.5 m (range 0.7-4.0 m). under the “longlengths” method, the proportion of recovered volume assigned to the clears grade decreased to 64%, but the average length of clear sections increased to 2.6 m (range 0.7-4.0 m). this method resulted in greater proportions of recovered volume being assigned to the cladding and structural grades, and the average board lengths for these grades being 0.3 m and 0.5 m longer, respectively, than under the “best” grading method. satchell & moore new zealand journal of forestry science (2023) 53:1 page 6 figure 1: relationship between sawn timber conversion and log diameter and taper. outturn category percentage of sawn timber in category by grading method “best” “long length” graded timber 93.9 94.7 defect end splits 2.4 2.4 defect knots and pruning wounds 3.2 2.4 defect crook and bow 0.5 0.5 defect pith <0.1 <0.01 total 100 100 table 3: recoveries of graded timber and defects as percentages of sawn timber volume for the two grading methods. relationships between timber defects and log characteristics there was a significant relationship between the incidence of knot and pruning wound defects and both log position (p<0.001) and log taper (p=0.001). there was no relationship with log diameter (p=0.781). the incidence of knot and pruning wound defects increased with increasing log position up the stem (figure 2) and with increasing log taper. boards cut from the butt log had approximately 0.5% of their volume downgraded due to these defects, while boards cut from the upper logs had more than 2% of their volume downgraded. no relationship was established between end-splits and log position (p=0.5). however, there was a weak positive relationship with log diameter (p=0.02) and suggestive but inconclusive evidence of a negative relationship with taper (p=0.08). these two variables were only able to explain approximately 13% of the variation in the volume downgraded due to end-splitting. bending strength and stiffness of structural grade boards the moisture content of the structural timber specimens at the time of testing ranged from 8.7% up to 12.1% with a mean of 10.8% (table 6). values of modulus of elasticity (unadjusted for moisture content) ranged from 6.88 gpa up to 13.04 gpa, with a mean value of 10.76 gpa. modulus of elasticity exhibited a moderate liner relationship with modulus of rupture (r2=0.47) and nominal density (r2=0.36). the characteristic values of modulus of elasticity and modulus of rupture were 10.58 gpa mm-2 and 45.90 mpa, respectively, which were sufficient for the timber to be assigned to the sg10 strength class. the bending strength exceeded the requirement for higher strength classes, with the assignment to sg10 determined by bending stiffness (modulus of elasticity). physical and mechanical properties of small defectfree specimens modulus of elasticity of the 140 small, defect-free specimens tested ranged from 5.65 gpa up to 11.17 gpa, with a mean of 9.02 gpa and a coefficient of variation of 13%. the mean moisture content of the specimens at the time of testing was 14.2%. there was a strong linear relationship between modulus of elasticity and modulus of rupture (r2=0.87; figure 3), and between nominal density and modulus of elasticity (r2=0.76). modulus of elasticity was significantly higher in specimens cut from the outer position in a log (p<0.001) and increased with log position up the stem (p<0.001; figure 4). there was also a significant effect of the interaction between radial position and log position up the stem (p<0.001). most notably, the difference in modulus of elasticity between specimens cut from the inner and outer positions was greatest in the butt log (i.e., log 1). here the difference was approximately 2.2 gpa between specimens cut from the inner and outer radial positions. overall, radial position, log number up the stem and their interaction explained approximately 70% of the variation in modulus of elasticity. both modulus of rupture and nominal density also exhibited similar patterns of radial and longitudinal variation within a tree as the radial position, log number and interaction terms in the models for these properties were all highly significant (p<0.001). the model for modulus of rupture was able to explain approximately 75% of the variation in this property, while the model for nominal density explained approximately 60% of the variation. satchell & moore new zealand journal of forestry science (2023) 53:1 page 7 knot/pruning wound defect category percentage of total volume large knot defect 20 large spike knot defect 33 knot degrade 12 spike knot degrade 3 large pruning wound defect 12 pruning wound degrade 19 total 100 table 4: frequency of different categories of knot and pruning wound defect/degrade. grade “best” grading method “long length” grading method grade recovery (%) average board length (m) grade recovery (%) average board length (m) clears 74 2.52 64 2.6 cladding 17 2.96 23 3.3 structural 9 2.71 13 3.2 overall 100 100 table 5: overall recovery of different timber grades and the average board length from application of the two different grading methods. boron retention and penetration boric acid equivalent (bae) ranged from 0.60% (m/m) up to 1.98%, with a mean of 1.03%. these values exceeded the threshold of 0.40%, which meant that all 16 samples passed the h1.2 boron retention and penetration requirements as per nzs 3640:2003. the samples tested also all achieved full cross section penetration of the preservation, not just penetration into the sapwood zone. discussion the recovery of sawn boards was in line with sawing studies reported in the literature for a range of species (cown et al. 2013; lin et al. 2011). the positive relationship between log diameter and sawn timber conversion is well established in wood processing, as is the negative relationship with log taper (moore & cown 2015; steele 1984). the same sawing pattern was used in earlier studies to process eucalyptus nitens and e. regnans logs with a portable sawmill, and in these studies slightly lower recoveries were obtained, likely due to increased movement (distortion) of the boards as they were cut (satchell & turner 2011; satchell 2015). previous experience with processing poplar logs in the united states recommended using the saw-dryrip method to help reduce timber distortion (maeglin 1985), however downgrade due to excessive distortion did not occur in this study which means that timber could be cut to its final dimensions in the green state. in the current study, logs were sawn to visually target grade recoveries when edging the boards. this is assumed to be best practice, but involves a trade-off between grade recoveries and sawn recoveries. the intention was to maximise timber value recovered from the log and minimise levels of residual defects in sawn boards. because the pruned logs had a large diameter over stubs (dos), measured at 25 cm on boards cut through the centre of butt logs, and debarking caused deep (2 cm) mechanical cutting damage through the log surface, achieving high grade recoveries required significantly higher levels of wastage than would be necessary if mechanical debarking had not occurred and if pruning had controlled dos to lower levels. nevertheless, these recoveries do still provide a conservative benchmark for an economic analysis based on a return-to-log approach (murphy & moore 2018). the modulus of elasticity of ‘kawa’ poplar structural timber was considerably higher than observed in previous studies on the same clone (i.e., jones 2016; wilkinson 2000). given the moderate positive relationship that was observed between modulus of elasticity and density in the current study, the modulus of elasticity reported satchell & moore new zealand journal of forestry science (2023) 53:1 page 8 figure 2: incidence of knots and pruning wounds in timber as a function of log height up the stem. moisture content (%) modulus of elasticity (gpa) modulus of rupture (mpa) density at test (kg m-3) nominal density (kg m-3) mean 10.8 ± 0.7 10.76 ± 1.33 59.13 ±11.25 476 ± 36 430 ± 32 range [8.7-12.1] [6.88 – 13.04] [33.37-74.34] [395-567] [358-511] characteristic value 10.58 45.09 471 426 assigned grade sg10 sg10 table 6: physical and mechanical properties determined from bending tests on full-size structural timber specimens. by wilkinson (2000) appeared unusually low given the density value given in their publication, although results in this early paper were based on samples from much younger trees (c.f. 8 years old and 28 years old). the sg10 grade assigned to the structural timber in this study means that the timber is in a higher strength class than is typically achieved by the main structural timber species in new zealand, radiata pine (cown 1999). however, it should be noted that in warmer regions of new zealand, such as northland, trees typically have higher wood density than cooler regions such as the central north island and central and southern parts of the south island (kimberley et al. 2017; palmer et al. 2013). given the positive relationship between density and modulus of elasticity, ‘kawa’ poplar grown on cooler sites may not achieve the requirements for the sg10 strength class. additional testing of material from a range of different regions would be required to confirm this. a larger dataset is required to provide more robust data on the characteristic strength properties for the species throughout new zealand, although the sample tested in this study does provide evidence showing the potential of this clone as a structural timber species. results from testing the small defect-free specimens provided data to compare with other poplar species including p. deltoides, which is grown in many other regions of the world where it is also hybridised with other species. the modulus of elasticity of ‘kawa’ in our study was higher than found in many other studies on poplar (balatinecz & kretschmann 2001; de boever et al. 2007; hernandez et al. 1998) but similar to the values found by jia et al. (2021) for a p. deltoides clone grown in china. the results obtained from the tests also gave some insight into the radial and longitudinal variation in wood properties that exists within ‘kawa’ poplar. the intention was to test the inner-most clearwood obtainable as 50 cm lengths, which tended to be very close to the pith, and to compare physical and mechanical properties with values obtained from specimens taken from the outer part of each log. while only two locations were sampled in the radial direction, the results were consistent with those from other studies which show that wood density, modulus of elasticity and modulus of rupture are higher in the outerwood than in the corewood (lachenbruch et al. 2011; zobel & sprague 1998). if more locations in the radial direction had been sampled and the ring number from the pith recorded for each specimen, models such as those developed for scots pine (pinus sylvestris l.) by auty et al. (2016) could have been developed. such models could then be used to predict the impact of rotation age on wood properties, as numerous studies have demonstrated the impact that rotation age has on the mechanical properties of structural timber (clark iii et al. 1996; cown & mcconchie 1982; duchesne 2006; moore et al. 2012). while much of the radial variation in modulus of elasticity and modulus of rupture is likely to be due to radial variation in density and microfibril angle, it was also observed that small defect-free specimens cut from the outer part of each log had straighter grain visible on the radial face than inner samples, which visually had more sloping grain caused by knots in the wood adjacent to the samples. in addition to this grain deviation caused by the presence of knots, spiral grain angle of many species is typically higher in the inner region of a tree and lower near the bark (harris 1989). regardless of its cause, grain deviations from the longitudinal axis of a board has a negative impact on timber mechanical properties (harris 1989). although it is not known how far the weaker corewood zone extends radially out from the pith in poplar, our results show that provided pith is excluded, structural satchell & moore new zealand journal of forestry science (2023) 53:1 page 9 figure 3: pairwise relationships between nominal density, modulus of rupture and modulus of elasticity for small defect-free specimens. the blue lines represent ordinary least squares regressions fitted to the variables. satchell & moore new zealand journal of forestry science (2023) 53:1 page 10 boards with acceptable strength characteristics can be produced. further investigations to confirm this result could be undertaken by cutting structural boards from different locations within the log and comparing their mechanical properties and distortion. the wood properties in the butt logs were also different than at equivalent radial positions in upper logs. it is unlikely that pruning of butt logs has adversely influenced wood properties. in addition to removing branches and the associated knots, pruning can also lead to small increases in wood density and a more rapid transition from corewood to outerwood (megraw 1996). it is more likely due to the zone of low wood density and high microfibril angle that is often found in the region near the pith at the bottom of the butt log in many tree species (e.g., moore et al. 2014). this could be further tested by comparing inner wood from pruned and unpruned trees as well as measuring the radial variation in density and microfibril angle at different heights up the stem. studies in other poplar species have shown that density and microfibril angle exhibit the typical radial pattern of variation that is observed in many tree species (fang et al. 2006; liu et al. 2022; sheng-zuo et al. 2004). regardless of the factors responsible for its occurrence, the lower stiffness clearwood produced from pruned butt logs could either be used for non-structural appearance applications, or it could be machine graded and allocated a stiffness class/grade. the boron diffusion process was successful at achieving full cross section penetration and retention to greater than 0.4% bae. because nzs 3640:2003 only requires full sapwood penetration and retention to 0.4% bae, if a method becomes available for determining the heartwood/sapwood boundary in ‘kawa’ poplar then lower concentrations of boron could potentially meet the retention requirements for the standard. further work may be required to prove that durability performance of ‘kawa’ poplar (a perishable hardwood) treated with boron to the h1.2 specifications satisfies the durability evaluation section of clause b2 of the new zealand building code. a similar materials test, for example using the third-party test procedure described in the awpc protocols (australasian wood preservation committee 2015), with h1.2 radiata pine as the reference preservative/material, would offer relative durability performance results within a short (1-2 years) time period. conclusions based on data from a 28-year-old ‘kawa’ poplar stand we found that sawlogs can be processed into appearance and structural products with minimal degrade using best practice portable sawmilling methods described here, followed by air-drying. in the warmer climate of northland, new zealand ‘kawa’ timber appears to be well suited to structural applications based on results for strength, stiffness and boron diffusion. there does not appear to be any significant differences between trees in terms of bending strength/stiffness and nominal density, suggesting that variation in mechanical properties may be negligible across the regional resource. however, it figure 4: variation in modulus of elasticity with radial position and log height class. satchell & moore new zealand journal of forestry science (2023) 53:1 page 11 should be noted that wood immediately adjacent to the pith has lower density, strength and stiffness, especially in butt logs. density, strength and stiffness all improve with height in the tree and processors could take advantage of this variation by selecting logs and cuts for production of higher-stiffness structural product. the sawn timber properties and grades achieved in this study indicate that farm production of sawn ‘kawa’ poplar timber from northland can meet market quality and regulatory requirements for structural and appearance products in new zealand. applications such as exposed beams, rafters, and glue-laminated products where appearance achieves a market premium would be appropriate applications for ‘kawa’ poplar sawn timber, however market studies to confirm this are required. competing interests the authors declare that they have no competing interests. authors' contributions ds initiated this study, oversaw the selection, felling and processing of trees and co-wrote the manuscript. jm analysed the data and co-wrote the manuscript. both authors approved the final version of the manuscript. acknowledgements northland regional council funded this research through the sustainable hill country and regional priorities (sharp) programme. the sharp programme was partly funded by the ministry for primary industries (mpi). many thanks to duncan kervell and john ballinger (northland regional council), who encouraged undertaking the research and understood its importance. special thanks to peter davies-colley who provided the trees, harvested them and loaded the log truck at no cost. without his enthusiastic approach and generous support this research would not have taken place. thanks to li legler who patiently undertook the sawmilling to the required specifications and assisted with the boron treatment. references american society for testing and materials. (2000). standard test methods for small clear specimens of timber. (astm d143-94). american wood protection association. (2008). standard method for the analysis of wood and wood treating solutions by inductively coupled plasma emission spectrometry. (awpa a21-08). american wood protection association. (2012). standard for wet ashing procedures for preparing wood for chemical analysis. (awpa a7-12). australasian wood preservation committee. (2015). protocols for assessment of wood preservatives. auty, d., achim, a., macdonald, e., cameron, a.d., & gardiner, b.a. (2016). models for predicting clearwood mechanical properties of scots pine. forest science, 62(4), 403–413. https://doi. org/10.5849/forsci.15-092 avery, t.e., & burkhart, h.e. (2015). forest measurements. long grove, il, usa: waveland press balatinecz, j.j., & kretschmann, d.e. (2001). properties and utilization of poplar wood. in d.i. dickmann, j.g. isebrands, j.e. eckenwalder, & j. richardson (eds.), poplar culture in north america (pp. 277– 291). ottawa, on, canada: nrc research press, national research council of canada. ball, j., carle, j., & del lungo, a. (2005). contribution of poplars and willows to sustainable forestry and rural development. unasylva, 221, 3–9. chen, p.y.s., puettmann, m.e., williams, l.h., & stokke, d.d. (1997). treatment of 8/4 hardwood lumber with borate preservatives. forest products journal, 47(6), 63–68. clark iii, a., mcalister, r.h., saucier, j.r., & reitter, k. (1996). effect of rotation age on lumber grade, yield, and strength of unthinned loblolly pine. forest products journal, 46, 63–68. cown, d.j. (1999). new zealand pine and douglas-fir: suitability for processing. rotorua: new zealand forest research institute. cown, d.j., marshall, h., silcock, p., & meason, d. (2013). sawn timber grade recovery from a planted coast redwood stand growing in new zealand. new zealand journal of forestry science, 43: 8. https:// doi.org/10.1186/1179-5395-43-8 cown, d.j., & mcconchie, d.l. (1982). rotation age and silvicultural effects on wood properties of four stands of pinus radiata. new zealand journal of forestry science, 12, 71–85. de boever, l., vansteenkiste, d., van acker, j., & stevens, m. (2007). end-use related physical and mechanical properties of selected fast-growing poplar hybrids (populus trichocarpa × p. deltoides). annals of forest science, 64(6), 621–630. https://doi.org/10.1051/ forest:2007040 duchesne, i. (2006). effect of rotation age on lumber grade yield, bending strength and stiffness in jack pine (pinus banksiana lamb.) natural stands. wood and fiber science, 38, 84–94. fang, s., yang, w., & tian, y. (2006). clonal and within-tree variation in microfibril angle in poplar clones. new forests, 31(3), 373–383. https://doi.org/10.1007/ s11056-005-8679-7 harris, j.m. (1989). spiral grain and wave phenomena in wood formation. berlin, heidelberg: springer berlin heidelberg. https://doi.org/10.1007/9783-642-73779-4 hematabadi, h., madhoushi, m., khazaeian, a., & https://doi.org/10.5849/forsci.15-092 https://doi.org/10.5849/forsci.15-092 https://doi.org/10.1186/1179-5395-43-8 https://doi.org/10.1186/1179-5395-43-8 https://doi.org/10.1051/forest:2007040 https://doi.org/10.1051/forest:2007040 https://doi.org/10.1007/s11056-005-8679-7 https://doi.org/10.1007/s11056-005-8679-7 https://doi.org/10.1007/978-3-642-73779-4 https://doi.org/10.1007/978-3-642-73779-4 ebrahimi, g. (2021). structural performance of hybrid poplar-beech cross-laminated-timber (clt). journal of building engineering, 44, 102959. https://doi.org/10.1016/j.jobe.2021.102959 hernandez, r.e., koubaa, a., beaudoin, m., & fortin, y. (1998). selected mechanical properties of fast growing hybrid poplar clones. wood and fiber science, 30(2), 138–147. jia, r., wang, y., wang, r., & chen, x. (2021). physical and mechanical properties of poplar clones and rapid prediction of the properties by near infrared spectroscopy. forests, 12(2), 206. https://doi. org/10.3390/f12020206 johansson, t., & hjelm, b. (2013). frequency of false heartwood of stems of poplar growing on farmland in sweden. forests, 4(1), 28–42. https://doi. org/10.3390/f4010028 jones, t. (2016). tree diameter growth and wood properties of poplar clones in new zealand. in poplars and other fast-growing trees renewable resources for future green economies. berlin, germany: international poplar commission (ipc). kimberley, m.o., mckinley, r.b., cown, d.j., & moore, j.r. (2017). modelling the variation in wood density of new zealand-grown douglas-fir. new zealand journal of forestry science, 47: 15. https://doi. org/10.1186/s40490-017-0096-0 lachenbruch, b., moore, j.r., & evans, r. (2011). radial variation in wood structure and function in woody plants, and hypotheses for its occurrence. in f.c. meinzer, b. lachenbruch, & t.e. dawson (eds.), sizeand age-related changes in tree structure and function (vol. 4, pp. 121–164). dordrecht: springer netherlands. accessed 27 june 2020 https://doi. org/10.1007/978-94-007-1242-3_5 lin, w., wang, j., wu, j., & devallance, d. (2011). log sawing practices and lumber recovery of small hardwood sawmills in west virginia. forest products journal, 61(3), 216–224. https://doi. org/10.13073/0015-7473-61.3.216 liu, y., zhou, l., guan, y., hu, j., zhao, z., gao, h., & liu, s. (2022). radial variation of wood anatomical and chemical properties in eight poplar clones. canadian journal of forest research, 52(1), 19–26. https://doi.org/10.1139/cjfr-2020-0465 maeglin, r.r. (1985). using poplar wood for structural lumber: the sdr process. presented at the 22nd annual meeting of the poplar council of the united states, lawrence, ks. megraw, r.a. (1996). effect of silvicultural practices on wood quality. raleigh, ca: tappi. ministry for primary industries. (2020). national exotic forest description as at 1 april 2020 (p. 49). wellington, new zealand. moore, j.r., & cown, d.j. (2015). processing of wood for wood composites. in wood composites (pp. 27– 45). elsevier. accessed 12 may 2022 https://doi. org/10.1016/b978-1-78242-454-3.00002-0 moore, j.r., cown, d.j., & mckinley, r.b. (2014). modelling microfibril angle variation in new zealand-grown radiata pine. new zealand journal of forestry science, 44: 25. https://doi.org/10.1186/s40490014-0025-4 moore, j.r., lyon, a.j., & lehneke, s. (2012). effects of rotation length on the grade recovery and wood properties of sitka spruce structural timber grown in great britain. annals of forest science, 69(3), 353–362. https://doi.org/10.1007/s13595-0110168-x murphy, g.e., & moore, j.r. (2018). segmod—a technoeconomic model for evaluating the impact of segregation based on internal wood properties. annals of forest science, 75(3), 73. https://doi. org/10.1007/s13595-018-0755-1 palmer, d.j., kimberley, m.o., cown, d.j., & mckinley, r.b. (2013). assessing prediction accuracy in a regression kriging surface of pinus radiata outerwood density across new zealand. forest ecology and management, 308, 9–16. https://doi. org/10.1016/j.foreco.2013.07.024 satchell, d., & turner, j. (2011). timber recovery and economics of short-rotation small-diameter eucalypt forestry. tree grower (february). https:// www.nzffa.org.nz/members/special-interestgroups/eucalypt-action-group/eucalyptus-actiongroup-reports/eucalyptus-report-2010/ satchell, d. (2015). evaluating profitability of solid timber production from 15-year-old pruned and thinned eucalyptus nitens (deane & maiden) in canterbury (master of forestry science thesis). school of forestry, university of canterbury. http://dx.doi. org/10.13140/rg.2.1.4586.0563 sheng-zuo, f., wen-zhong, y., & xiang-xiang, f. (2004). variation of microfibril angle and its correlation to wood properties in poplars. journal of forestry research, 15(4), 261–267. https://doi. org/10.1007/bf02844949 standards new zealand. (1988). new zealand timber grading rules. (nzs 3631:1988). standards new zealand. (1993). timber structures amendment 4 (nzs3603:1993). wellington, new zealand. standards new zealand. (2003a). timber and woodbased products for use in building. (nzs 3602:2003). standards new zealand. (2003b). chemical preservation of round and sawn timber. (nzs 3640:2003). standards new zealand. (2010a). characterization of structural timber part 1: test methods. (nzs4063.1:2010). satchell & moore new zealand journal of forestry science (2023) 53:1 page 12 https://doi.org/10.1016/j.jobe.2021.102959 https://doi.org/10.3390/f12020206 https://doi.org/10.3390/f12020206 https://doi.org/10.3390/f4010028 https://doi.org/10.3390/f4010028 https://doi.org/10.1186/s40490-017-0096-0 https://doi.org/10.1186/s40490-017-0096-0 https://doi.org/10.1007/978-94-007-1242-3_5 https://doi.org/10.1007/978-94-007-1242-3_5 https://doi.org/10.13073/0015-7473-61.3.216 https://doi.org/10.13073/0015-7473-61.3.216 https://doi.org/10.1139/cjfr-2020-0465 https://doi.org/10.1016/b978-1-78242-454-3.00002-0 https://doi.org/10.1016/b978-1-78242-454-3.00002-0 https://doi.org/10.1186/s40490-014-0025-4 https://doi.org/10.1186/s40490-014-0025-4 https://doi.org/10.1007/s13595-011-0168-x https://doi.org/10.1007/s13595-011-0168-x https://doi.org/10.1007/s13595-018-0755-1 https://doi.org/10.1007/s13595-018-0755-1 https://doi.org/10.1016/j.foreco.2013.07.024 https://doi.org/10.1016/j.foreco.2013.07.024 https://www.nzffa.org.nz/members/special-interest-groups/eucalypt-action-group/eucalyptus-action-group-reports/eucalyptus-report-2010/ https://www.nzffa.org.nz/members/special-interest-groups/eucalypt-action-group/eucalyptus-action-group-reports/eucalyptus-report-2010/ https://www.nzffa.org.nz/members/special-interest-groups/eucalypt-action-group/eucalyptus-action-group-reports/eucalyptus-report-2010/ https://www.nzffa.org.nz/members/special-interest-groups/eucalypt-action-group/eucalyptus-action-group-reports/eucalyptus-report-2010/ http://dx.doi.org/10.13140/rg.2.1.4586.0563 http://dx.doi.org/10.13140/rg.2.1.4586.0563 https://doi.org/10.1007/bf02844949 https://doi.org/10.1007/bf02844949 standards new zealand. (2010b). characterization of structural timber part 2: determination of characteristic values (as/nzs 4063.2:2010). standards new zealand. (2018a). methods for sampling and analysing timber preservatives and preservative-treated timber part 1: general requirements, sampling, and determination of sapwood and heartwood presence (as/nzs 1605.1:2018). standards new zealand. (2018b). methods for sampling and analysing timber preservatives and preservative-treated timber part 2: determination of preservative penetration by spot test (as/nzs 1605.2:2018). stanton, b., eaton, j., johnson, j., rice, d., schuette, b., & moser, b. (2002). hybrid poplar in the pacific northwest: the effects of market-driven management. journal of forestry, 100(4), 28–33. steele, p.h. (1984). factors determining lumber recovery in sawmilling. general technical report fpl39. madison, wi: united states department of agriculture, forest products laboratory. https:// doi.org/10.2737/fpl-gtr-39 timber preservation authority. (1980). timber preservation in new zealand: specifications. wellington, new zealand. truax, b., gagnon, d., fortier, j., & lambert, f. (2014). biomass and volume yield in mature hybrid poplar plantations on temperate abandoned farmland. forests, 5(12), 3107–3130. https://doi. org/10.3390/f5123107 wilkinson, a.g. (2000). introduced forest trees in new zealand recognition, role, and seed source 17. the poplars, populus spp. (no. fri bulletin 124) (p. 62). rotorua, new zealand: new zealand forest research institute. williams, d., simpson, i., & bier, h. (1986). properties of new zealand-grown poplar (no. fri bulletin 112). rotorua, new zealand: new zealand forest service, forest research institute. zobel, b.j., & sprague, j.r. (1998). juvenile wood in forest trees. berlin, heidelberg: springer berlin heidelberg. https://doi.org/10.1007/978-3-64272126-7 satchell & moore new zealand journal of forestry science (2023) 53:1 page 13 https://doi.org/10.2737/fpl-gtr-39 https://doi.org/10.2737/fpl-gtr-39 https://doi.org/10.3390/f5123107 https://doi.org/10.3390/f5123107 https://doi.org/10.1007/978-3-642-72126-7 https://doi.org/10.1007/978-3-642-72126-7 soil mycobiota under managed and unmanaged forests of nothofagus pumilio in tierra del fuego, argentina lorena a. elíades1, marta n. cabello1, verónica pancotto2, alicia moretto2, natalia a. ferreri1, mario c. n. saparrat1,3,4*, marcelo d. barrera5 1instituto de botánica carlos spegazzini, facultad de ciencias naturales y museo, universidad nacional de la plata (unlp), comisión de investigaciones científicas de la provincia de buenos aires (cic), 53 # 477, b1900avj, la plata, argentina. 2centro austral de investigaciones científicas consejo nacional de investigaciones científicas y técnicas (cadic-conicet), argentina. 3instituto de fisiología vegetal (infive), unlp, cct, la plata. conicet, diag. 113 y 61, cc 327, 1900, la plata, argentina. 4cátedra de microbiología agrícola, facultad de ciencias agrarias y forestales, unlp, 60 y 119, 1900, la plata, argentina. 5lisea, facultad de ciencias agrarias y forestales, unlp, cc 31, 1900, la plata, argentina. *corresponding author: masaparrat@yahoo.com.ar (received for publication 2 september 2017; accepted in revised form 15 june 2019) abstract background: management practices can modify the productivity of forests and the associated microbial diversity of soil. the soil mycobiota is considered a key factor in the ecological functions of forests. forests of nothofagus pumilio (poepp. & endl.) krasser (nothofagaceae) are the main source of timber and one of the most important economic resources in the province of tierra del fuego (argentina). however, there is no information on the impact of forest management interventions for the soil mycobiota, which can be reliable biological indicators of disturbance. methods: fungi were isolated from samples of soil collected under several nothofagus pumilio forests subjected to different types of management and periods of time since the intervention. types of management were represented by harvested forest with a shelter wood cutting, stockpile area and control forest without intervention and the periods of time since intervention were 1, 5–10 and 50 years. species richness, evenness and shannon’s diversity index of the mycobiota in each condition of management were calculated. additionally, the effect of seasonality was analysed. results: the soil mycobiota was represented by 70 taxa. richness and/or shannon’s diversity index of the mycobiota between undisturbed forest and stockpile area were higher in may (autumn) than in september or november. there were no differences in mycobiota diversity between dates in the harvested forest. conclusions: our results indicate that the forest intervention per se did not negatively affect the soil culturable mycobiota composition of n. pumilio forests in tierra del fuego (argentina). new zealand journal of forestry science elíades et al. new zealand journal of forestry science (2019) 49:7 https://doi.org/10.33494/nzjfs492019x53x e-issn: 1179-5395 published on-line: 10 july 2019 © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access forest-management practices affect the diversity of fungi and/or influence their spatial patterns is one of the central issues in soil microbial ecology (green and bohannan 2006). the whole soil microbiota is involved in the formation and stabilisation of organic matter but fungi play a greater role than bacteria in the metabolism and growth of trees. they are also involved introduction fungi are an important and highly diverse component of soil microbial communities (tedersoo et al. 2014). in forest ecosystems, they perform essential ecological functions including decomposition of organic matter and nutrient cycling and are involved in biotic interactions such as mycorrhizal symbioses. understanding whether keywords: biodiversity, forest management impact, soil fungi, sustainable forest management elíades et al. new zealand journal of forestry science (2019) 49:7 page 2 in most processes that occur in the forest soil, such as the ones related to soil formation, nutrient availability and recycling of organic matter (elíades et al. 2015). however, anthropic activities such as management practices can affect the forest productivity and the level of timber harvesting that the forest can sustain as well as the number and quality of habitats and of the associated biodiversity (martínez pastur et al. 2002). fifty-five percent of the area in the province of tierra del fuego (argentina) is currently covered by forests of nothofagus pumilio (poepp. & endl.) krasser (nothofagaceae), and, within the harvested forest, 2-3% are covered by stockpile (“canchón”) areas (martínez pastur et al. 2007). nothofagus pumilio is the most important source of timber in southern patagonia, and the main economic resource since the 19th century. these forests have been managed using several silvicultural systems ranging from light selective harvests to clearcuts (gea et al. 2004). selective cuttings with retained shelter wood or variable-retention cuttings have been the most common systems used over the last decade (martínez pastur et al. 2000, 2009). the period of time since intervention also affects the degree of tree cover in the forest. scorsetti et al. (2012) described the pathogenic and enzyme activities of the entomopathogenic fungus tolypocladium cylindrosporum (ascomycota: hypocreales) from n. pumilio forests in tierra del fuego, argentina. more recently, elíades et al. (2015) reported preliminary data on the growth and enzymatic abilities of the soil fungus humicolopsis cephalosporioides at different incubation temperatures collected under n. pumilio forests. however, studies comparing the soil fungal composition and diversity among forests dominated by the same tree species but under different management practices and periods of time since intervention are scarce. the decay rate in these forests is low and there is no information available on the role of these fungi in ecosystem stability so analysing the responses of the diversity of soil fungi to anthropological interventions is a high priority for understanding how forest management practices affect the ecology of n. pumilio forests and their components. therefore, forest management practices can be considered key factors in the ecology of the patagonian region. the type, intensity and frequency of management may affect soil microorganisms which are reliable biological indicators of disturbance because they react easily to environmental changes such as the soil chemical and physical changes related to timber harvesting (jurgensen et al. 1997). hartmann et al. (2014) reported less resistance and resilience of fungi in forest soils compared to bacteria, thus we hypothesised that management of n. pumilio forests in tierra del fuego causes a decrease in the species richness and diversity of fungi even following intervention. therefore, the aim of this study was to assess and compare the seasonal structure of soil fungal communities under n. pumilio forests subjected to different management practices and periods of time since the intervention in tierra del fuego, argentina. methods study area study sites are located in the forest-steppe ecotone of the central part of tierra del fuego island, argentina (54° 27’s, 67°27’w; fig. 1). the forests correspond to the sub-antarctic forest type (37°–60° south latitude) and are composed of nothofagus pumilio, n. antarctica (forster f.) oersted and n. betuloides (mirb.) oersted as dominant trees sparsely mixed with drymis winteri forster & forster f., maytenus magellanica (lam.) hooker f. and embothrium coccineum forster & forster f. (moore 1983; tuhkanem et al. 1990). the landscape occupied by forests has mostly acid brown soils of glacial origin with loess and alluvial materials in the foothills (frederiksen 1988; soil survey staff 1960). in this region, the climate is subpolar with short, cool summers and long, snowy winters, influenced by antarctic ice masses and cold oceanic currents. mean daily air temperatures above 0°c are found only during three months a year, and the growing season is restricted to approximately five months. rainfall, including snowfall, reaches up to 600 mm per year. annual average wind speed outside the forests is 8 km h−1, reaching up to 100 km h−1 during storms (barrera et al. 2000; martínez pastur et al. 2009). study sites comprised monospecific n. pumilio forests and were determined using satellite images from different years and the database of the natural resources forest area of the province. sites were selected based on their high similarity in soil type, slope, elevation, and land-use history (martínez pastur et al. 1997). the harvested forest stands were selected in the central zone of the island, where it is possible to find old and recent cuttings corresponding to the proposed treatments. two treatments (types of forest management and period of time since the intervention) were considered. three types of forest management were selected: (i) harvested forest (hf) with shelter wood cutting; (ii) stockpile area (sa), an area in the forest used during times of harvest to further process stems or trees extracted from the forest, store them, and then load out the logs. this is a designated area that is usually cleared of obstacles such as trees and stumps, and can vary in size depending on the processing, storage and loading-out requirements. this area is ca. 60 x 25 m2 and represents 2-3% of the harvested forest (martínez pastur et al. 2007); (iii) control forest (cf), i.e. without intervention. harvesting took place during the summer, and assessments were done 1 year, between 5 and 10 years, and 50 years after intervention. nine stands of forest types (i) and (ii) were selected with three stands for each time period after intervention. in addition, nine unharvested old-growth forests (type iii) without signs of intervention were selected near each harvested forest. these old-growth forests consist of trees with similar diameter at breast height and dominant stand height (28 m total height, 528 trees ha-1, 40.6 cm diameter at breast height, 65.0 m2 ha-1 basal area; lencinas et al. 2011), corresponding to sites of quality ii according to martínez pastur et al. (1997). the three hf sites 1 year after intervention selected were: ewan river (ew 1), los cerros ranch (lc 1) and lenga patagonia ranch (lp 1). ushuaia ranch a (eua 5–10), ushuaia ranch b (eub 5–10) and ewan river (ew 5–10) were the three hf sites 5–10 years after intervention, and ushuaia ranch (eu 50), lenga patagonia ranch (lp 50) and reserva corazón de la isla (rci 50) were the three hf sites 50 years after intervention, as shown in figure 1. the size of each selected site was between 30–60 hectares. soil sampling a transect with five points every 10 m was established within each site. at each point of the transect, three composite samples of soil (each consisting of four subsamples) were collected from the mineral horizon (0–10 cm) using a hole borer according to dick et al. (1996). sampling was carried out in november 2009, may and september 2010, which corresponded to late spring, autumn and early spring, respectively. after collection, samples (between 500 g and 1 kg) were stored in plastic bags at 4°c until processing and elíades et al. new zealand journal of forestry science (2019) 49:7 page 3 transported to the laboratory. a fraction of 5 g from each sample was used for fungal isolation while another fraction was oven-dried (105°c) overnight to determine the moisture content. fungal isolation and identification each soil sample was processed by the soil washing method according to parkinson and williams (1961). a total of 100 soil particles from each sample was used, placed on plates containing cornmeal agar medium supplemented with 0.05% streptomycin sulfate and 0.025% chloramphenicol at rate of five particles per plate (a total of 20 plates by sample) and incubated at 25°c for 10 days (elíades et al. 2008). daily observations of the plates were performed under the microscope and a representative of each taxon registered on the particles at each sampling date was isolated. stock cultures were kept at 4°c on 2% (w v-1) agar-malt extract (mea) slants, lyophilised and deposited in the culture collection of “instituto spegazzini”, unlp, la plata, argentina (lpsc). for morphological identification, mea slide cultures of each isolate were prepared and mounted with lactophenol cotton blue to observe the structures differentiated by hyaline fungi. original taxonomic papers based on cultural and morphological features and compendia (ellis 1971; carmichael et al. 1980; domsch et al. 1993; cabello and arambarri 2002) were used to identify sporulating fungi. statistical analyses the community structure of soil fungi was analysed by (i) frequency (%): number of particles bearing a specific fungus / total number of particles analysed × 100 (godeas 1983); (ii) species richness (s); (iii) evenness (e); and (iv) shannon’s diversity index (h’). h’= ∑pi ln p, where pi is the relative abundance of the i species compared to the abundance of all identified species in a sample (magurran 1988; cabello & arambarri 2002). a two-way anova was performed to analyse the effects of forest management type and period of time since intervention (between-subject effects) and of season (i.e., sampling dates, as the within-subject effect) on s, e and h’. prior to analysis, normality and homoscedasticity of the data were tested in order to confirm that they fulfilled the assumptions required for anova. principal component analysis (pca; digby & kempton 1987) was performed with the frequency data of all species using the multivariate statistical package mvsp 3.1. (kovach 1999). wilks’ lambda test was applied to verify if the sample units in the pca analysis were mainly separated in the two axes by the sampling time, years after intervention or forest-management type. results total soil mycobiota the total mycobiota recovered from all 80 soil samples collected was represented by 70 taxa whose higher frequency at each site and sampling time is shown in tables 1–3. a representative isolate obtained from the particles at each sampling date was obtained. of the figure 1: location of sample sites in tierra del fuego (54° 27’s, 67° 27’w). (1) 1 year (ewan river); (2) 1 year (los cerros ranch); (3) 1 year (lenga patagonia ranch); (4) 5–10 years (ushuaia ranch a); (5) 5–10 years (ushuaia ranch b); (6) 5–10 years (ewan river); (7) 50 years (ushuaia ranch); (8) 50 years (lenga patagonia ranch); (9) 50 years (reserva corazón de la isla). elíades et al. new zealand journal of forestry science (2019) 49:7 page 4 fu n ga l s pe ci es n ov . 20 09 m ay 20 10 se p. 20 10 fu n ga l s pe ci es n ov . 20 09 m ay 20 10 se p. 20 10 fu n ga l s pe ci es n ov . 20 09 m ay 20 10 se p. 20 10 ab si di a co er ul ea 36 fu sa ri um s ul ph ur eo us 1. 3 pe ni ci lli um fr eq ue nt an s 20 .4 38 .5 10 0 ab si di a cy lin dr os po ra 16 30 .2 58 ge om yc es p an no ru m 6. 2 pe ni ci lli um n ig ri ca ns 28 .3 40 .7 ab si di a ra m os a 7. 4 h um ic ol a fu sc oa tr a 60 .3 pe ni ci lli um o do ra tu m 6. 6 1 ab si di a sp in os a 1 29 h um ic ol a gr is ea 9. 3 14 .6 18 .4 pe ni ci lli um p ur pu ra sc en s 6. 6 19 20 .4 ac re m on iu m c er ea le s 11 .8 h um ic ol op si s ce ph al os po ri oi de s 77 .2 58 .8 84 .6 pe ni ci lli um r es tr ic tu m 1 ac re m on iu m s p. 23 .5 3. 4 ye as t s p. 1 23 .5 3. 7 9. 1 pe ni ci lli um r ub ru m 5. 8 al te rn ar ia a lt er na ta 1 5. 9 ye as t s p. 2 2. 4 9. 1 pe ni ci lli um s p. 6. 6 1. 9 ar th ri ni um p ha eo sp er m un 0. 8 ye as t s p. 3 6. 1 pe ni ci lli um th om ii 6. 6 1. 2 3. 6 as pe rg ill us n ig er 6. 6 1. 3 2 m el an os po ra fa lla x 7. 4 ph om a he rb ar um 1. 56 as pe rg ill us te rr eu s 2. 2 2 9. 1 m et ar rh iz iu m a ni so pl ia e 1. 08 pi pt oc ep ha lis s p. 1. 9 as pe rg ill us u st us 3. 7 m or ti er el la h um ili s 30 3. 6 30 pu rp ur eo ci lli um li la ci nu m 1. 75 4. 9 b as id io m yc ot a m yc el iu m 8. 6 m or ti er el la h ia lin a 1. 1 24 22 rh iz op us s to lo ni fe r 3. 1 be au ve ri a ba ss ia na 19 .1 m or ti er el la p ar vi sp or a 5. 8 rh od ot or ul a sp . 16 .2 be au ve ri a br on gn ia rt ii 33 .3 40 .9 1. 9 m or ti er el la r am os a 5 tr ic ho de rm a ha m at um 3. 5 19 .2 cl ad os po ri um h er ba ru m 3. 1 m or ti er el la s p. 1 5. 1 3. 7 tr ic ho de rm a ha rz ia nu m 1 cl ad os po ri um c la do sp or io id es 5. 9 3. 1 70 .6 m or ti er el la s p. 2 2 tr ic ho de rm a ko ni ng ii 3. 1 7. 8 50 .9 cl on os ta ch ys r os ea 2 m or ti er el la v in ac ea 47 .8 19 .5 72 .7 tr ic ho de rm a po ly sp or um 83 .3 64 .8 65 .3 co rd an a pa uc is ep ta ta 4. 1 m uc or g en ev en si s 35 .5 12 u lo cl ad iu m b ot ry ti s 2 cy lin dr oc ar po n s p. 3. 8 m uc or h ie m al is 22 .5 18 .8 55 .8 cy lin dr oc ar po n di dy m un 6 m uc or m uc ed o 10 .3 4 cy lin dr oc ar po n fu eg ui an um 15 .2 m uc or s ub ti lis si m us 55 .6 22 .5 34 .7 cy lin dr oc ar po n te nu e 1 n ig ro sp or a sp ha er ic a 0. 9 5. 9 d ac ty liu m d en dr oi de s 66 .6 pa ec ilo m yc es c ar ne ou s 1. 9 d em at ia ce ou s st er ile m yc el iu m 5. 4 14 .3 pa ec ilo m yc es s p. 1 55 .6 fu sa ri um o xy sp or um 4. 4 pe ni ci lli um c an es ce ns 1. 9 10 no v. , n ov em be r; s ep t., s ep te m be r. ta b le 1 . h ig he r pe rc en ta ge fr eq ue nc ie s of s oi l f un gi is ol at ed in th re e sa m pl in g ti m es ( n ov em be r 20 09 , m ay 2 01 0 an d se pt em be r 20 10 ), w hi ch a re c om bi ne d to ta ls fr om a ll 27 s it es st ud ie d. elíades et al. new zealand journal of forestry science (2019) 49:7 page 5 fu n ga l s pe ci es t im e si n ce in te rv en ti on ( ye ar s) fu n ga l s pe ci es t im e si n ce in te rv en ti on ( ye ar s) fu n ga l s pe ci es t im e si n ce in te rv en ti on ( ye ar s) 1 510 50 1 510 50 1 510 50 ab si di a co er ul ea 36 fu sa ri um s ul ph ur eo us 1 1. 3 1. 2 pe ni ci lli um fr eq ue nt an s 40 20 10 0 ab si di a cy lin dr os po ra 58 25 .3 42 ge om yc es p an no ru m 6. 2 pe ni ci lli um n ig ri ca ns 34 .3 30 .9 40 .7 ab si di a ra m os e 7. 4 2. 8 2. 1 h um ic ol a fu sc oa tr a 43 .8 3. 6 60 .3 pe ni ci lli um o do ra tu m 6. 6 1 ab si di a sp in os e 29 19 27 .1 h um ic ol a gr is ea 14 .6 18 .4 9. 3 pe ni ci lli um p ur pu ra sc en s 20 .4 19 ac re m on iu m c er ea ls 11 .8 h um ic ol op si s ce ph al os po ri oi de s 54 .9 66 84 .6 pe ni ci lli um r es tr ic tu m 1 0. 9 ac re m on iu m s p. 23 .5 3. 4 5. 2 ye as t s p. 1 23 .5 10 .9 1. 03 pe ni ci lli um r ub ru m 5. 8 1. 3 0. 8 al te rn ar ia a lt er na ta 1 5. 9 ye as t s p. 2 2. 4 9. 1 pe ni ci lli um s p. 1. 9 6. 6 ar th ri ni um p ha eo sp er m un 0. 8 ye as t s p. 3 6. 1 pe ni ci lli um th om ii 1. 2 6. 6 2 as pe rg ill us n ig er 2 6. 6 m el an os po ra fa lla x 7. 4 ph om a he rb ar um 1. 56 as pe rg ill us te rr eu s 2 9. 1 0. 9 m et ar rh iz iu m a ni so pl ia e 1. 08 pi pt oc ep ha lis s p. 1. 9 as pe rg ill us u st us 3. 7 m or ti er el la h um ili s 7. 5 26 .4 30 pu rp ur eo ci lli um li la ci nu m 1. 1 0. 9 4. 9 b as id io m yc ot a m yc el iu m 8. 6 m or ti er el la h ia lin a 24 7. 6 22 rh iz op us s to lo ni fe r 3. 1 1. 1 be au ve ri a ba ss ia na 19 .1 6. 6 5. 1 m or ti er el la p ar vi sp or a 5. 8 rh od ot or ul a sp . 16 .2 2. 2 3. 5 be au ve ri a br on gn ia rt ii 27 10 .9 40 .9 m or ti er el la r am os a 5 tr ic ho de rm a ha m at um 3. 5 19 .2 3 cl ad os po ri um h er ba ru m 3. 1 m or ti er el la s p. 1 5. 1 tr ic ho de rm a ha rz ia nu m 1 cl ad os po ri um c la do sp or io id es 5. 9 70 .6 3. 1 m or ti er el la s p. 2 2 tr ic ho de rm a ko ni ng ii 30 .6 50 .9 3. 4 cl on os ta ch ys r os ea 2 m or ti er el la v in ac ea 19 .5 72 .7 22 .5 tr ic ho de rm a po ly sp or um 83 .3 77 .3 68 .5 co rd an a pa uc is ep ta ta 4. 1 m uc or g en ev en si s 6. 4 35 .5 7. 8 u lo cl ad iu m b ot ry ti s 2 cy lin dr oc ar po n s p. 3. 8 m uc or h ie m al is 16 .5 55 .8 14 cy lin dr oc ar po n di dy m un 6 m uc or m uc ed o 10 .3 4 8. 2 cy lin dr oc ar po n fu eg ui an um 15 .2 2. 1 m uc or s ub ti lis si m us 34 .7 14 55 .6 cy lin dr oc ar po n te nu e 1 1 n ig ro sp or a sp ha er ic a 5. 9 d ac ty liu m d en dr oi de s 0. 9 66 .6 pa ec ilo m yc es c ar ne ou s 1. 9 d em at ia ce ou s st er ile m yc el iu m 5. 4 2. 2 14 .3 pa ec ilo m yc es s p. 55 .6 16 .8 8. 5 fu sa ri um o xy sp or um 4. 4 3. 1 pe ni ci lli um c an es ce ns 7. 3 10 ta b le 2 . h ig he r p er ce nt ag e fr eq ue nc ie s of s oi l f un gi is ol at ed in th re e ti m es s in ce in te rv en ti on a ve ra ge d ov er th e th re e sa m pl in g da te s (n ov em be r 2 00 9, m ay 2 01 0 an d se pt em be r 2 01 0) . elíades et al. new zealand journal of forestry science (2019) 49:7 page 6 fu n ga l s pe ci es cf sa h f fu n ga l s pe ci es cf sa h f fu n ga l s pe ci es cf sa h f ab si di a co er ul ea 36 8 6 fu sa ri um s ul ph ur eo us 1 pe ni ci lli um fr eq ue nt an s 18 .4 28 10 0 ab si di a cy lin dr os po ra 54 58 48 ge om yc es p an no ru m 6. 2 pe ni ci lli um n ig ri ca ns 30 .9 40 .7 30 ab si di a ra m os a 2. 8 1. 1 7. 4 h um ic ol a fu sc oa tr a 60 .3 41 .6 45 .5 pe ni ci lli um o do ra tu m 6. 6 ab si di a sp in os a 29 .8 14 .6 15 .5 h um ic ol a gr is ea 18 .4 9. 3 14 .6 pe ni ci lli um p ur pu ra sc en s 6. 6 20 .4 14 ac re m on iu m c er ea le s 11 .8 h um ic ol op si s ce ph al os po ri oi de s 50 84 .6 58 .8 pe ni ci lli um r es tr ic tu m 1 0. 9 ac re m on iu m s p. 8. 1 5. 2 23 .5 ye as t s p. 1 23 .5 pe ni ci lli um r ub ru m 5. 8 1. 3 al te rn ar ia a lt er na ta 5. 9 ye as t s p. 2 9. 1 pe ni ci lli um s p. 6. 6 1. 9 ar th ri ni um p ha eo sp er m un 0. 8 ye as t s p. 3 6. 1 pe ni ci lli um th om ii 6. 6 3. 6 1. 9 as pe rg ill us n ig er 1. 3 6. 6 m el an os po ra fa lla x 7. 4 ph om a he rb ar um 1. 56 as pe rg ill us te rr eu s 9. 1 2. 2 0. 9 m et ar rh iz iu m a ni so pl ia e 1. 08 pi pt oc ep ha lis s p. 1. 9 as pe rg ill us u st us 3. 7 m or ti er el la h um ili s 30 pu rp ur eo ci lli um li la ci nu m 4. 9 1, 1 b as id io m yc ot a m yc el iu m 8. 6 m or ti er el la h ia lin a 24 17 .8 1. 1 rh iz op us s to lo ni fe r 1. 1 3. 1 be au ve ri a ba ss ia na 6 2. 1 19 .1 m or ti er el la p ar vi sp or a 5. 8 rh od ot or ul a sp . 16 ,2 3. 5 be au ve ri a br on gn ia rt ii 27 40 .9 m or ti er el la r am os a 2. 8 1. 1 7. 4 tr ic ho de rm a ha m at um 19 .2 3. 5 cl ad os po ri um h er ba ru m 3. 1 2 m or ti er el la s p. 1 3. 7 tr ic ho de rm a ha rz ia nu m 1 cl ad os po ri um c la do sp or io id es 3. 1 70 .6 m or ti er el la s p. 2 2 tr ic ho de rm a ko ni ng ii 16 .4 50 .9 30 .8 cl on os ta ch ys r os ea 2 2 m or ti er el la v in ac ea 72 .7 4. 1 19 .5 tr ic ho de rm a po ly sp or um 30 .8 46 65 .3 co rd an a pa uc is ep ta ta 4. 1 m uc or g en ev en si s 7. 8 35 .5 14 .7 u lo cl ad iu m b ot ry ti s 2 cy lin dr oc ar po n s p. 3. 8 m uc or h ie m al is 38 .5 55 .8 36 .4 cy lin dr oc ar po n di dy m un 6 m uc or m uc ed o 4. 9 10 .3 7 cy lin dr oc ar po n fu eg ui an um 15 .2 5 m uc or s ub ti lis si m us 55 .6 34 .7 4. 2 cy lin dr oc ar po n te nu e 1 n ig ro sp or a sp ha er ic a 0. 9 5. 9 d ac ty liu m d en dr oi de s 0. 9 66 .6 pa ec ilo m yc es c ar ne ou s 1. 9 4 d em at ia ce ou s st er ile m yc el iu m 5. 4 pa ec ilo m yc es s p. 55 .6 24 .7 fu sa ri um o xy sp or um 4. 4 3. 1 pe ni ci lli um c an es ce ns 10 cf , c on tr ol fo re st ; s a , s to ck pi le d ar ea ; h f, h ar ve st ed fo re st . ta b le 3 . h ig he r pe rc en ta ge fr eq ue nc ie s of s oi l f un gi is ol at ed in t hr ee t yp es o f f or es t m an ag em en t av er ag ed o ve r th e th re e sa m pl in g da te s (n ov em be r 20 09 , m ay 2 01 0 an d se pt em be r 20 10 ). elíades et al. new zealand journal of forestry science (2019) 49:7 page 7 fungi identified, most were in the ascomycota phylum although some representatives of mucoromycota were also found. the s, j and h’ values of the soil mycobiota are shown in appendices 1–3. sampling time was the only factor that generated significant differences in the values of s and h’ (table 4). the h’ is a parameter that includes s so a fisher’s least-significant-difference test was performed using h’ data for the samples corresponding to each forest management situation at the three sampling times (fig. 2). there were no significant differences between dates in the harvested forest (p≤0.01), although in may h’ indices were significantly higher than at the other two sampling dates for both undisturbed forests and stockpile areas (fig. 2). the pca performed with the frequencies of all fungal species showed that the first two axes accounted for 50.7% of the total variance explained (fig. 3). a wilks’ lambda test was highly significant with sampling time (wilk’s lambda: 0.003, f: 180.3, p<0.001), which grouped soil samples according to seasonality and not significant between years after intervention (wilk’s lambda: 0.957, f: 0.21, p>0.001) and types of forest management (wilk’s lambda: 0.957, f: 0.85, p>0.001). september 2010 samples were mainly represented by humicola fuscoatra, penicillium frequentans, trichoderma koningii and t. polysporum, while the november 2009 samples included beauveria brongniarti and mortierella vinacea, and the may 2010 samples included aspergillus niger and a. terreus. soil mycobiota at undisturbed sites the 12 most frequently obtained species found in the nine control forests at each of the three sampling dates are shown in figure 4. among these ones, mortierella vinacea, mucor subtilissimus, humicolopsis cephalosporioides, penicillium frequentans and trichoderma polysporum occurred most frequently in november 2009 (late-spring). absidia cylindrospora, a. spinosa, paecilomyces sp., p. frequentans, penicillium nigricans and t. polysporum characterised the soil samples corresponding to may 2010 (autumn), while humicola fuscoatra and humicolopsis cephalosporioides exhibited a high frequency in september 2010 (earlyspring), together with a. cylindrospora, mucor hiemalis, and t. polysporum. in these undisturbed sites, the species richness was around 4 and 13, the evenness was between 0.40-0.63 and the species diversity was 0.882.13. effect of forest-management type and time since intervention even though the sampling time was the variable that explained the separation of the units according to their composition, some species were present in recently exploited sites (acremonium cerealis, melanospora fallax and mortierella ramosa) and others in ones intervened 50 years ago (cladosporium herbarum, dactylium dendroides and geomyces pannorum). averaged across all three sampling dates, h. cephalosporioides and t. polysporum exhibited the highest frequencies, with the former being more abundant in sites with the shortest periods of time since intervention (1 and 5–10 years) and t. polysporum being more abundant in soils at sites after 50 years of intervention. however, there were no differences in s, j and h’ in soils of different forest-management type or time since intervention. discussion in our project, we analysed and compared the community structure of soil fungi under forests subjected to different management practices and periods of time since the intervention. however, species richness and diversity of soil mycobiota associated with n. pumilio forests estimated here did not confirm the hypothesis that forest management decreases the mycobiota composition. silviculture practices can be a potential source of stress that influences both the ecophysiology of forests and the associated soil microbiota. it is wellknown that forest management practices generate specific microclimate conditions due to changes in the humidity and temperature in the soil and canopy as well as in sunlight availability (aussenac 2000). s j h’ source of variation d.f. f p value f p value f p value fm 2 1.231 0.315 0.014 0.986 0.492 0.620 yi 2 0.625 0.547 0.589 0.565 0.671 0.523 fm x yi 4 0.937 0.465 0.908 0.480 1.952 0.145 s 2 20.258 0.000 2.030 0.146 13.185 0.000 s x fm 4 0.243 0.912 1.971 0.120 1.179 0.336 s x yi 4 0.832 0.513 0.059 0.993 0.451 0.771 s x fm x yi 8 0.490 0.855 0.491 0.855 0.520 0.833 table 4. results of repeated-measures anova on specific richness (s), equitability (j) and diversity index (h’) indicating the effect of sampling time (s), forest management (fm) and years from intervention (yi). figure 2: h’ in control forest (a), stockpile areas (b) and harvested forest (c) at the three periods of time since intervention in november 2009 (nov), may 2010 (may) and september 2010 (sept). values are means that correspond to three forest sites (replicates). the asterisk (*) denotes significant differences between sampling dates for each period of time since the intervention according to anova and fisher’s lsd test (p ≤ 0.05). table 2: confusion matrix elíades et al. new zealand journal of forestry science (2019) 49:7 page 8 to date, there is no information about the impact of seasonal changes on the soil microfungi diversity in forests of n. pumilio in tierra del fuego subjected to different management practices and consequently to different degrees of tree cover. however, this information can be of key relevance for determining the forest productivity status and therefore contribute to the development of new sustainable management strategies (martínez pastur et al. 2009). higher s and/or h’ were found in undisturbed forests and stockpile areas in autumn (may 2010) compared to those from the other two seasons we analysed. similarly, differences in the diversity of microfungi associated with seasonality and temperature conditions in cold environments were reported by coleine et al. (2015) and rodolfi et al. (2015). voříšková et al. (2014) also found that the soil fungal community under a temperate oak forest was affected by seasonality, with the highest number of genera found in autumn. the two processes that probably contributed most to those differences were litter decomposition and allocation of photosynthates. since temperature can also affect nutrient recycling in n. pumilio forests and consequently influence soil biological activity (including microfungi and their enzymes), further analysis is necessary to demonstrate whether these differences in the diversity of microfungi are in fact related to the amount and availability of specific nutrients. preliminary studies on the in-vitro conditions of h. cephalosporioides, a fungus with a high frequency in forest soils of n. pumilio in tierra del fuego, revealed that its enzyme activity is affected by the temperature of incubation (elíades et al. 2015). all study sites were exposed to the same stressful conditions that prevailed due to the seasonal presence of snow (characteristic of the sub–antarctic climate) but this did not seem to have affected fungal diversity in the harvested forest with shelter wood. one possible reason could be the existence of a mosaic of vegetation types and floristic composition in disturbed ecosystems that minimised the regional effect of climate. bradley et al. (2001) compared the chemical and microbial properties of the forest floor between shelter wood, adjacent old-growth and clear-cut plots in the montane coastal western hemlock of british columbia (canada). these authors found that forest floor can develop under shelter wood plots with atypical properties of either clear-cut or old-growth plots. the absence of differences in the community of soil microfungi associated with different forest management practices and periods of time since intervention observed in the present study could be due to mechanisms of ecological compensation that mitigated the impact of the intervention. the presence of new substrates (such as wood particles and other organic remains), and of microhabitats equivalent to the original ones as a result of cutting and delimbing of trees, may contribute to the restoration of appropriate conditions for soil fungi to thrive, leading to the maintenance of a similar fungal community. yet, this also depends on the ability of each fungal species to survive in the modified environment. soil fungi are an ecological group composed of generalist representatives, as in the case of most aspergillus and penicillium, which are able to survive in both natural and man-made environments due to their ability to use a wide variety of carbon sources for growth (kowalczyk et al. 2014). ecological compensation is a mechanism that allows these fungi to survive different kinds of disturbances and therefore these organisms might play an important role in the sustainable development of a region (wang et al. 2007). they can establisher new a. b. c. 0 1 2 3 1. 5 to 10 50 h’ time from intervention (years) november 2009 may 2010 . september 2010 0 1 2 3 1 5 to 10 50 h’ time from intervention (years) november 2009 may 2010 . september 2010 0 1 2 3 1 5 to 10 50 h’ time from intervention (years) november 2009 may 2010 . september 2010 * * * * * * elíades et al. new zealand journal of forestry science (2019) 49:7 page 9 figure 3: pca of soil mycobiota data at three collection dates (november 2009, circles; may 2010, inverted triangles; september 2010, squares) with site codes above the symbols. the main species vectors are indicated with arrows. fungal names are abbreviated. red symbols indicate control samples. the code of each site is formed by: period of time (in years) since intervention (1, 5-10 or 50), followed by type of forest management (control forest, cf; stockpiled area, sa; harvest forest, hf) and season sampling (november 2009, n; may 2010, m; september 2010, s). the abbreviations of the taxa names are: abs coe, absidia coerulea; abs cyl, absidia cylindrospora; asp nig, aspergillus niger; asp ter, aspergillus terreus; bea bro, beauveria brongniartii; hum fus, humicola fuscoatra; hum cep, humicolopsis cephalosporioides; lev, yeast sp.1; mor hum, mortierella humilis; mor par, mortierella parvispora; mor vio, mortierella vinacea; muc hie, mucor hiemalis; muc sub, mucor subtilissimus; pen fre, penicillium frequentans; pen pur, penicillium purpurascens; pen tho, penicillium thomii; pip sp., piptocephalis sp.; tri kon, trichoderma koningii; tri pol, trichoderma polysporum; ulo bot, ulocladium botrytis. figure 4: accumulated frequency of the 12 most abundant fungal species in samples from the three control forests at each of the three sampling dates (november, november 2009; may, may 2010; september, september 2010) and their average values (ave). biotic interactions that often result in changes to the environment and activate inducible metabolic pathways that allow their growth under stressful conditions such as when nutrient resources are scarce (troncozo et al. 2015). this includes phenotypic plasticity according to available organic substrates and synthesis of lytic enzymes involved in soil formation, which have adaptive and ecological significance (franco et al. 2018). lin et al. (2016) reported that the soil mycobiota of a taiwanese cryptomeria japonica forest regained system stability and recovered from tree thinning disturbance in a relatively short period of time. forest management practices, including harvesting and forest conversion, could affect the soil microbial community in montane forest (chang et al. 2017). therefore, additional studies including other recently disturbed sites of n. pumilio forests in tierra del fuego are needed. conclusions we showed that the diversity of soil mycobiota associated with n. pumilio forests was not affected by silviculture practices and time since intervention. although 70 fungal taxa were recovered from these soils, a change in s and/or h’ was only found for undisturbed forests and stockpile areas in autumn compared to those from seasons more favourable to plant growth. therefore, our results indicate that forest harvesting per se does not affect the diversity of soil mycobiota in n. pumilio forests in tierra del fuego, since there were no changes in any of the structural parameters analysed associated with the harvested forest with sheltering wood cutting. competing interests the authors declare that they have no competing interests funding this study was partially supported by grants from agencia nacional de promoción científica y tecnológica (pict 2015-1620 for saparrat m.c.n.), comisión de investigaciones científicas de la provincia de buenos aires (cicpba for cabello m.n.) and universidad nacional de la plata (unlp, 11/n 773 for cabello m.n.), argentina. authors’ contributions lae conducted most of the planning, experimental design, sample processing, identification, data analysis and the writing of the manuscript. mnc advised on the identification of fungal isolates and was involved in the data analysis. vp conducted the field work. am helped with fieldwork. naf helped performing different tasks in laboratory. mcns was involved in the analysis and discussion of the data and assisted with the writing of the paper. mdb advised on the application of statistical tools and was involved in the statistical analysis of the data. all authors read and approved the final version of the manuscript. references aussenac, g. (2000). interactions between forest stands and microclimate: ecophysiological aspects and consequences for silviculture. annals of forest science, 57, 287-301. barrera, m., frangi, j., richter, l., perdomo, m., & pinedo, l. (2000). structural and functional changes in nothofagus pumilio forests along an altitudinal gradient in tierra del fuego, argentina. journal of vegetation science, 11, 179-188. bradley, r., titus, b., & hogg, k. (2001). does shelterwood harvesting have less impact on forest floor nutrient availability and microbial properties than clearcutting? biology and fertility of soils, 34, 162-169. cabello, m., & arambarri, a. (2002). diversity in soil fungi from undisturbed and disturbed celtis tala and scutia buxifolia forests in the eastern buenos aires province (argentina). microbiological research, 157, 115-125. carmichael, j., kendrick, w., conners, i., & sigler, l. (1980). genera of hyphomycetes. edmonton, alberta, canada: the university of alberta press. chang, e., tian, g., & chiu, c. (2017). soil microbial communities in natural and managed cloud montane forests. forests, 8(33), 1-9. coleine, c., selbman, l., ventura, s., d’aqui, l., onofri, s., & zucconi, l. (2015). fungal biodiversity in the alpine tarfala valley. microorganisms, 3, 612-624. dick, r., thomas, d., & turco, r. (1996). standardized methods, sampling and sampling treatment. in j. doran & a. jones (eds.), methods for assessing soil quality (pp. 107-121). madison, wi: soil science society of america. digby, p., & kempton, r. (1987). multivariate analysis of ecological communities. london – new york: chapman and hall. domsch, k., gams, w., & anderson, t. (1993). compendium of soil fungi. eching, germany: ihwverlag. elíades, l., voget, c., arambarri, a., & cabello, m. (2008). fungal communities on decaying saltgrass (distichlis spicata) in buenos aires province (argentina). sydowia, 59(2), 227-234. elíades, l., cabello, m., pancotto, v., moretto, a., rago, m., & saparrat, m. (2015). preliminary data on growth and enzymatic abilities of soil fungus humicolopsis cephalosporioides at different incubation temperatures. revista iberoamericana de micología, 32(1), 40-45. ellis, m. (1971). dematiaceous hyphomycetes. england: commonwealth mycological institute. franco, e., troncozo, m., baez, m., mirífico, m., robledo, g., balatti, p., & saparrat, m. (2018). fusarium elíades et al. new zealand journal of forestry science (2019) 49:7 page 10 equiseti lpsc 1166 and its in vitro role in the decay of heterostachys ritteriana leaf litter. folia microbiologica, 63, 169-179. frederiksen, p. (1988). soils of tierra del fuego, a satellite-based land survey approach. folia geographica danica, 18, 1-159. gea, g., martínez pastur, g., cellini, j., & lencinas, m. (2004). forty years of silvicultural management in southern nothofagus pumilio (poepp. et endl.) krasser primary forests. forest ecology and management, 201(2-3), 335-347. godeas, a. (1983). qualitative and quantitative studies of soil fungi from the nothofagus dombeyi forest. ciencia del suelo, 1(1), 21-31. green, j., & bohannan, b. (2006). spatial scaling of microbial biodiversity. trends in ecology & evolution, 21, 501-507. hartmann, m., niklaus, p., zimmermann, s., schmutz, s., kremer, j., abarenkov, k., lüscher, p., widmer, f., & frey, b. (2014). resistance and resilience of the forest soil microbiome to logging-associated compaction. the isme journal, 8(1), 226-244. jurgensen, m., harvey, a., graham r., page-dumroese, d., tonn, j., larsen, m., & jain, t. (1997). impacts of timber harvesting on soil organic matter, nitrogen, productivity, and health of inland northwest forests. annals of forest science, 43, 234-251. kovach, w. (1999). mvsp-a multivariate statistical package for windows, version 3.1. pentraeth: kovach computing services. kowalczyk, j., benoit, i., & de vries, r. (2014). regulation of plant biomass utilization in aspergillus. advances in applied microbiology, 88, 31-56. lencinas, m., pastur, g., gallo, e., & cellini, j. (2011). alternative silvicultural practices with variable retention to improve understory plant diversity conservation in southern patagonian forests. forest ecology and management, 262, 1236-1250. lin, w., wang, p., chen, w., lai, c., & winder, r. (2016). responses of soil fungal populations and communities to the thinning of cryptomeria japonica forests. microbes and environments, 31, 19-26. magurran, a. (1988). ecological diversity and its measurement. london: croom helm. martínez pastur, g., peri, p., vukasovic, r., vaccaro, s., & piriz carrillo, v. (1997). site index equation for nothofagus pumilio patagonian forests. phyton, 6(1-2), 55-60. martínez pastur, g., cellini, j., peri, p., vukasovic, r., & fernández, c. (2000). timber production of nothofagus pumilio forests by a shelter wood system in tierra del fuego (argentina). forest ecology and management, 134(1-3), 153-162. martínez pastur, g., peri, p., fernández, m., staffieri, g., & lencinas, m. (2002). changes in understory species diversity during the nothofagus pumilio forest management cycle. journal of forest research, 7, 165-174. martínez pastur, g., lencinas, m., peri, p., moretto, a., cellini, j., mormeneo, i., & vukasovic, r. (2007). harvesting adaptation to biodiversity conservation in sawmill industry: technology innovation and monitoring program. journal of technology management & innovation, 2, 58-70. martínez pastur, g., cellini, j., peri, p., lencinas, m., gallo, e., & soler, e. (2009). alternative silviculture with variable retention in timber management of south patagonia. forest ecology and management, 258, 436-443. moore, d. (1983). flora of tierra del fuego. london: anthony nelsonmissouri botanical garden. parkinson, d., & williams, s. (1961). a method for isolating fungi from soil microhabitats. plant and soil, 13, 347-355. rodolfi, m., oliveira longa, c., pertot, i., tosi, s., savino, e., guglielmineti, m., altobelli, e., del frate, g., picco, a. (2015). fungal biodiversity in the periglacial soil of dosde glacier (valtellina, northern italy). journal of basic microbiology, 55, 1-12. scorsetti, a., elíades, l., stenglein, s., cabello, m., pelizza, s., & saparrat, m. (2012). pathogenic and enzyme activities of the entomopathogenic fungus tolypocladium cylindrosporum (ascomycota: hypocreales) from tierra del fuego, argentina. revista de biología tropical, 60, 833-841. soil survey staff (1960). soil classification, a comprehensive system 7th approximation. washington, d.c.: usda. u.s. government printing office. tedersoo, l., bahram, m., põlme, s., kõljalg, u., yorou, n., wijesundera, r., villarreal ruiz, l., vasco-palacios, a., thu, p., suija, a., smith, m., sharp, c., saluveer, e., saitta, a., rosas, m., riit, t., ratkowsky, d., pritsch, k., põldmaa, k., piepenbring, m., phosri, c., peterson, m., parts, k., pärtel, k., otsing, e., nouhra, e., njouonkou, a., nilsson, r., morgado, l., mayor, j., may, t., majuakim, l., lodge, d., lee, s., larsson, k., kohout, p., hosaka, k., hiiesalu, i., henkel, t., harend, h., guo, l., greslebin, a,. grelet, g., geml, j., gates, g., dunstan, w., dunk, c., drenkhan, r., dearnaley, j., de kesel, a., dang, t., chen, x., buegger, f., brearley, f., bonito, g., anslan, s., abell, s., & abarenkov, k. (2014). global diversity and geography of soil fungi. science, 346 (6213), pp. 1078 and 1256688/1-1256688/10. troncozo, m., gómez, r., arambarri, a., balatti, p., bucsinszky, a., & saparrat, m. (2015). growth and oxidative enzymatic activity of in-vitro cultures of ciliochorella buxifolia. mycoscience, 56, 58-65. elíades et al. new zealand journal of forestry science (2019) 49:7 page 11 tuhkanen, s., kuokka, i., hyvonen, j., stenross, s., & niemela, j. (1990). tierra del fuego as a target for biogeographical research in the past and present. anales del instituto de la patagonia, serie ciencias naturales, 19(2), 1-107. voříšková, j., brabcová, v., cajthaml, t., & baldrian, p. (2014). seasonal dynamics of fungal communities in a temperate oak forest soil. new phytologist, 201, 269-278. wang, y., ding, x., qian, l., & wang, h. (2007). ecological compensation mechanism for urban green land and its application in shanghai, china. frontiers of environmental science & engineering in china, 1(3), 320-324. appendix 1. richness (s), evenness (j) and shannon weiner index (h’) from control forest (cf), stockpiled area (sa) and harvested forest (hf) of the studied sites in november 2009. cf sa hf site* s j h’ s j h’ s j h’ lp 1 6 0.54 1.41 12 0.41 1.47 12 0.48 1.72 lc 1 7 0.47 1.32 6 0.24 0.64 7 0.54 1.51 ew 1 9 0.47 1.51 9 0.35 1.11 6 0.62 1.62 ew 5-10 8 0.31 0.94 10 0.32 1.07 6 0.43 1.11 eua 5-10 9 0.50 1.61 5 0.34 0.80 8 0.42 1.27 eub 5-10 4 0.55 1.11 9 0.40 1.29 8 0.56 1.70 lp 50 8 0.52 1.56 9 0.52 1.67 10 0.36 1.19 eu 50 8 0.57 1.71 4 0.34 0.69 2 0.63 0.63 rci 50 8 0.37 1.13 10 0.44 1.47 7 0.38 1.08 *lp 1, lenga patagonia ranch; lc 1, los cerros ranch; ew 1, ewan river; ew 5–10, ewan river; eua 5–10, ushuaia ranch a; eub 5–10, ushuaia ranch b; eu 50, ushuaia ranch; lp 50, lenga patagonia ranch; rci 50, reserva corazón de la isla. for more information to see the section “methods”. appendix 2. richness (s), evenness (j) and shannon weiner index (h’) from control forest (cf), stockpiled area (sa) and harvested forest (hf) of the studied sites in may 2010. cf sa hf site* s j h’ s j h’ s j h’ lp 1 8 0.47 1.42 13 0.56 2.09 13 0.52 1.94 lc 1 10 0.63 2.11 10 0.53 1.77 11 0.52 1.80 ew 1 12 0.56 2.04 14 0.57 2.17 7 0.46 1.31 ew 5-10 11 0.53 1.84 14 0.57 2.19 5 0.51 1.20 eua 5-10 14 0.51 1.95 12 0.61 2.19 eub 5-10 12 0.55 1.98 10 0.60 2.02 8 0.49 1.47 lp 50 10 0.49 1.64 10 0.50 1.68 8 0.59 1.79 eu 50 11 0.57 1.98 13 0.61 2.28 10 0.40 1.35 rci 50 13 0.57 2.13 7 0.57 1.60 12 0.44 1.59 *lp 1, lenga patagonia ranch; lc 1, los cerros ranch; ew 1, ewan river; ew 5–10, ewan river; eua 5–10, ushuaia ranch a; eub 5–10, ushuaia ranch b; eu 50, ushuaia ranch; lp 50, lenga patagonia ranch; rci 50, reserva corazón de la isla. for more information to see the section “methods”. elíades et al. new zealand journal of forestry science (2019) 49:7 page 12 appendix 3. richness (s), evenness (j) and shannon weiner index (h’) from control forest (cf), stockpiled area (sa) and harvested forest (hf) of the studied sites in september 2010. cf sa hf site* s j h’ s j h’ s j h’ lp 1 5 0.40 0.94 7 0.52 1.47 7 0.43 1.21 lc 1 5 0.50 1.17 7 0.55 1.55 8 0.61 1.83 ew 1 4 0.49 0.99 4 0.52 1.04 5 0.47 1.09 ew 5-10 4 0.44 0.88 7 0.43 1.20 4 0.44 0.88 eua 5-10 6 0.48 1.24 6 0.43 1.13 2 0.54 1.95 eub 5-10 8 0.53 1.59 8 0.60 1.80 6 0.43 1.12 lp 50 7 0.45 1.28 6 0.51 1.32 1 0 0 eu 50 6 0.57 1.47 2 0.42 0.42 2 0.68 0.68 rci 50 5 0.55 1.29 8 0.61 1.84 6 0.56 1.46 *lp 1, lenga patagonia ranch; lc 1, los cerros ranch; ew 1, ewan river; ew 5–10, ewan river; eua 5–10, ushuaia ranch a; eub 5–10, ushuaia ranch b; eu 50, ushuaia ranch; lp 50, lenga patagonia ranch; rci 50, reserva corazón de la isla. for more information to see the section “methods”. elíades et al. new zealand journal of forestry science (2019) 49:7 page 13 strategic expansion of existing forest monitoring plots: a case study using a stratified gis-based modelling approach thai son le1,* and justin morgenroth2 1vietnam national university of forestry, hanoi, vietnam 2new zealand school of forestry, university of canterbury, christchurch 8140, new zealand *corresponding author: thaisonfuv@gmail.com (received for publication 10 february 2019; accepted in revised form 14 october 2020) abstract background: understanding the relationship between sites and the plant species they support is essential for effective vegetation management. site-species matching requires knowledge of the growth response of a given species to the full range of environmental conditions in potential planting sites. this can be achieved by repeatedly measuring species growth at a comprehensive network of sample plots that cover a range of environmental conditions, including topography, climate, and soil factors. the new zealand dryland forests initiative has established permanent sample plots (psps) of a plantation species, eucalyptus bosistoana f.muell., across new zealand. however, these psps do not cover the entire range of environmental conditions available for the species and hence there is a need to expand the network of sites. the aim of this study was to determine optimal locations for new psps to provide more unique information to support site-species matching studies for eucalyptus bosistoana in new zealand. methods: a geographic information system (gis) and stratified random sampling method were used to generate a model to identify optimal locations for e. bosistoana psp establishment. the variables used in this study included topography, climate, and soil data. redundancy between the initial set of potential explanatory variables was reduced by a multi-collinearity analysis. the potential habitat for the species was restricted to land with environmental conditions that could support e. bosistoana. all environmental variables were stratified and an initial priority index for each stratum in each variable was calculated. then a weighted-overlay analysis was conducted to create the final priority index, which was mapped to identify high-priority areas for targeted psp expansion. results: the existing psp network for e. bosistoana generally covers the environmental conditions in low-elevation new zealand dry lands, which are located alongside the east coast of the south island, and the southern part of the north island. the model identified high priority areas for psp expansion, including several large regions in the north island, especially in rangitikei and taupo districts. conclusions: the model successfully allowed identification of areas for a strategic expansion of permanent sample plots for e. bosistoana. newly identified areas expand upon the topographic, climatic, and soil conditions represented by the existing psp network. the new area for psp expansion has potential to provide valuable information for further site-species matching studies. the methodology in this paper has potential to be used for other plot networks of a different species, or even natural forests. new zealand journal of forestry science le and morgonroth new zealand journal of forestry science (2020) 50:9 https://doi.org/10.33494/nzjfs502020x41x e-issn: 1179-5395 published on-line: 05/11/2020 © the author(s). 2020 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access keywords: eucalyptus bosistoana; gis; habitat modelling; nzdfi; permanent sample plot network; site-species matching; stratified random sampling. http://creativecommons.org/licenses/by/4.0/), le and morgonroth new zealand journal of forestry science (2020) 50:9 page 2 introduction the use of forest monitoring plots is important for forest resource protection and management. for plantation forest development, permanent sample plots (psp) provide information on how trees grow, when and which silvicultural practices are required, and how effective they are if applied. this information is crucial for the decision making process of any plantation management plan and can be used to support decisions about whether more plantings of a given species should be established (millen et al. 2016). understanding the relationship between a species and its site requirements is essential to the success of the plantation. the new zealand dryland forest initiative (nzdfi) has used permanent sample plots to monitor the survival and growth of durable eucalypts not previously planted in large numbers in new zealand. one particular species being trialed by the nzdfi is eucalyptus bosistoana f.muell., which has its natural habitat in the southeastern coastal areas of australia (boland et al. 2006). this species has good potential as a plantation species in new zealand due to its abilities to produce highly durable timber, to coppice vigorously after fire and harvesting, and to provide nectar/pollen for native biodiversity (apiolaza et al. 2011; millen et al. 2016; nicholas & millen 2012). despite many positive characteristics, there are some concerns about its susceptibility to pests, namely defoliation from the eucalyptus variegated beetle (paropsisterna variicollis (chapuis)) (lin et al. 2017). with its hardness, wood density, straightness, and durability, the timber of this species has been used for numerous purposes such as farming fences, building materials, boat masts and railway ties. of particular relevance is that this species has high drought tolerance that satisfies the prerequisite for plantations in new zealand drylands (apiolaza et al. 2011; millen et al. 2016; nicholas & millen 2012). to determine whether this species can be commercially successful in new zealand, the nzdfi aims to ensure its trials are well distributed throughout new zealand’s drylands such that they can be used for growth and yield modelling. the growth and yield models will help determine the sites with the greatest growth rates and survival. as of 2015 the nzdfi had established and monitored 84 permanent sample plots (psps) of e. bosistoana in 30 sites across the north and south islands of new zealand (nzdfi 2015). however, there is a desire to expand the psp network to span a more complete range of environmental conditions suitable for eucalypt plantations in new zealand. the extended psp network will provide a greater understanding of how e. bosistoana grows in a variety of environmental situations, especially in marginal conditions (nzdfi 2015, 2017). to fill the gaps in the existing psp network, new psps should be established at sites with environmental conditions that are not well represented in the current network. habitat modelling is often used to examine the set of conditions within the habitat of a particular species or a group of species. in this approach, a habitat is considered as a group of separate factors (i.e. terrain, climate and soil). this approach was introduced by james and shugart (1970) for bird habitat modelling. in the early days of habitat modelling, the approach was often used to model the habitat or distribution of birds (e.g. johnston and temple (1986) or mammals (e.g. pereira (1989); pausas et al. (1995). these habitat models were typically created by collecting site description data in terms of site properties (i.e. model variables) to determine their statistical relationships with the distribution of a species. such statistical assessments usually require large amounts of empirical data that are not easily collected. as computing technology became more widely available, habitat modelling took advantage of these processing advances. a computer-based model, climex, was developed and used to describe the climatic favourability of a given location for a particular animal species (sutherst & maywald 1985). climex modelling requires climatic data and population data for the target species (sutherst & maywald 1985; sutherst et al. 2007). climex and its applications have enabled ecologists to utilise computing power for habitat modelling (pattison & mack 2008; shabani et al. 2012; taylor et al. 2012). recently, climex has been applied to examining the potential distribution of invasive plants and insects (aljaryian et al. 2016; hill et al. 2016; xuezhen et al. 2018). in new zealand, a form of habitat modelling has been widely used to assess the potential productivity of plantation forest species across a range of environmental conditions. two common indices used to predict the productivity of pinus radiata include the 300 index (defined as the stem volume mean annual increment at age 30 years (kimberley et al. 2005)) and site index (defined as the mean top height at age 20 years (goulding 2005)). watt et al. (2009) modelled the spatial distribution of cupressus lusitanica productivity. these indices reflect the potential growth of the target species in their habitat based on surrounding environmental conditions. environmental site descriptions are costly and time-consuming to undertake, especially in the case of large habitat areas. however, with the development of computer science and geographic information systems (gis), gis-based habitat models have become more widespread relative to traditional statistical habitat models because of their applicability on large spatial extents (basir 2014; reisinger & kennedy 1990; store & kangas 2001). in gis-based habitat modelling, each environmental variable is represented by a gis layer (i.e. either vector or raster). as a result, an environmental description of every spatial location in the habitat can be obtained (dettmers & bart 1999; shaw & atkinson 1988; wadge et al. 1993). gis-based habitat modelling has good potential to meet the nzdfi’s objective of strategically expanding their psp network for e. bosistoana. a challenge involved with this objective is how to allocate new sample plots optimally across a large study area to cover the wide range of environmental variability. simple random sampling tends to collect samples throughout the range of values in the population if those values appear with similar frequencies (green 1979; royall 1970). however, this method often involves a high risk of bias when applied on naturally distributed population, such as environmental objects on a large geographic surface (cawsey et al. 2002; danz et al. 2003; kohl et al. 2006). systematic sampling may solve the problem of the simple random sampling method but it cannot guarantee that samples from all important value gradients are taken into account unless the sampling density (i.e. the number of samples over a unit of area) is sufficiently high (j. e. austin et al. 2001; scott 1998). due to these limitations, neither random, nor systematic sampling are appropriate for expanding the e. bosistoana psp network. alternatively, to ensure the samples are distributed representatively across the entire range of values, the stratified random sampling method has been widely used, especially for low-density samples over an large areas (e.g. national survey) or an isolated natural area that restricts the ability to collect samples (esfahani & dougherty 2014; tomppo et al. 2014; yves & ecker 2014). indeed, if the environmental conditions within a specific area are stratified into several separate groups, taking samples from these groups will provide more representative information about the population (m. p. austin & heyligers 1991). stratified random sampling has been increasingly adopted in ecological and forestry studies (danz et al. 2003; knollova et al. 2005; wallenius et al. 2011; yves & ecker 2014). this study takes advantage of the merits of gis-based habitat modelling and stratified random sampling to find priority locations for a strategic expansion of the existing permanent sample plot network for e. bosistoana in new zealand. the resulting priority map highlights the environmental gaps in the existing plot network that need to be filled by the strategic expansion. although this study is specific to e. bosistoana, the methodology has general applicability as it involves a gis-based model of habitat that is easy to modify and adjust for other species. methods study area the study area includes the north and south islands of new zealand, which cover a total of approximately 268,000 km2 (fig. 1). new zealand has a wide range of topographic and climatic conditions, ranging from sea level to 3737 m above sea level (barringer et al. 2002). the annual average temperature ranges from -2.55 to 16.79 °c and annual precipitation ranges from 392 to 6807 mm year-1 (fick & hijmans 2017). tree species and permanent sample plots “coast grey box” or “gippsland grey box” (eucalyptus bosistoana) is a species naturally found on the southeast coast of australia (boland et al. 2006). it can reach a height of approximately 40 to 60 m at maturity. the habitat of this species is coastal mixed forests located in areas below 500 m of elevation within a range of latitudes from 33 to 37.5°s (apiolaza et al. 2011; boland et al. 2006). the climatic range is warm humid to cool, with a mean maximum monthly temperature of 24-29 °c in the hot season and a mean minimum temperature of 1-6 °c in the winter months. it can withstand 5-40 frost days per year. the mean average precipitation in its natural range is 700-1200 mm a year (boland et al. 2006), though it is currently planted on sites in nz with less than 700 mm of annual rainfall (nzdfi 2019). e. bosistoana has been planted at 30 sites across new zealand to support the nzdfi research programme. within these sites, 84 psps have been established, and together, these psps are considered as a psp network. the psps are further divided into 1095 sub-plots (fig. 1). description of data used in the study the environmental factors available for use in this study include climate, soil, and topography (table 1). climate data are from worldclim and include average temperatures and precipitation based on the period from 1970-2000 (fick & hijmans 2017). soil data are from the new zealand land resource inventory (newsome et al. 2008), while the digital elevation model (dem) was sourced from landcare research who interpolated the surface from land information new zealand’s 1:50,000 topographic spot heights and contours. le and morgonroth new zealand journal of forestry science (2020) 50:9 page 3 figure 1: the study area (source: statistics nz 2016) and existing permanent sample plots for e. bosistoana. general modelling approach the method applied in this study consisted of five steps, each of which is detailed below: (1) identification of variables for gis-based modelling, (2) building the dataset, (3) variable restriction, (4) variable stratification, and (5) cartographic modelling. identification of variables for gis-modelling according to studies by apiolaza et al. (2011) and prober et al. (2016) and based on the available sources of data, 17 variables were selected as potentially influencing the growth and distribution of e. bosistoana (table 2). the multi-collinearity analysis, which involved an assessment of variation inflation factors (vif), was undertaken to test the correlations between the variables to minimise information redundancy. vif values near 1 indicate that the variables were independent, while vifs exceeding 10 were indicative of multi-collinearity requiring correction (garcía et al. 2014; kutner et al. 2003). the next step was to determine whether a weighting should be applied to the remaining variables to highlight certain variables as having greater impact upon the growth and development of e. bosistoana. information from previous studies (apiolaza et al. 2011; boland et al. 2006; nzdfi 2015) and expert advice (eg mason pers. comm) contributed to the decision-making process. four le and morgonroth new zealand journal of forestry science (2020) 50:9 page 4 variables were deemed as the most influential factors on the target species, including annual average temperature, precipitation, soil ph and elevation. these variables were given a weight coefficient of 1.5 whereas all other variables were assigned a weight coefficient of 1. building the dataset in this study, data were derived from existing published data or through interpolation of field data in cases of large areas (table 2). all acquired data were processed into raster layers (one layer corresponding to a single model variable). all the environmental variables (table 2) were processed to raster-format layers with cell size of 25 × 25 m in the same projected coordinate system (i.e. new zealand transverse mercator) before being clipped to the study area boundary. the data processing stage was conducted using arcgis version 10.4 (esri, redlands, ca, usa) and saga-gis 4.2 (conrad et al. 2015). variable restriction potential trial planting sites in new zealand should have similar environmental conditions to the native habitat of e. bosistoana in australia, such that the species has a reasonable chance of surviving and providing growth and yield. variable value restriction was used to restrict category raw data unit source data type resolution climate monthly mean temperature °c worldclim version2 raster 1 × 1 km monthly minimum temperature °c monthly maximum temperature °c monthly precipitation mm soil potential rooting depth m new zealand land resource inventory (nzlri) vector n/a soil ph n/a soil salinity % soil temperature regime classified (*) profile available water in soil mm topography digital elevation model (dem) m landcare research raster 25 × 25 m boundaries territorial authority (2016 generalized version) n/a statistics new zealand vector n/a nz area units (2015 yearly pattern) n/a statistics new zealand vector n/a land cover database version 4.1 n/a landcare research vector n/a table 1: gis data used for habitat modelling including their sources, type of data and spatial resolution (*) soil temperature regime classes include: t thermic; wm warm mesic; mm mild mesic; cm cool mesic; dm cold mesic and c – cryic (webb & wilson 1995). the potential planting sites in new zealand to those having similar environmental conditions to those in the natura range of e. bosistoana. there were two reasons for restricting the site availability. first, some areas had current land uses that were not consistent with plantation forestry (e.g. settlement area, horticulture, and indigenous forests). these areas were identified using the new zealand land cover database (lcdb v4.1) and subsequently excluded from the analysis. second, areas were also excluded if they had environmental conditions that deviated considerably from the native habitat of the target species (e.g. permanent ice or extremely low precipitation that the species could not survive). restricting the study area based on environmental conditions considered the environmental conditions appearing in the natural habitat of e. bosistonana in south-eastern australia, those of existing trial plantations throughout nz, values found in previous studies (boland et al. 2006; grieve et al. 1999; webb & wilson 1995) and, where necessary, expert knowledge (eg mason, pers. comm.). ranges for each of the 17 environmental variables were inferred from one or more of these four sources; the specifics are detailed below. we obtained data, including 1596 recorded points of e. bosistoana occurrence from its natural range in southeastern australia (ala 2017). the occurrence locations were overlaid with environmental attribute data from the atlas of living australia (ala), worldclim, and csiro ecosystem sciences (ala 2017) to identify the ranges of environmental conditions (e.g. temperature, precipitation) that e. bosistoana tolerates in its native habitat. in addition to the natural range data, the species was successfully established in 30 sites throughout new zealand. these trial plots provided additional information about the range of conditions in which the species could survive. le and morgonroth new zealand journal of forestry science (2020) 50:9 page 5 variable category variable id variable unit source 1. climate v1.1 annual average temperature °c worldclim version2 (the determination of cold, hot, dry, and wet seasons based on the publication by leathwick et al. (1998)) v1.2 average monthly minimum temperature of the cold season °c v1.3 average monthly maximum temperature of the hot season °c v1.4 annual precipitation mm/year v1.5 monthly precipitation of the dry season mm 2. soil v2.1 potential rooting depth m new zealand land resource inventory (nzlri) v2.2 soil ph none v2.3 soil salinity % v2.4 soil temperature regime classified v2.5 profile available water in soil mm 3. topography v3.1 elevation m landcare research v3.2 slope degree derived from dem (burrough & mcdonell 1998) v3.3 aspect degree derived from dem (burrough & mcdonell 1998) v3.4 curvature none derived from dem (zeverbergen & thorne 1987) v3.5 terrain ruggedness index tri none derived from dem (riley et al. 1999) v3.6 topographic wetness index twi none derived from dem (beven & kirkby 1979) v3.7 wind exposition index none derived from dem (gerlitz et al. 2015) table 2: the environmental variables of the habitat model finally, a number of previous studies on eucalypts provided guidance to help determine the thresholds for environmental conditions that the species can survive (boland et al. 1984; grieve et al. 1999). in brief, the three sources of data together contributed to determining the recorded value range of each variable describing the habitat of e. bosistoana. to avoid being overly restrictive (i.e. causing loss of potential habitat area) and to create chances for the species to adapt to conditions in new areas that are marginally outside the current habitat conditions, the recorded value range was generally buffered by positive and negative 10% to create the “restriction value range” for each variable. some exceptions to buffering the recorded value range existed. the study did not apply an upper limit for the restriction value range of the variable average monthly min temperature of the cold season as higher values of this variable would be better for the development of the target species (millen et al. 2016). in practice, this restriction value range also contained values in australia’s conditions that did not exist within the study area (i.e. new zealand). an interference of the restriction value range in association with the range of values available in the study area was used to produce the “value range of interest” for each variable. once the value range of interest was defined for each variable layer, all pixels with values outside the range of interest, or those with unsuitable land covers, were replaced by no-data pixels, such that they would be excluded from subsequent analyses. variable stratification and standardisation the environmental raster layers used in this study contained continuous values. even though the full value range for any given environmental variable was restricted by the value ranges of interest, no sample plot system could effectively provide sufficient samples to cover every value in those ranges. the stratified random sampling approach allowed the study to conduct an analysis on a relatively small number of groups of values instead of a wide range containing continuous values. following the stratified random sampling approach, the value range of interest for each variable was stratified into non-overlapping compartments, also called strata. a stratum should include values reflecting similar conditions (e.g. cold weather or uneven surface) or a particular level of characteristics of the conditions at the corresponding locations (e.g. low slope or high soil ph). defining strata requires an understanding of the full range of values present for each variable, and then specifying breakpoints within the data; the breakpoints act simultaneously as an upper limit for one stratum and a lower limit for the next stratum. breakpoints for some of the variables used in this study were based on existing published values (webb and wilson 1995; newsome et al. 2008), while others were based on an assessment of the frequency distribution of a given variable. the stratification of soil variables, slope and aspect were carried out mainly based on descriptions of new zealand landscapes by webb and wilson (1995) and newsome et al. (2008). for example, the soil ph value range, in general including values from 0 to 14, was divided into six strata: very low (<4.9), low (4.9–5.4), moderately low (5.5–5.7), near neutral (5.8– 6.4), moderately high (6.5–7.5) and high (>7.5). for variables where published strata were not available, an alternative approach was required. climatic variables and topographic variables (other than slope and aspect), were stratified following the jenks natural breaks algorithm (jenks & caspall 1971). this stratification uses the frequency distribution of data for a particular variable and identifies natural groupings inherent in the data. breakpoints are identified to achieve groups with similar numbers of observations and to maximise the differences between strata. in this way, the variables were divided into strata that have relatively big differences in the data values (de smith et al. 2015). after stratification, values of variables from existing plot locations were extracted and then assigned into strata to count the frequency of occurrence within each stratum (i.e. the number of times that values from existing plots appear in each stratum). the study used this frequency as an indicator for priority. the priority indicator increases with decreasing frequency in a stratum. the lower the frequency, the greater the need for new plots in that stratum. the next step was to calculate a normalised priority indicator for each stratum. this ensured that all priorities were normalised on a scale of 0–100, using equation (1): pj = 100 × (1 – fj fmax -1) equation (1) where pj and fj are the normalised priority and the frequency of the stratum j respectively, fmax is the highest frequency in the variable. for example, assuming a variable was divided into three strata with the corresponding frequencies: f1 = 200, f2 = 160 and f3 = 20. thus, f1 = 200 is the maximum frequency and fmax = f1 = 200. then, the normalised priority for the strata was calculated following equation (1): p1 = 0, p2 = 20 and p3 = 90 respectively. this result was interpreted such that stratum 3, with normalised priority value of 90, had the highest priority. the example demonstrated that this normalisation enables a quantitative evaluation of priority in which the higher normalised priority values meant the higher priority to establish forest plots on locations with attribute values within the stratum. after the calculation of normalised priorities for all strata in all variables, the map layers were reclassified such that every pixel received the normalised priority value of the stratum which contained the pixel value. in other words, each map layer of a variable was converted to a priority layer that highlighted high-priority pixels on the map in respect to the particular environmental variable. cartographic modelling the last step was to produce the map of priority index (i.e. final priority point) for each pixel within the feasible area. this index was calculated on each pixel as the weighted sum of normalised priority values le and morgonroth new zealand journal of forestry science (2020) 50:9 page 6 from all variables by the priority function as shown in equation (2): equation (2) where p was the priority index, n was the number of environmental variables, pi was the normalised priority value of the variable i, and wi was the weight coefficient of the variable i. in particular, the priority index was calculated for each pixel by a weighted sum of, for example, 17 normalised priority values corresponding to the 17 selected variables (i.e. n = 17). the weighted sum was calculated by multiplying each normalised priority layer by its weight coefficient and then summing the weighted normalised priority values from all variable grids using an algebraic overlay analysis. in this stage, the coefficient of 1.5 was used for the four variables (i.e. annual average temperature, precipitation, soil ph and elevation) and a coefficient of 1 was used for all other variables. with each normalised priority value ranging between 0–100, the priority index in this example could have values between 0–1,900. le and morgonroth new zealand journal of forestry science (2020) 50:9 page 7 results model structure after all the variable layers were created, we estimated the dependence between model variables in each category to detect any multi-collinearity. vif values of all variables to identify multi-collinearity are shown in table 3. the vifs of variables 3.2 (slope) and 3.5 (terrain ruggedness index – tri) considerably exceeded 10.0 (i.e. approximately 31.2 and 32.2, respectively). this was interpreted as there being a high correlation between these variables, which could lead to information redundancy and overlapping in the model. to solve this multi-collinearity problem, tri was removed from further consideration because slope is easier to measure/calculate and is more common in forestry research as compared to tri. estimation of the priority index variable restriction value ranges of interest of the variables are presented in table 4. most of the variables, except topographic aspect and soil salinity, had ranges of interest narrower than table 3: multi-collinearity analysis of each variable category variable id detectiontolerance (%) variance inflation factor (vif) multi-collinearity present? (a) climatic variables annual average temperature 82.66 1.21 false average monthly min temperature of the cold season 76.12 1.31 false average monthly max temperature of the hot season 72.91 1.37 false annual precipitation 35.50 2.82 false monthly precipitation of the dry season 41.56 2.40 false (b) soil variables potential rooting depth 89.90 1.11 false soil ph 88.98 1.12 false soil salinity 96.11 1.04 false soil temperature regime 89.59 1.12 false profile available water in soil 97.64 1.02 false (c) topographic variables elevation 29.29 3.41 false slope 3.21 31.18 true aspect 54.80 1.82 false curvature 66.67 1.50 false terrain ruggedness index tri 3.20 31.24 true topographic wetness index twi 85.19 1.17 false wind exposition index 42.43 2.36 false le and morgonroth new zealand journal of forestry science (2020) 50:9 page 8 va ri ab le r an ge fr om n at iv e ha bi ta t r an ge fr om ex is ti n g p sp s fr om re fe re n ce s r ec or de d va lu e ra n ge r es tr ic ti on va lu e ra n ge st ud y ar ea r an ge r an ge o f in te re st a nn ua l a ve ra ge te m pe ra tu re 9. 00 – 1 8. 00 10 .1 0 – 14 .1 0 9. 00 – 1 8. 00 8. 10 – 1 8. 90 -2 .5 5 – 16 .7 9 8. 10 – 1 6. 79 av er ag e m on th ly m in te m pe ra tu re of th e co ld s ea so n -2 .0 – 8 .0 0. 5 – 5. 7 n o up pe r lim it -2 .0 – 8 .0 ≥ -3 .0 -1 1. 93 – 1 2. 2 -3 .0 – 1 2. 2 av er ag e m on th ly m ax te m pe ra tu re of th e ho t s ea so n 22 .2 0 – 29 .8 0 18 .9 0 – 23 .3 0 18 .9 0 – 29 .8 0 18 .1 4 – 30 .5 6 5. 97 – 2 4. 87 18 .1 4 – 24 .8 7 a nn ua l p re ci pi ta ti on 63 6 – 19 16 69 3 – 18 77 63 6 – 19 16 50 8 – 20 44 39 2 – 68 07 50 8 – 20 44 m on th ly p re ci pi ta ti on o f t he d ry se as on 39 .0 – 1 00 .0 42 .0 – 1 35 .0 39 .0 – 1 35 .0 32 .9 – 1 41 .1 28 .0 – 6 39 .0 32 .9 – 1 41 .1 po te nt ia l r oo ti ng d ep th 0. 40 – 1 .4 0 0. 29 – 1 .3 5 0. 29 – 1 .4 0 0. 19 – 1 .5 0 – 1. 35 0. 19 – 1 .3 5 so il ph 4. 0 – 5. 0 5. 2 – 8. 0 4. 0 – 8. 0 3. 9 – 8. 1 4. 7 – 8. 0 4. 7 – 8. 0 so il sa lin it y n /a = 0. 02 ≤ 2. 90 * 0. 02 – 2 .9 0 0. 02 – 2 .9 0 0. 02 – 0 .8 5 0. 02 – 0 .8 5 so il te m pe ra tu re r eg im e n /a cm – w m ≥ cm * * cm – w m cm – t c – t cm – t pr of ile a va ila bl e w at er in s oi l 32 .0 – 1 99 .0 35 .0 – 3 00 .0 32 .0 – 3 00 .0 15 .3 – 3 16 .7 0 – 30 0. 0 15 .3 – 3 00 .0 el ev at io n 0 – 11 96 .0 15 .0 – 6 40 .0 0 – 11 96 .0 -1 19 .6 – 1 31 5. 6 -6 0. 0 – 37 37 .0 -6 0 – 13 15 .6 sl op e 0 – 25 .5 7 0 – 33 .6 2 0 – 33 .6 2 -2 .5 6 – 36 .1 7 0 – 87 .3 2 0 – 36 .1 7 a sp ec t -1 – 3 59 .9 6 -1 – 3 51 .5 3 -1 – 3 59 .9 6 -3 7. 10 – 3 60 .0 0 -1 – 3 59 .9 4 -1 – 3 59 .9 4 cu rv at ur e -7 .5 4 – 36 .8 2 -3 .8 4 – 7. 52 -7 .5 4 – 36 .8 2 -1 1. 98 – 4 1. 26 -3 36 .0 0 – 13 50 .4 0 -1 1. 98 – 4 1. 26 to po gr ap hi c w et ne ss in de x t w i 1. 00 2 – 15 .8 05 7. 30 4 – 16 .0 12 1. 00 2 – 16 .0 12 -0 .4 78 – 1 7. 49 2 3. 37 7 – 26 .4 21 3. 37 7 – 17 .4 92 w in d ex po si ti on in de x 0. 72 2 – 1. 31 3 0. 81 5 – 1. 26 5 0. 72 2 – 1. 31 3 0. 66 3 – 1. 37 2 0. 75 0 – 1. 35 3 0. 75 0 – 1. 35 3 ta b le 4 : r es ul ts o f v ar ia bl e re st ri ct io n *( gr ie ve e t a l. 19 99 ; w eb b & w ils on 1 99 5) . ** (b ol an d et a l. 19 84 ; n zd fi 2 01 5) . the corresponding value ranges for the study area (i.e. all the values appeared in the study area). these conditions together determined available areas for the expansion of the existing psps (fig. 2). in the variable value restriction process, areas to be excluded from the variables were either areas with values out of the value range of interest or areas under undesirable land covers (e.g. urban area, indigenous forest and mangrove area). through this process, approximately 71.25% of the study area (i.e. 189,055.7 km2), primarily in the south island, was excluded, including high mountainous areas, indigenous forests, and a large natural reserve area in the north island (fig. 2). variable stratification and standardisation the results of variable stratification are in table 5. from the counted frequency of each stratum, the normalised priority values of strata in each variable were calculated as in table 6. overlay analysis the final result was a map representing priority index values over the study area (fig. 3). the maximum and minimum of the priority index were recorded at 1,506.5 and 20.5 respectively, with high values indicating areas with under-represented environmental conditions amongst the existing psp network. the locations of the 1095 existing psps were also added to the map. obviously, these locations distributed among low-priority areas as their environmental characteristics were well covered. in general, high index areas were mainly in high elevation areas where the conditions in terms of the three types of variables were quite different than those in existing psps. the highest priority areas for e. bosistoana were in rangitikei district (i.e. moawhango) and taupo district (i.e. broadlands, rangitaiki, and tongariro). the other high priority zones included northland, auckland and gisborne regions, and southeast-facing hillsides of the mountain chains in central south island. discussion the main objective of the model in this paper was to detect locations for new psps, which would enhance the effectiveness of the current psp network for eucalyptus bosistoana in new zealand. the modelling results successfully allowed us to identify areas for a strategic forest plot expansion in new zealand with the greatest potential to provide valuable information for site-species interaction. the capabilities of gis were used in this study in association with the stratified random sampling method to create a flexible habitat model that was used to identify le and morgonroth new zealand journal of forestry science (2020) 50:9 page 9 figure 2: map of the study area before (a) and after (b) variable restriction (value pixel: feasible for plot expansion; non-value pixel: excluded by variable restriction). the map inset in figure 2a provides an example of the large number of sub-plots within any given point on the map. le and morgonroth new zealand journal of forestry science (2020) 50:9 page 10 va ri ab le r an ge o f i n te re st m in . va lu e b re ak po in t ( up pe r li m it o f t he s tr at um ) r ef er en ce s fo r st ra ti fi ca ti on s1 s2 s3 s4 s5 s6 s7 s8 s9 a nn ua l a ve ra ge te m pe ra tu re 8. 1 – 16 .7 9 8. 1 8. 93 9. 68 10 .4 6 11 .2 4 12 .0 0 12 .7 7 13 .6 1 14 .5 6 16 .7 9 n at ur al b re ak s av er ag e m on th ly m in te m pe ra tu re of th e co ld s ea so n -3 – 1 2. 2 -3 -1 .6 7 -0 .4 0 0. 80 2. 03 3. 23 4. 47 5. 83 7. 57 12 .2 0 av er ag e m on th ly m ax te m pe ra tu re of th e ho t s ea so n 18 .1 4 – 24 .8 7 18 .1 4 18 .8 0 19 .4 7 20 .1 0 20 .7 3 21 .3 3 21 .9 0 22 .5 0 23 .1 3 24 .8 7 a nn ua l p re ci pi ta ti on 50 8 – 20 44 50 8 73 0 90 1 10 67 12 25 13 77 15 22 16 78 18 50 20 44 m on th ly p re ci pi ta ti on o f t he d ry se as on 32 .9 – 1 41 .1 32 .9 51 .0 61 .0 71 .0 82 .0 93 .0 10 4. 0 11 5. 0 12 7. 0 14 1. 1 po te nt ia l r oo ti ng d ep th 0. 19 – 1 .3 5 0. 19 0. 25 0. 45 0. 6 0. 9 1. 2 1. 35 (n ew so m e et a l. 20 08 ; w eb b & w ils on 1 99 5) so il ph 4. 7 – 8 4. 7 4. 9 5. 5 5. 8 6. 5 7. 6 8. 0 so il sa lin it y 0. 02 – 0 .8 5 0. 02 0. 05 0. 15 0. 3 0. 7 0. 85 so il te m pe ra tu re r eg im e cm – t cm cm m m w m t pr of ile a va ila bl e w at er in s oi l 15 .3 – 3 00 15 .3 30 .0 0 60 90 15 0 25 0 30 0 el ev at io n -6 0 – 13 15 .6 -6 0 11 0 23 5 36 5 50 0 64 1 79 1 95 1 11 25 13 15 n at ur al b re ak s sl op e 0 – 36 .1 74 27 0 4. 00 8. 00 12 .0 0 16 .0 0 21 .0 0 26 .0 0 36 .1 8 (w eb b & w ils on 19 95 ) a sp ec t -1 – 3 59 .9 4 -1 fl at n n e e se s sw w n w (* ) cu rv at ur e -1 1. 97 6 – 41 .2 56 -1 1. 97 6 -3 .5 2 -1 .1 2 0. 32 1. 76 3. 36 5. 12 7. 04 10 .0 8 41 .2 6 n at ur al b re ak s to po gr ap hi c w et ne ss in de x t w i 3. 37 7 – 17 .4 92 3. 37 7 7. 25 8. 06 8. 87 9. 77 10 .7 6 12 .0 2 13 .4 6 15 .1 7 17 .5 0 w in d ex po si ti on in de x 0. 75 1 – 1. 35 0 0. 75 1 0. 82 0. 88 0. 95 1. 02 1. 09 1. 15 1. 22 1. 29 1. 36 ta b le 5 : r es ul ts o f v ar ia bl e st ra ti fic at io n (* ) st ra ti fie d as fl at s ur fa ce a nd 8 c ar di na l-i nt er ca rd in al d ir ec ti on s: f la t ( -1 ); n ( 0 – 22 .5 ; 3 37 .5 – 3 59 .9 9) ; n e (2 2. 5 – 67 .5 ); e ( 67 .5 – 1 12 .5 ); s e (1 12 .5 – 1 57 .5 ); s ( 15 7. 5 – 20 2. 5) ; s w ( 20 2. 5 – 24 7. 5) ; w ( 24 7. 5 – 29 2. 5) ; n w (2 92 .5 – 3 37 .5 ). le and morgonroth new zealand journal of forestry science (2020) 50:9 page 11 va ri ab le st ra tu m 1 st ra tu m 2 st ra tu m 3 st ra tu m 4 st ra tu m 5 st ra tu m 6 st ra tu m 7 st ra tu m 8 st ra tu m 9 f p f p f p f p f p f p f p f p f p a nn ua l a ve ra ge te m pe ra tu re 0 10 0 0 10 0 2 10 0 0 10 0 38 4 24 50 7 0 19 9 61 3 99 0 10 0 av er ag e m on th ly m in te m pe ra tu re o f t he c ol d se as on 0 10 0 0 10 0 2 99 37 0 2 37 8 0 20 6 46 13 9 63 0 10 0 0 10 0 av er ag e m on th ly m ax te m pe ra tu re o f t he h ot s ea so n 0 10 0 4 99 79 78 85 76 35 4 0 20 2 43 35 4 0 11 97 6 98 a nn ua l p re ci pi ta ti on 27 3 24 35 9 0 27 6 23 15 6 57 16 96 2 99 8 98 3 99 2 99 m on th ly p re ci pi ta ti on o f t he dr y se as on 25 0 44 36 7 17 44 4 0 2 10 0 12 97 9 98 6 99 0 10 0 5 99 po te nt ia l r oo ti ng d ep th 0 10 0 51 8 0 15 0 71 37 9 27 9 98 39 92 so il ph 0 10 0 44 8 9 23 95 49 5 0 12 7 74 2 10 0 so il sa lin it y 10 95 0 0 10 0 0 10 0 0 10 0 0 10 0 so il te m pe ra tu re r eg im e 13 0 84 80 9 0 15 6 81 0 10 0 pr of ile a va ila bl e w at er in s oi l 0 10 0 35 2 39 12 3 79 57 5 0 43 93 2 10 0 el ev at io n 48 7 0 21 7 55 37 6 23 13 97 2 10 0 0 10 0 0 10 0 0 10 0 0 10 0 sl op e 46 7 0 15 0 68 18 2 61 14 6 69 11 7 75 31 93 2 10 0 a sp ec t 10 7 57 52 79 24 6 0 12 6 49 23 91 16 3 34 96 61 22 7 8 55 78 cu rv at ur e 1 10 0 16 98 87 2 0 18 7 79 13 99 5 99 0 10 0 1 10 0 0 10 0 to po gr ap hi c w et ne ss in de x tw i 0 10 0 10 7 53 22 7 0 17 5 23 19 0 16 12 9 43 21 8 4 48 79 1 10 0 w in d ex po si ti on in de x 1 10 0 14 97 29 7 26 40 2 0 21 4 47 51 87 11 2 72 4 99 0 10 0 ta b le 6 : f re qu en cy ( f) a nd n or m al is ed p ri or it y va lu es ( p) fo r ea ch c om bi na ti on o f v ar ia bl e an d st ra tu m . priority locations for psp expansion. gis was used here as a platform for contributing to the decision making process via variable data management, variable layer production, normalised priority calculation by means of spatial analysis, a weighted-overlaying combination of variables, and finally, mapping by means of cartographic modelling. the data processing and output of the spatial results prove the potential of gis-based modelling in generating information for large-scale studies. the most obvious advantage of the method used in this study is the possibility to generate priority indices for forest expansion in such a huge area as a whole country (i.e. new zealand nationwide) consuming reasonable time and labor resources. this time and money-saving method enabled the study to adopt available environmental data in combination with expert knowledge to build a habitat model for a species to be strategically planted. performing analyses in the habitat model allowed the study to spatially utilise normalised priority functions when dealing with different species and different existing forest plot systems in the same study area even if the area was huge with diverse environmental conditions. in the context of a new habitat for the species, the study set the habitat restriction based on actual occurrences of the species in its natural habitat and in the established sample plots within the study area. however, every species has its own tolerance ability to adapt to a wide range of environmental conditions. in this case, there were insufficient studies on the plasticity of the species, especially in respect to each environmental factor. we extended the range of each variable by 10% to reflect the potential for the species to adapt to a new habitat. however, the example of radiata pine (i.e. pinus radiata) raised the issue of how much adaptable e. bosistoana could be. indeed, radiata pine was introduced to new zealand in the twentieth century and it has developed well in a much broader range of conditions than those present in its natural habitat (mead 2013; mpi 2016; weston 1957). using an inappropriate value for the assumption of species’ plasticity may reduce the probability of identifying correct ecological thresholds. as a consequence, it is possible that variable buffering could result in either omitting significant suitable areas of potential new habitat (i.e. too small a buffer), or in adding unsuitable areas (i.e. too large a buffer) to the model. in this case study, the approach involved an assumption that the model input data was error-free. however, the evaluation processes used types of information that are often uncertain and imprecise. these problems may arise from errors from data measurement and processing (fischer & wang 2011; karger et al. 2017; pielou 1984). most of the information to create variable layers was collected from various providers who have different ways of managing data, and the rest resulted from spatial analysis. these sources of information had the most significant influence on the model quality because their errors were almost systematic and propagated all over the study area (m. p. austin & heyligers 1991; pielou 1984; sellars & jolls 2007). for example, the study by pearse et al. (2015) highlighted inaccuracies in the soil data used in the present study. however, the model in this paper had to use that soil data because it is the best dataset in existence for the study area. for gis-based modelling, it is possible to analyze the sensitivity of the model’s results to the uncertainties and uncertainty propagation of data. the solution should consider the analytical error propagation method suggested by store and kangas (2001) and the monte carlo simulation described by burrough and mcdonell (1998). the information from existing plots may also be affected by gps inaccuracy, but this type of errors were less than 10 m (nzdfi 2015) that were not crucial at the large scale of the study area. in general, the case study used a new flexible technique that can be applied to a wide range of contexts. the habitat model was built as a description of environmental conditions across new zealand. similarly, with the development of global environmental data by remote sensing, such habitat models can be easily built for other parts of the world. obviously, these models can be widely used for the expansion of any plot network of any species to be introduced to a new habitat. although le and morgonroth new zealand journal of forestry science (2020) 50:9 page 12 figure 3: the priority index map following the same procedure, each study subject requires a specific set of variables and its own criteria for variable restriction as well as particular methods for variable stratification. another potential use of this technique is to identify locations for the expansion of forest inventory plots, especially in natural forests. indeed, environmental conditions among natural forests were unevenly distributed which leads to heterogeneous forest compositions and structures. following the habitat modelling approach, an inventory plot system after expansion can better cover the whole range of environmental habitats of the forest statuses. this is crucial to achieve more precise and more accurate inventory result over the region of interest. conclusions the results indicate that it was inappropriate to plant e. bosistoana in some parts of the south island, such as the rainforest areas in westland, areas too high and too cold deep in the south of the island. the existing psp network for the species is over-represented with environmental conditions present in low-elevation new zealand dry lands, which are located alongside the east coast of the south island, and the southern part of the north island. moreover, high priority areas for further trial plots of this species included several large regions in the north island, such as northland, auckland and gisborne regions and especially some smaller parts in the center of the north island in taupo and rangitikei districts. plantations of this species should also be tested at higher elevation (e.g. in mountainous areas of the canterbury, marlborough and tasman regions). we have developed a new approach involving a priority index to determine strategic expansion of forest monitoring plots that uses gis-based models in association with stratified random sampling. the technique used in this study was illustrated by a case study that successfully identified optimal sites for new establishments of forest plots of eucalyptus bosistoana in new zealand. new plots in these sites, if established, will provide crucial information for the site-species matching programme of the nzdfi. we hope that, by describing the methodology in this study, a broader base will help forest resource professionals and researchers able to build gis-based habitat models and apply these models in creating more adequate and efficient plot network designs to monitor and assess forests and the surrounding conditions as well as relationships between them. for further studies, we recommend using more variables to better describe the environment surrounding psps, if possible, especially in terms of soil conditions that critically influence a plant’s growth. however, it is important to find references determining the restriction ranges regarding the new variables to be added. moreover, trial plantings of the target species should be established in areas with marginal environmental conditions to identify the species’ plasticity. this could provide more significant information to enhance the performance of the model. the model in this paper can also be supported by statistical methods to determine appropriate weight coefficients for the model variables in consideration. it is recommended to use the analytical hierarchy process method, which is one of the most popular methods to obtain variables’ weights in gisbased modelling (carver 1991; chen et al. 2010; marinoni et al. 2009). competing interests the authors declare that they have no competing interests. authors’ contributions tsl analysed the data and wrote the manuscript under the supervision of jm. both authors read and approved the final version of the manuscript. acknowledgements the authors are thankful to the nzdfi for the permission to use the data of psps. they also appreciate the useful discussions and knowledge provided by professor euan mason. finally, the authors are grateful to the reviewers and editor whose comments improved the manuscript. references aljaryian, r., kumar, l., & taylor, s. (2016). modelling the current and potential future distributions of the sunn pest eurygaster integriceps (hemiptera: scutelleridae) using climex. pest management science, 72(10), 1989-2000. https://doi. org/10.1002/ps.4247 apiolaza, l., mcconnochie, r., millen, p., van ballekom, s., & walker, j.c.f. (2011). introducing durable species to new zealand drylands: genetics of early adaption of eucalyptus bosistoana. paper presented at the developing a eucalypt resource-learning from australia and elsewhere, blenheim workshop proceedings, wood technology research centre. austin, j.e., buhl, t.k., guntenspergen, g.r., norling, w., & sklebar, h.t. (2001). duck populations as indicators of landscape condition in the prairie pothole region. environmental monitoring and assessment, 69(1), 29-47. https://doi. org/10.1023/a:1010748527667 austin, m.p., & heyligers, p.c. (1991). vegetation survey design, a new approach: gradsect sampling. in: c. r. margules & m. p. austin (eds.), nature conservation: cost effective biological surveys and data analysis (pp. 31-36). melbourne, australia: csiro. barringer, j.r.f., pairman, d., & mcneill, s.j. (2002). development of a high-resolution digital elevation model for new zealand. retrieved from https:// lris.scinfo.org.nz/document/9213/download/ le and morgonroth new zealand journal of forestry science (2020) 50:9 page 13 https://doi.org/10.1002/ps.4247 https://doi.org/10.1002/ps.4247 https://doi.org/10.1023/a:1010748527667 https://doi.org/10.1023/a:1010748527667 https://lris.scinfo.org.nz/document/9213/download/ https://lris.scinfo.org.nz/document/9213/download/ basir. (2014). gis-based approach to participatory land suitability analysis for tree plantations. (doctor of philosophy in natural resources and environmental sciences), university of illinois, urbana-champaign. beven, k.j., & kirkby, m.j. (1979). a physically based, variable contributing area model of basin hydrology. hydrological sciences bulletin, 24(1), 43-69. https://doi.org/10.1080/02626667909491834 boland, d.j., brooker, m.i.h., chippendale, g.m., hall, n., hyland, b.p.m., johnston, r.d., kleinig, d.a., mcdonald, m.w., & turner, j.d. (eds.) (2006). forest trees of australia (5th ed). melbourne, australia: csiro. https://doi.org/10.1071/9780643069701 boland, d.j., brooker, m.i.h., chippendale, g.m., hall, n., hyland, b.p.m., johnston, r.d., kleinig, d.a., & turner, j.d. (eds.) (1984). forest trees of australia. melbourne, australia: csiro. burrough, p.a., & mcdonell, r.a. (1998). principles of geographical information systems. new york: oxford university press. carver, s.j. (1991). integrating multi-criteria evaluation with geographical information systems. international journal of geographical information system, 5(3), 321-339. https://doi. org/10.1080/02693799108927858 cawsey, e.m., austin, m.p., & baker, b.l. (2002). regional vegetation mapping in australia: a case study in the practical use of statistical modelling. biodiversity and conservation, 11(12), 2239-2274. https://doi. org/10.1023/a:1021350813586 chen, y., yu, j., & khan, s. (2010). spatial sensitivity analysis of multi-criteria weights in gis-based land suitability evaluation. environmental modelling & software, 25(12), 1582-1591. https://doi. org/10.1016/j.envsoft.2010.06.001 conrad, o., bechtel, b., bock, m., dietrich, h., fischer, e., gerlitz, l., wehberg, j., wichmann, v., & böhner, j. (2015). system for automated geoscientific analyses (saga) v. 2.1.4. geoscientific model development, 8(7), 1991-2007. https://doi. org/10.5194/gmd-8-1991-2015 danz, n.p., regal, r.r., niemi, g.j., brady, v.j., hollenhorst, t., johnson, l.b., host, g.e., hanowski, j.m., johnston, c.a., brown, t., kingston, j., & kelly, j.r. (2003). environmentally stratified sampling design for the development of great lakes environmental indicators. environmental monitoring and assessment, 102(1), 41-65. https:// doi.org/10.1007/s10661-005-1594-8 de smith, m.j., goodchild, m.f., & longley, p.a. (2015). geospatial analysis. a comprehensive guide to principles, techniques and software tools (5th ed). kibworth harcourt, uk: troubador publishing ltd. dettmers, r., & bart, j. (1999). a gis modeling method applied to predicting forest songbird habitat. ecological applications, 9(1), 152-163. https://doi. org/10.1890/1051-0761(1999)009[0152:agmm at]2.0.co;2 esfahani, m.s., & dougherty, e.r. (2014). effect of separate sampling on classification accuracy. bioinformatics, 30(2), 242-250. https://doi. org/10.1093/bioinformatics/btt662 fick, s.e., & hijmans, r.j. (2017). worldclim 2: new 1-km spatial resolution climate surfaces for global land areas. international journal of climatology, 37(12), 4302-4315. https://doi.org/10.1002/joc.5086 fischer, m.m., & wang, j. (2011). spatial data analysis: models, methods and techniques. berlin/ heidelberg, germany: springer science & business media. garcía, c., garcía, j., lópez martín, m., & salmerón, r. (2014). collinearity: revisiting the variance inflation factor in ridge regression. journal of applied statistics, 42(3), 648-661. https://doi.org/ 10.1080/02664763.2014.980789 gerlitz, l., conrad, o., & bohner, j. (2015). large-scale atmospheric forcing and topographic modification of precipitation rates over high asia a neuralnetwork-based approach. earth system dynamics, 6(1), 61-81. https://doi.org/10.5194/esd-6-612015 goulding, c.j. (2005). measurement of trees. in m. colley (ed.), the nzif forestry handbook (4th ed.). christchurch, new zealand: new zealand institute of forestry. green, r.h. (1979). sampling design and statistical methods for environmental biologists. hoboken, new jersey, usa: john wiley & son. grieve, c.m., guzy, m.r., poss, j.a., & shannon, m.c. (1999). screening eucalyptus clones for salt tolerance. horticulture science, 34(5), 867-870. https://doi. org/10.21273/hortsci.34.5.867 hill, m.p., bertelsmeier, c., clusella-trullas, s., garnas, j., robertson, m.p., & terblanche, j.s. (2016). predicted decrease in global climate suitability masks regional complexity of invasive fruit fly species response to climate change. biological invasions, 18(1), 1105-1119. https://doi.org/10.1007/ s10530-016-1078-5 james, f.c., & shugart, h.h. (1970). a quantitative method of habitat description. audubon field notes by the national audubon society, 24(6), 727-736. jenks, g.f., & caspall, f.c. (1971). error on choroplethic maps: definition, measurement, reduction. annals of the association of american geographers, 61(2), 217-244. https://doi. org/10.1111/j.1467-8306.1971.tb00779.x johnston, r., & temple, s. (1986). assessing habitat quality for birds nesting in fragmented tallgrass prairies. in: wildlife 2000: modelling habitat relationships of terrestrial vertebrates, j. verner, le and morgonroth new zealand journal of forestry science (2020) 50:9 page 14 https://doi.org/10.1080/02626667909491834 https://doi.org/10.1071/9780643069701 https://doi.org/10.1080/02693799108927858 https://doi.org/10.1080/02693799108927858 https://doi.org/10.1023/a:1021350813586 https://doi.org/10.1023/a:1021350813586 https://doi.org/10.1016/j.envsoft.2010.06.001 https://doi.org/10.1016/j.envsoft.2010.06.001 https://doi.org/10.5194/gmd-8-1991-2015 https://doi.org/10.5194/gmd-8-1991-2015 https://doi.org/10.1007/s10661-005-1594-8 https://doi.org/10.1007/s10661-005-1594-8 https://doi.org/10.1890/1051-0761(1999)009%5b0152:agmmat%5d2.0.co;2 https://doi.org/10.1890/1051-0761(1999)009%5b0152:agmmat%5d2.0.co;2 https://doi.org/10.1890/1051-0761(1999)009%5b0152:agmmat%5d2.0.co;2 https://doi.org/10.1093/bioinformatics/btt662 https://doi.org/10.1093/bioinformatics/btt662 https://doi.org/10.1002/joc.5086 https://doi.org/10.1080/02664763.2014.980789 https://doi.org/10.1080/02664763.2014.980789 https://doi.org/10.5194/esd-6-61-2015 https://doi.org/10.5194/esd-6-61-2015 https://doi.org/10.21273/hortsci.34.5.867 https://doi.org/10.21273/hortsci.34.5.867 https://doi.org/10.1007/s10530-016-1078-5 https://doi.org/10.1007/s10530-016-1078-5 https://doi.org/10.1111/j.1467-8306.1971.tb00779.x https://doi.org/10.1111/j.1467-8306.1971.tb00779.x m.l. morrison, and c.j. ralph, eds. (pp. 245-249). madison, wisconsin, usa: university of wisconsin press. karger, d.n., conrad, o., bohner, j., kawohl, t., kreft, h., soria-auza, r.b., zimmermann, n.e., linder, h.p., & kessler, m. (2017). climatologies at high resolution for the earth’s land surface areas. scientific data, 4(170122). https://doi.org/10.1038/ sdata.2017.122 kimberley, m., west, g., dean, m., & knowles, l. (2005). the 300 index a volume productivity index for radiata pine. new zealand journal of forestry, 50(2), 13-18. knollova, i., chytry, m., tichy, l., & hajek, o. (2005). stratified resampling of phytosociological databases: some strategies for obtaining more representative data sets for classification studies. journal of vegetation science, 16(4), 363-372. h t t p s : / / d o i . o rg / 1 0 . 1 1 1 1 / j . 1 6 5 4 1 1 0 3 . 2 0 0 5 . tb02388.x kohl, m., magnussen, s., & marchetti, m. (2006). sampling methods, remote sensing and gis multiresource forest inventory. heidelberg, germany: springerverlag. https://doi.org/10.1007/978-3-54032572-7 kutner, h.m., nachtsheim, c.j., & neter, j. (2003). applied linear regression models. (4th ed.) new york, usa: mcgraw-hill companies. leathwick, j.r., wilson, g., & stephens, r.t.t. (1998). climate surfaces for new zealand. landcare research contract report: lc9798/126. 26 p. retrieved from https://www.landcareresearch.co.nz/uploads/ public/tools-and-resources/maps/lenz/climate_ surfaces_for_new_zealand.pdf lin, h., murray, t., & mason, e. (2017). incidence of and defoliation by a newly introduced pest, paropsisterna variicollis (coleoptera: chrysomelidae), on eleven durable eucalyptus species in hawke’s bay, new zealand. new zealand plant protection, 70, 45-51. https://doi.org/10.30843/nzpp.2017.70.26 marinoni, o., higgins, a., hajkowics, s., & collins, k. (2009). the multiple criteria analysis tool (mcat): a new software tool to support environmental investment decision making. environmental modelling & software, 24(2), 549-562. https://doi. org/10.1016/j.envsoft.2008.06.015 mead, d. (2013). sustainable management of pinus radiata. [fao forestry paper no. 170]. rome, italy: food and agriculture organization of the united nations. millen, p., burgess, j., mcconnochie, r., may, r., & buck, k. (2016). early height growth of durable eucalypt species. the measurement and analysis of nzdfi demonstration trials planted 2010-2014. retrieved from http://nzdfi.org.nz/wp-content/ uploads/2014/09/nzdfi-reportonassessmentof2 0 1 0 1 4 d u r a b l e e u c a l y p t d e m o t r i a l s 30june2016.pdf mpi. (2016). national exotic forest description as at 1 april 2016. retrieved from http://www. n o r t h l a n d w o o d c o u n c i l . c o . n z / d o w n l o a d s / national-exotic-forest-description-april-2016.pdf newsome, p.f.j., wilde, r.h., & willoughby, e.j. (eds.). (2008). land resource information system of spatial data layers. palmerston north, new zealand: landcare research new zealand ltd. nicholas, i., & millen, p. (2012). durable eucalypt leaflet series. [nzdfi project publication]. retrieved from http://nzdfi.org.nz/wp-content/ uploads/2015/01/e-bosistoana-informationleaflet.pdf nzdfi. (2015). nzdfi site-species matching research plan. retrieved from http://nzdfi.org.nz/wpcontent/uploads/2014/08/section-3-nzdfi-site_ species-matching-research-plan-.pdf nzdfi. (2017). breeding tomorrow’s trees today. retrieved from http://nzdfi.org.nz nzdfi. (2019). e. bosistoana: information for growers. retrieved from https://nzdfi.org.nz/growerinformation/growing-ground-durable-eucalypts/ growing-regimes/choosing-the-right-species/ebosistoana-grower-information/ pattison, r.r., & mack, r.n. (2008). potential distribution of the invasive tree triadica sebifera (euphorbiaceae) in the united states: evaluating climex predictions with field trials. global change biology, 14(4), 813-826. https://doi.org/10.1111/ j.1365-2486.2007.01528.x pausas, j., braithwaite, m., & austin, m. (1995). modelling habitat quality for arboreal marsupials in the south coastal forests of new south wales, australia. forest ecology management, 78(1-3), 39-49. https://doi.org/10.1016/0378-1127(95)03598-5 pearse, g., moltchanova, e., & bloomberg, m. (2015). assessment of the accuracy of profile available water and potential rooting depth estimates held within new zealand’s fundamental soil layers geodatabase. soil research, 53(7), 737-744. https:// doi.org/10.1071/sr14012 pereira, j.m.o.d. (1989). a spatial approach to statistical habitat suitability modeling: the mt. graham red squirrel case study. usa: university of arizona. pielou, e.c. (1984). the interpretation of ecological data: a primer of classification and ordination. john wiley & sons. prober, s.m., potts, b.m., bailey, t., byrne, m., dillon, s., harrison, p.a., hoffmann, a.a., jordan, r., mclean e.h., steane, d.a., stock, w.d., & vaillancourt, r.e. (2016). climate adaptation and ecological restoration in eucalypts. proceedings of the royal society of victoria, 128(1), 40-53. https://doi.org/10.1071/rs16004 le and morgonroth new zealand journal of forestry science (2020) 50:9 page 15 https://doi.org/10.1038/sdata.2017.122 https://doi.org/10.1038/sdata.2017.122 https://doi.org/10.1111/j.1654-1103.2005.tb02388.x https://doi.org/10.1111/j.1654-1103.2005.tb02388.x https://doi.org/10.1007/978-3-540-32572-7 https://doi.org/10.1007/978-3-540-32572-7 https://www.landcareresearch.co.nz/uploads/public/tools-and-resources/maps/lenz/climate_surfaces_for_new_zealand.pdf https://www.landcareresearch.co.nz/uploads/public/tools-and-resources/maps/lenz/climate_surfaces_for_new_zealand.pdf https://www.landcareresearch.co.nz/uploads/public/tools-and-resources/maps/lenz/climate_surfaces_for_new_zealand.pdf https://doi.org/10.30843/nzpp.2017.70.26 https://doi.org/10.1016/j.envsoft.2008.06.015 https://doi.org/10.1016/j.envsoft.2008.06.015 http://nzdfi.org.nz/wp-content/uploads/2014/09/nzdfi-reportonassessmentof-2010-14-durableeucalyptdemotrials-30june2016.pdf http://nzdfi.org.nz/wp-content/uploads/2014/09/nzdfi-reportonassessmentof-2010-14-durableeucalyptdemotrials-30june2016.pdf http://nzdfi.org.nz/wp-content/uploads/2014/09/nzdfi-reportonassessmentof-2010-14-durableeucalyptdemotrials-30june2016.pdf http://nzdfi.org.nz/wp-content/uploads/2014/09/nzdfi-reportonassessmentof-2010-14-durableeucalyptdemotrials-30june2016.pdf http://www.northlandwoodcouncil.co.nz/downloads/national-exotic-forest-description-april-2016.pdf http://www.northlandwoodcouncil.co.nz/downloads/national-exotic-forest-description-april-2016.pdf http://www.northlandwoodcouncil.co.nz/downloads/national-exotic-forest-description-april-2016.pdf http://nzdfi.org.nz/wp-content/uploads/2015/01/e-bosistoana-information-leaflet.pdf http://nzdfi.org.nz/wp-content/uploads/2015/01/e-bosistoana-information-leaflet.pdf http://nzdfi.org.nz/wp-content/uploads/2015/01/e-bosistoana-information-leaflet.pdf http://nzdfi.org.nz/wp-content/uploads/2014/08/section-3-nzdfi-site_species-matching-research-plan-.pdf http://nzdfi.org.nz/wp-content/uploads/2014/08/section-3-nzdfi-site_species-matching-research-plan-.pdf http://nzdfi.org.nz/wp-content/uploads/2014/08/section-3-nzdfi-site_species-matching-research-plan-.pdf http://nzdfi.org.nz https://nzdfi.org.nz/grower-information/growing-ground-durable-eucalypts/growing-regimes/choosing-the-right-species/e-bosistoana-grower-information/ https://nzdfi.org.nz/grower-information/growing-ground-durable-eucalypts/growing-regimes/choosing-the-right-species/e-bosistoana-grower-information/ https://nzdfi.org.nz/grower-information/growing-ground-durable-eucalypts/growing-regimes/choosing-the-right-species/e-bosistoana-grower-information/ https://nzdfi.org.nz/grower-information/growing-ground-durable-eucalypts/growing-regimes/choosing-the-right-species/e-bosistoana-grower-information/ https://doi.org/10.1111/j.1365-2486.2007.01528.x https://doi.org/10.1111/j.1365-2486.2007.01528.x https://doi.org/10.1016/0378-1127(95)03598-5 https://doi.org/10.1071/sr14012 https://doi.org/10.1071/sr14012 https://doi.org/10.1071/rs16004 reisinger, t., & kennedy, d. (1990). a spatial decision support system for opportunity area analysis on the jefferson national forest. paper presented at the annual conference and exposition gis/lis ‘90, anaheim, canada 5-10 november. riley, s.j., degloria, s.d., & elliot, r. (1999). a terrain ruggedness index that quantifies topographic heterogeneity. intermountain journal of sciences, 5(1-4), 23-27. royall, r.m. (1970). on finite population sampling theory under certain linear regression models. biometrika, 57(2), 377-387. https://doi.org/10.1093/ biomet/57.2.377 scott, c.t. (1998). sampling methods for estimating change in forest resources. ecological applications, 8(2), 228-233. https://doi.org/10.1890/10510761(1998)008[0228:smfeci]2.0.co;2 sellars, j.d., & jolls, c.l. (2007). habitat modeling for amaranthus pumilus: an application of light detection and ranging (lidar) data. journal of coastal research, 23(5), 1193-1202. https://doi. org/10.2112/04-0334.1 shabani, f., kumar, l., & taylor, s. (2012). climate change impacts on the future distribution of date palms: a modeling exercise using climex. plos one, 7(10). https://doi.org/10.1371/journal.pone.0048021 shaw, d., & atkinson, s. (1988). gis applications for golden-cheeked warbler habitat description. paper presented at the annual conference and exposition gis/lis ‘88, falls church, virginia, usa. store, r., & kangas, j. (2001). integrating spatial multicriteria evaluation and expert knowledge for gisbased habitat suitability modeling. landscape and urban planning, 55(2), 79-93. https://doi. org/10.1016/s0169-2046(01)00120-7 sutherst, r.w., & maywald, g.f. (1985). a computerized system for matching climates in ecology. agriculture, ecosystems & environment, 13(34), 281-299. https://doi.org/10.1016/01678809(85)90016-7 sutherst, r.w., maywald, g.f., & kristicos, d.j. (2007). climex version 3: user’s guide. south yarra, australia: hearne scientific software. taylor, s., kumar, l., reid, n., & kriticos, d.j. (2012). climate change and the potential distribution of an invasive shrub, lantana camara l. plos one, 7(10). https://doi.org/10.1371/journal.pone.0035565 tomppo, e., malimbwi, r., katila, m., mäkisara, k., henttonen, h.m., chamuya, n., zahabu, e., & otieno, j. (2014). a sampling design for a large area forest inventory: case tanzania. canadian journal of forest research, 44(8), 931-948. https://doi. org/10.1139/cjfr-2013-0490 wadge, g., wislocki, a., & pearson, e. (1993). spatial analysis in gis for natural hazard assessment. in m. goodchild, b. parks & l. steyaert (eds.), environmental modeling with gis (pp. 332-338). new york: oxford university press. wallenius, k., niemi, r.m., & rita, h. (2011). using stratified sampling based on pre-characterisation of samples in soil microbiological studies. applied soil ecology, 51(1), 111-113. https://doi. org/10.1016/j.apsoil.2011.09.006 webb, t.h., & wilson, a.d. (1995). a manual of land characteristics for evaluation of rural land. [landcare research science series 10]. lincoln, new zealand: landcare research. weston, g. (1957). exotic forest trees in new zealand. statement prepared for the 7th british commonwealth forestry conference. [australia & new zealand nz forest service bulletin]. xuezhen, g., shanyong, h., chenyi, z., tao, w., zhichun, x., & shixiang, z. (2018). projecting the current and future potential global distribution of hyphantria cunea (lepidoptera: arctiidae) using climex. pest management science, 75(1), 160-169. https://doi. org/10.1002/ps.5083 yves, t., & ecker, k. (2014). complex national sampling design for long-term monitoring of protected dry grasslands in switzerland. environmental and ecological statistics, 21(3), 453-476. https://doi. org/10.1007/s10651-013-0263-2 zeverbergen, l.w., & thorne, c.r. (1987). quantitative analysis of land surface topography. earth surface processes and landforms, 12(1), 47-56. https://doi. org/10.1002/esp.3290120107 le and morgonroth new zealand journal of forestry science (2020) 50:9 page 16 https://doi.org/10.1093/biomet/57.2.377 https://doi.org/10.1093/biomet/57.2.377 https://doi.org/10.1890/1051-0761(1998)008%5b0228:smfeci%5d2.0.co;2 https://doi.org/10.1890/1051-0761(1998)008%5b0228:smfeci%5d2.0.co;2 https://doi.org/10.2112/04-0334.1 https://doi.org/10.2112/04-0334.1 https://doi.org/10.1371/journal.pone.0048021 https://doi.org/10.1016/s0169-2046(01)00120-7 https://doi.org/10.1016/s0169-2046(01)00120-7 https://doi.org/10.1016/0167-8809(85)90016-7 https://doi.org/10.1016/0167-8809(85)90016-7 https://doi.org/10.1371/journal.pone.0035565 https://doi.org/10.1139/cjfr-2013-0490 https://doi.org/10.1139/cjfr-2013-0490 https://doi.org/10.1016/j.apsoil.2011.09.006 https://doi.org/10.1016/j.apsoil.2011.09.006 https://doi.org/10.1002/ps.5083 https://doi.org/10.1002/ps.5083 https://doi.org/10.1007/s10651-013-0263-2 https://doi.org/10.1007/s10651-013-0263-2 https://doi.org/10.1002/esp.3290120107 https://doi.org/10.1002/esp.3290120107 quantitative genetic parameters of heartwood and its chemical traits in a black pine (pinus nigra j.f.arnold) clonal seed orchard established in greece kostas ioannidis1* and polyxeni koropouli2 1 laboratory of forest genetics and biotechnology, institute of mediterranean and forest ecosystems, hellenic agricultural organization “demeter”, ilissia, 11528 athens, greece 2 research consultant, athens, greece *corresponding author: ioko@fria.gr (received for publication 6 july 2022; accepted in revised form 10 may 2023) abstract background: black pine (pinus nigra j.f.arnold) is one of the most productive conifers species for timber production in southern europe, the mediterranean region and greece. recently, the interest for its heartwood extractives content due to their medicinal properties has been renewed. black pine can be used to produce high added value products, such as bioactive compounds produced from wood and wood waste materials. methods: quantitative genetic parameters were estimated for heartwood chemical traits and heartwood percentage in a 44-year-old pinus nigra clonal seed orchard, established in peloponnese, greece. results: significant variation was found among clones and among provenances for all studied traits. heritability on a clone mean basis was extremely high for total acetone extractives, total resin acids as for all resin acids (≥0.85), except levopimaric acid (0.47) and very high for total stilbenes, pinosylvins as for dehydroabietic acid and heartwood percentage (0.69-0.79). on an individual basis, the genetic control was moderate to high (0.53-0.62) for total acetone extractives, total resin acids as for most of resin acids (≥0.85) with dehydroabietic acid presenting low value (0.39) while levopimaric acid very low (0.15). total stilbenes, pinosylvins and its ether derivatives as heartwood percentage exhibited low values of heritability on individual basis (0.31-0.43). the phenotypic correlation (rp) between total acetone extracts and total stilbenes was negatively weak (≤-0.173) and significant (p≤0.01) while the genetic correlation (rg) was moderate to strong (≤-0.502). the rp values between several pinosylvins were significantly (p≤0.01) moderate to strong (0.529-0.975) as were genetic correlations (0.583-0.975). between the studied resin acids, both rp and rg values were mostly medium to strong (rp≥0.8 and rg≥0.7) and significant (p≤0.01) in the case of phenotypic correlations, with minor exceptions (levopimaric acid). phenotypic and genetic correlations between heartwood percentage and its chemical traits were positive (being in most cases significant), except for dehydroabietic and levopimaric acid. conclusions: the studied clones, comprising the clonal seed orchard, can be used in clonal forestry and subsequent breeding cycles, indicating high potential for advanced breeding, especially for heartwood extractives that are of high pharmaceutical and economic value. new zealand journal of forestry science ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 https://doi.org/10.33494/nzjfs532023x249x e-issn: 1179-5395 published on-line: 12/06/2023 © the author(s). 2023 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access species has a wide geographical distribution, extending from the caucasian coast of the black sea to the atlas mountains in northwest africa. this terrestrial extent has led to the existence of many forms among the subspecies and high levels of genetic variation among and within populations (mirov 1967; vidakovic 1974, 1991). the introduction black pine (pinus nigra j.f.arnold) is one of the most economically important native conifers in southern europe and the mediterranean region (isajev et al. 2004), and is one of the main timber species in greece. the keywords: black pine; clonal seed orchard; stilbenes; pinosylvin; resin acids; heartwood; heritability. http://creativecommons.org/licenses/by/4.0/), ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 2 species’ importance is attributed to its ability to develop under dry and poor environmental conditions (as a low demanding species), on a variety of soil substrates, being wind resistant but shade intolerant, and to the natural durability of its wood (isajev et al. 2004). due to its ecological value and silvicultural significance, it has been planted in several regions outside its natural range, including new zealand, great britain, france, argentina and the united states (cown 1972; lee 1968; wheeler et al. 1976; wilcox & miller 1975). silviculturally, black pine has been extensively used in reforestation programs throughout greece because it is well-adapted to a wide range of sites, can grow on degraded land, is easy to establish even with minimal care, grows rapidly and produces high quality wood, which is in high demand (matziris 1989). to satisfy the demand for large quantities of quality seed, breeding efforts have focused on conducting provenance trials (varelides et al. 2001), progeny tests and establishing clonal seed orchards (matziris 2005). the aim has been to produce material that is genetically improved for traits of interest, mostly growth. recently, there has been renewed research interest in pine extracts, particularly in stilbenes, i.e., pinosylvin and its ether derivatives, as well as in resin acids, i.e., abietic, dehydroabietic, neoabietic, palustric, levopimaric, pimaric, isopimaric and sandaracopimaric acid. these bioactive compounds are high added value products which can be produced from wood and wood waste (ioannidis et al. 2017, 2019; pietarinen et al. 2006). stilbenes, also called phytoalexins, are polyphenolic compounds produced by various plants in response to environmental stresses such as weather changes, injuries from insects or infections by pathogens (celimene et al. 1999, 2001; dixon 2001; gref et al. 2000; hart 1981; kennedy et al. 1995; schultz et al. 1992; schultz et al. 1990) . due to their fungal toxicity and water repellent properties, they affect wood natural durability (gref et al. 2000; hart 1981; kennedy et al. 1995). pinosylvin is a stilbene with many medicinal properties, including antibacterial and antifungal (lee et al. 2005; lindberg et al. 2004), antioxidant (willför et al. 2003), antimetastatic (park et al. 2012), potential cancer chemopreventive activity (park et al. 2013), antiapoptotic and cardiovascular protective (jeong et al. 2013) and anti-inflammatory effects (laavola et al. 2015). in addition, pine resin and its derivatives have been widely applied in the chemical industry in uses such as polymer additives, emulsifiers in synthetic rubber, in adhesives, surface coatings, printing inks, chewing gums etc. and in the greek wine industry (langenheim 2003; rezzi et al. 2005). more importantly, several medicinal uses concerning the treatment of abscesses, boils, cancers, toothache and others have been reported even from antiquity (langenheim 2003). in recent years, a renewed scientific interest in pine resin extracts has arisen (reveglia et al. 2018) due to their medicinal properties, i.e. antimicrobial (spessard et al. 1995), antiinflammatory (fernandez et al. 2001; takahashi et al. 2003), cardiovascular (de oliveira et al. 2008; gonzalez et al. 2009), cytotoxic (schmeda-hirschmanna et al. 2005a) and others (schmeda-hirschmann et al. 2005b; talevi et al. 2007; ulusu et al. 2002). several genes are involved in the formation of both groups of bioactive compounds. rapidly activated stilbene synthase-encoding genes take part in the formation of stilbenes (chong et al. 2009; donnez et al. 2009; ebel 1986). respectively, diterpene synthaseencoding genes act on the geranylgeranyl diphosphate (ggpp) substrate to form diterpenes, following two independent transformations by other enzymes (keeling & bohlmann 2006). this explains why pinosylvin and its ether derivatives (fries et al. 2000; partanen et al. 2011) as well as the diterpene resin acids (ericsson et al. 2001; fries et al. 2000) are highly heritable properties. genetic control of extractive production has considerable economic potential in some species (taylor et al. 2002). the aim of the present work was to estimate for the first time, genetic effects on regulation of the traits of specific heartwood extractives in pinus nigra. we investigated the qualitative and quantitative variations in two major heartwood group of substances (stilbenes and resin acids) in 52 pinus nigra l. subsp. pallasiana plus-tree clones, originating from four provenances from the peloponnese region of southern greece (figure 1). the populations from which the selection of clones were made are the most southern or borderline populations of the species in greece as well as in europe. clonal repeatability, on individual and on clone mean basis, of the target bioactive compounds was evaluated, indicating the magnitude of their inheritance. the heartwood of these trees produced an exceptionally high extractive content, approximately 30% w/w (ioannidis et al. 2017), making it particularly interesting for the present analysis. the target compounds were the stilbenes pinosylvin, monomethyl and dimethyl ether of pinosylvin, the abietane type resin acids, abietic, dehydroabietic, neoabietic, palustric and levopimaric, and the pimarane type resin acids, pimaric, isopimaric and sandaracopimaric. methods plant material in greece, black pine’s significant effort has focused on selective breeding. in 1978, a 10-ha clonal seed orchard was established in the area of koumani in the western part of the peloponnese, greece. the orchard comprises 52 clones derived from intensively selected plus trees in the natural black pine forest of the peloponnese and a total number of 2,700 grafts (matziris 1989). the elite trees originated from four provenances (figure 1): mt. zarouhla, mt. feneos, mt. parnonas and mt. taigetos. ramets of each clone were randomly located within the clonal seed orchard under the condition that two ramets of the same clone were planted at least 30 m apart from each other (single tree plot design). sampling heartwood discrimination and orientation extraction protocol sampling, coring, heartwood discrimination and orientation and extraction protocol, 1h-nmr spectra analysis, calibration curves, validation method and quantitation of pinosylvins and resin acids are described extensively in ioannidis et al. (2017) and ioannidis et al. (2019). in brief, 12mm-diameter increment cores including the pith were extracted, 30 cm above ground and in a north-south orientation from a total of 260 healthy individuals, covering all 52 clones planted in the clonal seed orchard (five ramets per clone). heartwood was separated from the rest of the core based on a visual assessment and milled to produce ≤0.75 mm particles. pinosylvins and resin acids were extracted from freezedried ground heartwood using acetone as solvent. the extract from each sample was dissolved in 600 μl of deuterated chloroform (cdcl3) and the solution was transferred to a 5 mm nmr tube. 1h-nmr spectra were recorded at 400 mhz (bruker drx400). one-dimensional (1d) and two-dimensional (2d) quantitative nuclear magnetic resonance (qnmr) were used permitting rapid quantification of the analytes. concentrations were based on freeze-dried heartwood (dhw) and expressed in mg/gdhw. studied traits the following traits were studied in the koumani clonal seed orchard: heartwood concentration of total acetone extracts (tae), total stilbenes (ts), pinosylvin (p), monomethyl (pmme) and dimethyl ether (pdme) of pinosylvin, total resin acid (tra), the abietane-type resin acids, i.e. abietic acid (aa), dehydroabietic acid (daa), neoabietic acid (naa), palustric acid (pla) and levopimaric acid (lpma), the pimarane-type resin acids i.e. pimaric acid (pma), isopimaric acid (ipma) and sandaracopimaric acid (spma), and finally the heartwood proportion (hw). the heartwood proportion was calculated as the ratio of heartwood radius to total under bark radius and expressed as a percentage. statistical analysis analyses were based on values obtained from individual trees and were performed at a significance level of a=0.05. the following linear model was used in the analysis: yijk = μ+ pi + cj(pi) + eijk (1) where yijk is the phenotypic measurement for a trait measured on the kth tree, jth clone and ith provenance, μ is the fixed population mean of all trees averaged across the clonal seed orchard, pi is the random effect of the i th provenance (i=1,2,3,4), cj(pi) is the random effect of the jth clone (j=1,2,...,52) nested within the ith provenance, and eijk is the random residual error of k th tree, jth clone and ith provenance. the variance components were estimated by the restricted maximum likelihood (reml) method. descriptive statistics, analysis of variance (anova) as well as variance component estimates based on the 0.05 level of significance were calculated using spss v.20 software (ibm spss statistics 2011, ibm corp.). heritabilities the following variance components were estimated: σ2e the variance component due to error, σ 2 p the variance component due to provenances and σ2c(p) the variance component due to clones within provenances. estimates of broad sense heritability were obtained on an individual tree (h2i) and clone mean (h 2 c) basis, respectively, as follows: h2i = σ2c(p)/(σ2e + σ2c(p)) (2) ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 3 table 1: description of the study sites figure 1: map showing the location of the four black pine provenances (mt. zarouhla, mt. feneos, mt. parnonas and mt. taigetos) and the clonal seed orchard (cso). h2c = σ2c(p)/((σ2e / r) + σ2c(p)) (3) where h2i, h 2 c, σ 2 e, σ 2 p and σ 2 c(p) are as above, and r is the number of ramets sampled per clone, (i.e., r=5). standard errors for heritability estimates were calculated, where feasible, using dickerson’s approximation (dickerson 1969). phenotypic and genetic correlations for all the studied traits, phenotypic correlation coefficients and their significance level were assessed using pearson's correlation coefficient (snedecor & cohran 1980): rp = σpxy/(σpx × σpy) (4) where σpx and σpy are the square roots of the phenotypic variance from an analysis of variance of each trait, and σpxy, is the phenotypic component from analysis of covariance of x and y traits. broad sense genetic correlations (rg) between all paired traits were calculated as follows (falconer 1960): rg = σgxy/(σgx × σgy) (5) where σgx and σgy the square roots of the clonal variance components from the analysis of variance of each trait, and σgxy is the clonal component from analysis of covariance of x and y traits. these correlations as well as heritability estimates are biased if any genotype by environment interaction exists (lambeth et al. 1994). ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 4 results the overall mean values for all traits assessed in the koumani clonal seed orchard along with their minimum and maximum values, standard deviations and coefficients of variation are presented in table 1. more details about stilbenes and resin acids can be found in ioannidis et al. (2017) and ioannidis et al. (2019), respectively. the average heartwood content was 11.2% and ranged from 1.2% up to 27.0%. the visualised results of the traits assessed on the sampled trees in the clonal seed orchard are presented in the boxplots in additional files (figures a1 to a4). the variation among clones for the studied traits is evident from the analysis of variance. the analysis showed that there were statistically significant (p<0.001) differences among clones for all examined traits. concerning provenances, the analysis showed significant differences in concentrations of tae (p<0.05), p (p<0.01), pdme (p<0.05), tra (p<0.05), aa (p<0.01), naa (p<0.05), pma (p<0.01) and spma (p<0.01). the variance components for the studied traits are presented in figure 2. the individual (h2i) and clone (h 2 c) broad sense heritabilities and their standard errors for all studied traits are shown in figure 3 and table a1 in additional file. tae had the highest heritability on an individual tree basis (0.86) and moderate heritability on a clone mean basis (0.55). the heritability values for stilbene concentration on individual mean basis (h2i) were moderate to low and ranged from 0.33 (ts) to 0.79 (pdme). the genetic control on a clone mean basis (h2c) table 1: descriptive statistics of the traits assessed in the total clonal seed orchard sample (n=260). trait (units of measurement) minimum maximum mean std. dev. cv (%) total acetone extracts (mg/gdhw) 81.79 480.28 304.15 96.02 31.57 total stilbenes (mg/gdhw) 10.99 128.22 59.92 21.79 36.37 pinosylvin (mg/gdhw) 1.19 40.23 17.07 6.76 39.60 pinosylvin monomethyl ether (mg/gdhw) 8.94 94.28 40.32 15.55 38.58 pinosylvin dimethyl ether (mg/gdhw) 0.21 7.91 2.54 1.22 48.09 total resin acid (mg/gdhw) 30.05 424.70 219.98 96.20 43.73 abietic acid (mg/gdhw) 7.00 181.75 76.77 37.39 48.70 dehydroabietic acid (mg/gdhw) 2.56 38.59 11.69 5.73 49.02 neoabietic acid (mg/gdhw) 2.91 101.82 39.34 21.21 53.91 palustric acid (mg/gdhw) 9.76 105.22 47.94 23.31 48.62 levopimaric acid (mg/gdhw) 0.08 64.91 8.07 11.38 141.02 pimaric acid (mg/gdhw) 2.20 59.42 22.54 11.28 50.04 isopimaric acid (mg/gdhw) 0.50 34.09 10.91 6.53 59.85 sandaracopimaric acid (mg/gdhw) 0.16 6.67 2.72 1.49 54.78 heartwood proportion (%) 1.20 27.04 11.15 4.39 39.37 diameter at breast height (cm) 18.30 48.80 35.33 5.0351 14.25 total tree height (m) 15.20 21.90 18.64 1.3733 7.37 was low to moderate strong and ranged from 0.37 (p) to 0.71 (ts). the clone mean heritability values (h2c) for resin acids were higher comparing those for stilbenes, ranging from 0.47 (lpma) to 0.89 (naa and tra). the h2i values of the studied resin acids were lower and ranged from 0.15 (lpma) to 0.62 (naa and tra). heritability values on individual and clone mean basis for heartwood percentage were estimated at 0.44 and 0.77, respectively (figure 3 and additional file table a1). pearson's phenotypic correlation (rp) and genetic correlation (rg) coefficients between all pairs of traits are shown in table 2 and table 3. phenotypically, tae was negatively correlated with all types of stilbenes, i.e., with pinosylvin rp=-0.173 (p<0.01), with pmme rp=-0.236 (p<0.01) and with pdme rp=-0.359 (p<0.01). tae showed a slight positive correlation with heartwood percentage (rp=0.188, p<0.01). stilbenes were positively, strongly and significantly correlated with each other (rp=0.5290.770, p<0.01), and in most cases, negatively (range from -0.233 to -0.088) and significantly (p<0.01 or p<0.05) correlated with heartwood percentage, except for the very loose positive and non-significant correlation between p and heartwood percentage (rp=0.30). strong phenotypic correlations were identified between tae and tra (rp=0.918, p<0.01) and tae with all acids (rp=0.702-0.918, p<0.01) except lpma (rp=0.452, p<0.01). resin acids were positively, significantly and in most cases strongly (rp=0.622 to 0.964, p<0.01) correlated with each other. weak to medium phenotypic correlations were identified between all resin acids with lpma (0.172-0.502, p<0.01). some rp values, concerning correlations for ipma and spma with several resin acids, were moderate (table 2). ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 5 figure 2: results from the variance components analysis for the studied traits. the numbers in each section of the bars represent the variance estimates calculated by the restricted maximum likelihood method using the yijk = μ+ pi + cj(pi) + eijk general linear model. figure 3: broad sense heritability estimates at individual and clone mean basis and their standard errors (red whiskers) for the studied traits. full descriptions of the trait abbreviations are given in the text. total resin acids correlated slightly positively with heartwood percentage (rp=0.168, p<0.01). low positive correlations were estimated between abietane type resin acids with heartwood percentage (rp=0.136-0.224, p<0.01 and p<0.05 regarding pimaric acid), and in two cases i.e. daa and lpma, were negative (-0.059 and -0.034 respectively) but non-significant. likewise, weak positive correlations were estimated between pimarane type resin acids with heartwood percentage (rp=0.1290.150, p<0.05). the genetic correlations between tae and stilbenes were always negative and significant (rg=0.758 to -0.502, p<0.01), while they were positive with heartwood percentage (0.315, p<0.01). between pinosylvins, moderate to strong genetic correlations were observed (rp=0.583-0.975, p<0.01), while those between heartwood percentage and stilbenes were lower (0.062-0.315), and in the case of pdme, negative (-0.134). moderate to strong and significant genetic correlations were identified between tae and all resin acids (0.478-0.959) except for spma for which rg was negative (-0.131). resin acids were positively and strongly correlated with each other (r =0.692-0.989, p<0.01). however, the positive correlations between heartwood percentage and resin acids were much weaker (rg=0.151-0.340) and were negative with daa and lpma, i.e. -0.006 and -0.269 respectively (table 3). discussion despite the extensive spread and importance of black pine, few studies have been conducted internationally, either into its heartwood extractives or their proportion. especially for black pine’s stilbenes content, the literature is very poor. likewise, there are limited previous studies that have estimated the genetic control, heritabilities in broad and narrow sense, as well as phenotypic and genetic correlations between the mentioned traits. consequently, black pine is not considered a well-characterised species concerning heartwood stilbenes, resin acids and percentage content. heritabilities heartwood extractives the content of extractives is usually less than 10%, but it can vary from trace amounts up to 40% of the dry wood weight (sjöström 1993). miller (1999) reported that the ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 6 table 2: pearson’s phenotypic correlation coefficients (above diagonal) and broad sense genetic correlations (below diagonal) among all pairs of traits. phenotypic correlation coefficients tae ts p pmme pdme hw tae -0.233** -0.173** -0.236** -0.359** 0.176** ts -0.745** 0.838** 0.975** 0.737** -0.093 p -0.502** 0.864** 0.697** 0.529** 0.030 pmme -0.758** 0.975** 0.686** 0.770** -0.123* pdme -0.757** 0.791** 0.583** 0.747** -0.233** hw 0.315** 0.137* 0.310** 0.062 -0.134**g en et ic c or re la ti on co ef fi ci en ts correlation is significant at the ** p<0.01, * p<0.05 table 3: pearson’s phenotypic correlation coefficients (above diagonal) and broad sense genetic correlations below diagonal) among all pairs of traits. correlation is significant at the ** p<0.01, * p<0.05 pearson correlation coefficients tae tra aa naa daa pla lpma pma ipma spma hw tae 0.918** 0.848** 0.877** 0.702** 0.837** 0.452** 0.834** 0.732** 0.829** 0.176** tra 0.959** 0.907** 0.964** 0.748** 0.930** 0.502** 0.901** 0.802** 0.915** 0.168** aa 0.933** 0.980** 0.839** 0.650** 0.754** 0.172** 0.806** 0.693** 0.832** 0.197** naa 0.888** 0.985** 0.955** 0.622** 0.912** 0.472** 0.867** 0.807** 0.887** 0.224** daa 0.478** 0.861** 0.838** 0.795** 0.729** 0.485** 0.637** 0.462** 0.655** -0.059 pla 0.876** 0.971** 0.917** 0.958** 0.858** 0.530** 0.765** 0.722** 0.825** 0.136* lpma 0.571** 0.939** 0.887** 0.901** 0.912** 0.954** 0.480** 0.397** 0.457** -0.034 pma 0.730** 0.929** 0.921** 0.921** 0.751** 0.851** 0.796** 0.697** 0.857** 0.150* ipma 0.590** 0.889** 0.835** 0.868** 0.692** 0.868** 0.891** 0.796** 0.738** 0.133* spma -0.131* 0.989** 0.953** 0.983** 0.819** 0.969** 0.924** 0.949** 0.876** 0.129* hw 0.315** 0.225** 0.340** 0.291** -0.006 0.167* -0.269** 0.164* 0.151* 0.194**g en et ic c or re la ti on c oe ff ic ie n ts variation in extractives content depends on factors such as species, growth conditions, and the time of year when a tree is cut. while extractives are present in sapwood, they mainly occur in the heartwood and knots (ioannidis et al. 2017; partanen et al. 2011; taylor et al. 2002; venäläinen et al. 2003; willför et al. 2003; willför et al. 2004a; willför et al. 2004b). the concentrations of total acetone extractives, pinosylvin and its ether derivatives in black pine have described extensively by ioannidis et al. (2017), while information on resin acids is contained in ioannidis et al. (2019). the findings in the study presented here are important as the mean recorded amount of total stilbenes, for the studied genetic material, was five times higher than the highest ever reported result for black pine, although the number of studies focusing on the specific species and topic is rather limited. concerning resin acids, they predominate in conifer heartwood extracts, while a wide range of compounds occur in angiosperms, although the range varies within each species (gutiérrez et al. 2001). moreover, there are differences in resin content and composition between different parts of the tree, depending, apart from genetic and environmental factors, on the age of tree and the growing conditions (back 2000). in pinus sylvestris l. older trees had higher concentrations of extractives (hovelstad et al. 2006) while in picea abies (l.) h.karst. lower concentrations could be attributed to the young age of the clones (fengel & wegener 2003). in contrast, fries et al. (2001), in several pinus sylvestris l. experiments, showed that older trees generally had lower extractives content compared to younger trees, indicating that differences in environmental conditions and tree age may cause differences in the amounts of heartwood extracts. substantially, the differences among clones could be attributed to the different environments in which they are grown and evolve. generally, clones originating from the most southeastern origin (mt. parnon provenance), growing under xerothermic environmental conditions (tselepidakis & theoharatos 1989), predominated in most characteristics. conversely, clones from the zarouhla provenance, which is the most north-western origin of the four tested, had the lowest values. clones from feneos (northeastern origin) and taygetos (southwestern origin) provenances, had intermediate values. the high clonal repeatability estimates found for the studied traits is evident. both the high heritabilities of heartwood chemical traits and of its proportion demonstrate they are under strong genetic control. heritability on a clone mean basis was extremely strong, while it was moderate on an individual tree basis. high broad sense heritabilities on an individual and clone mean basis indicate high potential for advanced breeding, for stilbenes and resin acids, both being of high pharmaceutical and economic value. despite these high estimates, it should be emphasised that extractive content generally depends, apart from a genetic component, on factors such as species, growth conditions, and the specific period of the year where the analysis is done. heartwood extractives, as mentioned above, are strongly determined genetically (zobel & jett 1995) with a mild environmental effect, e.g. site quality (taylor et al. 2002). pinosylvin and its ether derivatives are highly heritable heartwood properties (fries et al. 2000; partanen et al. 2011). our results are in accordance to those of partanen et al. (2011) in scots pine (pinus sylvestris l.), who estimated high heritability values for pinosylvin concentration (0.81), total stilbenes (0.61) and total phenols (0.74). the heritability of pinosylvin monomethyl ether was lower (0.48), as in the genetic material of the peloponnese. in both studies, clones with high or low concentrations of pinosylvin could be clearly identified despite the within clone variation. strong genetic control is expected because the formation of typical pine species phytoalexins (e.g. pinosylvin) is based on the rapidly activated stilbene synthase-encoding genes (chong et al. 2009). pinosylvin formation in the genus pinus is catalysed by stilbene synthase (sts), while pinosylvin-o-methyltransferase gene (pmt), is involved in the pinosylvin metabolism and the formation of the pinosylvin monomethylether (chiron et al. 2000; kodan et al. 2002). it may be noted that heritability estimates are probably biased upward due to the consideration of only one experimental site, compared with a more general estimate for a set of environments (fries et al. 2000). heartwood percentage considering the age of the trees in the clonal seed orchard at the time of sampling (35 years old), the small proportion of heartwood was expected. heartwood formation varies with provenance and site, and also depends largely upon tree age (fries 1999). in pinus sylvestris l. provenance trials, the lowest heartwood percentage values were found in younger trials while the highest in trials that were close to harvest age. according to fries (1999), these data indicate that the time of initiation of heartwood formation may be a limiting factor for high heartwood production. this conclusion perhaps explains the small percentage of heartwood observed in the clonal seed orchard. heartwood formation in black pine depends largely upon tree age and the trees in the clonal seed orchard were relatively young. the rotation age of the black pine in greece can reach and exceed 120 years, depending on the objectives of the management, the site quality, etc., so higher amounts of heartwood would be expected from harvested stands than were found in this study. comparing black pine’s heartwood percentage (11.5% at age 35 years) to other species such as scots or radiata (pinus radiata d.don) pines has led to the assumption that the formation of black pine’s heartwood starts at an older age, increases as the age progresses, as it is formed at a constant rate of annual growth rings and is expanding at a slow rate (cown 1972). on the contrary, heartwood percentage in pinus sylvestris l. ranges at higher values depending on the origin, location and age of trees (fries 1999). the same applies to radiata pine (cown et al. 1992). heartwood size is generally under moderate genetic control and with some environmental influence (zobel & jett 1995). pâques (2001) observed high heartwood percentage in ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 7 larix sp., denoting statistically significant differences among the origins of families and clones. broad sense heritabilities at an individual basis of the peloponnesian pinus nigra l. heartwood percentage were quite similar to the findings in p. sylvestris l. (partanen et al. 2011) and p. radiata d. don (nicholls & brown 1974). on the other hand, venäläinen et al. (2006) estimated lower broad sense heritability on an individual basis in larix sibirica ledeb. phenotypic and genetic correlations total acetone extractives correlated positively but weakly to heartwood percentage, showing a combined selection potential for these traits. the correlations between different stilbene types are expected to be strongly positive because the formation of pinosylvin and its ether derivatives in pine species are due to rapidly activated encoding stilbene related genes that act in the heartwood and they are involved in their metabolism and formation (chiron 2000; chiron et al. 2000; ebel 1986; kodan et al. 2002; schanz et al. 1992). venalainen et al. (2003) found a positive and significant correlations between p and pmme in both outer and inner heartwood. ericsson et al. (2001), found that environmental and genetic correlations in p. sylvestris l., between stilbenes types, as well as with the resin acids were extremely high. sehlstedt-persson and karlsson (2010) estimated that phenolic concentration in p. sylvestris l. exhibits a positive phenotypic correlation with total extracts. in the same species, venalainen et al. (2003) estimated that although phenotypic correlations between resin acids and both stilbenes and total phenols were generally positive, they were not statistically significant. the results of our study in black pine from peloponnese, indicate an opposite relationship between total acetone extracts and stilbenes. the negative correlation of total acetone extractives with all types of stilbenes essentially indicates that stilbene output is independent of total extractives production. a similar result was obtained by ericsson et al. (2001) who observed negative genetic correlation between stilbene and total acetone extracts. negative environmental correlations of various extracts, including pinosylvin monomethyl ether and total pinosylvins, with heartwood proportion were found by ericsson et al. (2001) in p. sylvestris l., although genetic correlations between stilbenes and heartwood were not detected. ericsson et al. (2001) also estimated that there were positive environmental (pimaric and abietic acid) and genetic (pimaric acid) correlations with heartwood diameter as well as negative ones (fatty acids and sterols). essentially, the results concerning stilbenes and heartwood proportion indicate that black pine from the peloponnese could not be selected in combination for these traits, at least for the clones and the provenances that were in the koumani clonal seed orchard. on the contrary, when a positive correlation was estimated between some heartwood extractives and heartwood percentage, the selection of trees with more resistant wood should be efficient, resulting in increase of both the content of heartwood and the concentration of extractives. despite the negative correlations between several of the traits studied, clones representing correlation breakers are able to be identified, (i.e. clones that performed well for traits negatively correlated). in addition, exceptional clones for positively correlated traits were also identified. these clones could be selected and might be the base to produce improved reproductive material. conclusions the heartwood of the selected black pine phenotypes originating from peloponnese are rich in extracts such as stilbenes and resin acids. the great variation among clones and provenances for the studied traits led to the identification of specific clones that outperform for several chemical traits. the genetic control of the studied traits was moderate to high, indicating the high potential for effective selection and breeding, and especially for the pinosylvins and several types of resin acids that are of high pharmaceutical and economic value. the deployment of selected clones in clonal forestry plantations may yield immediate and significant revenues. list of abbreviations 1h-nmr: proton nuclear magnetic resonance anova: analysis of variance aa: abietic acid cso: clonal seed orchard daa: dehydroabietic acid dhw: dried heartwood ggpp: geranylgeranyl diphosphate hw: heartwood proportion ipma: isopimaric acid naa: neoabietic acid lpma: levopimaric acid mrt: multiple range test p: pinosylvin pma: pimaric acid pmme: monomethyl ether of pinosylvin pdme: dimethyl ether of pinosylvin pla: palustric acid pmt: pinosylvin-o-methyltransferase gene qnmr: quantitative nuclear magnetic resonance reml: restricted maximum likelihood tae: total acetone extracts spma: sandaracopimaric acid tra: total resin acid ts: total stilbenes competing interests the authors declare no competing interests. authors’ contributions ki designed the methodology, experimental work, performed the statistical analysis, and contributed to writing, reviewing and editing of the manuscript, pk carried out experimental work, data curation, helped ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 8 in formal analysis and participated in original draft preparation. all authors read, revised and approved the final manuscript. acknowledgements we appreciate, for their assistance, associate professor dr evi alizoti in data analysis, associate professor dr prokopios magiatis and dr eleni melliou in experimental work, and dr panagiotis kavvouras in methodology and sample preparation. the authors want to express their sincere thankfulness to the librarian of the forest research institute of athens, d. panayiotopoulou (msc library and information science), for information seeking and retrieving processes as well as her additional proof-reading services. funding this project received no external funding. additional file figures a1 to a4: boxplots of studied traits for the sampled trees of the cso. table a1: broad sense heritability estimates at individual and clone mean basis and their standard errors for the studied traits. references back, e.l. (2000). the location and morphology of resin components in the wood. in: back, e.l. and allen, l.h., (eds.), pitch control, wood resin and deresination. tappi press, pp. 1–35. celimene, c.c., micales, j.a., ferge, l., & young, r.a. (1999). efficacy of pinosylvins against white-rot and brown-rot fungi. holzforschung, 53(5), 491– 497. https://doi.org/10.1515/hf.1999.081 celimene, c.c., smith, d.r., young, r.a., & stanosz, g.r. (2001). in vitro inhibition of sphaeropsis sapinea by natural stilbenes. phytochemistry, 56(2), 161–165. https://doi.org/10.1016/s0031-9422(00)003848 chiron, h. (2000). molecular cloning and functional expression of a stress-induced multifunctional o-methyltransferase with pinosylvin methyltransferase activity from scots pine (pinus sylvestris l.). plant molecular biology, 44(6), 733– 745. https://doi.org/10.1023/a:1026507707186 chiron, h., drouet, a., lieutier, f., payer, h.-d., ernst, d., & sandermann, h. (2000). gene induction of stilbene biosynthesis in scots pine in response to ozone treatment, wounding, and fungal infection. plant physiology, 124(2), 865–872. https://doi. org/10.1104/pp.124.2.865 chong, j., poutaraud, a., & hugueney, p. (2009). metabolism and roles of stilbenes in plants. plant science, 177(3), 143–155. https://doi. org/10.1016/j.plantsci.2009.05.012 cown, d.j. (1972). physical properties of corsican pine grown in new zealand. new zealand journal of forestry science, 4(1), 76–93. cown, d.j., young, g.d., & burdon, r.d. (1992). variation in wood characteristics of 20-year-old half-sib families of pinus radiata. new zealand journal of forestry science, 22(1), 63–76. de oliveira, a.m., tirapelli, c.r., ambrosio, s.r., & da costa, f.b. (2008). diterpenes: a therapeutic promise for cardiovascular diseases. recent patents on cardiovascular drug discovery, 3(1), 1–8. https://doi.org/10.2174/157489008783331689 dickerson, g.e. (1969). techniques for research in quantitative animal genetics. in: techniques and procedures in animal science research. albany, new york: american society of animal science, pp. 3679. dixon, r.a. (2001). natural products and plant disease resistance. nature, 411(6839), 843–847. https:// doi.org/10.1038/35081178 donnez, d., jeandet, p., clément, c., & courot, e. (2009). bioproduction of resveratrol and stilbene derivatives by plant cells and microorganisms. trends in biotechnology, 27(12), 706–713. https:// doi.org/10.1016/j.tibtech.2009.09.005 ebel, j. (1986). phytoalexin synthesis: the biochemical analysis of the induction process. annual review of phytopathology, 24(1), 235–264. https://doi. org/10.1146/annurev.py.24.090186.001315 ericsson, t., fries, a., & gref, r. (2001). genetic correlations of heartwood extractives in pinus sylvestris progeny tests. forest genetics, 8(1), 73– 79. falconer, d.s. (1960). introduction to quantitative genetics. new york: the ronald press company. fengel, d., & wegener, g. (2003). wood-chemistry, ultrastructure, reactions. berlin and new york: walter de gruyter. fernandez, m., tornos, m., garcia, m., de las heras, b., villar, a., & saenz, m. (2001). anti-inflammatory activity of abietic acid, a diterpene isolated from pimenta racemosa var. grissea. journal of pharmacy and pharmacology, 53(6), 867–872. https://doi. org/10.1211/0022357011776027 fries, a. (1999). heartwood and sapwood variation in mature provenance trials of pinus sylvestris. silvae genetica, 48(1), 7–14. fries, a., ericsson, t., & gref, r. (2000). high heritability of wood extractives in pinus sylvestris progeny tests. canadian journal of forest research, 30(11), 1707–1713. https://doi.org/10.1139/x00-100 gonzalez, m.a., correa-royero, j., agudelo, l., mesa, a., & betancur-galvis, l. (2009). synthesis and biological ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 9 https://doi.org/10.1016/s0031-9422(00)00384-8 https://doi.org/10.1016/s0031-9422(00)00384-8 ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 10 evaluation of abietic acid derivatives. european journal of medicinal chemistry, 44(6), 2468–2472. https://doi.org/10.1016/j.ejmech.2009.01.014 gref, r., hakansson, c., henningsson, b., & hemming, j. (2000). influence of wood extractives on brown and white rot decay in scots pine heart-, lightand sapwood. material und organismen (germany), 33, 119-128. gutiérrez, a., del rıó, j.c., jesús martıńez, m., & martıńez, a.t. (2001). the biotechnological control of pitch in paper pulp manufacturing. trends in biotechnology, 19(9), 340–348. https://doi.org/10.1016/s01677799(01)01705-x hart, j.h. (1981). role of phytostilbenes in decay and disease resistance. annual review of phytopathology, 19(1), 437–458. https://doi. org/10.1146/annurev.py.19.090181.002253 hovelstad, h., leirset, i., oyaas, k., & fiksdahl, a. (2006). screening analyses of pinosylvin stilbenes, resin acids and lignans in norwegian conifers. molecules, 11(1), 103–114. https://doi. org/10.3390/11010103 ioannidis, k., melliou, e., alizoti, p., & magiatis, p. (2017). identification of black pine (pinus nigra arn.) heartwood as a rich source of bioactive stilbenes by qnmr. journal of the science of food and agriculture, 97(6), 1708–1716. https://doi. org/10.1002/jsfa.8090 ioannidis, k., melliou, e., & magiatis, p. (2019). highthroughput 1h-nuclear magnetic resonance-based screening for the identification and quantification of heartwood diterpenic acids in four black pine (pinus nigra arn.) marginal provenances in greece. molecules, 24(19), 3603. https://doi.org/10.3390/ molecules24193603 isajev, v., fady, b., semerci, h., & andonovski, v. (2004). euforgen technical guidelines for genetic conservation and use for european black pine (pinus nigra) rome, italy: international plant genetic resources institute, p. 6. jeong, e., lee, h.-r., pyee, j., & park, h. (2013). pinosylvin induces cell survival, migration and antiadhesiveness of endothelial cells via nitric oxide production. phytotherapy research: ptr, 27(4), 610–617. https://doi.org/10.1002/ptr.4770 keeling, c.i., & bohlmann, j. (2006). diterpene resin acids in conifers. phytochemistry, 67(22), 2415–2423. https://doi.org/10.1016/j. phytochem.2006.08.019 kennedy, m.j., drysdale, j.a., & brown, l. (1995). efficacy of some extractives from pinus heartwood for protection of pinus radiata sapwood against biodeterioation; i. fungal decay. [paper no. 9530072]. presented at the international research group on wood preservation. 26th annual meeting, stockholm. 10 p. kodan, a., kuroda, h., & sakai, f. (2002). a stilbene synthase from japanese red pine (pinus densiflora): implications for phytoalexin accumulation and down-regulation of flavonoid biosynthesis. proceedings of the national academy of science usa, 99(5), 3335–3339. https://doi.org/10.1073/ pnas.042698899 laavola, m., nieminen, r., leppänen, t., eckerman, c., holmbom, b., & moilanen, e. (2015). pinosylvin and monomethylpinosylvin, constituents of an extract from the knot of pinus sylvestris, reduce inflammatory gene expression and inflammatory responses in vivo. journal of agricultural and food chemistry, 63(13), 3445–3453. https://doi. org/10.1021/jf504606m lambeth, c., endo, m., & wright, j. (1994). genetic analysis of 16 clonal trials of eucalyptus grandis and comparisons with seedling checks. forest science, 40(3), 397–411. langenheim, j.h. (2003). plant resins: chemistry, evolution, ecology, and ethnobotany. portland, or, usa: timber press. lee, c.h. (1968). geographic variation in european black pine. silvae genetica, 17(5/6), 165–74. lee, s.k., lee, h.j., min, h.y., park, e.j., lee, k.m., ahn, y.h., cho, y.c., & pyee, j.h. (2005). antibacterial and antifungal activity of pinosylvin, a constituent of pine. fitoterapia, 76(2), 258–260. https://doi. org/10.1016/j.fitote.2004.12.004 lindberg, l.e., willför, s.m., & holmbom, b.r. (2004). antibacterial effects of knotwood extractives on paper mill bacteria. journal of industrial microbiology and biotechnology, 31(3), 137–147. https://doi.org/10.1007/s10295-004-0132-y matziris, d. (1989). variation in growth and branching characters in black pine (pinus nigra arnold) of peloponnesos. silvae genetica, 38, 3–4. matziris, d. (2005). genetic variation and realized genetic gain from black pine tree improvement. silvae genetica, 54(1–6), 96–104. https://doi. org/10.1515/sg-2005-0015 miller, r. (1999). structure of wood. wood handbook: wood as an engineering material. madison, wi, usa: usda forest service, forest products laboratory. mirov, n.t. (1967). the genus pinus. new york, ny, usa: ronald press company. nicholls, j.w.p., & brown, a.g. (1974). the inheritance of heartwood formation in pinus radiata d. don. silvae genetica, 23(5), 138–141. pâques, l.e. (2001). genetic control of heartwood content in larch. silvae genetica, 50(2), 69–75. park, e.-j., chung, h.-j., park, h.j., kim, g.d., ahn, y.-h., & lee, s.k. (2013). suppression of src/erk and gsk-3/β-catenin signaling by pinosylvin inhibits https://doi.org/10.1016/s0167-7799(01)01705-x https://doi.org/10.1016/s0167-7799(01)01705-x ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 11 the growth of human colorectal cancer cells. food and chemical toxicology, 55, 424–433. https://doi. org/10.1016/j.fct.2013.01.007 park, e.-j., park, h.j., chung, h.-j., shin, y., min, h.-y., hong, j.-y., et al. (2012). antimetastatic activity of pinosylvin, a natural stilbenoid, is associated with the suppression of matrix metalloproteinases. the journal of nutritional biochemistry, 23(8), 946–952. https://doi.org/10.1016/j.jnutbio.2011.04.021 partanen, j., harju, a.m., venäläinen, m., & kärkkäinen, k. (2011). highly heritable heartwood properties of scots pine: possibilities for selective seed harvest in seed orchards. canadian journal of forest research, 41(10), 1993–2000. https://doi.org/10.1139/ x11-116 pietarinen, s.p., willför, s.m., ahotupa, m.o., hemming, j.e., & holmbom, b.r. (2006). knotwood and bark extracts: strong antioxidants from waste materials. journal of wood science, 52(5), 436–444. https:// doi.org/10.1007/s10086-005-0780-1 reveglia, p., cimmino, a., masi, m., nocera, p., berova, n., ellestad, g., & evidente, a. (2018). pimarane diterpenes: natural source, stereochemical configuration, and biological activity. chirality, 30(10), 1115–1134. https://doi.org/10.1002/ chir.23009 rezzi, s., bighelli, a., castola, v., & casanova, j. (2005). composition and chemical variability of the oleoresin of pinus nigra ssp. laricio from corsica. industrial crops and products, 21(1), 71–79. https://doi.org/10.1016/j.indcrop.2003.12.008 schanz, s., schröder, g., & schröder, j. (1992). stilbene synthase from scots pine (pinus sylvestris). febs letters, 313(1), 71–74. https://doi.org/10.1016/00145793(92)81187-q schmeda-hirschmann, g., astudillo, l., rodríguez, j., theoduloz, c., & yáñez, t. (2005b). gastroprotective effect of the mapuche crude drug araucaria araucana resin and its main constituents. journal of ethnopharmacology, 101(1–3), 271–276. https:// doi.org/10.1016/j.jep.2005.04.027 schmeda-hirschmanna, g., astudillo, l., sepúlveda, b., rodríguez, j.a., theoduloz, c., yáñez, t., & palenzuela, j.a. (2005a). gastroprotective effect and cytotoxicity of natural and semisynthetic labdane diterpenes from araucaria araucana resin. zeitschrift für naturforschung c, 60(7–8), 511– 522. https://doi.org/10.1515/znc-2005-7-801 schultz, t. p., boldin, w.d., fisher, t.h., nicholas, d.d., mcmurtrey, k.d., & pobanz, k. (1992). structurefungicidal properties of some 3and 4-hydroxylated stilbenes and bibenzyl analogues. phytochemistry, 31(11), 3801–3806. https://doi.org/10.1016/ s0031-9422(00)97531-9 schultz, tor p., hubbard, t.f., jin, l., fisher, t.h., & nicholas, d.d. (1990). role of stilbenes in the natural durability of wood: fungicidal structure-activity relationships. phytochemistry, 29(5), 1501–1507. https://doi.org/10.1016/0031-9422(90)80109-t sehlstedt-persson, m., & karlsson, o. (2010). natural durability and phenolic content in dried scots pine heartwood. bioresources, 5(2), 1126–1142. sjöström, e. (1993). wood chemistry: fundamentals and applications (2nd ed.). san diego: academic press. snedecor, g.w., & cohran, w.g. (1980). statistical methods (7th ed.). ames, ia, usa: the iowa state university press. spessard, g.o., matthews, d.r., nelson, m.d., rajtora, t.c., fossum, m.j., & giannini, j.l. (1995). phytoalexinlike activity of abietic acid and its derivatives. journal of agricultural and food chemistry, 43(6), 1690–1694. https://doi.org/10.1021/ jf00054a051 takahashi, n., kawada, t., goto, t., kim, c.-s., taimatsu, a., egawa, k., et al. (2003). abietic acid activates peroxisome proliferator-activated receptor-γ (pparγ) in raw264. 7 macrophages and 3t3-l1 adipocytes to regulate gene expression involved in inflammation and lipid metabolism. febs letters, 550(1–3), 190–194. https://doi.org/10.1016/ s0014-5793(03)00859-7 talevi, a., cravero, m.s., castro, e.a., & bruno-blanch, l.e. (2007). discovery of anticonvulsant activity of abietic acid through application of linear discriminant analysis. bioorganic & medicinal chemistry letters, 17(6), 1684–1690. https://doi. org/10.1016/j.bmcl.2006.12.098 taylor, a.m., gartner, b.l., & morre, j.i. (2002). heartwood formation and natural durability a review. wood and fiber science, 34(4), 587–611. tselepidakis, i.g., & theoharatos, g.a. (1989). a bioclimatic classification of the greek area. theoretical and applied climatology, 40(3), 147– 153. https://doi.org/10.1007/bf00866177 ulusu, n.n., ercil, d., sakar, m.k., & tezcan, e.f. (2002). abietic acid inhibits lipoxygenase activity. phytotherapy research, 16(1), 88–90. https://doi. org/10.1002/ptr.983 varelides, c., brofas, g., & varelides, y. (2001). provenance variation in pinus nigra at three sites in northern greece. annals of forest science, 58(8), 893–900. https://doi.org/10.1051/forest:2001103 venäläinen, m., harju, a.m., kainulainen, p., viitanen, h., & nikulainen, h. (2003). variation in the decay resistance and its relationship with other wood characteristics in old scots pines. annals of forest science, 60(5), 409–417. https://doi.org/10.1051/ forest:2003033 venäläinen, m., m. harju, a., terziev, n., laakso, t., & saranpää, p. (2006). decay resistance, extractive https://doi.org/10.1016/0014-5793(92)81187-q https://doi.org/10.1016/0014-5793(92)81187-q https://doi.org/10.1016/s0031-9422(00)97531-9 https://doi.org/10.1016/s0031-9422(00)97531-9 https://doi.org/10.1016/0031-9422(90)80109-t https://doi.org/10.1016/s0014-5793(03)00859-7 https://doi.org/10.1016/s0014-5793(03)00859-7 content, and water sorption capacity of siberian larch (larix sibirica lebed.) heartwood timber. holzforschung, 60(1), 99–103. https://doi. org/10.1515/hf.2006.017 vidakovic, m. (1974). genetics of european black pine (pinus nigra arn.). annales forest academy of science et artium slavorum meridionalium, 6/3., 86 pp. vidakovic, m. (1991). conifers. morphology and variation. zagreb, yugoslavia: grafičko zavod hrvatske. wheeler, ν.c., kriebel, h.b., lee, c.h., read, r.a., & wright, j.w. (1976). 15-year performance of european black pine in provenance tests in north central united states. silvae genetica, (25), 1–7. wilcox, m.d., & miller, j.t. (1975). pinus nigra variation and selection in new zealand. silvae genetica, 24(5–6), 132–140. willför, s, ahotupa, m., hemming, j., reunanen, m., eklund, p., sjöholm, r., et al. (2003). antioxidant activity of knotwood extractives and phenolic compounds of selected tree species. journal of agricultural and food chemistry, 51(26), 7600–7606. https://doi. org/10.1021/jf030445h willför, s., nisula, l., hemming, j., reunanen, m., & holmbom, b. (2004a). bioactive phenolic substances in industrially important tree species. part 1: knots and stemwood of different spruce species. holzforschung, 58(4), 335–344. https:// doi.org/10.1515/hf.2004.052 willför, s, nisula, l., hemming, j., reunanen, m., & holmbom, b. (2004b). bioactive phenolic substances in industrially important tree species. part 2: knots and stemwood of fir species. holzforschung, 58(6), 650–659. https://doi. org/10.1515/hf.2004.119 zobel, b., & jett, j. (1995). genetics of wood production. springer series in wood science. new york, berlin heidelberg: springer. https://doi. org/10.1007/978-3-642-79514-5 ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 12 additional file ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 13 figure a3: boxplots of the studied resin acids (mg/gdhw) for the sampled trees of the cso. (the circles and the stars represent the outliers the extreme values respectively.) figure a1: boxplots of total acetone extracts (mg/gdhw) and total resin acid (mg/gdhw) for the sampled trees of the cso. figure a2: boxplots of total stilbenes (mg/gdhw), pinosylvin (mg/ gdhw), monomethyl (mg/gdhw) and dimethyl ether mg/gdhw) of pinosylvin for the sampled trees of the cso. (the circles present the outliers.) figure a4: boxplots of heartwood proportion (%), diameter at breast height (cm) and tree height (m) of the sampled trees of the cso. (the circles present the outliers.) ioannidis & koropouli new zealand journal of forestry science (2023) 53:7 page 14 es ti m at e ta e t s p p m m e p d m e t r a a a d a a n a a p la lp m a p m a ip m a sp m a h w h 2 i 0. 55 0. 32 0. 37 0. 31 0. 43 0. 62 0. 53 0. 39 0. 62 0. 59 0. 15 0. 53 0. 53 0. 59 0. 40 se η 2 i 0. 00 00 1 0. 00 01 9 0. 00 22 2 0. 00 03 7 0. 07 20 6 0. 00 00 1 0. 00 00 9 0. 00 31 2 0. 00 03 1 0. 00 02 5 0. 00 05 1 0. 00 09 7 0. 00 28 9 0. 05 95 8 0. 00 53 7 h 2 c 0. 86 0. 71 0. 75 0. 69 0. 79 0. 89 0. 85 0. 76 0. 89 0. 88 0. 47 0. 85 0. 85 0. 88 0. 77 se η 2 c 0. 00 00 3 0. 00 09 1 0. 00 89 2 0. 00 18 3 0. 24 52 2 0. 00 00 3 0. 00 02 3 0. 01 19 8 0. 00 06 4 0. 00 05 5 0. 00 49 3 0. 00 24 7 0. 00 73 6 0. 13 31 3 0. 01 95 9 ta b le a 1: b ro ad s en se h er it ab ili ty e st im at es a t i nd iv id ua l a nd c lo ne m ea n ba si s an d th ei r st an da rd e rr or s fo r th e st ud ie d tr ai ts . differentiating individuals of armillaria species in new zealand forests sarah dodd1*, farhat shah2, mark kimberley3, chanatda somchit4 and ian hood4 1 department of agriculture and fisheries queensland, ecosciences precinct, gpo box 267, brisbane qld 4001, australia 2 new zealand institute for plant and food research ltd., private bag 4704, christchurch, new zealand 3 environmental statistics ltd., 72 becroft drive, auckland, new zealand 4 scion (new zealand forest research institute ltd.), private bag 3020, rotorua 3046, new zealand *corresponding author: sarah.dodd@daf.qld.gov.au (received for publication 4 december 2021; accepted in revised form 13 march 2022) abstract background: armillaria novae-zelandiae and a. limonea occur naturally as wood decay fungi in native forests in new zealand. as pathogens they are responsible for significant root disease in trees and shrubs in plantations, crops and urban parks and gardens. a thorough understanding of their population dynamics entails knowledge of the spatial arrangement of their individual mycelia or genets. in previous work the distributions of vegetative compatibility groups (vcgs) of these fungi were mapped in an area of native forest prior to and after replacement by a young pinus radiata plantation. with the advent of molecular technology, it has become possible to test species identities made earlier using culture techniques and to verify whether or not their vcgs, determined by incompatibility reactions between paired cultures, represent distinct individual genets. methods: stock subcultures of isolates representing each vcg were recovered from storage in order to obtain dna. extracted dna was subjected to a polymerase chain reaction procedure (up-pcr) using 11 universal primers to assess genetic variation between subcultures. bands were scored as either present or absent for each primer-subculture combination and cluster analysis was undertaken by generating dendrogram trees to reveal genetic groupings among subcultures. results: dna cluster analysis divided subcultures of isolates into two species groups, a. novae-zelandiae and a. limonea, corresponding to identities determined through culture morphology. within species, subcultures grouped into clusters that matched vcgs determined by earlier culture pairing. there was little indication of genetic variation within vcgs, except for one of a. limonea, which comprised two sub-clusters. conclusions: the armillaria species and vcgs identified by culture techniques in the laboratory were verified by independent molecular methodology. in general, the vcgs represent discrete individual genets or colonies in the field. techniques that differentiate isolates based on differences in their dna sequence provide a quick alternative to timeconsuming laboratory culture methods for resolving population spatial structure. however, some complementary isolate pairing may be necessary when rationalising the significance of groupings in dendrogram trees. new zealand journal of forestry science dodd et al. new zealand journal of forestry science (2022) 52:9 https://doi.org/10.33494/nzjfs522022x203x e-issn: 1179-5395 published on-line: 06/04/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access known as somatic incompatibility, si, groups). vegetative compatibility groups are recognised by the formation of an interaction zone between paired field isolates in laboratory culture, denoted macroor microscopically by such features as a gap, a barrier zone or a region of hyphal dissolution between the two dissimilar mycelia. introduction to fully comprehend the nature of fungal populations in ecological studies it is necessary to investigate the spatial configuration of their individual mycelia. one approach for filamentous species is to map the distribution and range of their vegetative compatibility groups (vcgs; also keywords: armillaria limonea; armillaria novae-zelandiae; dna cluster analysis; fungal populations; genets; universally primed pcr; vegetative compatibility groups http://creativecommons.org/licenses/by/4.0/), dodd et al. new zealand journal of forestry science (2022) 52:9 page 2 however, while mutual incompatibility distinguishes mycelia that differ genotypically, isolates belonging to the same vcg may not be clones or genets of a single mycelial colony (tyson et al. 2002). work with ascomycete fungi has shown that compatibility is governed genetically by a series of het recognition loci (for heterokaryon compatibility; also known as vic loci, for vegetative incompatibility), each of which contains two or more alleles (glass et al. 2000; moll et al. 2016). isolates that share identical het loci alleles, and which are therefore mutually compatible in culture, may differ genetically at other sites. with basidiomycetes it appears that field isolates of the same vcg are more likely to be genetically identical (malik & vilgalys 1999), but even with these fungi there are examples of genetic variation among individuals of the same vcg (jacobson et al. 1993; matsumoto et al. 1996; worrall 1997; stenlid & vasiliauskas 1998). knowledge of incompatibility among fungi has been reviewed by a number of authors (leslie 1993; worrall 1997; malik & vilgalys 1999; burnett 2003; stenlid 2008; krnjaja et al. 2013; also, heinzelmann et al. 2019). identifying and plotting vcgs has been helpful in studying armillaria communities in both natural vegetation and artificial ecosystems in different parts of the world (e.g., shaw & roth 1976; korhonen 1978; kile 1983, 1986; hood & morrison 1984; rishbeth 1991; rizzo et al. 1995; guillaumin et al. 1996; legrand et al. 1996; abomo-ndongo & guillaumin 1997; bruhn et al. 1997; dettman & van der kamp 2001; ferguson et al. 2003; prospero et al. 2003; mihail & bruhn 2005; szewczyk et al. 2015; and others). similar work in new zealand has shown that armillaria novae-zelandiae (g. stevenson) herink and armillaria limonea (g. stevenson) boesewinkel occur in the form of numerous vcgs in native forests, plantations and orchards of kiwifruit (actinidia deliciosa (a.chev.) c.f.liang & a.r.ferguson), supporting other research demonstrating that basidiospore colonisation plays an important ecological role, particularly for armillaria novae-zelandiae (horner 1992; hood et al. 2002a,b, 2004; power et al. 2008). in one investigation, isolates representative of the armillaria vcgs identified within plots in an area of native forest in the central north island were placed in storage for possible future study (hood & sandberg 1987, 1989, 1993). with the subsequent development of newer techniques, it became possible to examine these stock isolates using a molecular approach (dodd et al. 2006). a comparison between their vcg identities, as determined earlier by the culture pairing procedure, and the uniqueness of their dna composition, would indicate how well the two procedures were in accord. it would, in addition, reveal the level of genetic variation, if any, among isolates of the same vcgs. undertaking this work would also determine if the molecular procedure was a simpler, quicker and more precise technique than culture methodology for studying armillaria populations. this paper describes how this was done and the outcomes that were obtained. materials field isolates from plot 3 (hood & sandberg 1987), stored individually under oil or water in phials, were selected for the study. most were held as two or more duplicate ‘subcultures’ of the original isolates (hood & sandberg 1993) and were between 15 and (mostly) 21 years old when cultured out from storage. altogether 69 subcultures were successfully recovered of 26 original armillaria novae-zelandiae isolates and 15 a. limonea isolates, as previously determined by culture procedures (hood & sandberg 1987). these isolates represented 16 vcgs of a. novae-zelandiae and 6 vcgs of a. limonea, established earlier using culture pairing (table s1). for this study, cultures were labelled according to the formula ‘axxx_a_b.c’, where ‘axxx’ indicates a. limonea (alim) or a. novae-zelandiae (anz), ‘a’ is the vcg number (as in the earlier publications), ‘b’ is a unique isolate number and ‘c’ is one of up to four (1–4) subcultures of the original isolate (table s1). methods dna extraction mycelium was harvested from 4to 6week-old potato dextrose broth cultures grown at 24 °c. dna was extracted from 100 mg fresh mycelium of each isolate with the gentra puregene plant dna extraction kit (progenz, auckland, new zealand) using a micro-pestle for cell disruption. up-pcr dna obtained from all isolate subcultures was subjected to universally primed-polymerase chain reaction methodology (up-pcr) with 11 primers to assess genetic variation between the isolates. in this method single primers are used in a reaction and these were 0.3-1, 3-2, aa2m2, as4, as15, as15inv, l15, l15/as19, l21, l45 and foki (lübeck et al.1998; lübeck & lübeck 2005). each 25 μl amplification reaction contained 20 mm tris-hcl (ph 8.4), 50 mm kcl, 1.5 mm mgcl2, 200 μm each of datp, dttp, dgtp, and dctp, 20 pmoles primer, 2.5 mm mgcl2, 25 ng genomic dna, and 0.25 u taq dna polymerase (roche diagnostics n.z. ltd). the pcr reactions consisted of an initial 5 min at 94°c followed by 45 cycles of 50 s at 94°c, 1 min at the specified annealing temperature for that primer as reported in tyson et al. (2002) or 52°c for fok1, and 1 min at 72°c, with a final extension of 72°c for 7 min. pcr products were separated by electrophoresis in 1.6% agarose gels (fig. 1). four gels were run for each up-pcr primer, with subcultures of the same vcgs placed in adjacent lanes (table s2; the order on each gel was the same for all primers). selected lanes across the gel were loaded with a 1kb ladder (invitrogen, auckland nz) to use for the gel normalisation process. bands were visualised using sybr gold nucleic acid gel stain (invitrogen, auckland, nz) to increase band resolution. for each primer the four gels were assessed together. gels were normalised and bands at a single distance migratory position scored as either present or absent for each primer-isolate combination using the software bionumerics (applied maths, https://www. applied-maths.com/bionumerics). note, faint bands were excluded from the analysis as preliminary studies showed their presence/absence varied in multiple reactions run for the same dna/primer combinations making them unreliable for pattern comparisons between isolates. analysis the up-pcr procedure yields an array of binary data suitable for cluster analysis, with bands at each unique distance migratory position from the loading combs treated as an independent character for each primer. however, not all primers produce meaningful groupings. in addition, small differences in band identification were found between some subcultures of the same isolate with different primers. in order to use only data from the most functional primers, including those with fewest discrepancies, 2  2 grid matrices were prepared for all primers combined (table s3), as well as for each individual primer. cladistic dendrograms were constructed with combined data from all primers and from just three selected primers (3-2, as4, 0.3-1; table s4) using ward’s minimum variance method (ward 1963), which at each step merges the two clusters that provide the smallest increase in the combined error sum of squares. this analysis was performed using the hclust function in r (https://www.rdocumentation. org/packages/stats/versions/3.6.2/topics/hclust). in addition, the r pvclust function (https://www. rdocumentation.org/packages/pvclust/versions/2.2-0/ topics/pvclust; suzuki & shimodaira 2006) was used to assign p values to each cluster using bootstrap resampling techniques. results species identities determined earlier by culture methods in the laboratory were confirmed. for all isolates, both species grouped separately using up-pcr and cluster analysis by means of the ward procedure with data for the three selected primers combined (fig. 2). in addition, there was a clear trend for most vcgs to separate out from one another in a systematic fashion using cluster analysis of the dna data even if these groupings were not always judged to be statistically significant (fig. 2). exceptions were the vcg pairs 22, 65; 16, 23; 21, 56 of a. novae-zelandiae, and 34, 35 of a. limonea, whose components were not distinguished in the cladogram (fig. 2). there was little evidence of genetic variation within vcgs using cluster analysis. one exception was vcg 33 of a. limonea, which separated out into two related subclusters (fig. 2). apparent variation within a. novaezealandiae vcgs 15 and 18 was not considered significant (fig. 2). likewise, the apparent separations of two subcultures of each of two isolates were not significant (fig. 2; anz_15_9.1, anz_15_9.2; and anz_18_12.1, anz_18_12.2). discussion armillaria species are important wood decomposer fungi in native forests in new zealand. four species are known in southern beech forests (fuscospora and lophozonia, nothofagaceae), of which at least two (armillaria novae-zelandiae and a. limonea) also occur in podocarp-hardwood forests (hood et al. 2004, 2019; dodd et al. 2010). armillaria novae-zelandiae and a. limonea are significant root pathogens of pinus radiata d.don and at one time caused severe mortality in young trees when plantations of this host were established on sites cleared of indigenous forest. losses were also formerly sustained in orchards of kiwifruit as a result of root and collar infection by a. novae-zelandiae. for these reasons, much research has been undertaken to understand and manage the disease in these crops. this work has included molecular studies to develop pcr primers facilitating the ready identification of individual armillaria species (dodd et al. 2010). although less important than previously, armillaria species still cause significant disease and mortality in trees and shrubs in plantations and orchards, as well as in urban and rural settings in new zealand. an awareness of the nature of the populations of the disease agents therefore remains relevant. the purpose of the present investigation was to examine and compare two approaches to recognising and distinguishing individual mycelia in order to be able to map their spatial distribution in ecological studies. the results of the dna-based technology confirmed the findings from the earlier laboratory culture work with both armillaria species. the previously determined species identities of the field isolates were verified by the molecular method, as was the distinctiveness of the individual vcgs. in addition, the substantial variation observed between the vcg genets, indicating that dodd et al. new zealand journal of forestry science (2022) 52:9 page 3 figure 1: early gel of as15 up-pcr products showing differences between vcg groups among isolates of armillaria limonea. lanes 2, 3, 4, 7 and 10 represent isolates of vcg 33; lanes 1 and 5, of vcg 32; lanes 6, 8 and 9, of vcg 34. m, size marker lane; ndc, no dna control lane. https://www.applied-maths.com/bionumerics https://www.applied-maths.com/bionumerics https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/hclust https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/hclust https://www.rdocumentation.org/packages/pvclust/versions/2.2-0/topics/pvclust https://www.rdocumentation.org/packages/pvclust/versions/2.2-0/topics/pvclust https://www.rdocumentation.org/packages/pvclust/versions/2.2-0/topics/pvclust recombination has occurred, corroborates the role of basidiospores as significant in dispersal, especially of a. novae-zelandiae. however, there was little evidence of genetic variation within vcgs, implying that each represented a distinct mycelial colony (diploid, as indicated by their dark, crustose, depressed culture morphology; haploid isolates tend to take on a different appearance). an earlier acceptance of variation within vcgs of both armillaria species is therefore here qualified (hood 2012; hood et al. 2019). however, this does not mean that such within-vcg variation does not occur with armillaria species (kile 1983; guillaumin et al. 1996; szewczyk et al. 2015), just that it appears to be less likely among isolates taken from the same individual vegetative field colonies. in the present study a difference was apparent between subcultures of two isolates (alim_33_31.1 and alim_33_34.1) and those of the other isolates within vcg 33 (a. limonea). both these isolates were obtained, two years apart, from a restricted zone, 4 m across, within the full area, ≥ 14 m wide, occupied by vcg 33 in plot 3. possible explanations include somatic mutation or perhaps the formation of adjacent sib-related colonies with a common parent derived from basidiospores from one local fruitbody cluster (kile 1983; rishbeth 1991; hood & sandberg dodd et al. new zealand journal of forestry science (2022) 52:9 page 4 1993; hood 2012). a change during storage appears less likely (hood & sandberg 1993). vcg 33 was represented by a comparatively large number of isolates. it may be possible that more isolates might have revealed similar variation within some other vcgs. likewise, four pairs of vcgs that each grouped together were represented by only one isolate per vcg. would these vcgs have been differentiated using the dna-based approach had there also been more isolates? in the up-pcr technique, the selected primer binds to regions of the entire genome of the same (or nearly the same) base pair sequence. in this procedure, the primers employed are designed to target non-coding genome regions that retain mutations and are hence more variable than coding segments. primers are longer and binding conditions more stringent than the similar rapd technique (bardakci 2001). nevertheless, despite this, these highly variable regions may or may not be phylogenetically informative. only some primers are successful in revealing meaningful groupings, depending on the genome regions being sampled. however, once obtained, useful data from appropriate primers can be combined to increase precision and reduce “noise”. in this study, data from some primers resulted in only partially workable, semi-rational patterns of clustering in figure 2: cluster analysis of subcultures of armillaria field isolates for primers 3_2, as4, 0.3_1 combined, using ward’s minimum variance method (ward 1963). label code: axxx_a_b.c, where axxx indicates a. novae-zelandiae or a. limonea (as identified in culture), a is the vcg number (determined by incompatibility between pairs of cultures), b is the assigned isolate number and c is a laboratory subculture of that isolate (1-4). results of bootstrap resampling using the r pvclust procedure (suzuki and shimodaira 2006): xx (red), approximately unbiased (au) p-value % (generally considered the best assessment of strength of support for the cluster), yy (green), bootstrap probability % (bp), zz (grey) rank of the cluster. red rectangles indicate clusters with au>95. which vcg clusters were, nonetheless, still recognisable. however, the three primers as4, 0.3-1 and 3-2, provided functional data suitable for cluster analyses, giving rise, when combined, to fully coherent results that matched those from the earlier research. the culture pairing and dna-based procedures proved complementary, and in combination were successful in identifying discrete mycelial field colonies of a. novaezelandiae and a. limonea previously mapped as vcgs. the methods for identifying species and vcgs in culture are laborious and time consuming, although a multiple pairing design procedure has been prescribed for reducing some of the effort (burgess et al. 2009). the uppcr technique was quicker and effective, using the three primers 3-2, as4, 0.3-1 followed by clustering analysis by means of the ward procedure. however, in this study, resolving which sub-groups represented authentic field colonies relied on information from pairing of cultures. one way to realise this goal in a future study would be to subject dna data to cluster analysis, “calibrating” resultant groups by pairings among a limited sample of isolates in the laboratory to support statistical estimates of group significance. authors' contributions sd and fs conducted the molecular and laboratory work with cultures supplied by ih. statistical analyses were performed by mk. cs undertook some preliminary analyses. the paper was written by ih, sd and mk and the final version accepted by all co-authors. acknowledgements judy gardner helped with the maintenance and revival of the stock cultures in preparation for this study. thanks, also, to ruth butler and ikram khan for earlier contributions to this study and to an anonymous referee. funding was provided by the former new zealand foundation for research, science and technology. additional file tables s1-s4 references abomo-ndongo, s., & guillaumin, j.-j. (1997). somatic incompatibility among african armillaria isolates. european journal of forest pathology, 27, 201206. https://doi.org/10.1111/j.1439-0329.1997. tb00863.x bardakci f. (2001). random amplified polymorphic dna (rapd) markers. turkish journal of biology, 25, 185-196. bruhn, j.n., brenneman, j.a., wetteroff jr., j.j., mihail, j.d., & leininger, t.d. (1997). spatial patterns of armillaria populations in the walker branch watershed throughfall displacement experiment, tennessee, usa. in s.g. pallardy, r.a. cecich, h.g. garrett, & p.s. johnson, eds. proceedings of the 11th central hardwood forest conference, columbia, missouri, march 23-26, 1997. general technical report nc-188, pp. 49-57. st. paul, mn: united states department of agriculture, forest service, north central experiment station. burgess, t., bihon, w., wingfield, m.j., & wingfield, b.d. (2009). a simple and rapid method to determine vegetative compatibility groups in fungi. inoculum, december 2009 (supplement to mycologia, 60(6)), 1-2. burnett, j.h. (2003). fungal populations and species. oxford, uk: oxford university press dettman, j.r., & van der kamp, b.j. (2001). the population structure of armillaria ostoyae and armillaria sinapina in the central interior of british columbia. canadian journal of botany, 79, 600-611. https:// doi.org/10.1139/b01-033 dodd, s.l., shah, f.h., & hood, i.a. (2006). molecular tools to determine the mode of spread of the root pathogen armillaria in commercial pine forests. poster abstract. proceedings of the joint meeting of the american phytopathological society, canadian phytopathological society, and the mycological society of america, 29 july – 2 august, 2006, québec city, québec, canada. phytopathology, 96(6: supplement), s30. https://apsjournals.apsnet. org/doi/epdf/10.1094/phyto.2006.96.6.s1 dodd, s.l., ramsfield, t.d., & marshall, j.w. (2010). pcr primers to distinguish armillaria species found in new zealand. australasian plant pathology, 39(6), 536-543. https://doi.org/10.1071/ap10026 ferguson, b.a., dreisbach, t.a., parks, c.g., filip, g.m., & schmitt, c.l. (2003). coarse-scale population structure of pathogenic armillaria species in a mixed-conifer forest in the blue mountains of northeast oregon. canadian journal of forest research, 33, 612-623. https://doi.org/10.1139/ x03-065 glass, n.l., jacobson, d.j., & shiu, p.k.t. (2000). the genetics of hyphal fusion and vegetative incompatibility in filamentous ascomycete fungi. annual review of genetics, 34, 165-186. https:// doi.org/10.1146/annurev.genet.34.1.165 guillaumin, j.-j., anderson, j.b., legrand, p., ghahari, s., & berthelay, s. (1996). a comparison of different methods for the identification of genets of armillaria spp. new phytologist, 133, 333-343. h t t p s : / / d o i . o rg / 1 0 . 1 1 1 1 / j . 1 4 6 9 8 1 3 7 . 1 9 9 6 . tb01900.x heinzelmann, r., dutech, c., tsykun, t., labbé, f., soularue, j.-p., & prospero, s. (2019). latest advances and future perspectives in armillaria research. canadian journal of plant pathology, 41, 1-23. https://doi.org/10.1080/07060661.2018.1 558284 dodd et al. new zealand journal of forestry science (2022) 52:9 page 5 https://doi.org/10.1111/j.1439-0329.1997.tb00863.x https://doi.org/10.1111/j.1439-0329.1997.tb00863.x https://doi.org/10.1071/ap10026 https://doi.org/10.1139/x03-065 https://doi.org/10.1139/x03-065 https://doi.org/10.1146/annurev.genet.34.1.165 https://doi.org/10.1146/annurev.genet.34.1.165 https://doi.org/10.1111/j.1469-8137.1996.tb01900.x https://doi.org/10.1111/j.1469-8137.1996.tb01900.x https://doi.org/10.1080/07060661.2018.1558284 https://doi.org/10.1080/07060661.2018.1558284 hood, i.a. (2012). fungi decaying fallen stems of rimu (dacrydium cupressinum, podocarpaceae) in southern westland, new zealand. new zealand journal of botany, 50, 59-69. https://doi.org/10.1 080/0028825x.2011.643474 hood, i.a., & morrison, d.j. (1984): incompatibility testing of armillaria isolates in a wood substrate. canadian forestry service research notes, 4(1), 8-9. hood, i.a., & sandberg, c.j. (1987). occurrence of armillaria rhizomorph populations in the soil beneath indigenous forests in the bay of plenty, new zealand. new zealand journal of forestry science, 17, 83-99. https://www.scionresearch. c o m / _ _ d a t a / a s s e t s / p d f _ f i l e / 0 0 0 4 / 5 9 8 7 2 / nzjfs1711987hood83_99.pdf hood, i.a., & sandberg, c.j. (1989). changes in soil populations of armillaria species following felling and burning of indigenous forests in the bay of plenty, new zealand. in d.j. morrison (ed.), proceedings of the seventh international conference on root and butt rots of forest trees. british columbia, canada, august 1988. iufro working party s2.06.01. pp. 288-296. hood, i.a., & sandberg, c.j. (1993). armillaria populations in a pinus radiata plantation on a former indigenous rainforest site. new zealand journal of forestry science, 23(1), 62-77. https://www.scionresearch. c o m / _ _ d a t a / a s s e t s / p d f _ f i l e / 0 0 0 4 / 5 9 7 8 2 / nzjfs2311993hood62_77.pdf hood, i.a., horner, i.j., gardner, j.f., & sandberg, c.j. (2002a). armillaria root disease of pinus radiata in new zealand: 1. basidiospore dispersal. new zealand journal of forestry science, 32, 94-102. https://www.scionresearch.com/__data/assets/ pdf_file/0010/59284/reprint-2803-01a.pdf hood, i.a., kimberley, m.o., gardner, j.f., & sandberg, c.j. (2002b). armillaria root disease of pinus radiata in new zealand: 3. influence of thinning and pruning. new zealand journal of forestry science, 32, 116-132. https://www.scionresearch.com/__data/assets/pdf_ file/0008/59282/reprint-2803-03a.pdf hood, i.a., beets, p.n., kimberley, m.o., gardner, j.f., oliver, g.r., & pearce, s. (2004). colonisation of podocarp coarse woody debris by decomposer basidiomycete fungi in an indigenous forest in the central north island of new zealand. forest ecology and management, 196, 311-325. https:// doi.org/10.1016/j.foreco.2004.03.024 hood, i.a., mcdougal, r.l., somchit, c., kimberley, m.o., lewis, a.s.r., & hood, j.o.l. (2019). fungi decaying the wood of fallen beech (nothofagus) trees in the south island of new zealand. canadian journal of forest research, 49, 1-17. https://doi.org/10.1139/ cjfr-2018-0179 horner, i.j. (1992). epidemiology of armillaria rootrot of kiwifruit. acta horticulturae (kiwifruit ii), 297, 573-578. https://doi.org/10.17660/ actahortic.1992.297.75 jacobson, k.m., miller jr., o.k., & turner, b.j. (1993). randomly amplified polymorphic dna markers are superior to somatic incompatibility tests for discriminating genotypes in natural populations of the ectomycorrhizal fungus suillus granulatus. proceedings of the national academy of sciences usa 90, 9159-9163. https://doi.org/10.1073/ pnas.90.19.9159 kile, g.a. (1983). identification of genotypes and the clonal development of armillaria luteobubalina watling & kile in eucalypt forests. australian journal of botany, 31, 657-671. https://doi. org/10.1071/bt9830657 kile, g.a. (1986). genotypes of armillaria hinnulea in wet sclerophyll eucalypt forest in tasmania. transactions of the british mycological society, 87, 312-314. https://doi.org/10.1016/s00071536(86)80036-5 krnjaja, v., lević, j., stanković, s., & vasić, t. (2013). the use of vegetative compatibility tests for identification of biodiversity of phytopathogenic fungi. pesticides and phytomedicine (belgrade), 28(3), 157-165. https://doi.org/10.2298/pif1303157k korhonen, k. (1978). interfertility and clonal size in the armillariella mellea complex. karstenia, 18, 31-42. https://doi.org/10.29203/ka.1978.135 legrand, p., ghahari, s., & guilaumin, j.-j. (1996). occurrence of genets of armillaria spp. in four mountain forests in central france: the colonization strategy of armillaria ostoyae. new phytologist, 133, 321-332. https://doi. org/10.1111/j.1469-8137.1996.tb01899.x leslie, j.f. (1993). fungal vegetative compatibility. annual review of phytopathology, 31, 127-150. https://doi. org/10.1146/annurev.py.31.090193.001015 lübeck, m., & lübeck, p.s. (2005). universally primed pcr (up-pcr) and its applications in mycology. in s.k. deshmukh & m.k. rai (eds.), biodiversity of fungi: their role in human life (pp. 409-438). enfield, nh, usa: science publishers, lübeck, p.s., alekhina, i.a., lübeck, m., & bulat s.a. (1998): up-pcr genotyping and rdna analysis of ascochyta pisi lib. journal of phytopathology, 146, 51-55. h t t p s : / / d o i . o rg / 1 0 . 1 1 1 1 / j . 1 4 3 9 0 4 3 4 . 1 9 9 8 . tb04749.x malik, m., & vilgalys, r. (1999). somatic incompatibility in fungi. in j.j. worrall (ed.), structure and dynamics of fungal populations. (pp. 123-138). population and community biology series volume 25. dordrecht, netherlands: springer. https://doi. org/10.1007/978-94-011-4423-0_6 dodd et al. new zealand journal of forestry science (2022) 52:9 page 6 https://doi.org/10.1080/0028825x.2011.643474 https://doi.org/10.1080/0028825x.2011.643474 https://doi.org/10.1016/j.foreco.2004.03.024 https://doi.org/10.1016/j.foreco.2004.03.024 https://doi.org/10.1139/cjfr-2018-0179 https://doi.org/10.1139/cjfr-2018-0179 https://doi.org/10.17660/actahortic.1992.297.75 https://doi.org/10.17660/actahortic.1992.297.75 https://doi.org/10.1073/pnas.90.19.9159 https://doi.org/10.1073/pnas.90.19.9159 https://doi.org/10.1071/bt9830657 https://doi.org/10.1071/bt9830657 https://doi.org/10.1016/s0007-1536(86)80036-5 https://doi.org/10.1016/s0007-1536(86)80036-5 https://doi.org/10.2298/pif1303157k https://doi.org/10.29203/ka.1978.135 https://doi.org/10.1111/j.1469-8137.1996.tb01899.x https://doi.org/10.1111/j.1469-8137.1996.tb01899.x https://doi.org/10.1146/annurev.py.31.090193.001015 https://doi.org/10.1146/annurev.py.31.090193.001015 https://doi.org/10.1111/j.1439-0434.1998.tb04749.x https://doi.org/10.1111/j.1439-0434.1998.tb04749.x https://doi.org/10.1007/978-94-011-4423-0_6 https://doi.org/10.1007/978-94-011-4423-0_6 matsumoto, n., uchlyama, k., & tsushima, s. (1996). genets of typhula ishikariensis biotype a belonging to a vegetative compatibility group. canadian journal of botany, 74, 1695-1700. https://doi. org/10.1139/b96-206 mihail, j.d., & bruhn, j.n. (2005). foraging behaviour of armillaria rhizomorph systems. mycological research, 109, 1195-1207. https://doi.org/10.1017/s0953756205003606 moll, j., hoppe, b., könig, s., wubet, t., buscot, f., & krüger, d. (2016). spatial distribution of fungal communities in an arable soil. plos one, 11(2), e0148130. https://doi.org/10.1371/journal. pone.0148130 power, m.w.p., ramsfield, t.d., & hood, i.a. (2008). detection of armillaria basidiospore dispersal. new zealand plant protection, 61, 35-40. https:// doi.org/10.30843/nzpp.2008.61.6830 prospero, s., rigling, d., & holdenrieder, o. (2003). population structure of armillaria species in managed norway spruce stands in the alps. new phytologist, 158, 365-373. https://doi. org/10.1046/j.1469-8137.2003.00731.x rishbeth, j. (1991). armillaria in an ancient broadleaved woodland. european journal of forest pathology, 21, 239-249. https://doi. org/10.1111/j.1439-0329.1991.tb00975.x rizzo, d.m., blanchette, r.a., & may, g. (1995). distribution of armillaria ostoyae genets in a pinus resinosa – pinus banksiana forest. canadian journal of botany, 73, 776-787. https://doi.org/10.1139/b95-085 shaw iii, c.g., & roth, l.f. (1976). persistence and distribution of a clone of armillaria mellea in a ponderosa pine forest. phytopathology, 66, 12101213. https://doi.org/10.1094/phyto-66-1210 stenlid, j. (2008): population biology of forest decomposer basidiomycetes. in l. boddy, j.c. frankland, & p. van west (eds.). ecology of saprotrophic basidiomycetes (pp. 105-122). british mycological society symposia series vol. 28. elsevier/academic press. https://doi. org/10.1016/s0275-0287(08)80008-2 stenlid, j., & vasiliauskas, r. (1998). genetic diversity within and among vegetative compatibility groups of stereum sanguinolentum. molecular ecology, 7, 1265-1274. https://doi.org/10.1046/j.1365294x.1998.00437.x suzuki, r., & shimodaira, h. (2006): pvclust: an r package for assessing the uncertainty in hierarchical clustering. bioinformatics, 22, 1540-2542. https:// doi.org/10.1093/bioinformatics/btl117 szewczyk, w., kwaśna, h., & behnke-borowczyk, j. (2015). populations of armillaria species in pine plantations in west-central poland. dendrobiology, 74, 95-108. https://doi.org/10.12657/denbio.074.010 tyson, j.l., ridgway, h.j., fullerton, r.a., & stewart, a. (2002). genetic diversity in new zealand populations of sclerotium cepivorum. new zealand journal of crop and horticultural science, 30, 3748. https://doi.org/10.1080/01140671.2002.951 4197 ward jr., j.h. (1963). hierarchical grouping to optimize an objective function. journal of the american statistical association, 58(301), 236-244. https:// doi.org/10.1080/01621459.1963.10500845 worrall, j.j. (1997). somatic incompatibility in basidiomycetes. mycologia, 89, 24-36. https://doi. org/10.1080/00275514.1997.12026751 dodd et al. new zealand journal of forestry science (2022) 52:9 page 7 https://doi.org/10.1139/b96-206 https://doi.org/10.1139/b96-206 https://doi.org/10.1017/s0953756205003606 https://doi.org/10.1371/journal.pone.0148130 https://doi.org/10.1371/journal.pone.0148130 https://doi.org/10.30843/nzpp.2008.61.6830 https://doi.org/10.30843/nzpp.2008.61.6830 https://doi.org/10.1046/j.1469-8137.2003.00731.x https://doi.org/10.1046/j.1469-8137.2003.00731.x https://doi.org/10.1111/j.1439-0329.1991.tb00975.x https://doi.org/10.1111/j.1439-0329.1991.tb00975.x https://doi.org/10.1139/b95-085 https://doi.org/10.1094/phyto-66-1210 https://doi.org/10.1016/s0275-0287(08)80008-2 https://doi.org/10.1016/s0275-0287(08)80008-2 https://doi.org/10.1046/j.1365-294x.1998.00437.x https://doi.org/10.1046/j.1365-294x.1998.00437.x https://doi.org/10.1093/bioinformatics/btl117 https://doi.org/10.1093/bioinformatics/btl117 https://doi.org/10.12657/denbio.074.010 https://doi.org/10.1080/01140671.2002.9514197 https://doi.org/10.1080/01140671.2002.9514197 https://doi.org/10.1080/01621459.1963.10500845 https://doi.org/10.1080/01621459.1963.10500845 https://doi.org/10.1080/00275514.1997.12026751 https://doi.org/10.1080/00275514.1997.12026751 tables s1–s4 new zealand journal of forestry science differentiating individuals of armillaria species in new zealand forests supplementary material table s1. details of isolates. isolate code1 species (by culture procedure) vcg (by culture pairing) source2 published reference (hood & sandberg 1987, 1989, 1993) anz_12_2.1 anz 12 00.08 1987 anz_12_2.2 anz 12 00.08 1987 anz_12_3.1 anz 12 04.16 1987 anz_12_3.2 anz 12 04.16 1987 anz_12_4.1 anz 12 06.20 1987 anz_13_5.1 anz 13 00.34 1987 anz_13_5.2 anz 13 00.34 1987 anz_13_6.1 anz 13 02.34 1987 anz_14_7.1 anz 14 06.28 1987 anz_14_7.2 anz 14 06.28 1987 anz_15_8.1 anz 15 10.22 1987 anz_15_8.2 anz 15 10.22 1987 anz_15_8.3 anz 15 10.22 1987 anz_15_9.1 anz 15 14.20 1989 anz_15_9.2 anz 15 14.20 1989 anz_16_10.1 anz 16 18.32 1987 anz_17_11.1 anz 17 18.26 1987 anz_17_11.2 anz 17 18.26 1987 anz_18_12.1 anz 18 18.34 1987 anz_18_12.2 anz 18 18.34 1987 anz_18_13.13 anz 18 28.24 1987 anz_18_14.13 anz 18 28.24 1987 anz_18_15.1 anz 18 34.22 1987 anz_19_16.2 anz 19 20.16 1987 anz_19_17.1 anz 19 24.16? 1987 anz_19_17.2 anz 19 24.16? 1987 anz_20_18.1 anz 20 28.10 1987 anz_20_18.2 anz 20 28.10 1987 anz_20_18.3 anz 20 28.10 1987 anz_20_18.4 anz 20 28.10 1987 anz_20_19.1 anz 20 32.10 1987 anz_20_19.2 anz 20 32.10 1987 anz_20_20.1 anz 20 34.06 1989 anz_21_21.1 anz 21 20.22 1987 anz_22_22.2 anz 22 22.32 1987 page 2 of 13 anz_23_23.1 anz 23 26.30 1987 anz_26_24.1 anz 26 f243 1987 anz_26_24.2 anz 26 f243 1987 alim_32_1.1 alim 32 04.26 1987 alim_33_25.1 alim 33 20.34 1987 alim_33_25.2 alim 33 20.34 1987 alim_33_26.1 alim 33 22.26 1987 alim_33_26.2 alim 33 22.26 1987 alim_33_26.3 alim 33 22.26 1987 alim_33_27.14 alim 33 22.30 1987 alim_33_27.24 alim 33 22.30 1987 alim_33_28.14 alim 33 22.30 1987 alim_33_29.1 alim 33 26.24 1987 alim_33_30.1 alim 33 30.30 1987 alim_33_30.2 alim 33 30.30 1987 alim_33_31.1 alim 33 30.32 1987 alim_33_32.1 alim 33 32.32 1987 alim_33_33.1 alim 33 32.34 1987 alim_33_34.1 alim 33 34.32 1989 alim_34_35.1 alim 34 28.00 1987 alim_35_36.1 alim 35 12.16 1989 alim_36_37.1 alim 36 26.12 1987 alim_36_37.2 alim 36 26.12 1987 alim_37_38.1 alim 37 f239 1987 alim_37_38.2 alim 37 f239 1987 anz_55_39.1 anz 55 s650 1993 anz_55_39.2 anz 55 s650 1993 anz_56_40.1 anz 56 s633 1993 anz_65_41.1 anz 65 s780 1993 1 code key: axxx_a_b.c, where axxx is alim (a. limonea) or anz (a. novae-zelandiae), determined in culture; a= vcg no., determined by culture pairing; b=assigned isolate no.; c=laboratory subculture no. 2 grid coordinates (in metres) locating sample position across plot 3 (from rhizomorphs), except (locations also known) 'f' from fruitbody tissue and 's' from pinus radiata d.don seedling. 3,4 separate isolates from rhizomorphs in the same soil cores. note: data from five subcultures, selected arbitrarily within their own isolate group to leave one remaining, were discarded where it was no longer clear whether they were separate isolates from the same core or subcultures from the one isolate (not shown in the above table). new zealand journal of forestry science table s2. arrangement of subcultures of isolates during electrophoresis runs. four gels were run for each up-pcr primer, with subcultures of the same vcgs placed in adjacent lanes (the order on each gel was the same for all primers). empty lanes were used for ladders. for each primer the four gels were assessed together. lane gel 1 gel 2 gel 3 gel 4 1 data omitted1 2 data omitted1 anz_12_2.1 anz_20_18.1 anz_19_16.2 3 alim_33_25.1 anz_12_2.2 anz_20_18.2 anz_19_17.1 4 alim_33_25.2 anz_12_3.1 anz_20_18.3 anz_19_17.2 5 alim_33_27.1 anz_13_5.1 anz_20_18.4 alim_33_28.1 6 alim_33_26.1 anz_12_4.1 anz_20_19.1 7 alim_33_26.2 anz_12_3.2 anz_20_19.2 alim_34_35.1 8 alim_33_26.3 anz_13_5.2 anz_20_20.1 alim_35_36.1 9 alim_33_27.2 anz_13_6.1 anz_15_8.1 alim_32_1.1 10 alim_37_38.1 11 alim_33_29.1 data omitted1 anz_15_8.2 alim_36_37.1 12 alim_33_31.1 data omitted1 anz_15_8.3 alim_36_37.2 13 alim_33_30.1 anz_23_23.1 anz_15_9.1 alim_37_38.2 14 alim_33_30.2 anz_14_7.1 anz_15_9.2 15 alim_33_32.1 anz_14_7.2 anz_18_12.1 anz_26_24.1 16 alim_33_33.1 data omitted1 anz_18_12.2 anz_26_24.2 17 alim_33_34.1 anz_16_10.1 anz_18_13.1 anz_65_41.1 18 anz_22_22.2 anz_17_11.1 anz_18_14.1 anz_56_40.1 19 anz_21_21.1 anz_17_11.2 anz_18_15.1 anz_55_39.1 20 anz_55_39.2 1 see note, table s1. n ew z ea la nd jo ur na l o f f or es tr y sc ie nc e ta bl e s 3. d is ta nc e m at ri x, c om bi ne d pr im er s (n um be rs o f b an d di ff er en ce s be tw ee n pa ir s of s ub cu ltu re s) . alim_32_1.1 anz_12_2.1 anz_12_2.2 anz_12_3.1 anz_12_3.2 anz_12_4.1 anz_13_5.1 anz_13_5.2 anz_13_6.1 anz_14_7.1 anz_14_7.2 anz_15_8.1 anz_15_8.2 anz_15_8.3 anz_15_9.1 anz_15_9.2 anz_16_10.1 anz_17_11.1 anz_17_11.2 anz_18_12.1 anz_18_12.2 anz_18_13.1 anz_18_14.1 anz_18_15.1 anz_19_16.2 al im _3 2_ 1. 1 0 49 49 50 50 59 43 42 47 50 62 65 62 62 62 52 54 50 59 57 55 55 55 60 62 an z_ 12 _2 .1 49 0 2 3 1 10 26 27 34 39 49 52 53 53 53 41 39 33 42 56 54 54 50 55 51 an z_ 12 _2 .2 49 2 0 3 1 10 24 25 32 37 47 50 51 51 51 39 37 31 40 54 52 52 48 53 49 an z_ 12 _3 .1 50 3 3 0 2 9 27 28 35 40 48 53 52 52 52 40 38 32 41 55 53 53 49 54 52 an z_ 12 _3 .2 50 1 1 2 0 9 25 26 33 38 48 51 52 52 52 40 38 32 41 55 53 53 49 54 50 an z_ 12 _4 .1 59 10 10 9 9 0 18 19 30 35 43 48 49 49 49 43 39 29 42 56 54 54 50 59 55 an z_ 13 _5 .1 43 26 24 27 25 18 0 1 12 27 37 42 43 43 43 37 37 25 40 50 48 48 44 53 45 an z_ 13 _5 .2 42 27 25 28 26 19 1 0 11 26 36 41 42 42 42 36 36 24 39 49 47 47 43 52 46 an z_ 13 _6 .1 47 34 32 35 33 30 12 11 0 17 45 48 51 51 51 35 35 31 46 54 52 52 48 51 47 an z_ 14 _7 .1 50 39 37 40 38 35 27 26 17 0 28 51 54 54 54 40 26 28 43 55 53 53 51 54 50 an z_ 14 _7 .2 62 49 47 48 48 43 37 36 45 28 0 27 28 28 28 58 42 36 25 41 39 39 37 56 52 an z_ 15 _8 .1 65 52 50 53 51 48 42 41 48 51 27 0 5 5 5 39 61 45 36 38 40 40 36 55 55 an z_ 15 _8 .2 62 53 51 52 52 49 43 42 51 54 28 5 0 0 0 34 62 42 35 33 35 35 35 54 52 an z_ 15 _8 .3 62 53 51 52 52 49 43 42 51 54 28 5 0 0 0 34 62 42 35 33 35 35 35 54 52 an z_ 15 _9 .1 62 53 51 52 52 49 43 42 51 54 28 5 0 0 0 34 62 42 35 33 35 35 35 54 52 an z_ 15 _9 .2 52 41 39 40 40 43 37 36 35 40 58 39 34 34 34 0 30 38 51 53 53 53 53 58 54 an z_ 16 _1 0. 1 54 39 37 38 38 39 37 36 35 26 42 61 62 62 62 30 0 34 43 55 53 53 53 58 52 an z_ 17 _1 1. 1 50 33 31 32 32 29 25 24 31 28 36 45 42 42 42 38 34 0 31 53 51 51 49 54 54 an z_ 17 _1 1. 2 59 42 40 41 41 42 40 39 46 43 25 36 35 35 35 51 43 31 0 28 26 26 24 53 53 an z_ 18 _1 2. 1 57 56 54 55 55 56 50 49 54 55 41 38 33 33 33 53 55 53 28 0 2 2 8 37 51 an z_ 18 _1 2. 2 55 54 52 53 53 54 48 47 52 53 39 40 35 35 35 53 53 51 26 2 0 0 6 35 49 an z_ 18 _1 3. 1 55 54 52 53 53 54 48 47 52 53 39 40 35 35 35 53 53 51 26 2 0 0 6 35 49 an z_ 18 _1 4. 1 55 50 48 49 49 50 44 43 48 51 37 36 35 35 35 53 53 49 24 8 6 6 0 29 51 p ag e 2 of 1 3 an z_ 18 _1 5. 1 60 55 53 54 54 59 53 52 51 54 56 55 54 54 54 58 58 54 53 37 35 35 29 0 24 an z_ 19 _1 6. 2 62 51 49 52 50 55 45 46 47 50 52 55 52 52 52 54 52 54 53 51 49 49 51 24 0 an z_ 19 _1 7. 1 64 49 47 50 48 53 43 44 45 48 50 53 54 54 54 56 50 54 53 53 51 51 49 22 2 an z_ 19 _1 7. 2 62 47 45 48 46 51 41 42 47 50 32 35 36 36 36 52 46 54 37 39 37 37 35 50 30 an z_ 20 _1 8. 1 63 46 46 47 47 52 52 53 58 51 33 38 37 37 37 53 53 53 38 36 34 34 34 49 53 an z_ 20 _1 8. 2 65 48 48 49 49 54 54 55 60 53 35 40 39 39 39 55 55 55 40 38 36 36 36 47 53 an z_ 20 _1 8. 3 64 47 47 48 48 53 53 54 59 52 34 37 38 38 38 54 54 54 39 37 35 35 35 50 54 an z_ 20 _1 8. 4 64 47 47 48 48 53 53 54 59 52 34 37 38 38 38 54 54 54 39 37 35 35 35 50 54 an z_ 20 _1 9. 1 63 48 48 49 49 54 54 55 58 51 35 36 39 39 39 55 55 55 40 38 36 36 36 51 55 an z_ 20 _1 9. 2 63 48 48 49 49 54 54 55 58 51 35 36 39 39 39 55 55 55 40 38 36 36 36 51 55 an z_ 20 _2 0. 1 65 50 50 51 51 54 54 55 54 47 61 62 65 65 65 57 57 51 64 60 58 58 58 55 59 an z_ 21 _2 1. 1 45 48 46 47 47 50 38 37 36 39 53 64 59 59 59 43 43 47 56 52 50 50 52 49 49 an z_ 22 _2 2. 2 40 35 33 36 34 41 33 32 33 32 46 57 54 54 54 36 28 36 47 45 43 43 45 50 42 an z_ 23 _2 3. 1 52 23 21 22 22 27 29 30 31 32 44 55 56 56 56 36 24 30 39 55 53 53 49 54 50 an z_ 26 _2 4. 1 66 53 53 52 54 57 57 58 57 46 48 57 56 56 56 54 50 54 47 55 53 53 53 38 34 an z_ 26 _2 4. 2 71 58 58 57 59 62 62 63 62 51 49 58 57 57 57 49 45 57 50 58 56 56 56 53 49 al im _3 3_ 25 .1 49 68 68 69 69 70 56 55 56 67 69 78 79 79 79 71 65 63 68 72 70 70 68 69 71 al im _3 3_ 25 .2 47 66 66 67 67 68 54 53 54 65 67 76 77 77 77 73 67 63 66 70 68 68 66 67 69 al im _3 3_ 26 .1 47 66 66 67 67 68 54 53 54 65 67 76 77 77 77 73 67 63 66 70 68 68 66 67 69 al im _3 3_ 26 .2 46 67 67 68 68 69 53 52 53 64 66 75 76 76 76 72 66 62 65 69 67 67 65 66 68 al im _3 3_ 26 .3 49 66 66 67 67 68 52 51 52 63 65 74 75 75 75 71 65 61 64 68 66 66 64 65 67 al im _3 3_ 27 .1 47 66 66 67 67 68 54 53 54 65 67 76 77 77 77 73 67 63 66 70 68 68 66 67 69 al im _3 3_ 27 .2 47 68 68 69 69 70 54 53 54 65 67 76 77 77 77 73 67 63 66 70 68 68 66 67 69 al im _3 3_ 28 .1 44 69 69 70 70 71 55 54 55 66 68 75 76 76 76 72 68 64 67 71 69 69 67 68 68 al im _3 3_ 29 .1 49 66 66 67 67 68 52 51 52 63 65 74 77 77 77 73 65 63 66 72 70 70 66 67 69 al im _3 3_ 30 .1 48 65 65 66 66 67 51 50 51 62 64 73 76 76 76 72 64 62 65 71 69 69 65 66 68 al im _3 3_ 30 .2 46 67 67 68 68 69 53 52 53 64 66 75 76 76 76 72 66 62 65 69 67 67 65 66 68 al im _3 3_ 31 .1 47 68 68 67 69 70 54 53 54 65 67 74 75 75 75 71 67 63 68 70 68 68 64 65 71 al im _3 3_ 32 .1 47 66 66 67 67 68 52 51 52 63 65 74 75 75 75 71 65 61 64 68 66 66 64 65 67 al im _3 3_ 33 .1 46 67 67 68 68 69 53 52 53 64 66 75 76 76 76 72 66 62 65 69 67 67 65 66 68 p ag e 3 of 1 3 al im _3 3_ 34 .1 43 64 64 67 65 68 50 49 50 59 63 70 73 73 73 73 65 59 62 66 64 64 60 61 65 al im _3 4_ 35 .1 62 83 83 84 84 85 73 72 71 76 70 79 80 80 80 86 80 76 71 81 79 79 79 72 72 al im _3 5_ 36 .1 46 75 75 78 76 81 71 70 69 72 70 75 74 74 74 74 70 74 61 65 63 63 65 66 66 al im _3 6_ 37 .1 46 79 79 78 80 79 63 62 61 68 60 63 62 62 62 68 68 68 63 65 63 63 61 60 64 al im _3 6_ 37 .2 45 82 82 81 83 82 66 65 64 71 63 66 63 63 63 69 71 69 64 64 62 62 62 61 65 al im _3 7_ 38 .1 48 81 81 80 82 81 69 68 67 70 64 65 62 62 62 64 68 72 69 65 63 63 63 60 62 al im _3 7_ 38 .2 53 84 84 83 85 82 70 69 66 69 67 68 65 65 65 67 69 71 72 66 64 64 66 59 57 an z_ 55 _3 9. 1 62 53 53 54 54 49 43 44 47 44 50 57 56 56 56 52 48 36 47 55 53 53 53 58 58 an z_ 55 _3 9. 2 61 52 52 53 53 48 44 45 50 47 51 58 55 55 55 53 51 35 46 54 52 52 54 61 59 an z_ 56 _4 0. 1 63 58 58 57 59 58 50 49 50 59 57 58 59 59 59 57 53 57 48 54 52 52 48 47 45 an z_ 65 _4 1. 1 52 51 49 50 50 49 41 40 43 38 40 43 42 42 42 44 40 42 39 43 41 41 37 42 36 n ew z ea la nd jo ur na l o f f or es tr y sc ie nc e anz_19_17.1 anz_19_17.2 anz_20_18.1 anz_20_18.2 anz_20_18.3 anz_20_18.4 anz_20_19.1 anz_20_19.2 anz_20_20.1 anz_21_21.1 anz_22_22.2 anz_23_23.1 anz_26_24.1 anz_26_24.2 alim_33_25.1 alim_33_25.2 alim_33_26.1 alim_33_26.2 alim_33_26.3 alim_33_27.1 alim_33_27.2 a lim _3 2_ 1. 1 64 62 63 65 64 64 63 63 65 45 40 52 66 71 49 47 47 46 49 47 47 a nz _1 2_ 2. 1 49 47 46 48 47 47 48 48 50 48 35 23 53 58 68 66 66 67 66 66 68 a nz _1 2_ 2. 2 47 45 46 48 47 47 48 48 50 46 33 21 53 58 68 66 66 67 66 66 68 a nz _1 2_ 3. 1 50 48 47 49 48 48 49 49 51 47 36 22 52 57 69 67 67 68 67 67 69 a nz _1 2_ 3. 2 48 46 47 49 48 48 49 49 51 47 34 22 54 59 69 67 67 68 67 67 69 a nz _1 2_ 4. 1 53 51 52 54 53 53 54 54 54 50 41 27 57 62 70 68 68 69 68 68 70 a nz _1 3_ 5. 1 43 41 52 54 53 53 54 54 54 38 33 29 57 62 56 54 54 53 52 54 54 a nz _1 3_ 5. 2 44 42 53 55 54 54 55 55 55 37 32 30 58 63 55 53 53 52 51 53 53 a nz _1 3_ 6. 1 45 47 58 60 59 59 58 58 54 36 33 31 57 62 56 54 54 53 52 54 54 a nz _1 4_ 7. 1 48 50 51 53 52 52 51 51 47 39 32 32 46 51 67 65 65 64 63 65 65 a nz _1 4_ 7. 2 50 32 33 35 34 34 35 35 61 53 46 44 48 49 69 67 67 66 65 67 67 a nz _1 5_ 8. 1 53 35 38 40 37 37 36 36 62 64 57 55 57 58 78 76 76 75 74 76 76 a nz _1 5_ 8. 2 54 36 37 39 38 38 39 39 65 59 54 56 56 57 79 77 77 76 75 77 77 a nz _1 5_ 8. 3 54 36 37 39 38 38 39 39 65 59 54 56 56 57 79 77 77 76 75 77 77 a nz _1 5_ 9. 1 54 36 37 39 38 38 39 39 65 59 54 56 56 57 79 77 77 76 75 77 77 a nz _1 5_ 9. 2 56 52 53 55 54 54 55 55 57 43 36 36 54 49 71 73 73 72 71 73 73 a nz _1 6_ 10 .1 50 46 53 55 54 54 55 55 57 43 28 24 50 45 65 67 67 66 65 67 67 a nz _1 7_ 11 .1 54 54 53 55 54 54 55 55 51 47 36 30 54 57 63 63 63 62 61 63 63 a nz _1 7_ 11 .2 53 37 38 40 39 39 40 40 64 56 47 39 47 50 68 66 66 65 64 66 66 a nz _1 8_ 12 .1 53 39 36 38 37 37 38 38 60 52 45 55 55 58 72 70 70 69 68 70 70 a nz _1 8_ 12 .2 51 37 34 36 35 35 36 36 58 50 43 53 53 56 70 68 68 67 66 68 68 a nz _1 8_ 13 .1 51 37 34 36 35 35 36 36 58 50 43 53 53 56 70 68 68 67 66 68 68 a nz _1 8_ 14 .1 49 35 34 36 35 35 36 36 58 52 45 49 53 56 68 66 66 65 64 66 66 a nz _1 8_ 15 .1 22 50 49 47 50 50 51 51 55 49 50 54 38 53 69 67 67 66 65 67 67 a nz _1 9_ 16 .2 2 30 53 53 54 54 55 55 59 49 42 50 34 49 71 69 69 68 67 69 69 p ag e 2 of 1 3 a nz _1 9_ 17 .1 0 28 51 51 52 52 53 53 57 51 44 48 32 47 71 69 69 68 67 69 69 a nz _1 9_ 17 .2 28 0 23 25 24 24 25 25 63 57 42 46 48 53 75 73 73 72 71 73 73 a nz _2 0_ 18 .1 51 23 0 2 1 1 2 2 40 64 53 53 43 48 80 78 78 77 76 78 78 a nz _2 0_ 18 .2 51 25 2 0 3 3 4 4 42 66 55 55 43 48 82 80 80 79 78 80 80 a nz _2 0_ 18 .3 52 24 1 3 0 0 1 1 39 65 54 54 44 49 81 79 79 78 77 79 79 a nz _2 0_ 18 .4 52 24 1 3 0 0 1 1 39 65 54 54 44 49 81 79 79 78 77 79 79 a nz _2 0_ 19 .1 53 25 2 4 1 1 0 0 38 66 55 55 45 50 80 78 78 77 76 78 78 a nz _2 0_ 19 .2 53 25 2 4 1 1 0 0 38 66 55 55 45 50 80 78 78 77 76 78 78 a nz _2 0_ 20 .1 57 63 40 42 39 39 38 38 0 52 55 55 49 58 70 68 68 67 66 68 68 a nz _2 1_ 21 .1 51 57 64 66 65 65 66 66 52 0 37 39 55 58 54 52 52 51 50 52 52 a nz _2 2_ 22 .2 44 42 53 55 54 54 55 55 55 37 0 22 56 53 53 55 55 54 53 55 55 a nz _2 3_ 23 .1 48 46 53 55 54 54 55 55 55 39 22 0 54 51 61 63 63 62 61 63 63 a nz _2 6_ 24 .1 32 48 43 43 44 44 45 45 49 55 56 54 0 23 77 75 75 74 73 75 75 a nz _2 6_ 24 .2 47 53 48 48 49 49 50 50 58 58 53 51 23 0 72 74 74 73 72 74 74 a lim _3 3_ 25 .1 71 75 80 82 81 81 80 80 70 54 53 61 77 72 0 2 2 3 6 2 4 a lim _3 3_ 25 .2 69 73 78 80 79 79 78 78 68 52 55 63 75 74 2 0 0 1 4 0 2 a lim _3 3_ 26 .1 69 73 78 80 79 79 78 78 68 52 55 63 75 74 2 0 0 1 4 0 2 a lim _3 3_ 26 .2 68 72 77 79 78 78 77 77 67 51 54 62 74 73 3 1 1 0 3 1 1 a lim _3 3_ 26 .3 67 71 76 78 77 77 76 76 66 50 53 61 73 72 6 4 4 3 0 4 4 a lim _3 3_ 27 .1 69 73 78 80 79 79 78 78 68 52 55 63 75 74 2 0 0 1 4 0 2 a lim _3 3_ 27 .2 69 73 78 80 79 79 78 78 68 52 55 63 75 74 4 2 2 1 4 2 0 a lim _3 3_ 28 .1 68 72 77 79 78 78 77 77 67 53 56 64 74 73 7 5 5 4 7 5 3 a lim _3 3_ 29 .1 67 71 78 80 79 79 78 78 68 52 55 61 73 72 6 4 4 3 6 4 2 a lim _3 3_ 30 .1 66 70 77 79 78 78 77 77 67 51 54 60 72 71 5 3 3 2 5 3 3 a lim _3 3_ 30 .2 68 72 77 79 78 78 77 77 67 51 54 62 74 73 3 1 1 0 3 1 1 a lim _3 3_ 31 .1 69 73 76 78 77 77 76 76 66 50 57 63 71 70 8 6 6 5 8 6 4 a lim _3 3_ 32 .1 67 71 76 78 77 77 76 76 66 50 53 61 73 72 4 2 2 1 2 2 2 a lim _3 3_ 33 .1 68 72 77 79 78 78 77 77 67 51 54 62 74 73 3 1 1 0 3 1 1 a lim _3 3_ 34 .1 63 67 74 76 75 75 74 74 64 52 51 61 71 72 10 8 8 7 10 8 8 a lim _3 4_ 35 .1 72 76 79 79 80 80 79 79 91 77 74 82 66 59 57 55 55 54 55 55 55 p ag e 3 of 1 3 a lim _3 5_ 36 .1 68 68 71 71 72 72 71 71 83 65 60 68 64 55 57 55 55 54 57 55 55 a lim _3 6_ 37 .1 64 66 71 71 72 72 71 71 83 61 62 70 58 49 49 49 49 48 49 49 47 a lim _3 6_ 37 .2 67 69 72 72 73 73 72 72 84 62 63 73 61 52 48 48 48 47 48 48 46 a lim _3 7_ 38 .1 64 68 71 71 72 72 71 71 77 59 62 74 56 43 55 55 55 54 55 55 55 a lim _3 7_ 38 .2 59 69 76 76 77 77 76 76 74 62 61 73 55 48 58 58 58 57 58 58 58 a nz _5 5_ 39 .1 58 62 55 57 54 54 55 55 51 51 52 48 48 49 71 71 71 70 67 71 69 a nz _5 5_ 39 .2 61 63 56 58 55 55 56 56 52 52 51 49 51 52 72 72 72 71 68 72 70 a nz _5 6_ 40 .1 43 49 54 54 53 53 54 54 60 60 61 59 43 52 60 58 58 57 58 58 58 a nz _6 5_ 41 .1 34 38 43 45 44 44 45 45 45 45 40 48 32 41 71 69 69 68 67 69 69 p ag e 4 of 1 3 alim_33_28.1 alim_33_29.1 alim_33_30.1 alim_33_30.2 alim_33_31.1 alim_33_32.1 alim_33_33.1 alim_33_34.1 alim_34_35.1 alim_35_36.1 alim_36_37.1 alim_36_37.2 alim_37_38.1 alim_37_38.2 anz_55_39.1 anz_55_39.2 anz_56_40.1 anz_65_41.1 a lim _3 2_ 1. 1 44 49 48 46 47 47 46 43 62 46 46 45 48 53 62 61 63 52 a nz _1 2_ 2. 1 69 66 65 67 68 66 67 64 83 75 79 82 81 84 53 52 58 51 a nz _1 2_ 2. 2 69 66 65 67 68 66 67 64 83 75 79 82 81 84 53 52 58 49 a nz _1 2_ 3. 1 70 67 66 68 67 67 68 67 84 78 78 81 80 83 54 53 57 50 a nz _1 2_ 3. 2 70 67 66 68 69 67 68 65 84 76 80 83 82 85 54 53 59 50 a nz _1 2_ 4. 1 71 68 67 69 70 68 69 68 85 81 79 82 81 82 49 48 58 49 a nz _1 3_ 5. 1 55 52 51 53 54 52 53 50 73 71 63 66 69 70 43 44 50 41 a nz _1 3_ 5. 2 54 51 50 52 53 51 52 49 72 70 62 65 68 69 44 45 49 40 a nz _1 3_ 6. 1 55 52 51 53 54 52 53 50 71 69 61 64 67 66 47 50 50 43 a nz _1 4_ 7. 1 66 63 62 64 65 63 64 59 76 72 68 71 70 69 44 47 59 38 a nz _1 4_ 7. 2 68 65 64 66 67 65 66 63 70 70 60 63 64 67 50 51 57 40 a nz _1 5_ 8. 1 75 74 73 75 74 74 75 70 79 75 63 66 65 68 57 58 58 43 a nz _1 5_ 8. 2 76 77 76 76 75 75 76 73 80 74 62 63 62 65 56 55 59 42 a nz _1 5_ 8. 3 76 77 76 76 75 75 76 73 80 74 62 63 62 65 56 55 59 42 a nz _1 5_ 9. 1 76 77 76 76 75 75 76 73 80 74 62 63 62 65 56 55 59 42 a nz _1 5_ 9. 2 72 73 72 72 71 71 72 73 86 74 68 69 64 67 52 53 57 44 a nz _1 6_ 10 .1 68 65 64 66 67 65 66 65 80 70 68 71 68 69 48 51 53 40 a nz _1 7_ 11 .1 64 63 62 62 63 61 62 59 76 74 68 69 72 71 36 35 57 42 a nz _1 7_ 11 .2 67 66 65 65 68 64 65 62 71 61 63 64 69 72 47 46 48 39 a nz _1 8_ 12 .1 71 72 71 69 70 68 69 66 81 65 65 64 65 66 55 54 54 43 a nz _1 8_ 12 .2 69 70 69 67 68 66 67 64 79 63 63 62 63 64 53 52 52 41 a nz _1 8_ 13 .1 69 70 69 67 68 66 67 64 79 63 63 62 63 64 53 52 52 41 a nz _1 8_ 14 .1 67 66 65 65 64 64 65 60 79 65 61 62 63 66 53 54 48 37 a nz _1 8_ 15 .1 68 67 66 66 65 65 66 61 72 66 60 61 60 59 58 61 47 42 a nz _1 9_ 16 .2 68 69 68 68 71 67 68 65 72 66 64 65 62 57 58 59 45 36 p ag e 5 of 1 3 a nz _1 9_ 17 .1 68 67 66 68 69 67 68 63 72 68 64 67 64 59 58 61 43 34 a nz _1 9_ 17 .2 72 71 70 72 73 71 72 67 76 68 66 69 68 69 62 63 49 38 a nz _2 0_ 18 .1 77 78 77 77 76 76 77 74 79 71 71 72 71 76 55 56 54 43 a nz _2 0_ 18 .2 79 80 79 79 78 78 79 76 79 71 71 72 71 76 57 58 54 45 a nz _2 0_ 18 .3 78 79 78 78 77 77 78 75 80 72 72 73 72 77 54 55 53 44 a nz _2 0_ 18 .4 78 79 78 78 77 77 78 75 80 72 72 73 72 77 54 55 53 44 a nz _2 0_ 19 .1 77 78 77 77 76 76 77 74 79 71 71 72 71 76 55 56 54 45 a nz _2 0_ 19 .2 77 78 77 77 76 76 77 74 79 71 71 72 71 76 55 56 54 45 a nz _2 0_ 20 .1 67 68 67 67 66 66 67 64 91 83 83 84 77 74 51 52 60 45 a nz _2 1_ 21 .1 53 52 51 51 50 50 51 52 77 65 61 62 59 62 51 52 60 45 a nz _2 2_ 22 .2 56 55 54 54 57 53 54 51 74 60 62 63 62 61 52 51 61 40 a nz _2 3_ 23 .1 64 61 60 62 63 61 62 61 82 68 70 73 74 73 48 49 59 48 a nz _2 6_ 24 .1 74 73 72 74 71 73 74 71 66 64 58 61 56 55 48 51 43 32 a nz _2 6_ 24 .2 73 72 71 73 70 72 73 72 59 55 49 52 43 48 49 52 52 41 a lim _3 3_ 25 .1 7 6 5 3 8 4 3 10 57 57 49 48 55 58 71 72 60 71 a lim _3 3_ 25 .2 5 4 3 1 6 2 1 8 55 55 49 48 55 58 71 72 58 69 a lim _3 3_ 26 .1 5 4 3 1 6 2 1 8 55 55 49 48 55 58 71 72 58 69 a lim _3 3_ 26 .2 4 3 2 0 5 1 0 7 54 54 48 47 54 57 70 71 57 68 a lim _3 3_ 26 .3 7 6 5 3 8 2 3 10 55 57 49 48 55 58 67 68 58 67 a lim _3 3_ 27 .1 5 4 3 1 6 2 1 8 55 55 49 48 55 58 71 72 58 69 a lim _3 3_ 27 .2 3 2 3 1 4 2 1 8 55 55 47 46 55 58 69 70 58 69 a lim _3 3_ 28 .1 0 5 6 4 7 5 4 11 56 58 46 45 54 57 70 71 59 68 a lim _3 3_ 29 .1 5 0 1 3 2 4 3 6 53 57 45 48 57 58 69 70 58 67 a lim _3 3_ 30 .1 6 1 0 2 3 3 2 5 52 56 46 49 56 57 70 71 57 66 a lim _3 3_ 30 .2 4 3 2 0 5 1 0 7 54 54 48 47 54 57 70 71 57 68 a lim _3 3_ 31 .1 7 2 3 5 0 6 5 8 55 59 43 46 55 56 67 68 56 65 a lim _3 3_ 32 .1 5 4 3 1 6 0 1 8 55 55 47 46 55 58 69 70 58 67 a lim _3 3_ 33 .1 4 3 2 0 5 1 0 7 54 54 48 47 54 57 70 71 57 68 a lim _3 3_ 34 .1 11 6 5 7 8 8 7 0 53 57 43 46 55 56 67 68 58 63 a lim _3 4_ 35 .1 56 53 52 54 55 55 54 53 0 42 40 43 50 51 76 73 69 70 p ag e 6 of 1 3 a lim _3 5_ 36 .1 58 57 56 54 59 55 54 57 42 0 42 39 38 45 76 71 63 62 a lim _3 6_ 37 .1 46 45 46 48 43 47 48 43 40 42 0 3 22 31 62 61 61 52 a lim _3 6_ 37 .2 45 48 49 47 46 46 47 46 43 39 3 0 19 30 63 60 62 55 a lim _3 7_ 38 .1 54 57 56 54 55 55 54 55 50 38 22 19 0 13 64 63 57 54 a lim _3 7_ 38 .2 57 58 57 57 56 58 57 56 51 45 31 30 13 0 67 66 62 55 a nz _5 5_ 39 .1 70 69 70 70 67 69 70 67 76 76 62 63 64 67 0 5 39 42 a nz _5 5_ 39 .2 71 70 71 71 68 70 71 68 73 71 61 60 63 66 5 0 44 43 a nz _5 6_ 40 .1 59 58 57 57 56 58 57 58 69 63 61 62 57 62 39 44 0 39 a nz _6 5_ 41 .1 68 67 66 68 65 67 68 63 70 62 52 55 54 55 42 43 39 0 new zealand journal of forestry science table s4. selection of primers for phylogenetic analysis. primer no. of subcultures that differ 3-2 1 as4 1 0.3-1 1 as15 3 l15 5 l45 5 l21 5.51 foki 6 l15/as19 8.51 aa2m2 15 1 decimals because occasionally halving triplets as well as pairs or quadruplet of subcultures. stress, psychosocial factors and the new zealand forest industry workforce: seeing past the risk of harm to the potential for individual and organisational wellbeing trevor best*, rien visser and david conradson university of canterbury, private bag 4800, christchurch 8140, new zealand *corresponding author: trevor.best@pg.canterbury.ac.nz (received for publication 21 january 2020; accepted in revised form 7 april 2021) abstract background: there is clear evidence that stress is having an impact on the health and wellbeing of the forest industry workforce in aotearoa new zealand. while this has legal ramifications under the national health and safety legislation, international research also shows that harm to mental health invariably leads to reductions in work force productivity and business profitability. the reverse is also true: improved mental wellbeing can lead to greater worker engagement and commitment, which in turn increases productivity and profitability. although these relationships are well substantiated, managers and leaders in the forest industry may not be aware of either the existence of a workplace stress problem or of its impact. methods: a critical review is undertaken of stress and psychosocial hazards research within the international forest industry or similar industries (e.g. construction), with particular attention given to the explanation of psychosocial hazards. results: international research on the forest industry largely confirms what we know about harmful aspects of job content and workplace conditions. however, it is argued that the focus within this research on job content and immediate workplace conditions obscures the impact of the wider social context. this limits the potential of management to move beyond seeing psychosocial factors simply as risks to be minimised at the workplace level. bringing an ecological perspective to the analysis of forestry workplaces makes it easier to identify the elements of forest management practice that may contribute to stress within the workforce. it also becomes easier to identify the interactions between family, community and workplaces that may either exacerbate or reduce workforce stress. conclusions: this paper highlights particular opportunities for reducing stress and enhancing wellbeing within the new zealand forest industry workforce. it suggests that the psychosocial conditions that contribute to mental ill-health can be reconfigured to promote mental health, with wellbeing benefits that extend beyond the workplace. psychosocial demands on a person can be motivating as long as the person has the resources to meet the challenge. successful stewardship of the psychosocial environment at the forest management level is thus an opportunity to increase value to both investors and other stakeholders. new zealand journal of forestry science best et al. new zealand journal of forestry science (2021) 51:5 https://doi.org/10.33494/nzjfs512021x93x e-issn: 1179-5395 published on-line: 21/06/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access environments in which our lives unfold (woodward 2015). work therefore has the potential to enhance a person’s sense of wellbeing (modini et al. 2016). unlike physical hazards, where wellbeing can be enhanced by the elimination of the hazard, psychosocial factors relating to the nature of work and the demands it places on the worker exist along a continuum from causing harm to promoting wellbeing (bentley et al. 2019; leka introduction at the core of the constitution of the world health organisation is the notion that health is a human right that goes beyond the absence of harm to include physical, mental and social wellbeing (world health organisation 2019). in the psychosocial domain, health can be understood as emerging from the relations between our physical and mental capabilities and the social keywords: work-related stress, psychosocial hazards, forest industry workforce, ecological perspective http://creativecommons.org/licenses/by/4.0/), best et al. new zealand journal of forestry science (2021) 51:5 page 2 et al. 2015). yet rather than managing those psychosocial factors to promote worker wellbeing (leka et al. 2015), the occupational health and safety frameworks in most developed countries seek to prevent harm by eliminating or minimising the risks to worker health represented by hazardous psychosocial conditions (chirico et al. 2019). recent changes in new zealand’s health and safety legislation are a good example of the limitations of this approach. the health and safety at work act 2015 contains a clear expectation that the work-related risks to a person’s mental health should be managed by the people in charge of that work or workplace (health and safety at work act 2015). in the interpretations (section 16) of the act, the definition of a hazard includes behaviour that has the potential to harm, “whether or not that behaviour results from physical or mental fatigue, drugs, alcohol, traumatic shock, or another temporary condition that affects a person’s behaviour”. the understanding of “health” in the act includes both mental and physical health. however, managing workplace factors that impact psychosocial health through a framework of psychosocial hazards may obscure the opportunity to enhance both individual and organisational health represented by those factors. health should be addressed as something more than harm elimination or reduction (leka et al. 2015). designing workplaces and work processes in ways that go beyond harm elimination and reduction can improve a worker’s quality of life and enhance productivity and sustainability. making such interventions can be a challenge when the workforce is largely employed through service contracts, however, as is the case with the new zealand forest industry. while the direct terms and conditions of the employment relationship are set by the contractor/employer, the scope of those conditions is largely controlled by the agreement between the contractor and the forest owner. the organisation of the work and the workplace is therefore not totally within the control of the employer. however, the health and safety at work act 2015 places the responsibility for the primary duty of care onto a “person conducting a business or undertaking” (pcbu). this means that the obligation for managing the risk of a negative health outcome arising from mental distress sits with whoever creates that risk, regardless of where in the process of work that risk arises (and irrespective of the nature of the employment relationship between the pcbu and the worker who suffers harm). this obligation to manage risk is more than just a legal and economic matter. the principles behind the international labour organisation’s health and safety standards are not just that work should take place in a safe and healthy working environment, but also that conditions of work should be consistent with workers’ wellbeing and human dignity. work should offer real possibilities for personal achievement, self-fulfilment and service to society (forastieri 2016). although the legal challenge for the forest industry is to design workplaces and processes that reduce mental harm across business boundaries, this expectation also presents a moral and economic opportunity. by focusing on wellbeing rather than harm reduction, the industry could positively impact both workers' quality of life and reap the potentially significant financial benefits of a more loyal, engaged, and productive workforce. what makes this challenging for forest owners and managers is that the feedback loops that could bring the impacts of mental distress to their attention are poorly developed. while noting signs of stress in the workforce, lovelock and houghton (2017) and nielsen (2015b) both concluded that more work was needed to increase awareness of the full range of risks faced by forestry workers and the health impacts of those risks. furthermore, the arm’s length nature of their service agreements means that forest owners and managers do not have the direct relationship with worker health and safety that would enable awareness of stress and its impacts. despite widespread annual physical health checks of workers by contractors, there is no consistent and centralised assessment process in operation within the industry (forest industry safety council 2018). the generally available investigation methods used to generate a learning feedback loop after an accident are unable to take psychosocial factors and any associated stress into account (van wassenhove & garbolino, 2008, as cited by leka et al. 2015). those charged with managing health and safety within the new zealand forest industry could well be operating somewhat unaware of the potential impacts of stress in its various expressions on workers’ wellbeing. against this backdrop, this paper examines the research on stress within the international forest industry workforce so as to identify opportunities for enhancing wellbeing amongst forest industry workers in new zealand. it begins by reviewing how work-related stress and its risk factors are generally explained. it then considers what the extant research on work-related stress from the world’s forest industries suggests about health impacts and psychosocial hazards within forestry. it finishes by questioning whether a focus on workplace psychosocial hazards is the most appropriate framework to address stress and wellbeing within the forest industry workforce in new zealand. an alternative approach is presented, based on ecological systems theory, and some of the implications for potential interventions are noted. explaining work-related stress work-related stress has generated a large body of academic research that focuses primarily on how a person fits or does not fit into his or her work environment (väänänen et al. 2014). in this framework, work-related stress is seen as psychological strain or a set of negative psychophysiological responses and reactions (chirico et al. 2019) that occur either when the demands of the work environment exceed the capabilities and resources of the worker or when the needs of the worker cannot be supplied by the work environment (dewe & cooper 2017; forastieri 2016). stress is thought to occur when that mismatch becomes chronic or unmanageable (leka et al. 2015). much research has sought to clarify the relationship between the work environment and the individual’s body and mind through investigations of the impact of work place characteristics on particular unhelpful behaviours and psychological and somatic symptoms (väänänen et al. 2014). measuring both the psychosocial hazards and the symptoms has required the development of self-rating scales (väänänen et al. 2014). this work has shown that chronic and unresolvable exposure to a number of work place characteristics can increase the likelihood that a proportion of the workforce will suffer a negative psychophysiological response as a result (leka & jain 2010; maslach & leiter 2016). both the new zealand workplace barometer (bentley et al. 2019) and the world health organisation review of psychosocial hazards at work (leka & jain 2010) use the following definition of psychosocial hazards: those aspects of work design and the organisation and management of work, and their social and environmental contexts, which have the potential for causing psychosocial or physical harm. (cox et al. 2003, p. 195) the world health organisation (2008) developed a summary of work-related psychosocial hazards (see table 1) for the european framework for psychosocial risk management. the framework identifies ten psychosocial domains, each of which can be thought of as a potential source of work-related stress (forastieri 2016). the domains are divided into two groups: work content (which includes psychosocial hazards related to the conditions, organisation and component tasks of the job), and work context (which includes psychosocial hazards related to workplace organisation) (cox & griffiths, 2005, as cited by forastieri 2016). the new zealand workplace barometer is closely based on this eu framework, in that it incorporates all ten domains and adds workplace bullying to the domain of interpersonal relationships at work (bentley et al. 2019). given that the new zealand workplace barometer lists forestry as one of the industries with the highest reported levels of bullying – with greater than 10% of respondents reported having been bullied – the addition of bullying is highly relevant to the industry. this is particularly the case as the definition of bullying used by the survey required the harassment to occur over a period of time and to involve one or more perpetrators (bentley et al. 2019). the world health organisation (2008) point out that while bullying can be considered a psychosocial risk, it should also be regarded as a consequence of a poor psychosocial work environment. the implication of this perspective is that if an organisation mitigates the risks listed in table 1, then the risk of bullying will also reduce. the findings of other research overlap considerably with this framework, albeit with different emphases. in their review of two decades of research on burnout and its causes and outcomes, maslach and leiter (2016) pointed to six key domains of psychosocial hazards: workload, control, reward, community, fairness and values. of these domains, the conceptualisation and best et al. new zealand journal of forestry science (2021) 51:5 page 3 table 1: description of the study sites psychosocial hazard explanation work content job content lack of variety or short work cycles, fragmented or meaningless work, underuse of skills, high uncertainty, continuous exposure to people through work. workload and work pace work overload or underload, machine pacing, high levels of time pressure, continually subject to deadlines. work schedule shift working, night shifts, inflexible work schedules, unpredictable hours, long or unsociable hours. environment and equipment inadequate equipment availability, suitability or maintenance, poor environmental conditions such as lack of space, poor lighting, excessive noise. control low participation in decision making, lack of control over workload, pacing, shift working. work context organisational culture and function poor communication, low levels of support for problem solving and personal development, lack of definition of, or agreement on, organisational objectives. interpersonal relationships at work social or physical isolation, poor relationships with superiors or co-workers, interpersonal conflict, lack of social support. role in organisation role ambiguity, role conflict, and responsibility for people. career development career stagnation and uncertainty, under promotion or over promotion, poor pay, job insecurity, low social value to work. home-work interface conflicting demands of work and home, low support at home, dual career problems. source: (cox et. al. 2000, as cited by forastieri 2016) table 1: work-related psychosocial hazards explanatory significance of workload, control and community are similar to the european framework. in contrast, maslach and leiter (2016) place much greater emphasis on the role of rewards (financial, institutional or social), fairness (the extent to which decisions are perceived as being fair and equitable) and values (the alignment between the individual’s values and those of the organisation they work for) in the development of burnout. nevertheless, these differences may be quite important to the health and wellbeing of the various actors within the new zealand forest industry. issues such as whose interests are represented in the service contracts that form the basis for employment of the workforce and how those contracts distribute risk and reward will shape the perception of whether those agreements are seen as fair and equitable. furthermore, with a workforce that is approximately 37% māori (ministry for primary industries 2020) there is a significant potential for differences in world views between a substantial part of the workforce and the forest industry, creating mis-alignment in values (b. hooper, personal communication, 13 august 2020). similarly, in a review of the epidemiological literature of work-related stress, pfeffer (2018) points to ten workplace exposures that affect human health through stress. as with the psychosocial hazards associated with burnout, this perspective is largely the same as that used in the european framework. however, there are key differences that are important in the context of the new zealand forest industry. job insecurity, whether for one’s own job or that of colleagues, is much more prominent in pfeffer’s framework. this is something that could be considered important in an industry which employs most of its workforce on a contractual basis. workers paid by piece rates or hourly rates are exposed to the risk of reduced hours or job loss resulting from contractual transgressions or downturns in the log market. job insecurity is also highlighted as a key work-related psychosocial stressor by other authors (e.g. dewe & cooper 2017). furthermore, pfeffer included access to health care as a significant stressor, reflecting the “us-centric” nature of the epidemiological literature. however, any industry reliant on a rurally located workforce in new zealand should be cognisant of reduced access to health care for those who live outside the urban centres, a pattern which reflects health service restructuring between 1980 – 2001 and the consequential differences in all-cause mortality rates between urban and rural regions (pearce et al. 2008). the differences between these frameworks highlight the contextual nature of psychosocial hazards and the need for psychosocial risks within the new zealand forest industry to be researched more thoroughly than is currently the case. however, there has been research undertaken within the forest industries of other countries that is relevant in the new zealand context. this research, which highlights potential psychosocial hazards in the new zealand forest industry and their impacts on health and wellbeing, is discussed in the next section. best et al. new zealand journal of forestry science (2021) 51:5 page 4 evidence of health impacts of work-related stress in the forest industry within the international forest industry, the study of known mental health conditions and their association with wellbeing and safety is centred on an 18 year prospective cohort study of workers at a finnish based multinational forest industry company (väänänen et al. 2008). this study assessed health and potential risk factors within the workforce, which included manual labourers and machine operators. research based on data from this study has highlighted the association of burnout with negative health and safety outcomes. burnout was assessed using the maslach burnout inventory (mbi, maslach, jackson & leiter, 1996, as cited in maslach & leiter 2016). this consists of three dimensions: overwhelming physical and emotional exhaustion arising from depleted emotional and physical resources with insufficient recovery (maslach & leiter 2017); feelings of cynicism that reflect a detached attitude towards work and increasing disregard towards one’s co-workers and clients (toppinen-tanner et al. 2002); and a reduced sense of accomplishment and effectiveness (seidler et al. 2014). assessments occurred at various times throughout the study period and could be correlated with a number of different health outcomes recorded by finland’s national population register centre and the company itself (väänänen et al. 2008). the health outcomes explored over the life of this research program are significant to the new zealand forest industry for a number of reasons: firstly, they involve a large number of participants (ranging from 3895 to 10062 employees) that are mostly men (greater than 76%) involved in manual work or machine operation (greater than 62%); and, secondly, burnout is correlated with clinically derived indicators of health (väänänen et al. 2008). these are considered more reliable than self-report measures (väänänen et al. 2014). the research facilitated by this program all points to burnout being associated with negative health outcomes. an increase in the mbi summary score of one unit was associated with a 35% increase in the risk of mortality among workers less than 45 years old (ahola et al. 2010). of the subscales, only exhaustion produced a statistically significant hazard ratio when adjusted for sociodemographic and baseline health factors. a similar study of the relationship between burnout and severe injuries by the same research group found a one unit increase in the burnout summary score to be related to a 10% increase in the risk of injury requiring hospitalisation or causing death (ahola et al. 2013). of the mbi subscales, emotional exhaustion was associated with a 9% increase in the risk of injury, while cynicism was related to a 10% increase. this suggests that having both energy and motivation to act safely is important to prevent workplace injury or death. toppinen-tanner et al. (2005) reported on burnout as an event prior to sickness absence for different medically certified causes of absence. they found that the mbi summary score was positively correlated with the risk of future medically certified absence (after adjustment for age, gender, occupation, and baseline absence). the increased risk of future illness was shown to include mental and behavioural disorders and diseases of the cardiovascular and musculoskeletal systems. burnout predicted future hospital admissions for mental health and cardiovascular disorders among participants who had not suffered the disorder prior to the start of the study (toppinen-tanner et al. 2009). although none of these studies defined a causal pathway between burnout and negative health outcomes, they do suggest that work-related stress conditions are associated with increased risk of injury, illness and early mortality within a male dominated, manual and machine operator workforce. such research is relevant to the new zealand forest industry. of concern, therefore, is that there are already indications that mental distress is having an impact on new zealand forest industry workers. the new zealand forest industry is part of an occupational group (forestry and farming) that comprises 6.8% of male suicide victims in new zealand (suicide mortality review committee 2016). if that percentage still holds, increases in male suicide levels in new zealand (coronial services 2020) suggest that deaths by suicide could have exceeded accidental workplace deaths for the farming and forestry occupational group in both 2018 and 2019. worksafe new zealand’s national health and safety attitudes and behaviours survey (nhabs) also noted that “stressrelated or mental illness was more likely to be identified as a long-term health problem by workers who had personally experienced a serious harm incident (22% compared with 12% of those who had not experienced an incident) or a near miss incident (19% compared with 11%)” (nielsen 2015a, p. 68). the same survey found that 27% of employees and 36% of employers experienced a serious harm near miss or actual incident in the preceding 12 months (nielsen 2015a). this is in line with international evidence that highlights (i) the interaction between exposure to actual and potential trauma and mental health disorders (tehrani 2004) and, more specifically, (ii) the relationship between exposure to risks and hazards and mental distress (nahrgang et al. 2011). furthermore, the lovelock and houghton (2017) review of the industry highlighted the prevalence of health conditions among workers such as hypertension and diabetes, poor lung function due to high levels of smoking, and high levels of substance abuse. all of these conditions have some association with stress as lifestyle responses to mental distress (forastieri 2016; leka & jain 2010; solar & irwin 2010). the 2014 nhabs (nielsen 2015a) also lists fatigue, ill health, stress and addictions as barriers to improvement in health and safety outcomes and notes that emotional and physical stress is of high concern to those working in the industry. mental distress and strain are also known to have significant negative impacts on business profitability and sustainability (leka et al. 2015; pfeffer 2018; world economic forum 2008). presenteeism (presenting for work while sick or injured) has been shown to reduce worker productivity, with a cost impact four times greater than that of directly treating the condition (edington & burton 2003). the latest nhabs reported that 53% of forest workers surveyed had worked while sick or injured and 46% had worked while overtired (nielsen 2018). similarly, a reduction in psychological health has also been associated with the sort of risky and dangerous behaviour that can lead to both accidents and quality loss arising from adverse events (du plessis et al. 2013; forastieri 2016; leka et al. 2015). the same study that identified exposure to risks and hazards as a risk factor for mental distress also found an association between mental distress and risky and dangerous behaviour (nahrgang et al. 2011). finally, workers exposed to hazardous psychosocial environments are less likely to engage in re-training or further learning (leka et al. 2015). this should be of concern to an industry looking to adapt to the physical safety risks through the introduction of mechanised harvesting systems (steepland harvesting programme 2018) and increase its workforce to take advantage of growth opportunities (harris 2017; moore 2017). health impacts of psychosocial workplace conditions within the international forest industry there are several studies that examine the relationship between psychosocial workplace conditions and workers’ health in forest and logging operations. although none employ the psychosocial hazard framework outlined in table 1, all consider factors that fit within that framework. elements such as psychological demand, intellectual discretion and exposure to risks and hazards have been associated with disorders of the neck / shoulders and lower back (hagen et al. 1998), mental strain (inoue 1996), and reduced job or life satisfaction (mylek & schirmer 2015). the international findings fit with the work of lilley et al. (2002) on fatigue, work / rest patterns and recent injury and near injury experience. this found that 78% of participants reported experiencing fatigue sometimes, often, or always, with 19% experiencing fatigue often or always. there was also a significant association between self-reported near misses in the previous 12 months and the reported level of fatigue experienced at work. getting eight hours sleep and taking breaks was associated with reduced fatigue, but the majority of participants reported having seven hours or less sleep per night (and almost 25% reported six hours or less). these are all psychosocial conditions associated with reduced mental wellbeing. despite the paucity of research, there is enough evidence to suggest forestry workplaces contain psychosocial hazards that are harmful to mental health and that these hazards fit within the european framework (see table 1). research in the new zealand forest industry (lilley et al. 2002; lovelock & houghton 2017; nielsen 2015b) suggests these conditions also apply to local forestry workplaces. however, some research within the forest industry highlights a key difference between physical hazards and the psychosocial domains listed as hazards in table 1. as noted above, the risk management objective for a physical hazard is to reduce the potential of that hazard negatively impacting the health condition of the worker (health and safety at work act 2015). the goal is for best et al. new zealand journal of forestry science (2021) 51:5 page 5 the worker to go home at the end of the day in the same health state as when they arrived. on the other hand, effective management of psychosocial risks creates the potential for the worker to go home with enhanced wellbeing (bentley et al. 2019; leka et al. 2015). in their study of the impact of job content (see table 1) on logging machine operator wellbeing, hanse and winkel (2008) found that daily task variety, job rotation and access to breaks when required were all positively associated with job satisfaction to a statistically significant degree. they also found a statistically significant positive association between job control and job rotation with reduced musculoskeletal symptoms, and between job rotation and access to breaks with reduced headaches and sleeping problems. overall, job rotation – defined as operating a shift system that broke up machine operating hours, altering tasks to reduce machine operating tasks, and restricting or controlling the number of machine operating hours – had a positive impact across all three measures of wellbeing in the study. similarly, in their survey of australian forestry managers and workers, mylek and schirmer (2015) found a number of work context elements (see table 1) were associated with improved wellbeing. participants who felt they had more control over their work, reported a better work-life balance and were more satisfied with their income also reported higher life satisfaction and general health. other psychosocial conditions that were significantly associated with higher life satisfaction included job security, a positive workplace culture (defined as confidence in being able to express views), a felt level of social support, higher work efficacy and a positive work-related social identity. interestingly, only job control, work-life balance, income, a positive culture, and work-related efficacy were positively associated with general health. what these results reflect is that psychosocial factors can be managed, not just to reduce work-related stress but to promote worker engagement, “a persistent, positive affective-motivational state of fulfilment that is characterised by the three components of vigour, dedication and absorption” (maslach & leiter 2016, p. 104) as a state of wellbeing. furthermore, nahrgang et al. (2011) found engagement was positively associated with reductions in risky and unsafe behaviour, adverse events and accidents and injuries. if psychosocial risk management is approached with engagement as the goal, psychosocial factors can switch from hazards to be eliminated to protective factors that can be pursued, not only to protect workers from harm but also to promote wellbeing. changing psychosocial conditions – from harmful to “well-ful” while the health and safety at work act 2015 might make it clear that any forest owner or manager must ensure, to a reasonably practicable extent, the mental and physical health of those working in the forest, the nature of their relationship with the workforce does not easily fit with the ‘employee’ focus of the psychosocial risk framework. a forest owner or manager could easily be forgiven for thinking that psychosocial risks exist only within the organisations and workplaces in which the workers are directly employed. that highlights a specific weakness of this approach to thinking about workrelated stress. the weakness is that the model assumes that all of the stress experience captured in the research originates within the worker’s immediate work context (theorell et al. 2015). this is of concern for the forest industry, as it ignores the broader social structures and systems (e.g. piece rate contracts) that may drive those risk factors in the immediate work context. it also overlooks the ways people exist within adjacent systems that may have highly permeable boundaries. with such a narrow view of context, the focus goes onto the individual and what can be done to enable individual coping (harkness et al. 2005). as a result, workplace wellbeing interventions typically seek either to modify micro-organisational factors (e.g. decision latitude and social support) that surround the individual (väänänen et al. 2014) or to enhance the individual’s ability to cope through counselling or stress management techniques (harkness et al. 2005). macro-organisation and wider social system issues are often not addressed and the opportunity to eliminate stress through removal of the stressors in the wider context is not considered (dewe et al. 2010). in considering how psychosocial factors could be managed to the benefit of both individuals and the organisations in which they work, it is important to recognise that workplaces sit within an ecological system where they exist in relationship with all other parts of that ecological system. ecological systems theories, such as that proposed by bronfenbrenner (1977), explain human behaviour by recognising that individuals always act within these larger social and ecological systems (figure 1). to understand behaviour it is also important to understand the nature of the institutions and social structures within each level of the system and the ways in which those levels interact and may reinforce each other (golden & earp 2012). stokols (1992) argues that the social, physical, and cultural aspects of this multilayered environment each have a cumulative effect on health. there are consequently multiple influences on specific health behaviours and outcomes, and multiple opportunities to intervene. achieving change will require interventions at a number of different points within the system (sallis et al. 2008). unfortunately, interventions for worker wellbeing within the new zealand forest industry, as guided by the legislation and the traditional conception of workplace mental health, are focused almost entirely on the specific settings in which people work. yet what the ecological perspective shows is that health is determined as much by what goes on in the mesosystem (where those settings interact, see figure 1) and by the social, political and cultural settings of the exoand macrosystems, as by what goes on within the specific work setting. the significance of this point for designing wellbeing interventions can be illustrated by considering the interactions between the various industry players. lilley et al. (2002) confirmed that the total workday length for forestry workers in new zealand was increasing, best et al. new zealand journal of forestry science (2021) 51:5 page 6 that there were substantial groups of workers whose break times were compromised, and that there had been a reduction in the number of workers getting two consecutive days off in every seven days in the preceding ten years. hide et al. (2010) study of cable logging operations noted the inconsistent break times, and that work pace and workload were often driven by the pace of the adjacent workstations. these are all factors directly controllable within the workplace (the microsystem, in bronfenbrenner’s framework, see figure 1). however, they also pointed to the impact of elements beyond the direct control of the contractor. the challenge of achieving daily piece rate targets, working on sites with limited operating and storage space, and bottlenecks in the downstream supply chain all directly impacted the working day length. these conditions arise from the mesosystem (interactions within the supply chain) and the exosystem (outsourcing operations using piece rate contracts). furthermore, long commutes were found to increase the length of the workday, suggesting that urbanisation, a macrosystem change, was adding to the problem. ecological systems theory also helps explain the impact of interactions between work, family, the community, and wider societal issues such as gender and socio-economic status. lovelock and houghton (2017) identified that the poor health and safety outcomes in the new zealand forest industry may originate with psychosocial stressors outside of the workplace. these included high drug use in worker families and communities, insecure and overcrowded accommodation, and conflict with unemployed family members. studies from outside the industry have also highlighted the potential of family conflict to reduce the cognitive resources available to an employee at work (du et al. 2017). while confirming partner conflict as a predictor of wellbeing (in this case, using burnout as the measure of wellbeing), rössler et al. (2015) also found an association with never having been married. this suggests that it is not only what goes on in families that impacts worker performance and wellbeing (kinnunen et al. 2006) but also the structure of the family itself. as ecological systems theory indicates, work can also impact wellbeing within settings outside work. what appear to be unhealthy lifestyle choices (e.g. smoking, drug and alcohol use, a carbohydrate dense diet associated with obesity, diabetes and hypertension) could be, in part, a coping response to stress arising from work or from the situations workers find themselves in as a result of the way their work is organised (forastieri 2016; leka & jain 2010). construction workers in australia have linked several personal health issues, including an increased use of alcohol, to the pressures of long working hours (mckenzie, 2008, as cited by du plessis et al. 2013). evaluations of health promotion programs within male dominated industries in canada and australia have also found that while workers recognise the importance of healthy lifestyle choices on their physical and mental health, they also face a number of obstacles in making those choices (lingard & turner 2015; seaton et al. 2019). low socio-economic status, long work hours that interfere with family commitments, best et al. new zealand journal of forestry science (2021) 51:5 page 7 figure 1: bronfennbrenner’s ecological systems theory (adapted from bronfenbrenner 1977) macrosystems exosystems mesosystems microsystems individual system levels macro – overarching institutional patterns of the culture or subculture, such as economic, social, educational, legal and political structures (e.g. hegemonic masculinity, business orthodoxy) exo – specific social structures that impinge upon or encompass the immediate settings in which people are found (e.g. the forest industry, worksafe, fisc) meso – interactions between the major settings an individual inhabits (e.g. specific log supply chains) micro – specific settings such as place, time, physical features, activity, participant, and role (e.g. specific logging crew) and socio-cultural constructions of masculinity that emphasise material success can contribute to a culture that inadvertently promotes unhealthy diets, alcohol misuse, and risk taking and stoicism in the face of difficulties (du plessis et al. 2013; iacuone 2005; kolmet et al. 2006; lingard & turner 2015; seaton et al. 2019). lovelock and houghton (2017) pointed to a similar conflict between the imposition of safety rules on new zealand logging crews and the socio-cultural constructs operating within those crews (e.g. the role of experience in establishing crew hierarchy). lingard and turner (2015) concluded that the underlying environmental causes of construction workers’ unhealthy behaviours may be structural and that health promotion initiatives designed to change workers’ health behaviour will consequently be of limited effectiveness. this could well apply to the new zealand forest industry. as stated above, an ecological approach suggests that improving workers’ health outcomes will require intervening in multiple places within the system (sallis et al. 2008). poor mental health at work will most likely reflect multiple psychosocial factors, some of which will be located outside of the direct relationship with the employee or, indeed, outside of the workplace entirely, in families, the community and society more generally (forastieri 2016; lingard & turner 2015; sallis et al. 2008). however it is also important to recognise that benefits from successful interventions are also likely to accrue in multiple places within the system. leka et al. (2015) argue that successful psychosocial risk management can result in benefits to organisational productivity and quality. a study of 7000 polish machine operators using the european framework for psychosocial risks set out in table 1 highlighted the inverse relationship between the level of the psychosocial risk reported by the participants and their reported levels of commitment to and enjoyment of the work and their workplace (mościcka-teske et al. 2017). while the target setting for intervention may be in the forest, benefits such as improvements to productivity, quality and worker commitment will flow beyond the immediate employer to the forest owner and industry level. similarly, psychosocial protective factors experienced at work also have the potential to spill over into the family environment through enhanced mood and skills such as time management (kinnunen et al. 2006) or self-esteem and social support (ten brummelhuis & bakker 2012). given this complexity, improving the psychosocial factors within forestry workplaces will mean looking beyond the day-to-day work settings and workplaces in which forest workers are engaged and considering the forest management practices and operations that impact the way work is organised and completed. figure 2 sets out some aspects of forest management practices that have the potential to influence the psychosocial risk factors for stress. they represent risk factors because of their potential impact on the relationship with the contractor, particularly with respect to the contractor’s profitability, the balance of power within the contract and its impact on business risk. examples of the way in which risk is transferred to the contractor, through the contract, include the setting of a production target as the basis for payment, and the forest owner / manager’s engineering of the work site, particularly the quality of the access and, for harvesting, the setting layout, the maximum and average haul distances, and the skid size. some of these elements of risk involve decisions made with information gathered for the forest owner’s uses but which may not be fit for purpose for managing the contractor’s risk (e.g. inventory data). some of the key decisions may be made in the absence of data or evidence (e.g. estimating production targets without prior productivity measurement evidence). the forest owner / manager may still have control of the sources of risk despite the consequences of the risk having been handed over to the contractor (e.g. establishing piece rates using production when the payment is actually based on uplift and the trucking and delivery is directly contracted and managed by the forest owner). elements of the forest owner’s / manager’s risk can be mitigated by passing some of that risk to the contractor (e.g. the need for layoffs during market downturn). risk is also imposed on the contractor though the terms of the contract, including the crew day rate used as the basis for the piece rate and the way in which perceived transgressions against the contract conditions are dealt with (e.g. stand downs). the allocation of risk between the forest owner / manager and their contractors can be thought of as an expression of the forest owner / manager’s psychosocial safety climate. the psychosocial safety climate refers to the “shared perceptions of organisational policies, practices and procedures for the protection of worker psychological health and safety that are largely driven from senior management” (idris et al. 2012, p. 19). the terms and conditions of the contract have a material impact on the demands made on the contractors and their workers (e.g. work pressure resulting from target or throughput) and the resources they have available to them (e.g. profitability, cashflow, skills, machinery, work study data, control over site lay out). the job demand – resources model (bakker & demerouti 2007) describes how work-related stress is constructed in the balancing of demands and resources. when demands outweigh resources, stress results. being inherently motivational, resources can overcome the costs associated with demands and generate engagement. the potential for the work contained in that contract to have a positive impact on wellbeing is established, essentially, through the process of generating the relationship between the forest owner / manager and the contractor, and then capturing that relationship within the contracting processes. figure 2 also suggests that working directly with communities may be required to ensure interventions in the workplace are successful (sallis et al. 2008). there is a need to engage with the workforce and their communities in a socially and culturally aligned manner (wold & mittelmark 2018). the new zealand forestry industry is dominated by men who often conform to the dominant constructs of working class masculinity in aotearoa new zealand, irrespective of whether best et al. new zealand journal of forestry science (2021) 51:5 page 8 that helps or hinders the industry’s efforts to mitigate the health risks of stress. working with that dominant construct means involving those men in the design, decision making and implementation of any efforts to mitigate mental health risks. fortunately, there are some good examples of successful mental health initiatives (mostly focused on suicide prevention) centred on male participant empowerment, such as the community response to eliminating suicide (cores) programme developed in rural tasmania (jones et al. 2015) and the mates in construction initiative developed in the queensland construction industry (martin et al. 2016). technology is also being used in mental health prevention and care to overcome obstacles to accessing help services (luxton et al. 2011). it also means recognising that the community’s contribution to health and wellbeing involves infrastructure and services such as housing, schools and health centres (solar & irwin 2010) and that business are increasingly playing a role in the development of community capability as a community partner (lee 2011). implications if the new zealand forest industry accepts that its workers operate in conditions that pose a mental health risk, then ecological systems theory can be used as a basis for turning that risk into an opportunity to enhance the industry’s value and social licence. however, it has not been the intention of this paper to be specific about recommended interventions. while the little research that does exist about the psychosocial conditions within new zealand’s forestry workplaces suggest they can be understood through the internationally recognised frameworks, lovelock and houghton (2017) show that even well-intentioned initiatives such as the imposition of greater controls around safe practice can be met with resistance if they do not fit with the socio-cultural constructs in operation for this particular workforce. socio-cultural constructs of gender have also been implicated in the resistance to making healthier eating choices by australian construction workers (lingard & turner 2015). as the research reviewed here indicates, proceeding with worker wellbeing interventions in the absence of an ecological perspective carries some risk. further research that aims to understand what those who work in the bush perceive as their biggest threats and challenges, and what they regard as their coping resources and obstacles, is required before interventions can be prescribed with confidence. conclusions this paper has summarised what the extant research can tell forest managers in new zealand about stress and its various expressions in the workplace (where it is both a potential risk and a potential opportunity). it has assessed that risk by looking at the health and safety consequences of mental distress and by examining what is known about psychosocial hazards within forestry workplaces. it then suggested that mitigating those risks will require going beyond harm reduction as a strategy to thinking about psychosocial factors as possible drivers of a more engaged and committed workforce. interventions aimed at taking advantage of those opportunities within forest management practice and the environments outside the workplace will require thinking beyond the contracts engaging the workforce and instead focusing best et al. new zealand journal of forestry science (2021) 51:5 page 9 figure 2: solution map best et al. new zealand journal of forestry science (2021) 51:5 page 10 on the risk factors inherent in forest management practice and the communities in which workers reside. the rewards for doing so go beyond compliance with health and safety legislation. at its heart, the provision of safe and healthy work environments is a moral and ethical issue. as noted earlier, the international labour organisation’s principles are that work should take place not only in a safe and healthy working environment but also in an environment that offers real possibilities for personal achievement, self-fulfilment and service to society (forastieri 2016). in other words, the imperative with health and safety management is to go beyond ensuring workers survive to enabling workers to thrive. the new zealand forest industry has an opportunity to go beyond the harm reduction focus of the current legislation through promoting worker health and wellbeing, and this should enhance both the industry’s economic performance and its environmental sustainability. competing interests the authors declare that they have no competing interests. authors’ contributions tb undertook most of the work for this manuscript, including establishing the project concept, reviewing the literature, and writing the text. the development of the project and revisions to the manuscript were extensively supported by rv and dc. acknowledgements the authors wish to express their gratitude for the contributions of keith raymond, harvesting programme leader, forest growers research ltd, and brionny hooper, human factors scientist, scion, who commented on an earlier version of the manuscript. funding this research project that informs this article is being carried out with the financial support of forest growers research ltd, the forest industry safety council, the new zealand institute of forestry, the ministry for primary industries and the school of forestry at the university of canterbury. we gratefully acknowledge their support. references ahola, k., salminen, s., toppinen-tanner, s., koskinen, a., & väänänen, a. (2013). occupational burnout and severe injuries: an eight-year prospective cohort study among finnish forest industry workers. journal of occupational health, 55(6), 450-457. https://doi.org/10.1539/joh.13-0021-oa ahola, k., väänänen, a., koskinen, a., kouvonen, a., & shirom, a. (2010). burnout as a predictor of allcause mortality among industrial employees: a 10year prospective register-linkage study. journal of psychosomatic research, 69(1), 51-57. https://doi. org/10.1016/j.jpsychores.2010.01.002 bakker, a.b., & demerouti, e. (2007). the job demandsresources model: state of the art. journal of managerial psychology, 22(3), 309-328. https:// doi.org/10.1108/02683940710733115 bentley, t., tappin., d., blackwood, k., bone, k., forsyth, d., gardner, d., d’souza, n., cately, b., port, z., piplani, r., brougham, d., & ashby, l. (2019). the new zealand workplace barometer: a report on findings from the 2018 survey of the new zealand workplace barometer. auckland, nz: massey university. retrieved from https://www.massey. ac.nz/massey/fms/school_of_management/hwg/ the%20nz%20workplace%20barometer_2018_ final.pdf ?c6c774e163837648c1fc91cf543fc fc3 bronfenbrenner, u. (1977). toward an experimental ecology of human development. american psychologist, 32(7), 513-531. https://doi. org/10.1037/0003-066x.32.7.513 chirico, f., heponiemi, t., pavlova, m., zaffina, s., & magnavita, n. (2019). psychosocial risk prevention in a global occupational health perspective. a descriptive analysis. international journal of environmental research and public health, 16(14), 2470. https://doi.org/10.3390/ijerph16142470 coronial services. (2020). annual provisional suicide statistics for deaths reported to the coroner between 1 july 2007 and 30 june 2020. wellington, nz: ministry of justice. retrieved from https://coronialservices.justice.govt.nz/ assets/documents/publications/2020-annualprovisional-suicide-statistics.pdf cox, t., griffiths, a., & randall, r. (2003). a risk management approach to the prevention of work stress. in m. j. shabracq, j. a. m. winnubst & c. cooper (eds.), the handbook of work and health psychology (2nd ed., pp. 191-208). chichester, uk: wiley. https://doi.org/10.1002/0470013400. ch10 dewe, p., & cooper, c. (2017). work stress and coping: forces of change and challenges (1st ed.). london: sage. https://doi.org/10.4135/9781526421678 dewe, p., o’driscoll, m., & cooper, c. (2010). coping with work stress: a review and critique. chichester, uk: wiley-blackwell. https://doi. org/10.1002/9780470711712 du, d., derks, d., & bakker, a.b. (2017). daily spillover from family to work: a test of the work–home resources model. journal of occupational health psychology. https://doi.org/10.1037/ocp0000073 du plessis, k., cronin, d., corney, t., & green, e. (2013). australian blue-collar men’s health and well-being: contextual issues for workplace health promotion interventions. health promotion practice, 14(5), 715-720. https://doi. org/10.1177/1524839912464046 https://doi.org/10.1539/joh.13-0021-oa https://doi.org/10.1016/j.jpsychores.2010.01.002 https://doi.org/10.1016/j.jpsychores.2010.01.002 https://doi.org/10.1108/02683940710733115 https://doi.org/10.1108/02683940710733115 https://www.massey.ac.nz/massey/fms/school_of_management/hwg/the%20nz%20workplace%20barometer_2018_final.pdf?c6c774e163837648c1fc91cf543fcfc3 https://www.massey.ac.nz/massey/fms/school_of_management/hwg/the%20nz%20workplace%20barometer_2018_final.pdf?c6c774e163837648c1fc91cf543fcfc3 https://www.massey.ac.nz/massey/fms/school_of_management/hwg/the%20nz%20workplace%20barometer_2018_final.pdf?c6c774e163837648c1fc91cf543fcfc3 https://www.massey.ac.nz/massey/fms/school_of_management/hwg/the%20nz%20workplace%20barometer_2018_final.pdf?c6c774e163837648c1fc91cf543fcfc3 https://www.massey.ac.nz/massey/fms/school_of_management/hwg/the%20nz%20workplace%20barometer_2018_final.pdf?c6c774e163837648c1fc91cf543fcfc3 https://doi.org/10.1037/0003-066x.32.7.513 https://doi.org/10.1037/0003-066x.32.7.513 https://doi.org/10.3390/ijerph16142470 https://doi.org/10.1002/0470013400.ch10 https://doi.org/10.1002/0470013400.ch10 https://doi.org/10.4135/9781526421678 https://doi.org/10.1002/9780470711712 https://doi.org/10.1002/9780470711712 https://doi.org/10.1037/ocp0000073 https://doi.org/10.1177/1524839912464046 https://doi.org/10.1177/1524839912464046 best et al. new zealand journal of forestry science (2021) 51:5 page 11 edington, d.w., & burton, w.n. (2003). health and productivity. in r. j. mccunney (ed.), a practical approach to occupational and environmental medicine. (3rd ed.), pp.140-152. boston, ma, usa: little, brown & co. forastieri, v. (2016). prevention of psychosocial risks and work-related stress. international journal of labour research, 8(1-2), 11-34. forest industry safety council. (2018). fisc’s kynd wellness health pilot. 5. retrieved from https:// www.fisc.org.nz/uploads/6/6/2/5/66257655/ fisc_kynd_pilot_2018.pdf golden, s.d., & earp, j.a.l. (2012). social ecological approaches to individuals and their contexts: twenty years of health education & behavior health promotion interventions. health education & behavior, 39(3), 364-372. https://doi. org/10.1177/1090198111418634 hagen, k.b., magnus, p., & vetlesen, k. (1998). neck/shoulder and low-back disorders in the forestry industry: relationship to work tasks and perceived psychosocial job stress. ergonomics, 41(10), 1510-1518. https://doi. org/10.1080/001401398186243 hanse, j.j., & winkel, j. (2008). work organisation constructs and ergonomic outcomes among european forest machine operators. ergonomics, 51(7), 968-981. https://doi. org/10.1080/00140130801961893 harkness, a.m.b., long, b.c., bermbach, n., & patterson, k. (2005). talking about work stress: discourse analysis and implications for stress interventions. work and stress, 19(2), 121-136. https://doi. org/10.1080/02678370500160068 harris, c. (2017). nz forestry industry cries out for young workers, dominion post. retrieved from: https:// www.stuff.co.nz/business/industries/90013570/ skills-shortage-in-forestry-more-problematicthan-drugs-fica-says health and safety at work act (2015). retrieved from: https://www.legislation.govt.nz/act/ public/2015/0070/latest/dlm5976660.html hide, s.a., tappin, d.c., & parker, r.j. (2010). optimising work organisation for maximum performance. harvesting technical note, 3(6). rotorua, new zealand: future forests research. 6 p. retrieved from: https://fgr.nz/documents/optimising-workorganisation-maximum-performance-2/ iacuone, d. (2005). “real men are tough guys”: hegemonic masculinity and safety in the construction industry. journal of men’s studies, 13(2), 247-266. https://doi.org/10.3149/ jms.1302.247 idris, m.a., dollard, m.f., coward, j., & dormann, c. (2012). psychosocial safety climate: conceptual distinctiveness and effect on job demands and worker psychological health. safety science, 50(1), 19-28. https://doi.org/10.1016/j.ssci.2011.06.005 inoue, k. (1996). operators’ mental strain in operating the high proficient forestry machine. journal of forest research, 1(4), 195-197. https://doi. org/10.1007/bf02348324 jones, s., walker, c., miles, a., de silva, e., & zimitat, c. (2015). a rural, community-based suicide awareness and intervention program. rural & remote health, 15(2). https://doi.org/10.22605/ rrh2972 kinnunen, u., feldt, t., geurts, s., & pulkkinen, l.e.a. (2006). types of work-family interface: wellbeing correlates of negative and positive spillover between work and family. scandinavian journal of psychology, 47(2), 149-162. https://doi. org/10.1111/j.1467-9450.2006.00502.x kolmet, m., mariño, r., & plummer, d. (2006). angloaustralian male blue-collar workers discuss gender and health issues. international journal of men’s health, 5(1), 81-91. https://doi.org/10.3149/ jmh.0501.81 lee, l. (2011). business-community partnerships: understanding the nature of partnership. corporate governance: the international journal of business in society, 11(1), 29-40. https://doi. org/10.1108/14720701111108826 leka, s., & jain, a. (2010). health impact of psychosocial hazards at work: an overview. geneva: world health organisation. retrieved from https://apps.who.int/iris/bitstream/ handle/10665/44428/9788393608102_pol.pdf leka, s., van wassenhove, w., & jain, a. (2015). is psychosocial risk prevention possible? deconstructing common presumptions. safety science, 71, 61-67. https://doi.org/10.1016/j. ssci.2014.03.014 lilley, r., feyer, a.-m., kirk, p., & gander, p. (2002). a survey of forest workers in new zealand: do hours of work, rest, and recovery play a role in accidents and injury? journal of safety research, 33(1), 53-71. https://doi.org/10.1016/s0022-4375(02)000038 lingard, h., & turner, m. (2015). improving the health of male, blue collar construction workers: a social ecological perspective. construction management & economics, 33(1), 18-34. https://doi.org/10.108 0/01446193.2014.997259 lovelock, k., & houghton, r. (2017). health on the outside, sick on the inside. work related health in forestry. wellington, new zealand: worksafe new zealand. retrieved from: http://www. worksafe.govt.nz/worksafe/research/researchreports/2016-forestry-research/2016-forestryresearch-documents/2016-forestry-researchreport-5.pdf. luxton, d.d., june, j.d., & kinn, j.t. (2011). technologybased suicide prevention: current applications and https://www.fisc.org.nz/uploads/6/6/2/5/66257655/fisc_kynd_pilot_2018.pdf https://www.fisc.org.nz/uploads/6/6/2/5/66257655/fisc_kynd_pilot_2018.pdf https://www.fisc.org.nz/uploads/6/6/2/5/66257655/fisc_kynd_pilot_2018.pdf https://doi.org/10.1177/1090198111418634 https://doi.org/10.1177/1090198111418634 https://doi.org/10.1080/001401398186243 https://doi.org/10.1080/001401398186243 https://doi.org/10.1080/00140130801961893 https://doi.org/10.1080/00140130801961893 https://www.stuff.co.nz/business/industries/90013570/skills-shortage-in-forestry-more-problematic-than-drugs-fica-says https://www.stuff.co.nz/business/industries/90013570/skills-shortage-in-forestry-more-problematic-than-drugs-fica-says https://www.stuff.co.nz/business/industries/90013570/skills-shortage-in-forestry-more-problematic-than-drugs-fica-says https://www.stuff.co.nz/business/industries/90013570/skills-shortage-in-forestry-more-problematic-than-drugs-fica-says https://doi.org/10.3149/jms.1302.247 https://doi.org/10.3149/jms.1302.247 https://doi.org/10.1016/j.ssci.2011.06.005 https://doi.org/10.1007/bf02348324 https://doi.org/10.1007/bf02348324 https://doi.org/10.22605/rrh2972 https://doi.org/10.22605/rrh2972 https://doi.org/10.1111/j.1467-9450.2006.00502.x https://doi.org/10.1111/j.1467-9450.2006.00502.x https://doi.org/10.3149/jmh.0501.81 https://doi.org/10.3149/jmh.0501.81 https://doi.org/10.1108/14720701111108826 https://doi.org/10.1108/14720701111108826 https://apps.who.int/iris/bitstream/handle/10665/44428/9788393608102_pol.pdf https://apps.who.int/iris/bitstream/handle/10665/44428/9788393608102_pol.pdf https://doi.org/10.1016/j.ssci.2014.03.014 https://doi.org/10.1016/j.ssci.2014.03.014 https://doi.org/10.1016/s0022-4375(02)00003-8 https://doi.org/10.1016/s0022-4375(02)00003-8 https://doi.org/10.1080/01446193.2014.997259 https://doi.org/10.1080/01446193.2014.997259 http://www.worksafe.govt.nz/worksafe/research/research-reports/2016-forestry-research/2016-forestry-research-documents/2016-forestry-research-report-5.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/2016-forestry-research/2016-forestry-research-documents/2016-forestry-research-report-5.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/2016-forestry-research/2016-forestry-research-documents/2016-forestry-research-report-5.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/2016-forestry-research/2016-forestry-research-documents/2016-forestry-research-report-5.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/2016-forestry-research/2016-forestry-research-documents/2016-forestry-research-report-5.pdf future directions. telemedicine and e-health, 17(1), 50-54. https://doi.org/10.1089/tmj.2010.0091 martin, g., swannell, s., milner, a., & gullestrup, j. (2016). mates in construction suicide prevention program: a five year review. journal of community medicine & health education, 6(465). https://doi. org/10.4172/2161-0711.1000465 maslach, c., & leiter, m.p. (2016). understanding the burnout experience: recent research and its implications for psychiatry. world psychiatry, 15(2), 103-111. https://doi.org/10.1002/wps.20311 maslach, c., & leiter, m.p. (2017). new insights into burnout and health care: strategies for improving civility and alleviating burnout. medical teacher, 39(2), 160-163. https://doi.org/10.1080/014215 9x.2016.1248918 ministry for primary industries. (2020). human capability in the primary industries data: part one. wellington, new zealand: nz government .retrieved from https://www.mpi.govt.nz/ dmsdocument/17638-human-capability-in-theprimary-industries-part-1-2002-to-2016-anoverview. modini, m., joyce, s., mykletun, a., christensen, h., bryant, r.a., mitchell, p.b., & harvey, s.b. (2016). the mental health benefits of employment: results of a systematic meta-review. australasian psychiatry, 24(4), 331-336. https://doi. org/10.1177/1039856215618523 moore, r. (2017). forestry worker shortage, youths sought out, the southland times. retrieved from https://www.stuff.co.nz/southland-times/ news/95190310/forestry-worker-shortageyouths-sought-out mościcka-teske, a., sadłowska-wrzesińska, j., butlewski, m., misztal, a., & jacukowicz, a. (2017). stressful work characteristics, health indicators and work behavior: the case of machine operators. international journal of occupational safety and ergonomics, 23(4), 510. https://doi.org/10.1080/ 10803548.2016.1197577 mylek, m.r., & schirmer, j. (2015). beyond physical health and safety: supporting the wellbeing of workers employed in the forest industry. forestry: an international journal of forest research, 88(4), 391-406. https://doi.org/10.1093/forestry/ cpv011 nahrgang, j.d., morgeson, f.p., & hofmann, d.a. (2011). safety at work: a meta-analytic investigation of the link between job demands, job resources, burnout, engagement, and safety outcomes. journal of applied psychology, 96(1), 73-94. https://doi. org/10.1037/a0021484 nielsen. (2015a). health and safety attitudes and behaviours in the new zealand workforce. a survey of workers and employers. 2014 baseline survey. forestry report. wellington, new zealand. retrieved from http://www.worksafe.govt.nz/ worksafe/research/research-reports/reports/ a t t i t u d e s a n d b e h av i o u r s s u r ve y b a s e l i n e forestry-report.pdf nielsen. (2015b). health and safety attitudes and behaviours in the new zealand workforce: a study of workers and employers. 2014 qualitative research. forestry report. (a report for worksafe new zealand). wellington, new zealand. retrieved from http://www.worksafe.govt.nz/worksafe/ research/research-reports/reports/attitudesand-behaviours-survey-qualitative-forestryreport.pdf nielsen. (2018). health and safety attitudes and behaviours survey. worker engagement and participation report 2017. (a report for worksafe new zealand). wellington, new zealand. retrieved from https://worksafe.govt.nz/research/ attitudes-and-behaviours-survey-2017/ pearce, j., tisch, c., & barnett, r. (2008). have geographical inequalities in cause-specific mortality in new zealand increased during the period 1980-2001? new zealand medical journal, 121(1281). pfeffer, j. (2018). dying for a paycheck: how modern management harms employee health and company performance and what we can do about it. new york: harper collins. rössler, w., hengartner, m.p., ajdacic-gross, v., & angst, j. (2015). predictors of burnout: results from a prospective community study. european archives of psychiatry and clinical neuroscience, 265(1), 1925. https://doi.org/10.1007/s00406-014-0512-x sallis, j.f., owen, n., & fisher, e.b. (2008). ecological models of health behaviour. in k. glanz, b. k. rimer & k. viswanath (eds.), health behavior and health education: theory, research, and practice (pp. 465486). hoboken, usa: wiley. seaton, c.l., bottorff, j.l., oliffe, j.l., medhurst, k., & deleenheer, d. (2019). mental health promotion in male-dominated workplaces: perspectives of male employees and workplace representatives. psychology of men & masculinities, 20(4), 541-552. https://doi.org/10.1037/men0000182 seidler, a., thinschmidt, m., deckert, s., then, f., hegewald, j., nieuwenhuijsen, k., & riedel-heller, s.g. (2014). the role of psychosocial working conditions on burnout and its core component emotional exhaustion – a systematic review. journal of occupational medicine and toxicology, 9(1), 10. https://doi.org/10.1186/1745-6673-910 solar, o., & irwin, a. (2010). a conceptual framework for action on the social determinants of health. social determinants of health discussion paper 2 (policy and practice). geneva: world health organisation. retrieved from https://apps.who.int/iris/rest/ bitstreams/52952/retrieve steepland harvesting programme. (2018). steepland harvesting programme post-programme report best et al. new zealand journal of forestry science (2021) 51:5 page 12 https://doi.org/10.1089/tmj.2010.0091 https://doi.org/10.4172/2161-0711.1000465 https://doi.org/10.4172/2161-0711.1000465 https://doi.org/10.1002/wps.20311 https://doi.org/10.1080/0142159x.2016.1248918 https://doi.org/10.1080/0142159x.2016.1248918 https://www.mpi.govt.nz/dmsdocument/17638-human-capability-in-the-primary-industries-part-1-2002-to-2016-an-overview https://www.mpi.govt.nz/dmsdocument/17638-human-capability-in-the-primary-industries-part-1-2002-to-2016-an-overview https://www.mpi.govt.nz/dmsdocument/17638-human-capability-in-the-primary-industries-part-1-2002-to-2016-an-overview https://www.mpi.govt.nz/dmsdocument/17638-human-capability-in-the-primary-industries-part-1-2002-to-2016-an-overview https://doi.org/10.1177/1039856215618523 https://doi.org/10.1177/1039856215618523 https://www.stuff.co.nz/southland-times/news/95190310/forestry-worker-shortage-youths-sought-out https://www.stuff.co.nz/southland-times/news/95190310/forestry-worker-shortage-youths-sought-out https://www.stuff.co.nz/southland-times/news/95190310/forestry-worker-shortage-youths-sought-out https://doi.org/10.1080/10803548.2016.1197577 https://doi.org/10.1080/10803548.2016.1197577 https://doi.org/10.1093/forestry/cpv011 https://doi.org/10.1093/forestry/cpv011 https://doi.org/10.1037/a0021484 https://doi.org/10.1037/a0021484 http://www.worksafe.govt.nz/worksafe/research/research-reports/reports/attitudes-and-behaviours-survey-baseline-forestry-report.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/reports/attitudes-and-behaviours-survey-baseline-forestry-report.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/reports/attitudes-and-behaviours-survey-baseline-forestry-report.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/reports/attitudes-and-behaviours-survey-baseline-forestry-report.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/reports/attitudes-and-behaviours-survey-qualitative-forestry-report.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/reports/attitudes-and-behaviours-survey-qualitative-forestry-report.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/reports/attitudes-and-behaviours-survey-qualitative-forestry-report.pdf http://www.worksafe.govt.nz/worksafe/research/research-reports/reports/attitudes-and-behaviours-survey-qualitative-forestry-report.pdf https://worksafe.govt.nz/research/attitudes-and-behaviours-survey-2017/ https://worksafe.govt.nz/research/attitudes-and-behaviours-survey-2017/ https://doi.org/10.1007/s00406-014-0512-x https://doi.org/10.1037/men0000182 https://doi.org/10.1186/1745-6673-9-10 https://doi.org/10.1186/1745-6673-9-10 2018. rotorua, nz. retrieved from https://fgr.nz/ documents/download/7206?999267066 stokols, d.s. (1992). establishing and maintaining healthy environments: toward a social ecology of health promotion. american psychologist, 47(1), 6-22. https://doi.org/10.1037/0003-066x.47.1.6 suicide mortality review committee. (2016). ngā rāhui hau kura: suicide mortality review committee feasibility study 2014-15. report to the ministry of health, 31 may 2016. wellington, nz: health quality & safety commission. retrieved from https:// www.hqsc.govt.nz/our-programmes/mrc/sumrc/ publications-and-resources/publication/2471/ tehrani, n. (2004). workplace trauma: concepts, assessment and interventions: london: brunner-routledge. https://doi. org/10.4324/9780203500620 ten brummelhuis, l.l., & bakker, a.b. (2012). a resource perspective on the work–home interface: the work–home resources model. american psychologist, 67(7), 545-556. https://doi. org/10.1037/a0027974 theorell, t., hammarström, a., aronsson, g., träskman bendz, l., grape, t., hogstedt, c., marteinsdottir, i., skoog, i., & hall, c. (2015). a systematic review including meta-analysis of work environment and depressive symptoms. bmc public health, 15: 738, 14 p. https://doi.org/10.1186/s12889-015-19544 toppinen-tanner, s., ahola, k., koskinen, a., & väänänen, a. (2009). burnout predicts hospitalization for mental and cardiovascular disorders: 10-year prospective results from industrial sector. stress & health: journal of the international society for the investigation of stress, 25(4), 287-296. https://doi. org/10.1002/smi.1282 toppinen-tanner, s., ojajärvi, a., väänänen, a., kalimo, r., & jäppinen, p. (2005). burnout as a predictor of medically certified sick-leave absences and their diagnosed causes. behavioral medicine, 31(1), 1827. https://doi.org/10.3200/bmed.31.1.18-32 toppinen-tanner, s., kalimo, r., & mutanen, p. (2002). the process of burnout in white-collar and blue-collar jobs: eight-year prospective study of exhaustion. journal of organizational behavior, 23(5), 555-570. https://doi.org/10.1002/job.155 väänänen, a., koskinen, a., joensuu, m., kivimäki, m., vahtera, j., kouvonen, a., & jäppinen, p. (2008). lack of predictability at work and risk of acute myocardial infarction: an 18-year prospective study of industrial employees. american journal of public health, 98(12), 2264-2271. https://doi. org/10.2105/ajph.2007.122382 väänänen, a., murray, m., & kuokkanen, a. (2014). the growth and stagnation of work stress: publication trends and scientific representations 1960–2011. history of the human sciences, 27(4), 116-138. https://doi.org/10.1177/0952695114525168 wold, b., & mittelmark, m.b. (2018). health-promotion research over three decades: the social-ecological model and challenges in implementation of interventions. scandinavian journal of public health, 46(20_suppl), 20-26. https://doi. org/10.1177/1403494817743893 woodward, k. (2015). psychosocial studies: an introduction. new york, usa: taylor & francis. https://doi.org/10.4324/9781315867823 world economic forum. (2008). working towards wellness: the business rationale. geneva. retrieved from https://www.pwc.pl/en/publikacje/ business_rationale.pdf world health organisation. (2008). prima-ef guidance on the european framework for psychosocial risk management: a resource for employers and workers representatives. geneva. retrieved from http:// www.prima-ef.org/uploads/1/1/0/2/11022736/ prima-ef_brochure_english.pdf world health organisation. (2019). constitution of the world health organisation. retrieved from https:// www.who.int/about/who-we-are/constitution best et al. new zealand journal of forestry science (2021) 51:5 page 13 https://fgr.nz/documents/download/7206?999267066 https://fgr.nz/documents/download/7206?999267066 https://doi.org/10.1037/0003-066x.47.1.6 https://www.hqsc.govt.nz/our-programmes/mrc/sumrc/publications-and-resources/publication/2471/ https://www.hqsc.govt.nz/our-programmes/mrc/sumrc/publications-and-resources/publication/2471/ https://www.hqsc.govt.nz/our-programmes/mrc/sumrc/publications-and-resources/publication/2471/ https://doi.org/10.4324/9780203500620 https://doi.org/10.4324/9780203500620 https://doi.org/10.1037/a0027974 https://doi.org/10.1037/a0027974 https://doi.org/10.1186/s12889-015-1954-4 https://doi.org/10.1186/s12889-015-1954-4 https://doi.org/10.1002/smi.1282 https://doi.org/10.1002/smi.1282 https://doi.org/10.3200/bmed.31.1.18-32 https://doi.org/10.1002/job.155 https://doi.org/10.2105/ajph.2007.122382 https://doi.org/10.2105/ajph.2007.122382 https://doi.org/10.1177/0952695114525168 https://doi.org/10.1177/1403494817743893 https://doi.org/10.1177/1403494817743893 https://doi.org/10.4324/9781315867823 https://www.pwc.pl/en/publikacje/business_rationale.pdf https://www.pwc.pl/en/publikacje/business_rationale.pdf http://www.prima-ef.org/uploads/1/1/0/2/11022736/prima-ef_brochure_english.pdf http://www.prima-ef.org/uploads/1/1/0/2/11022736/prima-ef_brochure_english.pdf http://www.prima-ef.org/uploads/1/1/0/2/11022736/prima-ef_brochure_english.pdf https://www.who.int/about/who-we-are/constitution https://www.who.int/about/who-we-are/constitution species-specific basic stem-wood densities for twelve indigenous forest and shrubland species of known age, new zealand michael marden1,*, suzanne lambie2 and larry burrows3 1 31 haronga road, gisborne 4010, new zealand 2 manaaki whenua – landcare research, private bag 3127, hamilton 3240, new zealand 3 manaaki whenua – landcare research, po box 69041, lincoln 7640, new zealand *corresponding author: mardenm@landcareresearch.co.nz (received for publication 19 july 2019; accepted in revised form 26 january 2021) abstract background: tree carbon estimates for new zealand indigenous tree and shrub species are largely based on mean basic stem-wood densities derived from a limited number of trees, often of unspecified age and from a limited number of sites throughout new zealand. yet stem-wood density values feed directly into new zealand’s international and national greenhouse gas accounting. we augment existing published basic stem-wood density data with new agespecific values for 12 indigenous forest and shrubland species, including rarely obtained values for trees <6-years old, across 21 widely-distributed sites between latitudes 35° and 46° s, and explore relationships commonly used to estimate carbon stocks. methods: the volume of 478 whole stem-wood discs collected at breast height (bh) was determined by water displacement, oven dried, and weighed. regression analyses were used to determine possible relationships between basic stem-wood density, and tree height, root collar diameter (rcd), and diameter at breast height (dbh). unbalanced anova was used to determine inter-species differences in basic stem-wood density in 5-yearly age groups (i.e. 0–5 years, 6–10 years etc.) (p<0.05). as specific taxa of kunzea ericoides (myrtaceae) has only been identified at some study sites we combine the data from each site, and use the term kunzea spp. we compare our ageand species-specific results with existing published data where age is specified versus non-age-specific values. results: kunzea spp. and leptospermum scoparium exhibited positive correlations between basic stem-wood density and tree height, rcd, and dbh. no relationships were established for melicytus ramiflorus, coprosma grandiflora, weinmannia racemosa ≥6-years old, or for podocarpus totara, agathis australis, vitex lucens, and alectryon excelsus <6-years old. dacrydium cupressinum and prumnopitys ferruginea <6-years old exhibited a significant positive relationship with dbh only, while for dacrycarpus dacrydioides, each correlation was negative. irrespective of age, basic stem-wood density is not different between the hardwood species l. scoparium and kunzea spp. but is significantly greater (p=0.001) than that of the remaining, and predominantly softwood species of equivalent age. for kunzea spp., l. scoparium, coprosma grandiflora, weinmannia racemosa, and melicytus ramiflorus ≥6-years old there was no evidence that basic stem-wood density increased with tree age, and values were within the range of published and unpublished data. for naturally reverting stands of kunzea spp. located between latitudes 35° to 46° s, basic stem-wood density values tended to increase with decreased elevation and increased temperature. conclusions: increasing basic wood density values in kunzea spp. with decreased elevation and increased temperature suggest that where local data are available its use would improve the accuracy of biomass estimates both locally and nationally. furthermore, refining biomass estimates for existing communities of mixed softwood species, stands of regenerating shrubland, and new plantings of indigenous species will require additional basic stem-wood density values for scaling from stem wood volume to total stand biomass. new zealand journal of forestry science marden et al. new zealand journal of forestry science (2021) 51:1 https://doi.org/10.33494/nzjfs512021x121x e-issn: 1179-5395 published on-line: 15/02/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access keywords: basic stem-wood density, allometric relations, 12 indigenous forest and shrubland species, new zealand. http://creativecommons.org/licenses/by/4.0/), marden et al. new zealand journal of forestry science (2021) 51:1 page 2 introduction the variability in basic stem-wood density and age are critical factors influencing estimates of wood biomass and carbon storage capability (chave et al. 2004, dale 2013). stem-wood density values feed directly into new zealand’s international greenhouse gas accounting of forest carbon stocks, and for internal schemes such as the emissions trading scheme (ets) (ministry for primary industries 2017), and the 1 billion trees programme (1bt) (ministry for primary industries 2018). previously, new zealand studies have estimated the biomass of indigenous forest stands for tree carbon stocks and sequestration using diameters and height measurements of individual trees in forest inventory plots (carswell et al. 2012, scott et al. 2000, trotter et al. 2005, beets et al. 2014, schwendenmann & mitchell 2014, dale 2013, holdaway et al. 2014). when basic wood density values are available for only a limited number of species and locations, wood volume is converted to carbon stocks using generic (as opposed to speciesspecific) functions based on the basic density of stemwood (oven-dry mass/ ‘green’ volume). where speciesspecific and/or regional basic stem-wood density values are unavailable, congeneric values are used instead, or in their absence, the mean of all published values e.g. beets et al. (2012 and unpublished data1). while most early studies in new zealand collected basic stem-wood density data from sites of wellestablished indigenous shrubs and trees, age-specific and species-specific stem-wood density data for the early growth period of many species remain elusive. the absence of taxon-specific stem-wood density and age-class distribution data of a wide variety of species over a range of geographic sites introduces uncertainty in the accuracy of new zealand’s national carbon budget calculations (scott et al. 2000, chave et al. 2004, holdaway 2014). the use of taxon-specific stem-wood density to scale tree volume, as yield or growth, to stem biomass, and from stem biomass to total biomass will improve the accuracy of species-specific allometric equations for estimating tree carbon storage, and avoid potential bias to national carbon budgets. furthermore, basic stem-wood density values for a few widespread indigenous species (entrican et al. 1951, hall et al. unpublished data2), and for specific species with a more restricted geographic range (wardle 1991), can vary depending on geographic location, though no relationships have been verified with respect to climate or site factors (hall et al. unpublished data2). clifton (1990) suggests that basic stem-wood density varies according to the age of the tree, the location of the wood within a tree (outer-wood/inner-wood, base or top of a tree), and while densities have been determined for some of new zealand’s historically important merchantable wood species (hinds & reid 1957, beets et al. 2012), the age of the trees and variations in basic stem-wood density were not determined, the sample size was generally small, the methods uncertain, and the location vague. stand basic stem-wood densities will also change with time, influenced by climatic variability and site-specific physical factors, including soil type, slope, aspect, elevation and rainfall regime, all of which can affect growth rates, plant survival, and carbon sequestration rates. furthermore, as the area of indigenous species plantings and their diversity increases with age, age-specific and speciesspecific stem-wood density data, will be relevant for afforestation/reforestation reporting, for updating the national carbon inventory system (land use and carbon accounting system – lucas), and policy, to reduce net greenhouse gas emissions as required under the kyoto protocol (ministry of the environment 2010), and for comparison with pre-calculated forest carbon stocks (includes stem, bark, branch, leaves, litter, woody debris, stumps and roots expressed in units of tonnes of co2 ha-1), by age, for given forest types in the emissions trading scheme (ministry for primary industries 2017). we augment existing published basic stem-wood density data with new age-specific values for 12 of new zealand’s indigenous forest and shrubland species from 21 widely distributed sites located between latitudes 35° to 46° s. we explore relationships between basic stem-wood density and tree parameters commonly used to estimate stem carbon stocks, and applicable to future efforts to reduce the uncertainty of carbon stock estimates for forest and shrubland communities where basic stem-wood density values for different age classes of many species is currently missing. methods study sites basic wood density data was collected from 14 sites located in the north island and from 7 sites in the south island of new zealand with a latitudinal range between 35° and 46° s (fig. 1). details of species, elevation, and substrate characteristics are summarised in table 1, and presented in more detail in appendix table a1. species nomenclature since this study began, there has been a taxonomic revision of the new zealand kunzea ericoides (myrtaceae) complex in new zealand (de lange 2014). ten kunzea species endemic to new zealand are now recognised, seven of which are new. where we have some confidence in the identification of new taxa these are presented in table 1 and appendix table a1. as specific taxa have not been identified for all sites we have not attempted to analyse for possible inter-specific variations in basic stem-wood density for this genus but 1 beets, p.n., oliver, g.r, kimberley., m.o, pearce, s.h. (2008). allometric functions for estimating above ground carbon in native forest trees, shrubs and ferns. scion report 12679 prepared for the ministry for the environment 63 p. 2 hall, g., wiser, s., allen, r., moore, t., beets, p., goulding, c. (1998). estimate of the carbon stored in new zealand’s indigenous forest and scrub vegetation for 1990. landcare research contract jnt9798/147 prepared for ministry for the environment, wellington, new zealand. 36 p. instead we combine data for all sites where present and use the generic term kunzea spp. wood sampling and density there are many methods of sampling wood and determining wood density (chave 2005, williamson & wiemann 2010). in this study, wood density is defined as the ratio of the oven-dry mass of a stem-wood disc sampled at a standard height divided by the mass of water displaced by its green volume to give wood specific gravity (wsg). wsg is described as basic wood density or stem-wood density throughout the text. discs cut from the stem account for the change in density from pith to bark (williamson & wiemann 2010, beets et al. 2012). basic stem density measurements of discs were sourced from trees located in areas of naturally regenerating kunzea spp. (sites 2–7, 9, 12– 21), regenerating leptospermum scoparium (sites 1, 2, 9, 14, 16, and 21), a lowland shrub community (site 11), a species growth trial of indigenous softwood and hardwood species (site 8), and from an area of low-density plantings of l. scoparium (site 10). as the purpose of the research undertaken at each site differed, 256 of the basic stem-wood density measurements were of discs with the bark intact (cornelissen et al. 2003) (e.g. sites 2, 4-11 & 21) and 222 measurements were of discs with the bark removed (e.g. sites 1, 3, 1220). all discs were sampled at breast height (bh) (1.4 m above ground-level). the fresh volume of each wood disc was determined by water displacement, then oven dried at 105°c (cornelissen et al. 2003) and weighed. for multiple-stemmed trees, a disc was cut from each stem, and the density averaged for the tree. tree age in naturally regenerating stands was based on ring counts of the single oldest stem. the age of the species established in the plant growth trial (site 8) was based on the known date that seedlings were ‘pricked-out’ into seed trays in the nursery. for the site established in l. scoparium for honey production (site 10), the year in which 1-year-old, nursery-raised seedlings were planted was known. for melicytus ramiflorus and coprosma grandiflora (site 11), discs were collected in the field at bh and transported in a sealed container to avoid moisture loss. in the laboratory, discs were soaked before the volume was determined by water displacement. discs were dried at 80°c until dry and weighed (cornelissen et al. 2003). tree height was based on the tallest single stem. tree age was based on ring counts of a disc cut from a representative stem of the tree. for kunzea spp. and l. scoparium collected from sites 2, 9, and 21, discs were collected at bh and frozen at –20°c. the discs were thawed at room temperature and soaked in water for 2 days before their volume was assessed. as l. scoparium and kunzea spp. tend to split during drying making ring counting and measuring difficult, the discs were partially dried at 35°c, the rings counted, and then dried at 80°c and weighed. for kunzea spp., and l. scoparium, tree parameters were predominantly measured in regenerating shrubland >6-years old. other regenerating shrubland species including melicytus ramiflorus, coprosma grandiflora, and weinmannia racemosa include measurements for a wide range of ages both < and >6-years old while regressions for plot-based alectron excelsus, podocarpus totara, agathis australis, dacrydium cupressinum, prumnopitus ferruginea, dacrycarpus dacrydioides and vitex lucens include only data for trees <6-years old. statistical analyses linear regression analysis best fitted the data and was used within each tree species to determine the possible relationship between basic stem-wood density and tree height, root collar diameter (rcd), diameter at breast height (dbh), and tree age. unbalanced anova with least significant differences (lsd) was used to determine differences in basic stemwood density between species and for kunzea spp. to assess if densities differed between 17 sites located throughout new zealand. density values were grouped into 5-yearly age classes (e.g. 0–5-years, 6–10 years etc.). only data sets within a species, and within an age class with three or more replicates (irrespective of the geographical position) were used in the analysis. the average basic stem-wood densities for younger (<6-years old) and older (≥6-years old) trees are compared with published values. for the marden et al. new zealand journal of forestry science (2021) 51:1 page 3 figure 1: location of 21 new zealand indigenous forest, shrubland, and experimental trial sites where discs were collected for analysis of basic stem-wood density. marden et al. new zealand journal of forestry science (2021) 51:1 page 4 figure 2: ta b le 1 : lo ca ti on s, tr ee s pe ci es p re se nt , e le va ti on a bo ve s ea le ve l ( as l) , a nd s ub st ra te c ha ra ct er is ti cs fo r 21 s tu dy s it es s am pl ed fo r ba si c w oo d de ns it y. m ap c oor di na te s ar e n ew z ea la nd m ap g ri d (n zm g ) si te n um be r an d na m e g ri d re fe re nc e sp ec ie s el ev at io n as l ( m ) su bs tr at e 1: t au to ro 17 3° 5 0´ 1 3 15 e , 3 5° 2 8´ 5 2 00 s le pt os pe rm um s co pa ri um 10 0– 14 0 gr ey w ac ke a rg ill it es a nd s an ds to ne s 2: w ai ta ke re r an ge 17 4° 3 5´ 1 4 42 e , 3 7° 0 0´ 1 0 17 s ku nz ea s pp . a nd l . s co pa ri um 40 vo lc an ic a nd es it ic la va , c on gl om er at es , a nd b re cc ia 3: n ik au v al le y 17 6° 5 8´ 2 3 85 e , 3 8° 0 1´ 2 5 27 s ku nz ea r ob us ta 40 –1 00 un di ffe re nt ia te d gr ey w ac ke 46: t ol ag a b ay 17 8˚ 1 2΄ 1 9 29 e , 3 8˚ 2 0΄ 42 5 8 s ku nz ea r ob us ta 64 ca lc ar eo us s an dy s ilt st on es w it h ba nd ed s an ds to ne s 7: w ai m at a va lle y 17 8° 0 3΄ 1 3 66 e , 3 8˚ 2 8΄ 3 3 84 s ku nz ea r ob us ta 20 7 ca lc ar eo us s an dy s ilt st on es w it h ba nd ed s an ds to ne s 8: g is bo rn e 17 8° 0 0´ 1 6 02 e , 3 8° 3 8´ 4 4 82 s ag at hi s au st ra lis , p ru m no pi ty s fe rr ug in ea , p od oc ar pu s to ta ra , d ac ry ca rp us d ac ry di oi de s, d ac ry di um cu pr es si nu m , a le ct ry on e xc el su s, an d vi te x lu ce ns . 5 al lu vi al g ra ve ls a nd s ilt . 9: t ur an gi 17 5° 4 7´ 1 1 53 e , 3 9° 0 9´ 1 9 20 s l. s co pa ri um a nd k un ze a sp p. 80 0 rh yo lit ic a nd a nd es it ic v ol ca ni cs 10 : l ak e tu ti ra 17 6° 5 4´ 1 0 44 e , 3 9° 1 4´ 0 0 44 s l. s co pa ri um 20 0– 37 5 m ud st on e, s an ds to ne , a nd li m es to ne 11 : w ai nu io m at a & ca nn on s cr ee k 17 4° 5 7´ 1 9 75 e , 4 1° 1 7´ 4 5 29 s co pr os m a gr an di flo ra , w ei nm an ni a ra ce m os a, a nd m el ic yt us r am ifl or us 11 7 al te rn at in g da rk g re y ar gi lli te a nd g re yw ac ke s an ds to ne 12 : l on g g ul ly 17 4° 4 0´ 5 5 30 e , 4 1° 1 8´ 3 4 82 s ku nz ea a m at hi co la 30 0– 40 0 ar gi lli te a nd g re yw ac ke s an ds to ne w it h ra re li m es to ne an d vo lc an ic s 13 : r iv er sd al e 17 5° 2 5´ 5 3 49 e , 4 1° 3 0´ 5 7 74 s ku nz ea r ob us ta 60 –2 00 gr ey w ac ke -l ik e da rk g re y m ud dy s ilt st on e w it h m in or co ng lo m er at es a nd s pi lit ic la va 14 : c oa tb ri dg e 17 3° 3 9´ 2 3 16 e , 41 ° 2 9´ 0 8 99 s l. s co pa ri um a nd k un ze a sp p. 20 0– 30 0 m et am or ph os ed s ed im en ta ry li th ol og ie s an d vo lc an ic s 15 : l on g sp ur 17 5° 3 2´ 0 9 01 e , 4 1° 2 7´ 2 2 12 s ku nz ea r ob us ta 40 –2 00 sa nd st on e an d m ud st on e, m in or c on gl om er at es a nd vo lc an ic s 16 : p eg gi oh 17 4° 0 1´ 1 3 67 e , 4 1° 5 1´ 3 1 57 s l. s co pa ri um a nd k un ze a ro bu st a 20 0– 30 0 gr ey w ac ke a nd a rg ill it e w it h m in or v ol ca ni cs , co ng lo m er at es , a nd r ar e lim es to ne 17 : s he na nd oa h 17 2° 1 5´ 0 5 30 e , 4 1° 5 3´ 3 6 00 s ku nz ea e ri co id es 20 0– 30 0 lim es to ne a nd c al ca re ou s si lt st on e, lo ca l s an ds to ne a nd co al m ea su re s 18 : a vo ca s ta ti on 17 1° 5 3´ 2 3 31 e , 43 ° 1 1´ 4 9 51 s ku nz ea s er ot in a 42 0– 54 0 gr ey w ac ke a nd a rg ill it e w it h m in or v ol ca ni cs , co ng lo m er at es , a nd r ar e lim es to ne 19 : e yr ew el l. 17 2° 1 1´ 4 1 76 e , 4 3° 2 2´ 5 9 35 s ku nz ea s er ot in a 20 0 po st -g la ci al a llu vi um a nd g la ci al o ut w as h gr av el s 20 : h in ew ai . 17 3° 0 2´ 1 8 74 e , 4 3° 4 9´ 0 2 85 s ku nz ea r ob us ta 20 –4 50 ba sa lt tu ff , a nd a ss oc ia te d in tr us iv e ro ck s 21 : d un ed in 17 0° 3 6´ 3 7 14 e , 4 5° 4 5´ 1 1 19 s l. s co pa ri um a nd k un ze a ro bu st a 20 0– 30 0 lo es s, b as al t a nd p ho no lit e earliest of the published data (kirk 1889, but mostly by entrican et al. 1951, and republished by hinds & reid 1957, harris 1986, and clifton 1990), tree age is rarely specified, and variations in basic stem-wood density values derived from merchantable-sized trees after removal of the bark is not given. for comparative purposes we use these few available published values (appendix table a2) together with a larger data set of mean age-specific/non-age-specific wood density values (bark removed) collected from carbon monitoring system (lucas) plots (20m x 20 m) across a wide range of well-established and pre-defined natural forest and shrubland types (table a2) indicative of advanced succession toward indigenous forest (hall et al. unpublished data2, peltzer & payton unpublished data3, beets et al. 2012 and unpublished data1). we did not attempt to analyse for the influence of bark thickness on basic stem wood density values (i.e. inclusive versus exclusive of bark), as for the age-range (3to 105-years old) of the shrubland species presented in this paper, all values were expected to fall well within the range of the published data. in the absence of reliable basic stem-wood density values for individual stems, often determined for only a small sample size of trees with widely varying, or of unknown age, and variability in basic stem-wood density values, the values in this paper are presented as means (appendix tables a3–a5). all statistical analyses were undertaken using genstat (vsn international, hemel hempstead, uk) and were considered significant if p<0.05. results basic stem-wood density-allometric relationships for >6-year-old regenerating kunzea spp. basic stemwood density was significantly, positively correlated with tree height, as was also the case for l. scoparium (table 2). of the plot-based species <6-years old, the correlation for basic stem-wood density with tree height was strongest (and positive) for prumnopitys ferruginea (table 2) but was only just statistically significant, probably due to the small sample size (n=7). interestingly, dacrycarpus dacrydioides exhibited a significant negative correlation with about 30% of the variation in basic stem-wood density explained by tree height. there were no other significant relationships between basic stem-wood density and tree height for the remaining plot-based or regenerating shrubland species. basic stem-wood density and rcd were positively correlated for regenerating l. scoparium and kunzea spp. >6-years old (table 2). root collar diameter and density values were negatively correlated for plotbased dacrycarpus dacrydioides (table 2). there were no significant correlations between basic stem-wood density and rcd for the remaining plot-based and regenerating shrubland species <6-years old. basic stem-wood density and dbh were positively correlated for regenerating l. scoparium, kunzea spp., plot-based dacrydium cupressinum and prumnopitys ferruginea (table 2) with dbh explaining 17–73% of the variation in density. basic stem-wood and dbh were negatively correlated for dacrycarpus dacrydioides (table 2). there were no significant correlations between basic stem-wood density and dbh for the remaining plot-based and regenerating species. basic stem-wood density was not correlated with tree age for low-density plantings of l. scoparium (site 10) between ages 4and 6-years and increased with increasing tree age (data not shown). conversely, for naturally reverting stands of l. scoparium, kunzea spp., coprosma grandiflora, melicytus ramiflorus and weinmannia racemosa, basic stem-wood density values of ≥6-years-old trees were not significant. comparisons of mean basic wood densities by ageclass basic stem-wood density of l. scoparium was greater than for the remainder of the plot-based species trialled for trees <6-years of age (fig. 2a). basic stem-wood density was as follows for the various species in this age group: l. scoparium > alectryon excelsus > dacrycarpus dacrydioides = podocarpus totara = prumnopitys ferruginea = dacrydium cupressinum > agathis australis = vitex lucens. for naturally regenerating stands between 6–10 and 11–15 years old, kunzea spp. had greater basic stemwood density than melicytus ramiflorus (fig. 2b). basic stem-wood density of kunzea spp. was also greater than coprosma grandiflora and melicytus ramiflorus in the 16– 20 (fig. 2b) and 21–25-year-old age class (fig. 2c). there was no difference in basic stem-wood density between coprosma grandiflora and melicytus ramiflorus between 16–20 (fig 2b) and 21–25-year-old age classes (fig 2c). in the age classes 26–30, 31–35, 36–40 (fig. 2c), and 46-50, 51–70 years (fig. 2d) there were no differences in basic stem-wood density between kunzea spp. and l. scoparium. however, for the oldest of the age classes their respective densities were significantly greater (p<0.05) than for weinmannia racemosa of the same age (fig. 2d). irrespective of age, the basic stem-wood density values for both kunzea spp. and l. scoparium were not significantly different from each other but were significantly greater than that for all other species for which age-specific data was available. comparisons of basic stem-wood density values with published data basic stem-wood densities for ≥6-year-old specimen trees of l. scoparium, kunzea spp., melicytus ramiflorus, coprosma grandiflora, and weinmannia racemosa derived from natural stands indicative of advanced succession toward indigenous forest, fall within the range of these published values (fig. 3a). conversely, the mean basic stem-wood density values for trees <6-years old were either bordered on the lower limit of published means of older trees or significantly lower than published values (fig. 3b). marden et al. new zealand journal of forestry science (2021) 51:1 page 5 3peltzer, d.a., & payton, i.j. (2006). analysis of carbon monitoring system data for indigenous forests and shrublands collected in 2002/03. landcare research contract report lc0506/099. 55 p. marden et al. new zealand journal of forestry science (2021) 51:1 page 6 sp ec ie s lo ca ti on n o. tr ee s si te ty pe * h ei gh t r cd d b h a ge r² p r² p r² p r² p ku nz ea s pp . tu ra ng i 22 r s 0. 12 0 <0 .0 01 0. 20 1 <0 .0 01 0. 06 7 0. 00 1 0. 00 4 0. 39 2 w ai m at a 32 r s to la ga b ay 13 r s d un ed in 11 r s w ai ta ke re 6 r s co at br id ge 3 r s lo ng g ul ly 5 r s r iv er sd al e 21 r s ey re w el l 2 r s n ik au v al le y 56 r s le pt os pe rm um s co pa ri um tu ra ng i 24 r s 0. 20 9 0. 00 3 0. 17 9 0. 00 6 0. 16 6 0. 01 0 0. 13 4 0. 00 2 d un ed in 2 r s al ec tr yo n ex ce ls us g is bo rn e 13 pb 0. 08 1 0. 34 5 0. 24 7 0. 08 4 0. 21 6 0. 10 9 d ac ry ca rp us d ac ry di oi de s g is bo rn e 30 pb 0. 29 5 0. 00 2 0. 43 3 <0 .0 01 0. 40 2 <0 .0 01 po do ca rp us to ta ra g is bo rn e 9 pb 0. 02 6 0. 67 6 0. 37 9 0. 07 7 0. 3 0. 12 7 ag at hi s au st ra lis g is bo rn e 8 pb 0. 14 5 0. 35 2 0. 08 1 0. 49 5 0. 02 3 0. 72 1 d ac ry di um c up re ss in um g is bo rn e 14 pb 0. 00 6 0. 8 0. 10 9 0. 24 9 0. 42 6 0. 01 1 pr um no pi ty s fe rr ug in ea g is bo rn e 7 pb 0. 57 9 0. 04 7 0. 27 1 0. 23 1 0. 73 2 0. 01 4 vi te x lu ce ns g is bo rn e 8 pb 0. 44 6 0. 07 1 0. 30 1 0. 15 9 0. 27 8 0. 17 9 co pr os m a gr an di flo ra w el lin gt on 10 r f 0. 07 1 0. 48 7 0. 00 8 0. 80 6 0. 01 2 0. 75 9 0. 15 8 0. 25 8 m el ic yt us r am ifl or us w el lin gt on 30 r f 0. 00 6 0. 68 8 0. 00 9 0. 61 5 0 0. 96 1 0. 03 0 0. 09 7 w ei nm an ni a ra ce m os a w el lin gt on 10 r f 0. 18 3 0. 21 8 0. 12 7 0. 31 1 0. 17 6 0. 22 7 0. 00 2 0. 91 2 ta b le 2 : li ne ar r eg re ss io ns b et w ee n st em -w oo d de ns it y an d tr ee h ei gh t ( m ), ro ot c ol la r di am et er (r cd ; m m ), di am et er a t b re as t h ei gh t ( d b h ; m m ) a nd a ge ( ye ar s) fo r 12 o f n ew z ea la nd ’s in di ge no us s pe ci es . r eg re ss io ns fo r bo th k un ze a sp p. a nd l . s co pa ri um in cl ud ed d at a fr om c ol le ct iv e si te s. v al ue s in b ol d w er e st at is ti ca lly s ig ni fic an t (p <0 .0 5) . *r s = re ge ne ra ti ng s hr ub la nd , p b = p lo tba se d gr ow th tr ia l, r f = re ge ne ra ti ng fo re st marden et al. new zealand journal of forestry science (2021) 51:1 page 7 figure 2: stem-wood density values for: a) species <6-years old from plot-based growth trials; b) 6–10-year-old kunzea spp. and leptospermum scoparium, for 11–15-year-old kunzea spp. and melicytus ramiflorus, and for 16–20-year-old kunzea spp., melicytus ramiflorus, and coprosma grandiflora older than 6-years collected from regenerating shrubland or forest; c) 21–25-year-old melicytus ramiflorus and coprosma grandiflora, and for 26–30-year-old, 31–35-year-old, and 36–40-year-old kunzea spp. and l. scoparium collected from regenerating shrubland or forest ≥6-years old; and d) 46–50 and 51–70-year-old kunzea spp., l. scoparium and weinmannia racemosa collected from ≥6-years-old regenerating shrubland or forest. error bars represent the standard error of the mean. sample numbers shown at base of each grey bar. bars with different letters were significantly different (p<0.05). figure 3: comparison of: a) age-specific mean basic stem-wood density values for kunzea spp. and leptospermum scoparium ≥6-years old with densities sourced from published and unpublished literature. density data for trees of known age was analysed separate to that for trees where age was not specified (see table a2); and b) comparison of mean basic wood densities for trees <6-years old (grey bars) with mean densities of ≥6-yearold trees (dots) as sourced from published and unpublished literature (see table a2). for melicytus ramiflorus, coprosma grandiflora, and weinmannia racemosa, age-specific mean basic stem-wood density values (white bars in fig. 3a) are compared with mean densities (dots) sourced from published and unpublished literature where age was not specified. sample numbers shown at base of each bar. error bars represent the standard error of the mean. geographic distribution in kunzea spp. and l. scoparium basic stem-wood density while there is considerable variation in mean basic stem-wood values within naturally regenerating stands of kunzea spp. and l. scoparium, there is no supporting evidence that their density is significantly different between locations within either the north or south island of new zealand, between these islands, or between latitudes 35° to 46°s (fig. 4). for all remaining species there was insufficient basic stem-wood density data to support a similar statistical analysis. discussion basic wood density is one of the largest sources of variation in estimates of biomass and in the calculation of carbon sequestration (holdaway et al. 2014), yet these estimates are essential for new zealand’s international and national reporting of ghg budgets. to date, allometric functions have largely been based on limited stem-wood density data, and where species-specific and/or regional basic stem-wood density values are table 2: confusion matrix unavailable, congeneric values have been used instead, or, in their absence, the mean of all published values have been used (peltzer & payton unpublished data3, beets et al. unpublished data1). however, given that the earliest of the published values of basic stem wood density for merchantable timber trees were likely determined following the removal of the bark, a comparison with the means of all age-specific stem-wood densities, whether determined with the bark intact or after the removal of bark, might be considered invalid. nonetheless, as has been shown in this paper, the basic stem-wood densities of ≥6-year-old trees comprising natural stands indicative of advanced succession toward indigenous forest fall well-within the range of the earlier published values. furthermore, given the dearth of available data for many of the dominant and larger tree components of new zealand’s indigenous forests, the diversity of species, and the difficulty of accessing them in remote locations, where species-specific wood density values obtained for indigenous species harvested for timber exist, they serve as valuable reference points. marden et al. new zealand journal of forestry science (2021) 51:1 page 8 figure 4: mean basic stem-wood density values for kunzea spp. (17 locations) and leptospermum scoparium (4 locations) trees from naturally regenerating stands distributed throughout the north and south islands between latitudes 35° and 46°s. site locations are shown in figure 1. annotated site details are tabulated in table 1 and presented in greater detail in table a1. error bars represent the standard error of the mean. bars with different letters were significantly different (p<0.05). at the younger end of the age spectrum, for species typically associated with the early phase of shrubland regeneration, tall statured shrubland classes, and mixed species forests, insufficient basic wood density data together with simple field measurements are a limitation to the development of appropriate allometric functions for improving estimates of biomass and carbon stocks. furthermore, the use of different methods in the measurement of basic stem wood density (over bark versus under bark) has necessitated the development of equations that account for related variations in basic wood density in the calculation of tree biomass and changes in carbon stocks over time (hall et al. unpublished data2). however, until additional basic stem-wood density data can be collected for a sufficiently diverse range of specimen trees comprising a wide range of indigenous shrubland, forest types, and ages, the continued use of the mean of all available basic stem-wood density values will likely give the best estimate of stem carbon stocks. although the basic stem-wood densities of kunzea spp. and l. scoparium (both widely distributed shrubland species and a dominant component of regenerating forest on extensive areas of marginal hill country), are not significantly different from each other, they are both significantly higher than those of most of new zealand’s oldest indigenous forest and other shrubland species typically falling between 400 and 600 kg m3 (allen et al. 1992). therefore, using functions based on the stemwood density of either kunzea spp. or l. scoparium to scale tree volume, as yield or growth, to stem biomass, and from stem biomass to total biomass for different mixed-species indigenous forest communities is likely to overestimate total biomass. for kunzea spp., while there is variation in intraspecific mean basic stem-wood density values at different sites, there is no evidence from our data that stem-wood density is significantly different between the 17 locations where this species occurs as naturally regenerating shrubland. trends of increasing wood density values with decreased elevation (lassen & okkonen 1969) and increased temperature (filipescu et al. 2014) have been reported for new zealand-grown douglas-fir (kimberley et al. 2017), and for p. radiata basic wood density values show a gradual decrease from sea level to higher elevations, and from north to south (clifton 1990, palmer et al. 2013). for kunzea spp., however, while the results support a correlation between decreasing basic wood densities from sea level to higher elevations, there remains little evidence in support of wood densities decreasing north to south. other environmental influences, including intolerance to salt (esler & astridge 1974), soil fertility (cown & mcconchie 1981), soil moisture retention and stress (smale 1994), variations in genetics (de lange 2014) and rainfall distribution, are also likely to affect growth strategies (wardle 1969), tree form, and ultimately basic stem-wood density of many of new zealand’s indigenous shrubland and forest species. a site-by-site analysis of these factors was considered beyond the scope of this paper. mean basic stem-wood densities of trees <6-years old were either significantly lower, or at the lower end of published values (fig. 3b), but that within ca. ≥6 years after establishment, basic stem-wood density values approach that of older trees, and differs little thereafter (fig. 3a). we therefore concur with beets et al. (unpublished data1) on the strength of this relationship. differences in basic stem-wood density values between trees <6-years old and older are therefore likely to be primarily a function of their age. deng et al. (2014) found that stem-wood density of pinus massoniana stems was significantly influenced by tree age, relative heights, and social class, while beets et al. (2012) confirmed that stem-wood density at each relative height in older trees (age unspecified) was significantly higher than that of younger trees. iida (2012) found that low stem-wood density was linked to the propensity of some species to select for vertical growth (tall and thin stemmed with narrow and shallow canopies) and may therefore underlie the interspecific trade-off between effective height gain and a persistent life in the understorey (kohyama 1987, 1993; kohyama & hotta 1990). furthermore, relationships between stem-wood density and tree height may be related to differences in stand density. for example, l. scoparium <6-years old in densely-stocked, naturally reverting stands are tall and thin-stemmed and contrast markedly with the shorter and thicker-stemmed trees that develop when planted at low densities (marden et al. 2020). perhaps, as has been shown in studies across a range of conifer species (watt et al. 2011), the basic stem-wood density of l. scoparium would be expected to be lower in wider-spaced (planted) stands than in fully stocked stands that have reverted naturally. unfortunately, insufficient wood density data for l. scoparium <6-years old from naturally reverting stands precluded such an analysis. to reduce net greenhouse gas emissions, as required under the kyoto protocol (ministry of the environment 2010), a number of government-funded schemes (e.g. afforestation grant schemes (ministry for primary industries 2015a) and the permanent forest sink initiative (ministry for primary industries 2015b) have been introduced to facilitate natural regeneration of shrubland, and the planting of new areas of forest (exotic and indigenous). together with the recently announced government goal to plant one billion trees over the next 10 years (1 bt programme) (ministry for primary industries 2018), ca 1.45 million ha of steep, erosionprone pastoral hill country considered marginal for long-term agriculture will be targeted for transitioning to a permanent indigenous shrubland or forest (trotter et al. 2005). in such high-risk areas woody indigenous shrubland largely comprising kunzea spp. and l. scoparium has in the past played a significant role in mitigating erosion (marden & rowan 1993; ministry for primary industries 2015a, 2015b, 2016). together with increasing interest in high umf (unique mānuka factor) values associated with honey produced by l. scoparium, the establishment of low-density plantings averaging ca 825 to 1100 stems ha–1 (mcpherson & newstrommarden et al. new zealand journal of forestry science (2021) 51:1 page 9 lloyd 2018) is seen as an alternative and viable land management option for erosion prone steeplands (ministry for primary industries 2015c). using linear regression analyses based on mean wood density values measured for leptospermum scoparium <6-years old, new plantings at the recommended planting density, would by year 5 amass a forest carbon stock of 6.1 t co2 ha−1 (excluding coarse woody debris and fine litter on the forest floor) (marden & lambie 2016). alternatively, a mixed planting of successional broadleaved and conifer species would within the same time frame potentially amass a carbon stock of ~3.8 t co2 ha -1 (marden et al. 2018), while plantings consisting of a mix of early colonising seral species would amass a forest carbon stock of 8.8 t co2 ha -1 (unpublished). thus, the establishment of early colonising seral species on marginal land would amass an additional ~1 t co2 ha−1 over and above the 7.8 t co2 ha −1 estimated for the 5-year period from the date of planting (ministry for primary industries 2017). conversely, the planting of mixed indigenous broadleaved and coniferous species at the same density would amass ~4 t co2 ha -1 less, and plantings of leptospermum scoparium ~1.7 t co2 ha −1 less. by implication, to achieve a similar level of carbon stock for new plantings of broadleaved and conifer species within this time frame would require an increase in planting density to ~2000 stems ha-1 and for areas planted and managed for mānuka honey production, a planting density of 1200–1300 stems ha-1 would be required. these estimates of carbon stocks are however based on only a few studies of indigenous species that comprise the many shrubland and forest communities present within new zealand. with the pending conversion of extensive areas of former pastoral land to indigenous shrubland and forest through passive reversion, and by planting, therein lies an opportunity to validate and/or improve the accuracy of current estimates of biomass and carbon stocks during their early growth period, and for a wider range of species, by developing further allometric functions based on species-specific, basic stem-wood density values. conclusions this study presents an analysis of a significant database of previously unpublished basic wood-density values collected for a range of new zealand’s indigenous shrubland and forest species of varying age, and from sites located throughout both north and south islands. the findings indicate that for the most geographically widespread shrubland species, kunzea spp., differences in local site factors may affect tree parameters including basic wood density to a greater extent than wide differences in latitude within the normal growing range of the species. the data do however support trends showing that basic wood density values increase with decreased elevation, and increased temperature and where local data are available its use would improve the accuracy of biomass estimates both locally and nationally. insufficient site-specific information precludes further comment on other factors (e.g. soil fertility, plant spacing) that likely contribute to variability in basic stem-wood density values. for each of the species <6-years old for which basic stem-wood densities were collected, their mean values were significantly lower, or at the lower end of published values for trees ≥6-years old after which basic stemwood density values remain unchanged. age-specific basic stem-wood density data is scarce for shrubland communities dominated by mixed softwood species that comprise 90% of the national live tree biomass stock. furthermore, as their stem-wood density is considerably lower than for hardwood species, additional stem-wood density data are needed for use in combination with species-abundance information from lucas plots to update allometric functions applicable to areas of naturally reverting shrubland and to areas of former pastoral land pending their conversion to indigenous shrubland. as shown for the few indigenous species for which biomass and/or wood density data has been collected, at a planting density of 1000 stems ha-1, early colonising seral species would within 5-years amass a higher carbon stock of 8.8 t co2 ha −1 than would plantings of leptospermum scoparium ~6.1 t co2 ha −1 or a mixedspecies planting of indigenous broadleaved and coniferous species ~3.8 t co2 ha -1. to account for the variability in densities between outer-wood (and bark) and inner-wood with tree height, estimates of the mean density of whole stems will require the collection of stem-wood data from discs at intervals along the stem, as opposed to just breast height or by coring. list of abbreviations dbh diameter at breast height bh breast height 1bt one billion trees programme ets emission trading scheme ghg greenhouse gas lsd least significant difference rcd root collar diameter wsg wood specific gravity competing interests the authors declare that they have no competing interests. authors’ contributions mm was the primary author. sl compiled the data into spreadsheets and completed the statistical analyses. lb contributed data. all authors read and approved the manuscript. acknowledgements we acknowledge the support of the tairāwhiti polytechnic rural studies unit, gisborne, on whose land the plot-based softwood and hardwood plant marden et al. new zealand journal of forestry science (2021) 51:1 page 10 trial was located and to other landowners for allowing access to their respective properties at the time these studies were undertaken. we thank interns claire butty (france), sandra viel (germany), kaisa valkonen (finland), and landcare research colleague’s dr chris phillips, alex watson, richard hemming and scott bartlam for assistance with data collection. hawke’s bay regional council provided stevie and jack smidt to assist with data collection at lake tutira. john dando and ted pinkney assisted with the collection of discs and growthring counts. graphics were drawn by nic faville. anne austin edited the script and gis support was provided by anne sutherland of landcare research, nz, ltd. this paper was reviewed by dr mark smale, thanks also to the anonymous external reviewers for their valuable comment. over the years, research has been supported by funding from the ministry of business, innovation and employment, the sustainable land use research initiative (sluri) to plant and food, the ministry for primary industries, and the landcare research capability fund. availability of data and materials please contact the corresponding author for data requests. references allen, r.b., partridge, t.r., lee, w.g., efford, m. (1992). ecology of kunzea ericoides (a. rich.) j. thompson (kānuka) in east otago, new zealand. new zealand journal of botany, 30, 135-149. https://doi.org/10. 1080/0028825x.1992.10412894 beets, p.n., kimberley, m.o., oliver, g.r., pearce, s.h., graham, j.d., brandon, a. (2012). allometric equations for estimating carbon stocks in natural forest in new zealand. forests, 3, 818-839. https:// doi.org/10.3390/f3030818 beets, p.n., kimberley, m.o., paul, t.s.h., oliver, g.r., pearce, s.h., buswell, j.m. (2014). the inventory of carbon stocks in new zealand’s post-1989 natural forest for reporting under the kyoto protocol. forests, 5(9), 2230-2252. https://doi. org/10.3390/f5092230 bier, h. (1983). the strength properties of small clear specimens of new zealand-grown timber. [forest research institute bulletin 41]. rotorua, new zealand: new zealand forest research institute/ new zealand forest service, 29 p. bier, h. & r.a.j. britton (1999). strength properties of small clear specimens of new zealand-grown timbers. [forest research institute bulletin 41 (revised)]. rotorua, new zealand: new zealand forest research institute/new zealand forest service, 27 p. carswell, f.e., burrows, l.e., hall. g.m.j., mason, n.w.h., allen, r.b. (2012). carbon and plant diversity gain during 200 years of woody succession in lowland new zealand. new zealand journal of ecology, 36(2), 191-202. chave j., condit, r., aguilar, s., hernandez, a., lao, s., perez, r. (2004). error propagation and scaling for tropical forest biomass estimates. philosophical transactions of the royal society of london series b 359: 409-420. https://doi.org/10.1098/ rstb.2003.1425 chave, j. (2005). measuring wood density for tropical forest trees. a field manual. 7 p. http://www. ra i n f o r. n e t / u p l o a d / m a n u a l s e n g l i s h / w o o d _ density_english%5b1%5d.pdf clifton, n.c. (1990). new zealand timbers exotic and indigenous. the complete guide. wellington, new zealand: government printing office. 170 p. cornelissen, j.h.c., lavorel, s., garnier, e., díaz, s., buchmann, n., gurvich, d.e., reich, p.b., ter steege, h., morgan, h.d., van der heijden, m.g.a., pausas, j.g., poorter, h. (2003). a handbook of protocols for standardised and easy measurement of plant functional traits worldwide. australian journal of botany, 51, 335-380. https://doi.org/10.1071/ bt02124 cown, d.j., & mcconchie, d.l. (1981). effects of thinning and fertiliser application on wood properties of pinus radiata. new zealand journal of forestry science, 11, 79-91. dale, m.j. (2013). evaluation of methods for quantifying carbon storage of urban trees in new zealand. auckland, new zealand: university of auckland. http://itreetools.org/documents/36/evaluation_ of_methods_for_quantifying_carbon_storage_of_ urban_trees_in_new_zealand_dale2013.pdf de lange, p.j. (2014). a revision of the new zealand kunzea ericoides (myrtaceae) complex. phytokeys, 40, 1-185. https://doi.org/10.3897/ phytokeys.40.7973 deng, x., zhang, l., lei, p., xiang, w., yan, w. (2014). variations of wood basic density with tree age and social classes in the axial direction within pinus massoniana stems in southern china. annals of forest science, 71, 505-516. https://doi. org/10.1007/s13595-013-0356-y entrican, a.r., ward, w.c., reid, js. (1951). the physical and mechanical properties of the principal indigenous woods of new zealand. wellington: new zealand forest service, government printer, 83 p. esler, a.e., astridge, s.j. (1974). tea tree (leptospermum) communities of the waitakere range, auckland, new zealand. new zealand journal of botany, 12, 485-501. https://doi.org/10.1080/002882 5x.1974.10428635 marden et al. new zealand journal of forestry science (2021) 51:1 page 11 https://doi.org/10.1080/0028825x.1992.10412894 https://doi.org/10.1080/0028825x.1992.10412894 https://doi.org/10.3390/f3030818 https://doi.org/10.3390/f3030818 https://doi.org/10.3390/f5092230 https://doi.org/10.3390/f5092230 https://doi.org/10.1098/rstb.2003.1425 https://doi.org/10.1098/rstb.2003.1425 http://www.rainfor.net/upload/manualsenglish/wood_density_english%5b1%5d.pdf http://www.rainfor.net/upload/manualsenglish/wood_density_english%5b1%5d.pdf http://www.rainfor.net/upload/manualsenglish/wood_density_english%5b1%5d.pdf https://doi.org/10.1071/bt02124 https://doi.org/10.1071/bt02124 http://itreetools.org/documents/36/evaluation_of_methods_for_quantifying_carbon_storage_of_urban_trees_in_new_zealand_dale2013.pdf http://itreetools.org/documents/36/evaluation_of_methods_for_quantifying_carbon_storage_of_urban_trees_in_new_zealand_dale2013.pdf http://itreetools.org/documents/36/evaluation_of_methods_for_quantifying_carbon_storage_of_urban_trees_in_new_zealand_dale2013.pdf https://doi.org/10.3897/phytokeys.40.7973 https://doi.org/10.3897/phytokeys.40.7973 https://doi.org/10.1007/s13595-013-0356-y https://doi.org/10.1007/s13595-013-0356-y https://doi.org/10.1080/0028825x.1974.10428635 https://doi.org/10.1080/0028825x.1974.10428635 filipescu, c.n., lowell, e.c., koppenaal, r., mitchell, a.k. (2014). modeling regional and climatic variation in wood density and ring width in intensively managed douglas-fir. canadian journal of forestry research, 44, 3, 48-53. https://doi. org/10.1139/cjfr-2013-0275 geological map of new zealand (1967). scale 1:2 000 000. wellington, new zealand: new zealand geological survey/department of scientific and industrial research,. harris, j.m. (1986). wood properties of important new zealand commercial tree species. in h. levack (ed.), forestry handbook. (pp. 112-114). wellington, new zealand: new zealand institute of foresters (inc.) hayward, b.w. (1983). sheet q 11 waitakere. geological map of new zealand 1:50 000. map (1 sheet) and notes (28 p). wellington, new zealand: new zealand department of scientific and industrial research. hessell, j.w.d. (1980). the climate and weather of the gisborne region. new zealand meteorological service [miscellaneous publication 115(8)]. wellington, new zealand: government printer. hewitt, a.e. (2010). new zealand soil classification. landcare research science series no. 1 (3rd ed.). lincoln, new zealand: manaaki whenua press hinds, h.v., reid, j.s. (1957). forest trees and timbers of new zealand. [new zealand forest service bulletin, 12]. 221 p. https://cdm20044.contentdm.oclc. org/digital/api/collection/p20044coll6/id/421/ download holdaway, r.j., mcneill, s.j., mason, n.w., carswell, f.e. (2014). propagating uncertainty in plot-based estimates of forest carbon stock and carbon stock change. ecosystems, 17, 627-640. https://doi. org/10.1007/s10021-014-9749-5 iida, y., poorter, l., sterck, f.j., kassim, a.r., kubo, t., potts, m.d., kohyama, t.s. (2012). wood density explains architectural differentiation across 145 co-occurring tropical tree species. functional ecology, 26, 274-282. https://doi.org/10.1111/ j.1365-2435.2011.01921.x jessen, m.r., crippen, t.f., page, m.j., rijkse, m.c., harmsworth, g.r., mcleod, m. (1999). land use capability classification of the gisborne-east coast region: a report to accompany the second edition new zealand resource inventory. [landcare research science series 21] lincoln, new zealand: manaaki whenua press. kelliher, f.m., marden, m., watson, a.j., arulchelvam, i.m. (1995). estimating the risk of landsliding using historical extreme river flood data (note). journal of hydrology (nz), 33(2), 123-129. kimberley, m.o., mckinley, r.b., cown, d.j., moore, j.r. (2017). modelling the variation in wood density of new zealand-grown douglas-fir. new zealand journal of forestry science, 47: 15. https://doi. org/10.1186/s40490-017-0096-0 kingma, j.t. (1965). geological map of new zealand 1:250 000 sheet 6, east cape (1st edition). wellington, new zealand: department of scientific and industrial research. kingma, j.t. (1967). geological map of new zealand 1:250 000 sheet 12, wellington (1st edition). wellington, new zealand: department of scientific and industrial research. kirk, t. (1889). the forest flora of new zealand. wellington, new zealand: government printer. kohyama, t. (1987). significance of architecture and allometry in saplings. functional ecology, 1, 399404 https://doi.org/10.2307/2389797 kohyama, t. (1993). size-structured tree populations in gap-dynamic forest the forest architecture hypothesis for the stable coexistence of species. journal of ecology, 81, 131-143. https://doi. org/10.2307/2261230 kohyama, t. & hotta, m. (1990). significance of allometry in tropical saplings. functional ecology, 4, 515-521. https://doi.org/10.2307/2389319 lassen, l.e., & okkonen, e.a. (1969). effect of rainfall and elevation on specific gravity of coast douglas-fir. wood and fiber science, 1, 227-235. marden, m., & lambie, s. (2015). plot-based growth performance of space-planted mānuka (leptospermum scoparium) on marginal land, and vulnerability to erosion. [ministry for primary industries technical report no: 2015/19] https:// eastcoastmanuka.files.wordpress.com/2015/08/ lcr-final-report-on-manuka-trials-on-erodiblehill-country.pdf marden, m., & lambie, s. (2016). plot-based growth performance of space-planted mānuka (leptospermum scoparium) on marginal land, and vulnerability to erosion. [ministry for primary industries technical report no: 2016/20] https:// www.mpi.govt.nz/dmsdocument/13915/direct marden, m., lambie, s., phillips, c. (2018). biomass and root attributes of eight of new zealand’s most common indigenous evergreen conifer and broadleaved forest species during the first 5 years of establishment. new zealand journal of forestry science, 48: 9. https://doi.org/10.1186/s40490018-0113-y marden, m., lambie, s., phillips, c. (2020). potential effectiveness of low-density plantings of mānuka (leptospermum scoparium) as an erosion mitigation strategy in steeplands, northern hawke’s bay, new zealand new zealand journal of forestry science, 50: 10. https://doi.org/10.33494/ nzjfs502020x82x marden et al. new zealand journal of forestry science (2021) 51:1 page 12 https://doi.org/10.1139/cjfr-2013-0275 https://doi.org/10.1139/cjfr-2013-0275 https://cdm20044.contentdm.oclc.org/digital/api/collection/p20044coll6/id/421/download https://cdm20044.contentdm.oclc.org/digital/api/collection/p20044coll6/id/421/download https://cdm20044.contentdm.oclc.org/digital/api/collection/p20044coll6/id/421/download https://doi.org/10.1007/s10021-014-9749-5 https://doi.org/10.1007/s10021-014-9749-5 https://doi.org/10.1111/j.1365-2435.2011.01921.x https://doi.org/10.1111/j.1365-2435.2011.01921.x https://doi.org/10.1186/s40490-017-0096-0 https://doi.org/10.1186/s40490-017-0096-0 https://doi.org/10.2307/2389797 https://doi.org/10.2307/2261230 https://doi.org/10.2307/2261230 https://doi.org/10.2307/2389319 https://eastcoastmanuka.files.wordpress.com/2015/08/lcr-final-report-on-manuka-trials-on-erodible-hill-country.pdf https://eastcoastmanuka.files.wordpress.com/2015/08/lcr-final-report-on-manuka-trials-on-erodible-hill-country.pdf https://eastcoastmanuka.files.wordpress.com/2015/08/lcr-final-report-on-manuka-trials-on-erodible-hill-country.pdf https://eastcoastmanuka.files.wordpress.com/2015/08/lcr-final-report-on-manuka-trials-on-erodible-hill-country.pdf https://www.mpi.govt.nz/dmsdocument/13915/direct https://www.mpi.govt.nz/dmsdocument/13915/direct https://doi.org/10.1186/s40490-018-0113-y https://doi.org/10.1186/s40490-018-0113-y https://doi.org/10.33494/nzjfs502020x82x https://doi.org/10.33494/nzjfs502020x82x marden, m, rowan, d. (1993). protective value of vegetation on tertiary terrain before and during cyclone bola, east coast, north island, new zealand. new zealand journal of forestry science, 23(3), 255-263. mcpherson, a., & newstrom-lloyd, l. (2018). handbook for manuka plantations and farms: profitable and sustainable manuka honey harvesting. 34 p. ministry for primary industries, wellington, new zealand. www.treesforbeesnz.org ministry for primary industries (2015a). a guide to the afforestation grant scheme. wellington: ministry for primary industries http://www.mpi.govt.nz/ funding-andprogrammes/forestry/afforestationgrant-scheme/ accessed 21 may 2017. ministry for primary industries (2015b). a guide to the permanent forest sink initiative. wellington: ministry for primary industries http://www. mpi.govt.nz/funding-and-programmes/forestry/ permanent-forest-sink-initiative/afforestation accessed 2 aug 2016. ministry for primary industries. (2015c). high performance mānuka plantations. http:// www.mpi.govt .nz/funding-and-programmes/ primary-growth-partnership/primary-growthpartnership-programmes/high-performancemanuka-plantations/ accessed 10 june 2015. ministry for primary industries (2016). a guide to the erosion control funding programme (east coast). wellington: ministry for primary industries http:// www.mpi.govt .nz/funding-and-programmes/ forestry/erosion-control-fundingprogramme/ accessed 2 aug 2016. ministry for primary industries (2017). a guide to carbon look-up tables for forestry in the emissions trading scheme. wellington: ministry for primary industries http://www.mpi.govt.nz/growing-andproducing/forestry/forestry-in-theemmissionstrading-scheme/emissions-returns/ accessed 11 sept 2017. ministry for primary industries (2018). the one billion trees programme: our future, our billion trees. http://www.teururakau.govt.nz ministry of the environment. (2010). the kyoto protocol. h t t p : / / w w w. m f e . g o v t . n z / i s s u e s / c l i m a t e / international/kyoto-protocol.html new zealand geological survey (1972). south island (first edition) geological map of new zealand 1:1 000,000. wellington, new zealand: department of scientific and industrial research. new zealand meteorological service. (1966). summaries of climatological observations at new zealand stations to 1960. new zealand meteorological service miscellaneous publications, 122, 59 p. new zealand meteorological service. (1973). rainfall normals for new zealand 1941-1970. [miscellaneous publications, no. 145]. wellington, new zealand: meteorological service. new zealand meteorological service. (1984). meteorological observations for 1984. [miscellaneous publications, no. 109]. wellington, new zealand: meteorological service. palmer, d.j., kimberley, m.o., cown, d.j., mckinley, r.b. (2013). assessing prediction accuracy in a regression kriging surface of pinus radiata outerwood density across new zealand. forest ecology and management, 308, 9-16. https://doi. org/10.1016/j.foreco.2013.07.024 scott, n.a., white, j.d., townsend, j.a., whitehead, d., leathwick, j.r., hall, g.m.j., marden, m., rogers, g.n.d., watson, a.j., whaley, p.t. (2000). carbon and nitrogen distribution and accumulation in a new zealand scrubland ecosystem. canadian journal of forest research-revue canadienne de recherche forestiere, 30(8), 1246-1255. https:// doi.org/10.1139/x00-048 schwendenmann, l., & mitchell, n.d. (2014). carbon accumulation by native trees and soil in an urban park, auckland. new zealand journal of ecology, 38(2), 213-220. smale, m.c. (1994). structure and dynamics of kanuka (kunzea ericoides var. ericoides) heaths on sand dunes in bay of plenty, new zealand. new zealand journal of botany, 32, 441-452. https://doi.org/10. 1080/0028825x.1994.10412931 tomlinson, p.r., & leslie, d.m. (1978). soils of dunedin city and environs, new zealand. [new zealand soil survey report 37]. wellington, new zealand: soil bureau. trotter, c., tate, k., scott, n., townsend, j., wilde, h., lambie, s., marden, m., pinkney, t. (2005). afforestation/reforestation of new zealand marginal pasture lands by indigenous shrublands: the potential for kyoto forest sinks. annals of forestry science, 62, 865-871. https://doi. org/10.1051/forest:2005077 wardle, p. (1969). biological flora of new zealand: 4. phyllocladus alpinus hook f. (podocarpaceae) mountain toatoa, celery pine. new zealand journal of botany, 7, 76-95. https://doi.org/10.1080/0028 825x.1969.10429103 wardle, p. (1991). vegetation of new zealand. cambridge uk: cambridge university press. watson, a.j., marden m., rowan, d. (1994). tree species performance and slope stability. in: d.h. barker (ed.), proceedings of the institute of civil engineers conference, ‘vegetation and slopes stabilisation, protection and ecology’, university museum, oxford, 29-30 september 1994. pp.161-170. https://doi. org/10.1680/vasspae.20313.0018 marden et al. new zealand journal of forestry science (2021) 51:1 page 13 http://www.treesforbeesnz.org http://www.mpi.govt.nz/funding-andprogrammes/forestry/afforestation-grant-scheme/ http://www.mpi.govt.nz/funding-andprogrammes/forestry/afforestation-grant-scheme/ http://www.mpi.govt.nz/funding-andprogrammes/forestry/afforestation-grant-scheme/ http://www.mpi.govt.nz/funding-and-programmes/forestry/permanent-forest-sink-initiative/afforestation http://www.mpi.govt.nz/funding-and-programmes/forestry/permanent-forest-sink-initiative/afforestation http://www.mpi.govt.nz/funding-and-programmes/forestry/permanent-forest-sink-initiative/afforestation http://www.mpi.govt.nz/funding-and-programmes/primary-growth-partnership/primary-growth-partnership-programmes/high-performance-manuka-plantations/ http://www.mpi.govt.nz/funding-and-programmes/primary-growth-partnership/primary-growth-partnership-programmes/high-performance-manuka-plantations/ http://www.mpi.govt.nz/funding-and-programmes/primary-growth-partnership/primary-growth-partnership-programmes/high-performance-manuka-plantations/ http://www.mpi.govt.nz/funding-and-programmes/primary-growth-partnership/primary-growth-partnership-programmes/high-performance-manuka-plantations/ http://www.mpi.govt.nz/funding-and-programmes/primary-growth-partnership/primary-growth-partnership-programmes/high-performance-manuka-plantations/ http://www.mpi.govt.nz/funding-and-programmes/forestry/erosion-control-fundingprogramme/ http://www.mpi.govt.nz/funding-and-programmes/forestry/erosion-control-fundingprogramme/ http://www.mpi.govt.nz/funding-and-programmes/forestry/erosion-control-fundingprogramme/ http://www.teururakau.govt.nz http://www.mfe.govt.nz/issues/climate/international/kyoto-protocol.html http://www.mfe.govt.nz/issues/climate/international/kyoto-protocol.html https://doi.org/10.1016/j.foreco.2013.07.024 https://doi.org/10.1016/j.foreco.2013.07.024 https://doi.org/10.1139/x00-048 https://doi.org/10.1139/x00-048 https://doi.org/10.1080/0028825x.1994.10412931 https://doi.org/10.1080/0028825x.1994.10412931 https://doi.org/10.1051/forest:2005077 https://doi.org/10.1051/forest:2005077 https://doi.org/10.1080/0028825x.1969.10429103 https://doi.org/10.1080/0028825x.1969.10429103 https://doi.org/10.1680/vasspae.20313.0018 https://doi.org/10.1680/vasspae.20313.0018 watt, m.s., zoric, b., kimberley, m.o., harrington, j. (2011). influence of stocking on radial and longitudinal variation in modulus of elasticity, microfibril angle, and density in a 24-year-old pinus radiata thinning trial. canadian journal of forest research, 41(7), 1422-1431. https://doi.org/10.1139/x11-070 williamson, g.b, wiemann, m.c. (2010). measuring wood specific gravity correctly. american journal of botany, 97(3), 519-524. https://doi.org/10.3732/ ajb.0900243 marden et al. new zealand journal of forestry science (2021) 51:1 page 14 https://doi.org/10.1139/x11-070 https://doi.org/10.3732/ajb.0900243 https://doi.org/10.3732/ajb.0900243 marden et al. new zealand journal of forestry science (2021) 51:1 page 15 appendix table a1: location and physical characteristics of 21 sample sites throughout new zealand. site 1: tautoro. 8 km south of kaikohe, northland (173° 50´ 13 15 e, 35° 28´ 52 00 s). regenerating leptospermum scoparium (table a4) stand on gently, southwest-facing slope 100-140m above sea level. bedrock consists of greywacke argillites and sandstones (geological map of new zealand, 1967). soils are deeply weathered and classified as altic soils (hewitt, 2010). site 2: waitakere range. within the waitakere range (174° 35´ 14 42 e, 37° 00´ 10 17 s) stem-wood discs were collected from naturally reverting stands of well-established kunzea spp. (table a3) and l. scoparium (table a4) of unknown age. at an elevation of ca. 40 m, slopes ranged between 0 and 35°. the geology comprises volcanic andesitic lava, conglomerates, and breccia of the waitemata and waitakere groups of early miocene (late otaian-middle altonian) age. soils comprise weathered volcanics consisting of yellow-brown granular clay grading to a compact yellow brown to brown subsoil (hayward, 1983). the climate is relatively mild and moist with annual rainfall of ca 1250 mm increasing to over 2000 mm in the higher central parts at elevations of ca 460 m (new zealand meteorological service, 1966). site 3: nikau valley. 8km south of whakatane, bay of plenty (176° 58´ 23 85 e, 38° 01´ 25 27 s). managed, dense, east-facing kunzea robusta (table a3) stands of all ages 40–100 m above sea level. bedrock consists of undifferentiated greywacke (geological map of new zealand, 1967). pumice soils consisting of tarawera and whakatane ash overly bedrock on rolling hill country (hewitt, 2010). sites 4–6: tolaga bay and site 7: waimata valley. sites 2–4 are located approximately 8 km inland of tolaga bay (178˚ 12΄ 19 29 e, 38˚ 20΄42 58 s), and site 5 is located 15 km inland of gisborne city (178° 03΄ 13 66 e, 38˚ 28΄ 33 84 s). each site represents an even-canopied stand of naturally reverting kunzea robusta (table a3) at a different stage of development, the age of which was determined by the history of vegetation clearance, and verified by growth ring counts (watson et al., 1994). the tolaga bay sites occur on slopes between 23° and 32°, have a nw (300o) to ne (60o) aspect, and are at elevations between ca 64 m and 160 m above sea level. the waimata site is on a sw aspect at an elevation of 207 m. the underlying bedrock at these sites consists of pliocene-age calcareous sandy siltstones with banded sandstones and thick tuffaceous horizons (kingma, 1965). soils are a stony colluvium varying from orthic recent soils and their intergrades to brown soils (on well-drained sites) and gley soils (on poorly drained sites) typical of slopes being eroded or has received sediment mainly as a result of slope processes (hewitt, 2010). the climate is warm temperate maritime, with moist summers and cool wet winters. mean annual rainfall varies from about 700 mm at the coast to 2500 mm at higher elevations (new zealand meteorological service, 1973). lengthy periods of little or no rainfall are common during january to april (mid-summer to late autumn). this region has a history of extreme rainfall events (kelliher et al., 1995), often associated with storms of tropical origin (e.g. cyclone bola in 1988). site 8: gisborne. five indigenous softwood (agathis australis, prumnopitys ferruginea, podocarpus totara, dacrycarpus dacrydioides, dacrydium cupressinum) and two hardwood species (alectryon excelsus and vitex lucens) (table a5) were established as a planting trial to establish their relative growth performance, above-and below-ground, over a 5-year period (marden et al., 2018). the trial site was located on a low-lying (5 m above sea level), even-surfaced alluvial terrace adjacent to the taraheru river, in gisborne city (178° 00´ 16 02 e, 38° 38´ 44 82 s). the soil is a naturally fertile, free draining, typic sandy brown soil of the te hapara soil series (hewitt, 2010) with no physical or chemical impediments. temperatures over summer average 23° c and over winter 12° c and mean annual rainfall is ca 1200 mm (hessell, 1980). site 9: turangi. stands of 25-, 35and 55-year-old kunzea spp. (table a3) and l. scoparium (table a4) were selected in tongariro national park near turangi township in the central north island (175° 47´ 11 53 e, 39° 09´ 19 20 s) at an elevation of 800 m, approaching the maximum elevation at which these species are found (scott et al., 2000). mean annual temperature is 11.1c°, and mean annual precipitation is ca 1610 mm. soils derived from a series of rhyolitic and andesitic volcanic eruptions are classified as podzolic orthic pumice soils of the rangipo series (hewitt, 2010). site 10: lake tutira. l. scoparium (table a4) was planted at lake tutira (176° 54´ 10 44 e, 39° 14´ 00 44 s) in 2011 and 2012 at a spacing (3 m × 3 m, ca 1100 stems ha–1) more typical of an exotic plantation forest. nine permanent sample plots (20 m × 20 m) were established in 2015 (marden & lambie, 2015 & 2016). the terrain is 7e3 (jessen et al., 1999) consisting of pliocene-age mudstone, sandstone, and limestone subjected to extreme shallow landsliding during storm events. slight tunnel gullying is also present. slopes are predominantly west facing, between 21° and 35⁰, and occur at an elevation of 200°–375 m. soils are typic immature pallic (hewitt, 2010). site 11: wainuiomata and cannons creek. this site consists of well-established indigenous hardwoods and lowland shrub communities dominated by mixed hardwood coprosma grandiflora, weinmannia racemosa, and melicytus ramiflorus (table a5) shrubs indicative of advanced succession progressing toward indigenous forest. three plots were installed (174° 57´ 19 75 e, 41° 17´ 45 29 s) on slopes ranging between 17° and 28°, with a southwest aspect between 200° and 240°, and at an elevation of ca 117 m. the geology consists of complexly deformed alternating dark grey marden et al. new zealand journal of forestry science (2021) 51:1 page 16 argillite and greywacke sandstone, rare limestone and minor spilitic lava of triassic age (kingma, 1967). soils are a stony colluvium derived from greywacke bedrock and vary from orthic recent soils and their intergrades to brown soils (on well-drained sites) and gley soils (on poorly-drained sites) typical of slopes being eroded or has received sediment mainly as a result of slope processes (hewitt, 2010). site 12: long gully. 6 km southwest of wellington (174° 40´ 55 30 e, 41° 18´ 34 82 s). regenerating wind shorn stands of kunzea amathicola (table a3) on south-facing slope 300-400 m above sea level. bedrock consists of alternating argillite and greywacke sandstone with rare limestone and volcanics (kingma, 1967). soils are a stony colluvium derived from greywacke bedrock and vary from orthic recent soils to brown soils and gley soils typical of slopes being eroded or has received sediment mainly as a result of slope processes (hewitt, 2010). site 13: riversdale. near white rock on the se coast of wairarapa (175° 25´ 53 49 e, 41° 30´ 57 74 s). wide range of kunzea robusta (table a3) stands at different stages of development on mainly southwest-facing slopes 60–200m above sea level. bedrock consists of greywacke-like dark grey muddy siltstone with minor conglomerates and spilitic lava (kingma, 1967). soils are a stony colluvium derived from greywacke bedrock and vary from orthic recent soils to brown soils and gley soils typical of slopes being eroded or has received sediment mainly as a result of slope processes (hewitt, 2010). site 14: coatbridge. 12 km west of renwick, marlborough (173° 39´ 23 16 e, 41° 29´ 08 99 s) dense regenerating kunzea spp. (table a3) and l. scoparium (table a4) on moderate to steep south facing slopes 200–300m above sea level. bedrock consists of metamorphosed sedimentary lithologies and volcanics (new zealand geological survey, 1972). soils are derived from greywacke bedrock and vary from brown soils to orthic recent and gley soils typical of slopes being eroded or has received sediment mainly as a result of slope processes (hewitt, 2010). site 15: long spur. 9 km south of tururumuri near the southeast coast of wairarapa (175° 32´ 09´ 01 e, 41° 27´ 22 12 s). dense regenerating stands of kunzea robusta (table a3) on slopes on a range of aspects 40–200 m above sea level. bedrock consists of graded bedded, fine-grained, sandstone and mudstone, minor conglomerates and volcanics (kingma, 1967). soils are a stony colluvium derived from greywacke bedrock and vary from orthic recent soils to brown soils and gley soils typical of slopes being eroded or has received sediment mainly as a result of slope processes (hewitt, 2010). site 16: peggioh. 10 km west of ward (174° 01´ 13 67 e, 41° 51´ 31 57 s). dense, regenerating kunzea robusta (table a3) and l. scoparium (table a4) stands 200–300 m above sea level. on south-facing slopes. bedrock consists of interbedded greywacke and argillite with minor volcanics, conglomerates, and rare limestone (new zealand geological survey, 1972). soils are a stony colluvium derived from greywacke bedrock and vary from brown soils to orthic recent and gley soils typical of slopes being eroded or has received sediment mainly as a result of slope processes (hewitt, 2010). site 17: shenandoah. 20 km south of murchison, buller (172° 15´ 05 30 e, 41° 53´ 36 00 s). regenerating kunzea ericoides (table a3) stand on west-facing slope 200–300 m above sea level. bedrock consists of mainly limestone and calcareous siltstone, local sandstone and coal measures (new zealand geological survey, 1972). soils are classed as brown and melanic soils (hewitt, 2010). site 18: avoca station 22 km south of cass, canterbury (171° 53´ 23 31e, 43° 11´ 49 51 s). regenerating stands of kunzea serotina (table a3) on north-facing slopes 420–540 m above sea level. bedrock consists of interbedded greywacke and argillite with minor volcanics, conglomerates, and rare limestone (new zealand geological survey, 1972). soils are a stony colluvium derived from greywacke bedrock and vary from orthic recent soils to brown soils and gley soils typical of slopes being eroded or has received sediment mainly as a result of slope processes (hewitt, 2010). site 19: eyrewell. 10km south of oxford and 6km north of waimakariri river, canterbury plains (172° 11´ 41 76 e, 43° 22´ 59 35 s). fenced remnant kunzea serotina (table a3) stand 200 m above sea level. flat floodplain, well drained postglacial alluvium and glacial outwash gravels (new zealand geological survey, 1972). soils are classed as stony brown soils (hewitt, 2010). site 20: hinewai. 5km east of akaroa, banks peninsula above otanerito bay (173° 02´ 18 74 e, 43° 49´ 02 85 s). wide range of kunzea robusta (table a3) stands at different stages of development on steep southeast-facing slopes 20–450 m above sea level. bedrock consists of basalt tuff and associated intrusive rocks (new zealand geological survey, 1972). soils are classed as melanic soils (hewitt, 2010). site 21: dunedin. this study site consists of a ca 130 ha mosaic of 2–70 year old stands of kunzea robusta (table a3) and l. scoparium (table a4) forest located on the western side of the purakanui inlet catchment (170° 36´ 37 14 e, 45° 45´ 11 19 s), 16 km north of dunedin. soils are described as brown granular loams and clays derived from loess, basalt and phonolite (tomlinson & leslie, 1978). slopes are ne-e facing between 2° and 35°, and at 200–300 m elevation. annual rainfall is about 680 mm (new zealand meteorological service, 1984). table a1 continued... marden et al. new zealand journal of forestry science (2021) 51:1 page 17 table a2: non-age specific basic mean wood density values for old growth indigenous forest and shrubland species tree species wood density (kg m–3) location/number/age reference leptospermum scoparium (mānuka) 695-714 woodhill, cms plot bb114 payton (pers. comm.) 906-1042 kirk (1989) 892 cms plots, n=1573 peltzer & payton unpublished dataa 720 puketi forest, northland jager et al. (2014) kunzea spp. (kānuka) 671-720 camp creek, woodhill payton (pers. comm.) 757 clifton (1990), bier (1983) 642 akaroa, n=40, <50 years carswell et al. (2012) 680 auckland schwendenmann (2014) 772 cms plots, n=1708 peltzer & payton unpublished dataa alectryon excelsus (titoki) 622 woodhill payton (pers. comm.) 837 cms plots, n=4 peltzer & payton unpublished dataa 854 bier & britton (1999) dacrycarpus dacrydioides (kahikatea) 465 gray county, n=5 kirk (1889) 410 gray country, 152-310 years entrican (1951) 390 hinds & reid (1957) in harris (1986) 450 clifton (1990) 420 cms plots, n=118 peltzer & payton unpublished dataa 440 maungatautari (n=1) beets et al. unpublished datab 389 whirinaki (n=20) beets et al. unpublished datab 410 puketi forest, northland jager et al. (2014) 429 bier & britton (1999) podocarpus totara (totara) 443 14-110 years steward (pers. comm.) 559 kirk (1889) 430 taupo county, n=5, 408612 years entrican (1951) 410 hinds & reid (1957) in harris (1986) 480 taupo county clifton (1990) 480 cms plots, n=80 peltzer & payton unpublished dataa 383-407 whirinaki, n-14 beets et al. unpublished datab 435 bier & britton (1999) agathis australis (kauri) 449 10-69 years steward (pers. comm.) 489 126-240 years steward (pers. comm.) 498-595 kirk (1889) 520 waitamata county, n=5 entrican (1951) 480 waitamata hinds and reid (1957) in harris (1986) 520 cms plots, n=1 peltzer & payton unpublished dataa 470 puketi forest, northland jager et al. (2014) 441 taranaki, n=20 beets et al. unpublished datab 495 bier & britton (1999) marden et al. new zealand journal of forestry science (2021) 51:1 page 18 table a2: continued... * note: mean basic wood density values from beets et al. unpublished datab are from breast height outer wood at 5–15 cm (measured from bark). bbeets, p.n., oliver, g.r, kimberley., m.o, pearce, s.h. (2008). allometric functions for estimating above ground carbon in native forest trees, shrubs and ferns. scion report 12679 prepared for the ministry for the environment 63 p. tree species wood density (kg m–3) location/number/age reference dacrydium cupressinum (rimu) 575 payton (pers. comm.) 550-644 kirk (1889) 520 raurimu, kaitieke county, n=5, 330-443 years entrican (1951) 490 central north island hinds & reid (1957) in harris (1986) 560 bier (1983) 595 clifton (1990) 558 cms plots, n=456 peltzer & payton unpublished dataa 461-466 whirinaki, n=30 beets et al. unpublished datab 460 puketi forest, northland jager et al. (2014) 504 bier & britton (1999) prumnopitys ferruginea (miro) 787 raurimu, kaitieke county, n=5, 248-363 yrs kirk (1889) 520 kaitieke county entrican (1951) 510 hinds & reid (1957) in harris (1986) 625 clifton (1990) 568 cms plots, n=151 peltzer & payton unpublished dataa 592 maungatautari, n=1 beets et al. unpublished datab 527-531 whirinaki, n=26 beets et al. unpublished datab 510 puketi forest, northland jager et al. (2014) vitex lucens (puriri) 573 auckland dale (2013) 633 cms plots, n=8 peltzer & payton unpublished dataa 730 puketi forest, northland jager et al. (2014) melicytus ramiflorus (mahoe) 396 maungatautari, n=6 beets et al. unpublished datab 585 cms plot, n=638 peltzer & payton unpublished dataa 445 woodhill forest payton (pers. comm.) 464 cms plot au146 payton (pers. comm.) weinmannia racemosa (kamahi) 484 maungatautari, n=21 beets et al. unpublished datab 619 cms plot, n=4175 peltzer & payton unpublished dataa 542 cms plot az118 payton (pers. comm.) 520 cms plot q171 payton (pers. comm.) 553 cms plot bf117 payton (pers. comm.) 572 bier & britton (1999) coprosma grandiflora (coprosma) 368 maungatautari, n=1 beets et al. unpublished datab 583 cms plot, n=208 peltzer & payton unpublished dataa marden et al. new zealand journal of forestry science (2021) 51:1 page 19 location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) riversdale 53 9.5 16.0 707 52 11.9 15.7 698 24 5.9 7.4 706 22 6.1 7.3 707 22 4.9 7.0 691 17 4.8 5.7 721 22 4.4 6.0 792 22 5.0 7.0 742 31 5.9 9.0 710 30 6.7 9.3 730 21 5.7 7.8 742 29 9.7 10.3 763 43 8.2 11.6 717 50 8.7 11.0 741 31 5.8 8.7 737 30 7.5 12.1 741 31 6.9 8.9 749 68 9.9 19.5 715 56 10.9 18.8 765 45 9.6 19.2 732 53 12.3 24.6 699 60 12.2 20.4 772 30 6.5 9.7 760 29 6.6 9.3 713 26 7.1 8.7 738 8 2.3 1.0 687 6 2.4 1.0 640 32 8.2 11.2 698 24 7.3 9.0 755 25 8.6 10.3 725 40 8.5 16.4 753 41 9.1 16.2 727 40 6.5 12.6 749 43 10.9 23.3 759 32.3 718 23 4.3 5.0 796 23 4.1 4.8 812 table a3: basic stem-wood densities, tree age, height, rcd and dbh of individual kunzea spp. from areas of natural regeneration at: riversdale (site 13), turangi (site 9), shenandoah (site 17), eyrewell (site 19), waimata (site 7), long gully (site 12), hinewai (site 20), tolaga bay (sites 4-6), dunedin (site 21), waitakere (site 2), avoca (site 18), coatbridge (site 14), peggioh (site 16), long spur (site 15), and nikau valley (site 3). location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) riversdale 20 4.4 5.1 802 20 3.6 3.8 815 21 4.6 4.2 783 70 8.8 23.3 756 742 shenandoah 661 673 691 689 743 694 676 799 700 716 eyrewell 42 11.0 759 46 9.8 642 turangi 53 6.6 53 43 650 41 5.6 30 26 670 89 101.4 113 98 650 80 8.4 87 77 640 76 9.1 68 58 710 69 9.0 57 48 670 105 12.0 186 141 680 61 7.6 69.5 58 678 67 9.4 100 83 765 28 6.3 65 48 712 34 5.4 42 37 682 29 5.5 44 35 681 35 7.1 97 80 711 37 6.3 122 70 667 63 6.6 43.5 43 622 77 8.4 76 65 721 27 4.0 38.5 38 658 34 5.1 68 45 628 28 7.0 29 26 665 32 6.5 53 36 662 42 9.1 82 73 748 50 9.6 121 110 768 location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) waimata 15 6.5 110 710 21 12.0 120 734 16 10.7 113 698 18 10.8 119 747 20 12.0 141 777 19 9.8 130 753 16 9.5 142 721 20 11.3 143 750 18 10.2 141 806 17 9.5 120 722 19 8.3 108 655 19 9.7 104 738 16 11.0 134 805 15 9.2 103 732 31 9.5 136 107 704 26 13.0 154 129 734 23 11.5 120 108 729 29 16.4 181 143 786 31 16.4 187 158 828 29 16.4 165 132 789 30 13.3 147 111 647 29 12.1 138 141 720 31 13.3 151 119 774 26 10.6 153 120 731 26 11.9 155 142 743 31 12.8 122 100 766 37 11.4 117 94 86 22 12.2 143 125 811 24 12.4 108 96 722 35 14.2 149 134 849 29 13.7 98 91 771 34 13.9 154 139 757 table a3: continued location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) tolaga bay 14 6.6 80 67 758 13 5.7 74 61 674 15 7.2 78 60 703 21 7.2 100 93 790 15 6.6 54 49 694 6 5.7 60 50 686 4 4.6 40 33 647 6 6.2 60 47 652 7 6.8 48 37 704 8 7.4 67 57 730 4 2.1 24 2.4 605 3 1.9 15 1.5 660 4 2.6 36 3.6 714 long gully 12 781 15 762 20 690 22 760 26 793 hinewai 13.5 747 6.5 699 dunedin 35 10.5 108 97 699 45 10.5 119 101 702 38 9.6 81 75 699 43 11.4 150 130 730 48 10.9 168 150 671 29 8.6 101 84 681 25 9.2 95 72 602 16 7.5 44 35 619 34 9.3 114 102 706 27 7.5 98 76 722 18 6.0 72 62 704 waitakere 75 18.7 325 247 693 59 14.7 222 177 724 37 8.2 72 62 748 35 8.7 1145 109 704 42 9.1 120 97 700 16 7.5 48 44 704 avoca station 748 811 737 770 834 748 742 829 778 751 marden et al. new zealand journal of forestry science (2021) 51:1 page 20 location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) coatbridge 15 758 15 697 12 782 peggioh 764 722 765 810 711 686 713 753 735 749 778 800 779 729 731 746 744 758 689 736 764 713 long spur 708 753 723 768 764 744 701 783 711 786 table a3: continued location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) nikau valley 56 15.5 26.9 725 43 12.6 14.3 711 40 12.8 13.2 739 47 11.4 13.4 734 40 12.6 12.8 737 39 11.6 12.1 736 44 13.7 14.5 724 51 11.5 20.1 673 62 12.4 20.5 694 13 3.7 2.7 603 13 3.2 2.7 624 11 3.7 2.5 662 50 13.7 27.7 731 26 7.8 15.5 677 27 8.8 15.2 670 44 12.5 15.9 696 78 16.0 43.2 732 46 12.5 14.2 735 48 12.6 15.5 765 10 4.2 3.5 629 10 3.5 4.6 687 13 8.5 9.9 687 14 8.2 10.0 687 29 12.3 10.0 658 25 11.5 9.2 592 15 7.2 11.4 602 31 9.6 8.8 683 72 16.9 35.5 612 11 3.4 2.6 608 11 3.4 3.2 668 9 3.5 2.6 626 7 3.2 2.5 604 70 14.0 31.0 703 69 12.1 31.5 714 70 12.6 29.5 690 37 11.5 14.4 755 44 13.0 15.5 673 32 12.5 14.8 721 marden et al. new zealand journal of forestry science (2021) 51:1 page 21 location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) nikau valley 46 11.9 20.8 709 47 13.8 19.3 710 42 5.8 6.0 701 13 4.3 5.2 668 11 1.2 661 7 1.5 667 9 2.3 637 6 7.0 5.6 698 12 6.8 5.9 682 13 7.0 6.3 653 16 45 5.4 664 12 4.5 5.0 629 11 4.0 5.0 669 40 8.2 10.5 581 29 8.2 10.9 630 30 8.7 10.7 706 28 7.8 9.7 772 36 10.9 10.4 757 29 8.8 10.5 710 table a3: continued marden et al. new zealand journal of forestry science (2021) 51:1 page 22 marden et al. new zealand journal of forestry science (2021) 51:1 page 23 table a4: basic stem-wood densities, tree age, height, rcd and dbh of individual leptospermum scoparium from areas of natural regeneration at turangi (site 9), dunedin (site 21), coatbridge (site 14), peggioh (site 16), tautoro (site 1), and from planted stands at lake tutira (site 10) location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) turangi 29 5.4 38 25 690 32 5.9 38 28 770 33 6.1 44 37 660 39 6.4 42 33 870 40 5.8 64 46 870 27 5.5 31 25 720 39 6.2 47 37 760 47 5.6 45 37 720 60 7.5 51 43 720 21 4.9 36 31 721 25 5.1 81 36 778 26 6.3 90 52 756 28 6.2 42 38 642 29 5.6 55 37 682 30 4.6 49 45 684 31 4.0 54 40 710 34 6.1 68 59 739 40 6.7 36 29 672 48 8.0 100 55 718 51 6.7 49 44 667 53 8.4 94 80 710 55 7.6 96 72 712 27 5.8 46 31 824 68 7.8 73 59 716 dunedin 10 4.0 34 26 665 11 4.5 42 30 680 lake tutira 3 2.9 8 6 680 3 2.1 10 6 690 3 2.1 3 1 610 3 2.2 3 3 650 3 2.8 5 1 590 3 2.5 5 700 3 2.0 6 5 650 4 1.8 5 4 660 4 2.3 7 4 680 6 4.2 15 33 697 6 4.8 96 42 681 6 2.9 36 7 634 6 2.3 33 6 724 location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) coatbridge 15 709 11 710 9 723 8 684 10 715 14 719 11 690 14 743 11 682 14 742 13 696 13 734 14 742 10 624 16 692 14 707 13 694 13 637 14 686 8 689 18 704 19 718 20 700 11 680 16 744 10 735 23 718 23 744 23 611 23 666 17 704 marden et al. new zealand journal of forestry science (2021) 51:1 page 24 table a4: continued location age (y) ht (m) rcd (mm) dbh (mm) wood density (kg m-3) peggioh 724 756 645 669 666 696 684 705 732 727 656 730 662 679 724 733 729 607 610 739 689 683 660 tautoro 656 780 662 679 724 733 729 607 610 739 689 683 660 marden et al. new zealand journal of forestry science (2021) 51:1 page 25 table a5: basic stem-wood densities, age, height, table a5: basic stem-wood densities, age, height, root collar diameter (rcd), and rcd), and diameter at breast height (dbh) dbh) of individual hardwood species from areas of natural regeneration at wainuiomata (site 11), and hardwood and of individual hardwood species from areas of natural regeneration at wainuiomata (site 11), and hardwood and softwood species from plot trials based at gisborne (site 8) softwood species from plot trials based at gisborne (site 8) species location age (y) height (m) rcd (mm) dbh (mm) wood density (kg m–3) melicytus ramiflorus wainuiomata 12 3.7 64 34 541 melicytus ramiflorus wainuiomata 14 4.7 67 55 571 melicytus ramiflorus wainuiomata 18 5.3 114 68 446 melicytus ramiflorus wainuiomata 12 5.3 101 47 484 melicytus ramiflorus wainuiomata 10 3.9 41 28 536 melicytus ramiflorus wainuiomata 13 3.6 56 37 556 melicytus ramiflorus wainuiomata 16 7.2 66 62 423 melicytus ramiflorus wainuiomata 18 7.0 125 106 453 melicytus ramiflorus wainuiomata 11 3.3 39 24 480 melicytus ramiflorus wainuiomata 9 2.9 39 26 490 melicytus ramiflorus wainuiomata 18 6.5 125 51.5 529 melicytus ramiflorus wainuiomata 25 5.4 103 65 533 melicytus ramiflorus wainuiomata 58 9.1 350 175 496 melicytus ramiflorus wainuiomata 51 8.1 258 150 476 melicytus ramiflorus wainuiomata 37 8.8 154 115 513 melicytus ramiflorus wainuiomata 35 7.1 133 92 509 melicytus ramiflorus wainuiomata 36 7.1 178 128 495 melicytus ramiflorus wainuiomata 27 6.5 198 128 519 melicytus ramiflorus wainuiomata 41 7.5 135 96 521 melicytus ramiflorus wainuiomata 23 6.1 137 90 480 melicytus ramiflorus wainuiomata 15 6.4 84 54 429 melicytus ramiflorus wainuiomata 19 5.4 143 82 467 melicytus ramiflorus wainuiomata 19 1.7 92 69 474 melicytus ramiflorus wainuiomata 14 4.5 117 47 430 melicytus ramiflorus wainuiomata 25 5.1 154 102 444 melicytus ramiflorus wainuiomata 17 — 165 65 405 melicytus ramiflorus wainuiomata 17 5.0 130 64 465 melicytus ramiflorus wainuiomata 5 — 47 11 457 melicytus ramiflorus wainuiomata 9 4.7 77 31 464 melicytus ramiflorus wainuiomata 15 4.2 70 45 495 coprosma grandiflora wainuiomata 20 — 128 83 476 coprosma grandiflora wainuiomata 21 6.1 130 88 433 coprosma grandiflora wainuiomata 23 6.1 121 61 460 coprosma grandiflora wainuiomata 17 7.2 106 40 493 coprosma grandiflora wainuiomata 23 6.5 107 68 485 coprosma grandiflora wainuiomata 20 6.0 124 55 465 coprosma grandiflora wainuiomata 17 6.8 109 57 442 coprosma grandiflora wainuiomata 23 5.8 71 53 512 coprosma grandiflora wainuiomata 19 4.0 45 27 426 coprosma grandiflora wainuiomata 18 6.6 92 71 412 marden et al. new zealand journal of forestry science (2021) 51:1 page 26 table a5: continuedtable a5: continued species location age (y) height (m) rcd (mm) dbh (mm) wood density (kg m–3) weinmannia racemosa wainuiomata 44 6.9 152 115 544 weinmannia racemosa wainuiomata 38 7.6 120 97 576 weinmannia racemosa wainuiomata 26 6.2 89 67 573 weinmannia racemosa wainuiomata 69 10.2 220 143 543 weinmannia racemosa wainuiomata 47 10.3 160 113 551 weinmannia racemosa wainuiomata 30 7.7 107 80 548 weinmannia racemosa wainuiomata 61 6.8 150 105 614 weinmannia racemosa wainuiomata 51 8.9 215 108 541 weinmannia racemosa wainuiomata 63 10.2 158 128 538 weinmannia racemosa wainuiomata 46 7.7 159 120 509 alectryon excelsus gisborne 5 2.4 34 4 391 alectryon excelsus gisborne 5 2.4 34 13 426 alectryon excelsus gisborne 5 1.9 36 6 429 alectryon excelsus gisborne 5 2.5 50 18 464 alectryon excelsus gisborne 5 2.2 44 9 533 alectryon excelsus gisborne 5 2.2 44 10 478 alectryon excelsus gisborne 5 2.2 43 15 590 alectryon excelsus gisborne 5 2.5 36 11 556 alectryon excelsus gisborne 5 2.5 36 12 467 alectryon excelsus gisborne 5 2.0 28 7 500 alectryon excelsus gisborne 5 2.0 28 6 529 alectryon excelsus gisborne 5 2.7 52 17 618 alectryon excelsus gisborne 5 2.7 52 13 605 dacrycarpus dacrydioides gisborne 4 2.3 36 7 385 dacrycarpus dacrydioides gisborne 4 2.3 36 8 360 dacrycarpus dacrydioides gisborne 4 2.8 39 11 386 dacrycarpus dacrydioides gisborne 4 2.8 39 14 413 dacrycarpus dacrydioides gisborne 4 2.3 32 10 385 dacrycarpus dacrydioides gisborne 4 1.9 25 4 375 dacrycarpus dacrydioides gisborne 4 1.9 25 5 444 dacrycarpus dacrydioides gisborne 4 2.4 31 9 394 dacrycarpus dacrydioides gisborne 4 1.9 34 3 400 dacrycarpus dacrydioides gisborne 4 1.9 34 5 417 dacrycarpus dacrydioides gisborne 4 2.9 42 16 382 dacrycarpus dacrydioides gisborne 4 2.1 34 6 417 dacrycarpus dacrydioides gisborne 4 2.1 23 7 389 dacrycarpus dacrydioides gisborne 4 2.7 43 13 432 dacrycarpus dacrydioides gisborne 4 2.2 30 9 450 dacrycarpus dacrydioides gisborne 4 2.2 30 6 444 dacrycarpus dacrydioides gisborne 5 3.4 63 27 331 dacrycarpus dacrydioides gisborne 5 3.0 57 22 362 marden et al. new zealand journal of forestry science (2021) 51:1 page 27 table a5: continuedtable a5: continued species location age (y) height (m) rcd (mm) dbh (mm) wood density (kg m–3) dacrycarpus dacrydioides gisborne 5 3.0 57 13 376 dacrycarpus dacrydioides gisborne 5 3.1 48 18 328 dacrycarpus dacrydioides gisborne 5 3.1 49 21 338 dacrycarpus dacrydioides gisborne 5 1.6 47 12 375 dacrycarpus dacrydioides gisborne 5 2.4 43 13 340 dacrycarpus dacrydioides gisborne 5 2.9 47 15 338 dacrycarpus dacrydioides gisborne 5 2.9 47 13 317 dacrycarpus dacrydioides gisborne 5 2.9 47 11 366 dacrycarpus dacrydioides gisborne 5 2.7 45 17 365 dacrycarpus dacrydioides gisborne 5 2.6 54 18 328 dacrycarpus dacrydioides gisborne 5 2.6 54 16 329 dacrycarpus dacrydioides gisborne 5 2.7 32 13 361 podocarpus totara gisborne 5 2.2 43 23 359 podocarpus totara gisborne 5 3.0 57 27 397 podocarpus totara gisborne 5 3.3 60 24 362 podocarpus totara gisborne 5 3.2 64 33 345 podocarpus totara gisborne 5 3.0 50 16 375 podocarpus totara gisborne 5 2.6 50 18 383 podocarpus totara gisborne 5 2.4 49 17 452 podocarpus totara gisborne 5 2.8 41 16 526 podocarpus totara gisborne 5 3.2 49 24 436 agathis australis gisborne 5 1.5 24 7 318 agathis australis gisborne 5 1.5 28 9 286 agathis australis gisborne 5 1.6 21 8 294 agathis australis gisborne 5 1.5 19 5 250 agathis australis gisborne 5 1.7 20 13 210 agathis australis gisborne 5 1.9 25 10 220 agathis australis gisborne 5 1.6 26 10 268 agathis australis gisborne 5 1.8 24 18 326 dacrydium cupressinum gisborne 5 2.3 34 8 484 dacrydium cupressinum gisborne 5 2.2 36 10 463 dacrydium cupressinum gisborne 5 2.2 36 8 484 dacrydium cupressinum gisborne 5 2.0 40 9 417 dacrydium cupressinum gisborne 5 2.0 40 6 385 dacrydium cupressinum gisborne 5 1.5 34 11 482 dacrydium cupressinum gisborne 5 2.3 41 12 471 dacrydium cupressinum gisborne 5 2.5 44 13 483 dacrydium cupressinum gisborne 5 2.4 38 6 385 dacrydium cupressinum gisborne 5 2.4 38 13 462 dacrydium cupressinum gisborne 5 1.9 39 8 435 dacrydium cupressinum gisborne 5 2.2 30 11 455 marden et al. new zealand journal of forestry science (2021) 51:1 page 28 species location age (y) height (m) rcd (mm) dbh (mm) wood density (kg m–3) dacrydium cupressinum gisborne 5 2.1 42 10 424 dacrydium cupressinum gisborne 5 2.1 42 8 414 prumnopitys ferruginea gisborne 5 1.5 11 2 333 prumnopitys ferruginea gisborne 5 1.6 30 5 429 prumnopitys ferruginea gisborne 5 1.7 17 3 286 prumnopitys ferruginea gisborne 5 1.5 25 3 250 prumnopitys ferruginea gisborne 5 1.9 28 6 500 prumnopitys ferrugínea gisborne 5 1.5 15 2 333 prumnopitys ferrugínea gisborne 5 1.9 24 7 500 vitex lucens gisborne 4 1.9 44 8 302 vitex lucens gisborne 4 1.6 45 5 145 vitex lucens gisborne 5 1.8 95 8 227 vitex lucens gisborne 5 2.2 95 13 354 vitex lucens gisborne 5 2.2 95 10 340 vitex lucens gisborne 5 2.2 95 13 324 vitex lucens gisborne 5 3.3 82 34 335 vitex lucens gisborne 5 3.2 84 40 348 table a5: continuedtable a5: continued more timber from fewer trees – determining what tree density optimises silver beech merchantable yield based upon a long-term thinning trial tomás a. easdale1*, robert b. allen2, larry e. burrows1, david henley3 and dudley a. franklin4 1 manaaki whenua – landcare research, po box 69040, lincoln 7640, new zealand 2 8 roblyn place, lincoln 7608, new zealand 3 scion, po box 29237, christchurch 8440, new zealand 4 7 mt thomas rd, rangiora 7471, new zealand *corresponding author: easdalet@landcareresearch.co.nz (received for publication 24 august 2021; accepted in revised form 22 may 2022) abstract background: the tree stem density which optimises merchantable timber yield (volume per unit area) is unknown for most of new zealand’s indigenous tree species. while moderate thinning of even-aged stands can promote yield, intense thinning may decrease yield by creating space that cannot be filled by residual trees, increasing tree mortality or reducing tree height. we quantified the effects of density on silver beech (lophozonia menziesii (hook.f.) heenan & smissen) tree growth, height and mortality, identified the density leading to optimal merchantable yield and assessed if this density varied with stand age. methods: tree stem diameter growth, height, and mortality responses to density were determined using tagged individuals monitored over time on a long-term thinning trial combined with flexible, multilevel, non-linear models. empirical stand yield responses to density were determined and compared to yield–density relationships in simulated stands. the stand simulations projected beyond the monitored stand ages using the tree-level responses fitted to empirical data. results: low densities (≤400 stems ha-1) sustained fast tree growth for longer than high densities (≥700 stems ha-1) after thinning, but density did not consistently affect merchantable tree heights. the probability of tree mortality increased after intense thinning, but only temporarily, and never exceeding c. 0.01 year−1. a regression of yield–density relationships identified an empirical optimum of c. 570 stems ha–1 for stand ages of 48 and 58 years. at this density, merchantable yield at 58 years was seven-fold greater than that in unthinned stands. the simulations suggested moderately higher densities for optimal yield than our empirical optimum, a moderate increase in optimal densities with stand age, and that c. 90 % of potential cumulative yield was attained at 80 years. conclusions: because thinning increased tree growth, but had minimal effect on tree mortality, our results alleviate concerns about the stability and productivity of thinned stands. densities that optimise yield are about two-fold greater than those previously recommended for silver beech and they remain relatively stable as stands age. this suggests that a single density will be adequate for a range of harvest ages, although harvest should take place before a stand age of 80 years. such conclusions are relevant to managing regeneration within coupes harvested under existing legislation and to areas planted with silver beech. new zealand journal of forestry science easdale et al. new zealand journal of forestry science (2022) 52:16 https://doi.org/10.33494/nzjfs522022x179x e-issn: 1179-5395 published on-line: 07/06/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access and ecosystem loss have revitalised the long-standing pursuit of efficient production from natural resources (trewavas 2001; keating et al. 2010). there are introduction human population growth and per capita consumption, land-use impacts on soils and concerns about biodiversity keywords: individual-based simulation; mixed-effects models; nothofagaceae; resource-use efficiency; southern beeches; tree growth; tree mortality mailto:easdalet@landcareresearch.co.nz http://creativecommons.org/licenses/by/4.0/), easdale et al. new zealand journal of forestry science (2022) 52:16 page 2 compelling reasons for an eco-efficient transformation where more is produced from limited land, water, nutrients and energy (keating et al. 2010). this raises a need for information, practices and technologies that can maintain or increase yield via efficient use of resources (keating et al. 2010). a key contribution from plant breeding research has been the selection of crops that allocate a higher fraction of plant productivity to harvestable yield and that reduce resources wasted in plant competition (weiner 2003). by examining the effects of thinning on tree density and yield, forestry science has long been making a similar contribution. much has been learned but challenges remain (see zeide 2001; pretzsch 2005; deng et al. 2012). once a forest canopy is closed and trees make full use of available resources, the volume of wood produced per unit area (hereafter yield) by even-aged stands is nearly constant across a wide range of tree stem densities (hereafter densities), regardless of whether there are many small trees or a few large trees (langsaeter 1941; zeide 2001). this can be seen in thinned, evenaged stands (nishizono 2010) or natural self-thinning stands (xue & hagihara 1998) and is consistent with the constant final yield of undisturbed monocultures (weiner & freckleton 2010). thus, thinning dense even-aged stands is a means of redistributing resources to promote growth of residual trees and increase merchantable yield of timber (zeide 2001). there are, however, associated trade-offs; although moderate thinning will increase merchantable yield, intense thinning will not. excessive thinning can compromise yield by: (1) creating large canopy spaces that cannot be filled by the crowns of residual trees (smith et al. 1997; zeide 2001); (2) promoting lateral growth that produces short trees with less usable timber (clutter et al. 1983; smith et al. 1997); and, (3) increasing residual tree mortality (harrington & reukema 1983; kariuki 2008) leading to further canopy gaps and inefficient use of space. while it is well established that merchantable yield has a hump-shaped relationship with density, it is generally unknown what density leads to maximum merchantable yield for a given species and site condition (zeide 2001; pretzsch 2005) as well as how the density giving optimal merchantable yield changes with stand age (fig. 1; pretzsch 2005; zeide 2008). the cost of thinning interventions, and how long it takes for trees to reach harvesting size, underscores the importance of determining what density is optimal for yield. we investigate how density influences treeand stand-level responses by silver beech (lophozonia menziesii (hook.f.) heenan & smissen) using a long-term thinning trial. more harvest volume is produced from silver beech than any other indigenous tree species yet this is a species that tolerates competition and grows slowly in natural forests (richardson et al. 2011). the effects of thinning on residual trees are somewhat contentious. although a shade-tolerant species with low apical dominance, such as silver beech, can become short-statured at low densities (wardle 1984), the small diameters of trees in dense stands can lead observers to under-estimate tree age and over-estimate height growth at high densities (smith et al. 1997). while strong competition in dense, even-aged stands leads to intense self-thinning driven tree mortality (e.g., osawa & allen 1993), intense thinning of such stands can raise mortality through physiological shock or by destabilising residual trees and increasing the risk of windfall (e.g., harrington & reukema 1983; kariuki 2008). the first aim of our study was to determine how density influences treelevel stem diameter growth, height, and mortality as well as how this varies with stand age and tree size. predictions are also contentious at a stand-level. since tree competition can become more intense as tree size increases, we might expect that densities for optimal yield will decrease in older stands. however, long-term assessments of thinned fagus sylvatica stands indicate that optimal merchantable yield shifts to higher densities during stand development (pretzsch 2005). the second aim was to distinguish between these alternative standlevel responses by answering the following questions (see fig. 1 for graphical representation): (1) what density optimises merchantable (harvestable ‘crop’) yield?; (2) what is the resulting yield?; (3) how does this yield compare with that of unthinned stands?; and, (4) how does the density giving optimal yield change with stand age? we address these questions with a combination of empirical analyses and simulations of yield and density relationships over time. finally, we consider the management implications for new zealand beech forests. density m er ch an ta bl e yi el d ?1 ?2 ?3 ?4 time1 time2 figure 1: although the relationship between merchantable timber yield and density is of a known shape, key questions relate to the density that optimises merchantable yield for a given species, site condition and stand age (?1), ensuing yield (?2), yield gains relative to untended stands (?3), and temporal changes in these responses (?4). all four questions are addressed here for silver beech. methods study site our study focuses on a silver beech forest in the alton valley (46°02’s 167°37’ e and 150-190 m elevation), southland, new zealand. the site has orthic brown soils (dystrudept in us soil taxonomy; hewitt 2010) and the terrain is mostly flat. the area receives, on average, 1250 mm of annual rainfall and the mean annual temperature is c. 9.5˚ c. the original old-growth forest dominated by silver beech was felled in 1951 but c. 40 seed trees per hectare were retained. seed trees were poisoned 12 years after logging. subsequent natural regeneration was variable but typically led to a high density of saplings (franklin 1981; easdale et al. 2009). silver beech dominated the regenerating forest although mountain beech (fuscospora cliffortioides (hook.f.) heenan & smissen) was present locally and there were also low numbers of other indigenous angiosperm and podocarp trees (easdale et al. 2009). silver beech silvics silver beech is a long-lived (up to 600 years) evergreen tree species that occurs from c. 37˚ 30’ to 46˚ 30’ s (wardle 1984). trees up to 42.7 m tall and 3.0 m diameter have been measured. silver beech is restricted to montane and subalpine forests in the north of its range (up to 1400 m elevation), but in the south it is found near sea-level. it spans a wide range of annual rainfall from >8000 mm in western parts of the southern alps, where it can be the dominant species, down to c. 600 mm in south-eastern parts of the south island, where it comes close to forming the driest new zealand beech forests (wardle 1984). this species seeds prolifically, and in 33% of years >2000 seeds m−2 are produced which commonly lead to an abundance of seedlings on the forest floor (burrows and allen 1991; wardle 1984). even seedlings of this shade-tolerant species which have been suppressed for decades are capable of responding to increased resource availability (wardle 1984). asymmetric competition for light is the major factor controlling diameter growth of small silver beech trees throughout new zealand’s south island, whereas physical environment (e.g., elevation) is relatively more important to large tree growth (easdale et al. 2012), thus thinning of regenerating stands is expected to enhance growth of residual trees. experimental design and sampling protocols a thinning trial of 15 contiguous 0.2 ha stands was established in the alton valley site in 1971. stand treatments comprised different combinations of thinning interventions implemented in 1971 and 1980 to give a wide range of densities by 1980, with one unthinned control stand (fig. 2; franklin 1981). other than this unthinned stand with c. 8000 stems ha–1 (see below), total stand-level densities after the second thinning in 1980 were 3000 (1 stand), 1500 (1), 730 (1), 390-415 (4), 325 (1), 285-295 (3) and 190-200 (3) stems ha–1 (fig. 2). trees were thinned mostly from below, aiming to retain dominant ‘crop’ trees with an even spacing while discarding forked trees where possible. selected trees were ≥ 3 cm dbh in 1971 and ≥ 5 cm dbh in 1980 but smaller stems were retained at densities ≥ 1500 stems ha–1. crop trees were pruned in 1975, 1977 and 1979 to a final height of 5.5 m. easdale et al. new zealand journal of forestry science (2022) 52:16 page 3 table 1: description of the study sites figure 2: layout of the alton valley thinning trial. fourteen 0.2-ha stands were thinned in 1971 and 1980 to densities in stems ha–1 shown in the figure, with one left unthinned. 190 190 200 410 415 390 290 395 295 325 730 285 50 m crop trees (n = 1431, including 54 mountain beech trees) were tagged in 1971 with tags drawn-out by copper wires and those surviving were remeasured in 1974, 1977, 1980, 1982, 1985, 1995, 1999 and 2009. the selected crop trees had initial diameters at breast height (dbh) of 8.2 ± 3.1 cm (mean ± sd) and total heights of 5–8 m in 1971, as well as a dbh of 14.1 ± 4.9 cm at the time of the second thinning in 1980. only a subset of crop trees, with large dbh and a regular spacing, were initially tagged in the two stands with highest final densities and in the control plot, but all trees ≥10 cm dbh were measured in all stands in 1999 and 2009. even though we used metal detectors to relocate fallen tags in 2009, 133 tags were not found and these trees could not be matched to prior measurements. a subset of crop trees had merchantable height (to the lowest major fork, bend, or the point where tapering decreased to c. 50-60% of dbh) and total height measured with ultrasonic hypsometers (vertex iii, häglof, sweden) in 1999 and 2009. the height measurements were made on five randomly selected trees within each of four dbh classes to allow for calibration of height–dbh curves across the range of tree sizes found in each stand. data analysis to assess the extent to which thinning affected treelevel growth and whether low density compromised the height and survivorship of residual trees, we (step 1) assessed empirical responses of dbh, merchantable height, and mortality to 1980 densities using the repeated measurements of crop trees and available height measurements. we then (step 2) assessed empirical stand-level yield responses to 1980 thinning densities using the full measurements of stems ≥10 cm dbh in 1999 and 2009 (stand ages of 48 and 58-years; details below). although the trial spanned densities of 190 to c. 8000 stems ha–1 with intermediate densities, 38 years of monitoring, and eight remeasurements, only eight densities were trialled and not all stands had attained merchantable yields by the last measurement. thus, as a last step (3), we combined the tree-level models developed in step 1 into simulations of stand development to predict merchantable yield across a wider range of densities and stand ages than those measured. in assessing yield, we followed local industry standards (ministry for primary industries 2013) that define merchantable yield as the combined volume per unit area of clear tree stems ≥ 30 cm dbh up to a height that excludes major forks or stem tapering. tree-level responses we modelled the cumulative growth (dbh as a function of stand age) for silver beech crop trees that survived to 2009. this encompassed a total of 1153 trees with eight or nine repeated measurements and, except for one intensively thinned stand with 14 trees, included 32 to 157 modelled trees per stand (stems with unmatched tags or with <7 dbh measurements were excluded from growth analyses). tree growth analyses relied on non-linear multilevel (‘mixed’) models to account for the covariance structure that results from repeated easdale et al. new zealand journal of forestry science (2022) 52:16 page 4 measurement of crop trees within stands (gelman & hill 2007). we modelled the effects of density on dbh in three steps. first, we fitted tree dbh (cm) as a gompertz function of stand age (years): (1) with a = (a0 + aj + aij) and d = (d0 + dj + dij) where the parameters a, b and d respectively define the asymptote, location and slope of the curve (sit & poulin-costello 1994). multilevel growth curves were simultaneously fitted for individual trees i and for the average tree in each stand j by allowing a and d to vary both at stand-level (aj and dj) and at individual treelevel (aij and dij) via nested random effects and assuming normally distributed residuals εij. here, the slope of the fitted curves represents dbh growth (husch et al. 2003). second, we extracted the stand-level parameters aj and dj fitted in the first step and, from a candidate set of models, identified two models that best explained their respective relationship with density after second thinning (1980). third, we substituted parameters a and d into the original gompertz equation with the models resulting from the second step, as shown below, and fitted this modified gompertz model with random (tree and stand-level) effects on the parameters controlling the asymptote and slope of the curve (i.e. dbh growth). this three-step procedure allowed us to explicitly model and calibrate the non-linear relationship between diameter curves and tree density within stands. we modelled merchantable height (hmerch) using the 320 silver beech crop trees measured in 1999 and/or 2009. to this end, we first used multilevel models to identify the height–dbh function having the strongest fit among candidate functions given by husch et al. (2003) and then tested the effects of density on fitted standlevel parameters (second step above). inspection of crop tree mortality rates computed for each density and census interval showed that the probability of mortality increased exponentially towards lower densities (heavy thinning) and also increased at first (up to 19 years after the second thinning) and then decreased with time up to 29 years after the second thinning (fig. s1, supplemental information). these also suggested that timing since thinning may be more important than stand age in determining tree mortality. thus, we modelled the probability of tree mortality (pa) as an exponential effect of density with a flexible lognormal function to account for the humped response to t_thin, time since second thinning: (2) where a, b, c, f and g are estimated parameters. since the data consisted of multi-year census intervals (t) of different duration (i.e. periodic observations of survivorship/mortality), and periodic (multi-year) mortality (pm) relates to annual mortality pa as: (3) we modelled annual probabilities of mortality by substituting pa in equation 3 with equation 2: (4) in this way, equation 4 allowed us to estimate parameters for annual mortality (a, b, c, f and g) from periodic mortality data of variable duration. we accounted for all survivorship/mortality records (including any mountain beech selected as crop trees) so long as tagged stems were relocated in consecutive measurements and accounted for mortality events only where a stem was confirmed dead (i.e. stem measurements with tags ‘not found’ were excluded from analysis). maximum likelihood methods were used to estimate the parameters most likely to have produced the data, given the models. for dbh growth and merchantable height, parameters were estimated with non-linear multilevel models and a gaussian distribution using the “nlme” library in r (r development core team, 2013). mortality was modelled with non-linear models and a binomial distribution based on simulated annealing, a robust global optimisation algorithm (goffe et al. 1994) implemented using the “likelihood” package in r (murphy 2012). alternative models for each response variable were compared using their associated akaike information criterion (aic) values, where lower aic values indicate greater empirical support for a model. differences in aic values (∆aic) < 2, between 4 and 7, and >10 respectively indicate negligible, moderate and strong empirical support between alternative models (burnham & anderson 2002). as the 1980 density in the unthinned control stand was unavailable, we: (i) fitted a regression for the relationship between tree density post-thinning in 1980 and density of live stems ≥10 cm dbh in 2009 (df =13, r2 = 0.92) for all but the unthinned stand and then: (ii) back-estimated tree density in 1980 from the fitted model and the estimated density of the unthinned stand in 2009. stand-level responses merchantable yield was first predicted for each tree from merchantable height and dbh using equations calibrated for nothofagaceae (hereafter southern beech) found in new zealand (ellis 1979) and then added up by stand. only trees ≥ 30 cm dbh were used in calculating merchantable yield (ministry for primary industries 2013). to identify the tree density that optimised yield, we first fitted a flexible third-degree polynomial curve to describe the relationship between merchantable yield and density. we then identified the density which corresponded with the peak of the merchantable yield curve using the general-purpose “optim” optimizer function in r (r development core team, 2013). lastly, to assess how the density giving optimal yield varied over time, we developed a stochastic simulation model of stand development that combined the empirical growth and mortality responses derived for individual trees. this involved simulating the cumulative stem growth of individual trees at various stem densities and, for each simulated stand, estimating merchantable volumes as the combined volume of all live trees at a given stand age. simulated growth curves varied as a function of stem density and stand age and built-in growth deviations around the ‘mean curve’, based upon our previously calibrated empirical responses. incorporating growth deviations was important since trees only contribute to yield once they reach merchantable size (≥30-cm dbh) and small fluctuations in dbh distribution can cause large changes in yield (smith et al. 1997). merchantable heights were determined by a calibrated height-dbh allometry function and both height and dbh then informed the corresponding merchantable volume for each tree, according to a published allometry (ellis 1979). the model also accounted for tree mortality, with trees stochastically surviving or dying based on previously estimated probabilities of mortality, as a function of stand age and time after thinning. for consistency with the field trial, we assumed that final thinning took place at 29 years. we: (1) ran simulations for post-thinning densities spanning 100 to 10000 stems ha−1, with multiple runs for each density, and concluded by: (2) identifying what post-thinning density optimised merchantable yield for each stand age. our simulations were conditional upon the empirical tree-level results so that further specifics of the simulation approach are presented below, after the corresponding empirical results. results tree-level responses tree growth growth curves fitted to the empirical data showed that progressively larger dbhs resulted from gradually lower densities (fig. 3a). intense thinning led to longer-lasting growth responses, with responses to ≤400 stems ha–1 only starting to become discernible from each other after some 10 years from second thinning (39 years of age) (fig. 3a). at 58 years, the mean dbh of crop trees in two of the lowest stocked stands was twice that of the unthinned control stand (37 cm dbh at 190 stems ha–1 vs. 17 cm dbh at c. 8000 stems ha–1). tree growth was positively autocorrelated, with dbh at 20 years predicting dbh at 58 years (cross-stand mean r = 0.72). starting with equation 1, we progressively built in the effects of density on tree dbh growth. parameter a was best described by a power function of density at 29 years, namely, the age at second thinning (∆aic = −24 relative to a null model). adding a term for log density at 20 years, the age at first thinning, did not improve predictions (∆aic = −25 relative to a null model). parameter d was best described as a logarithmic function of density at 29 years (∆aic = −22 relative to a null model). adding a term easdale et al. new zealand journal of forestry science (2022) 52:16 page 5 for log density at 20 years only moderately improved predictions (∆aic = −26 relative to a null model) so we simply modelled d as a function of density at 29 years. substituting parameter a and d in equation 1 with power and logarithmic functions: and gave a modified gompertz function for tree growth based on stand age and density after second thinning: (5) we parameterised equation 5 with nested tree(i) and stand-level (j) random effects on parameters a2 and d2 (i.e., and ) and obtained an unbiased model with the parameter values given in table 1. the fitted model explained substantial variance when only accounting for the fixed parameters (r2 = 0.63) and captured nearly all of the variance with both the fixed and random parameters (r2 = 0.99). this meant that the random tree-level effects captured much of the dbh variability that could not be explained by density. inspection of tree-level random parameters extracted from the “nlme” model showed that a2 and d2 were negatively correlated (r = −0.329) and that a2 had a negatively skewed distribution, which we overcame by a logarithmic transformation: where the negative sign turned a left skew into a right skew, the internal constant minimised the skew and the external constant centred the transformed values. further assessment indicated that the standard deviation (sd) of tree-level parameters tra2 and d2 increased towards stands with lower stem density. after testing simple linear and non-linear functions (via “nls” in r), we modelled the standard deviation (sd) of tra2and d2 as a power function of ln density with fitted parameters given in table 1. these models described much of the ‘random’ within-stand variance that remained unaccounted for by the ‘fixed’ parameters. tree height of the various height–dbh functions tested, a simple logarithmic function gave the best combination of balanced residuals and strength of support (∆aic = −18 relative to a null model (i.e. fixed height estimates)). the resulting model was unbiased but had minimal explanatory value (r2 = 0.04) and gave a 0.7 m increase in merchantable height between a 30-cm and a 55-cm-dbh tree (about the largest recorded at last measurement). the fitted model and estimated parameter values are given in table 1. a logarithmic model with variable slope parameters for stands had stronger support than the above (∆aic = −34 relative to a null model), with up to 1.6 m differences in merchantable height among stands for a 30-cm-dbh tree (fig. 3b). however, the fitted slopes (fig. 3b) were unrelated to density at 20 or 29 years. easdale et al. new zealand journal of forestry science (2022) 52:16 page 6 10 20 30 40 stand age (years) d b h (c m ) 20 30 40 50 60 7915 3000 1500 730 415 290 325 200 395 410 295 390 285 190 190 (stems ha−1) density at 29 yr(a) 6 7 8 9 dbh (cm) m er ch an ta bl e he ig ht (m ) 30 40 50 60 415 325 190 285 410 295 390 200 190 1500 3000 7915 290 730 395 (stems ha−1) density at 29 yr(b) figure 3: cumulative dbh–age growth curves (a); and merchantable height–dbh curves (b); fitted for the average crop trees across a range of densities at second thinning (29 years). vertical dotted lines indicate the timing of the two thinning interventions in (a). the grey saturation of curves scales with the logarithm of density at 29 years. tree mortality the probabilities that trees died were low. only 26 of 1379 crop trees with original tags found were confirmed to have died within 29 years of monitoring after the second thinning. mortality of ‘crop’ trees was lowest in dense stands and highest in stands with the lowest densities (fig. s1, supplemental information). the highest annual probability of mortality was 0.0106 per year for stands with 190 stems ha–1 between 15 and 19 years after second thinning. a model with an exponential effect for density and a flexible response to t_thin after second thinning (fig. 4) had strong support relative to a constant probability of mortality (∆aic = −30) and captured the empirical pattern of mortality effectively (fig. s1, supplemental information). estimated parameter values are given in table 1. stand-level responses empirical density and optimal yield a regression of merchantable yield calculated from field measurements (for all trees ≥30 cm dbh) showed that yield peaked at 62 m3 ha–1 with 571 stems ha–1 in 48-year-old stands (fig. 5a). subsequently, the optimal yield became more pronounced with an estimated peak yield of 114 m3 ha–1 at c. 567 stems ha–1 in the 58-yearold stands (fig. 5a). at this age, merchantable yield was c. seven times greater than in the unthinned control stand. an assessment of stand data using all crop and noncrop trees ≥10 cm dbh (table 2) showed some variation from the above results for crop trees only. mean stem diameters were consistently larger at low density relative to high density but differences were partly dampened due to stem recruitment in the more heavily thinned stands (note an increase in stem densities from 1999 to 2009 at low densities). mean top heights had differences of up to 4 m between stands and a weak positive association with density, with heights of 15.118.0 m for densities of 400 stems ha−1 or less and heights easdale et al. new zealand journal of forestry science (2022) 52:16 page 7 table 1: set of fitted models and their associated parameter estimates, as calibrated from repeated measurement of tagged silver beech trees at the alton valley trial. these models and parameters where subsequently employed to simulate merchantable yield across a broad range of stand ages and densities. ra2-d2 is the correlation coefficient between tree-level random parameters a2 and d2. variable model equation parameter estimated value standard error dbh (5) a1 138.0 23.7 a2 −0.2369 0.0282 b 1.680 0.00570 d1 0.01962 0.00792 d2 0.007785 0.00130 dbh variance (6) a 0.5917 0.430 b −0.5915 0.404 (7) c 0.003343 0.00177 d −0.4177 0.294 ra2-d2 −0.329 – merchantable height (8) a 3.124 0.845 b 1.142 0.256 probability of mortality (2) a 1.327 1.51 b 0.2424 0.101 c 0.3850 1.29 f 14.98 2.37 g 0.5181 0.121 30 25 20 15 10 5 0 density at 29 years (stems ha−1) ye ar s af te r s ec on d th in ni ng 5e−04 0.001 0.0015 0.002 0.0025 0.003 200 500 1000 2000 5000 density at 29 years (stems ha−1) thinning figure 4: contour plot for probability of tree mortality as a function of years after the second thinning and density after the second thinning (29 years). both contour lines and grey saturation of the background follow the ‘topography’ of mortality probabilities, with higher saturation for higher probabilities of mortality. of 17.1-19.4 m for densities over 700 stems ha−1. basal areas generally corresponded with the merchantable yields reported above, with a maximum of 46.1 m2 ha−1 noted at 750 stems ha−1 at 58 years. stand simulations and changes in optimal density with stand age with tree-level responses understood, we then assembled a stochastic simulation model. for a given density, mean dbh growth was given by equation 5 (table 1). growth variability was incorporated by: (a) sampling parameters tra2 and d2 from a gaussian distribution with sample size given by density, mean value of 0, and the standard deviation defined by eqs. 6 and 7 (table 1): (b) back-transforming tra2 into a2; (c) shuffling values of d2 until they reached the correlation with a2 detected in the field data (r = −0.329) and (d) adding up the values obtained from the previous step to the mean values of a2 and d2 given in equation 5. each set of parameters described the cumulative growth of each tree in a simulated stand (fig. s2, supplemental information). merchantable heights were defined from dbh by equation 8. simulated trees died if a random number between 0 and 1 drawn each year for each tree was lower than the fitted probability of mortality given by equation 2 (table 1) for a given density and t_thin, time since second thinning. predictions of merchantable yield in simulated stands corresponded relatively well with the empirical analyses (fig. 5). as in the field estimates, differences in such yield across a range of densities were small early on and became more pronounced with stand age (fig. s3 and s4, supplemental information). simulations indicate that a density of 650 stems ha–1 gives maximum merchantable yield for 40 to 53 year-old stands; the density giving optimal yield shifts to 850 stems ha–1 for 54 to 64 yearold stands and stabilises at 950 stems ha–1 over 65 years (fig. s3, supplemental information). at a density giving optimal yield, simulations predict a mean merchantable yield of 126 m3 ha–1 at 58 years (fig. 5b) and 190 m3 ha–1 at 80 years. the dome in the yield response curve was however, sufficiently flat to predict similar yields across a range of densities (fig. s4). the temporal increase in merchantable yield is asymptotic with c. 90 % of the aggregate yield attained at 80 years. merchantable yield is predicted to be four to four and a half times greater when compared to unthinned stands with c. 8000 stems ha–1 (fig. 5b). discussion tree-level responses the implications of controlling density in even-aged, silver beech stands are clear and confirm initial assessments that the species responds well to thinning (baker & benecke 2001). growth trajectories overlap early on as crop trees grow freely but they begin to diverge once neighbouring trees begin to compete. the higher the density of trees, the sooner competition is manifested in reduced growth (weiner & freckleton 2010) and this shade-tolerant species does not self-thin up to at least 58 years. competition can affect growth at surprisingly low densities (clutter et al. 1983) but the effects only become evident later on (>40 years at 190 stems ha–1). this supports a view that assessing easdale et al. new zealand journal of forestry science (2022) 52:16 page 8 ● ● ● ● ● ● ● ● ● ● ● ● ● 100 200 500 1000 2000 5000 10000 0 50 10 0 15 0 density at 29 years (stems ha−1) m er ch an ta bl e yi el d ( m 3 ha −1 ) ●● ●● ●● ●● ●● ●● ●● ●● ●● ●● ●●●● ●● 48 yrs 58 yrs ● ●● r ec or de d fi tte d empirical results simulation results(a) (b) ● ● ● ● ● ● ● ● ● ● ● ● ● 100 200 500 1000 2000 5000 10000 0 50 10 0 15 0 density at 29 years (stems ha−1) m er ch an ta bl e yi el d ( m 3 ha −1 ) ●● ●● ●● ●● ●● ●● ●● ●● ●● ●● ●●●● ●● 48 yrs 58 yrs ● ●● r ec or de d s im ul at ed figure 5: empirical estimate of merchantable yield for the last two measurements (48 and 58 years) (a). optimum yields with respect to density at second thinning (29 years) were fitted with third-order polynomial regressions. dotted lines are 95% confidence intervals for the regressions. simulated yield results (lines) for 48and 58-year-old stands and corresponding empirical estimates (points) (b). solid lines represent the mean of 100 simulation runs and dotted lines the 95% confidence intervals. easdale et al. new zealand journal of forestry science (2022) 52:16 page 9 ta b le 2 : s um m ar y of s ta nd m et ri cs fo r al l l iv e st em s ≥1 0 cm d b h in t he a lt on t ri al a t 48 a nd 5 8 ye ar s (1 99 9 an d 20 09 r es pe ct iv el y) . m ea n to p he ig ht w as e st im at ed fr om m od el le d he ig ht s fo r th e 10 0 la rg es t st em s pe r he ct ar e. m ea n d b h a nd m ea n to p he ig ht a re p re se nt ed w it h th ei r as so ci at ed 9 5% c on fid en ce in te rv al s. p lo ts a re so rt ed fr om lo w es t t o hi gh es t p os tth in ni ng d en si ty in 1 98 0. st em d en si ty ( st em s ha −1 ) m ea n d b h ( cm ) m ea n to p he ig ht ( m ) b as al a re a (m 2 h a− 1 ) cr op tr ee s st em s ≥1 0 cm d b h st em s ≥1 0 cm d b h st em s ≥1 0 cm d b h st em s ≥1 0 cm d b h p lo t 19 71 19 80 19 99 20 09 19 99 20 09 19 99 20 09 19 99 20 09 4 30 00 19 0 17 0 38 0 35 .4 5 (3 3. 2 3 7. 9) 27 .8 2 (2 4. 6 3 1. 2) 15 .3 2 (1 5. 2 1 5. 5) 16 .5 7 (1 6. 5 1 6. 7) 17 .4 29 .1 5 15 00 19 0 17 0 60 0 31 .5 5 (2 8. 9 3 4. 1) 20 .3 1 (1 8. 6 2 2. 8) 14 .0 8 (1 3. 9 1 4. 2) 15 .2 ( 15 .1 15 .3 ) 14 .1 25 .8 8 75 0 20 0 20 0 52 0 28 .9 8 (2 6. 3 3 1. 2) 20 .7 6 (1 8. 7 2 3. 1) 15 .1 8 (1 5. 1 1 5. 3) 16 .3 6 (1 6. 3 1 6. 4) 14 .1 22 .6 2 79 15 28 5 26 5 43 0 27 .6 5 (2 5. 8 2 9. 3) 24 .7 3 (2 2. 5 2 7) 14 .1 ( 14 14 .3 ) 15 .1 ( 15 15 .3 ) 16 .8 24 .5 14 30 00 29 0 30 0 52 5 28 .1 6 (2 6. 4 3 0. 3) 23 .0 5 (2 0. 9 2 5. 4) 15 .8 5 (1 5. 7 1 6) 16 .3 6 (1 6. 2 1 6. 5) 20 .2 27 .6 15 15 00 29 5 30 0 59 5 26 .9 7 (2 5. 6 2 8. 3) 23 .2 5 (2 1. 4 2 5. 3) 14 .7 2 (1 4. 6 1 4. 8) 17 .9 ( 17 .8 18 ) 17 .8 30 .2 13 75 0 32 5 31 0 55 5 27 .3 5 (2 5. 6 2 9. 2) 23 .4 4 (2 1. 4 2 5. 5) 15 .1 4 (1 5 1 5. 3) 17 .9 5 (1 7. 8 1 8. 1) 19 .5 29 .3 12 15 00 39 0 37 0 50 5 27 .5 5 (2 6. 1 2 9. 2) 26 .5 6 (2 4. 4 2 8. 5) 15 .2 3 (1 5. 1 1 5. 4) 17 .5 3 (1 7. 4 1 7. 7) 23 .4 32 .4 11 30 00 39 5 37 0 45 0 28 .8 ( 27 .4 30 .4 ) 28 .5 9 (2 6. 7 3 0. 8) 16 .7 1 (1 6. 6 1 6. 9) 17 .9 4 (1 7. 8 1 8. 1) 25 .5 32 .6 1 30 00 41 0 41 0 50 0 27 .6 8 (2 6. 8 2 8. 7) 27 .8 7 (2 6. 1 2 9. 3) 15 .2 8 (1 5. 2 1 5. 3) 16 .0 8 (1 6 1 6. 1) 25 .3 32 .9 7 15 00 41 5 40 5 52 0 25 .5 7 (2 4. 4 2 6. 8) 25 .4 6 (2 4 2 7) 14 .1 ( 14 14 .2 ) 15 .3 4 (1 5. 3 1 5. 5) 21 .8 29 .3 3 30 00 73 0 72 5 93 0 23 .4 8 (2 2. 5 2 4. 7) 23 .6 4 (2 2. 5 2 4. 9) 16 .1 4 (1 6 1 6. 2) 17 .5 8 (1 7. 5 1 7. 7) 34 46 .1 9 15 00 15 00 83 0 88 0 22 .4 7 (2 1. 5 2 3. 4) 24 .0 3 (2 3 2 5. 1) 16 .8 8 (1 6. 8 1 7) 19 .4 1 (1 9. 3 1 9. 5) 35 .8 43 .8 10 30 00 30 00 12 85 13 60 17 .8 2 (1 7. 1 1 8. 6) 18 .7 5 (1 8 1 9. 6) 15 .7 ( 15 .6 15 .8 ) 17 .0 6 (1 7 1 7. 2) 35 .8 42 .4 6 ~ 80 00 ~ 80 00 19 25 20 55 14 .9 4 (1 4. 5 1 5. 4) 15 .8 4 (1 5. 4 1 6. 3) 14 .5 ( 14 .4 14 .6 ) 18 .3 3 (1 8. 2 1 8. 5) 36 .9 44 .2 easdale et al. new zealand journal of forestry science (2022) 52:16 page 10 thinning responses requires long-term monitoring. at 58 years, differences in mean dbh, between highest and lowest densities, were two-fold (17 cm to 37 cm) and, with growth at 190 stems ha–1 little constrained, the ratio might expand further in older stands. for some hardwoods, total height growth can be significantly less at lower densities than at moderate or higher densities (clutter et al. 1983) despite a potential increase in mean height derived from thinning shaded and suppressed trees. we found only moderate statistical support for reduced total heights at low densities (result not shown) but no significant change in merchantable heights as a function of density. this relieves concern about costs from possible shifts in stem allometry for silver beech at low density (wardle 1984) and supports a view that dominant tree heights are least affected by thinning (lanner 1985; skovsgaard & vanclay 2007). the basis of thinning is to concentrate growth on selected trees either by avoiding untimely timber losses in stands that self-thin (smith et al. 1997) or by transferring growth from many small slow-growing trees in stands that are slow to self-thin (e.g. farnden & herring 2002). stands of silver beech tend to be dense and slow to self-thin, leading to stagnant growth both in regenerating stands (wardle 1984) and in old-growth forests (richardson et al. 2011; easdale et al. 2012). the low mortality recorded in the unthinned stand was unforeseen. in part, this may reflect that the selected crop trees being monitored were the larger individuals, whereas the mortality occurred in the untagged smaller trees, although it is also not at odds with densitydependent mortality and simply signals that, up to 58 years, mortality of silver beech is low among the wave of ‘crop’ trees. this slow rate of self-thinning does limit timber yield and accounts for the potential need for silvicultural interventions in silver beech stands. conversely, the implications of thinning leading to mortality of residual trees need not be of concern. consistent with thinned eucalypts (kariuki 2008), we found that intense thinning increased mortality in retained trees but the effects were transient and of minor impact. in silver beech the probability of mortality decreased for c. 16 years after a second thinning, but rates were generally low (≤ 0.007 year–1). the low mortality associated with heavy thinning corresponds with findings that patch cuts (<0.2 ha) in southern beech forests do not elevate silver beech tree mortality at the edge of cuts (wiser et al. 2005). given that many factors can influence tree growth, it is notable that stand age and density alone explained up to 63 % of the recorded variation in cumulative growth for silver beech at our study site. yet, accounting for the remaining within-stand variability in growth was vital for realistic predictions of yield (fig. 5). simulations that assumed zero or constant growth variability across stand densities produced opposite and markedly biased predictions at high density (fig. s5 vs s3, supplemental information). we found larger growth variability at low density, which differs from a common observation that size inequality increases at high densities in evenaged stands (e.g. nord-larsen et al. 2006). this can be explained by a confounding effect of tree size. plant density is only meaningful when related to mean plant size so that the performance of large trees at low density can resemble that of small trees at high density (weiner et al. 2001). the differences in tree genetics, time of establishment and microsites can take a long time to be manifested in slow growing trees in dense stands, but rapidly become evident in larger trees at low density. also worthy of attention was a strong correlation (r = 0.72) between dbh at 20 and 58 years. this signals auto-correlated growth, with early establishment or fast initial growth determining later growth because of competitive advantage, as previously documented by tree ring studies (e.g., brienen et al. 2006) and serves to identify candidate crop trees at an early stage. stand-level responses although it seems indisputable that thinning can promote merchantable yield, it is usually unknown what density leads to maximum yield for a given species, site and stand age (zeide 2001). initial advice suggested thinning to 200–300 stems ha–1 (franklin 1981; wardle 1984) but both empirical evidence (fig. 5a) and simulations (fig. 5b and fig. s3, esm1) reveal that those densities are far too low and compromise yield due to unused space and, to lesser extent, increased tree mortality. regressions with empirical data indicate that, up to 58 years, yield peaks at an optimum density of 570 stems ha–1 for silver beech. simulation results suggest a higher optimum of 850 stems ha–1 for 54to 64-yearold stands and 950 stems ha–1 for older stands. both a quantitative study (pretzsch 2005) and a theoretical review (zeide 2008) have indicated that while low densities foster diameter growth early on, it is high densities that promote ingrowth of smaller stems into a merchantable dbh later on (see fig. s2). so even though somewhat counterintuitive, the density for optimal yield is expected to be higher for older stands. the simulations confirm this prediction but indicate that the increase in density is relatively minor and the density for optimal yield is fairly stable across a breadth of stand ages (fig. s4). the stand age for optimal yield in thinned stands appears promisingly shorter than the c. 135 years estimated for maximum yield in untended silver beech stands (williams & chavasse 1951). since the simulation model was calibrated with the original selection of tagged crop trees in dense stands (and those could have outperformed untagged trees), the simulation optimum ought to be interpreted with some caution. our simulations of stand development predicted an optimal merchantable yield of 190 m3 ha–1 at 80 years for 950 stems ha–1, which corresponds to clear logs at least 30-cm dbh and merchantable heights not exceeding 8 m. when all trees above 7-cm dbh are instead accounted for (a standard given by smith et al. 1997) using the same volume function, the yield at 80 years is estimated to be 665 m3 ha–1 for 950 stems ha–1. this generates a mean annual increment of 8.31 m3 ha–1 year–1 which is comparable in magnitude to an earlier estimate of 5.6 to 6.3 m3 ha–1 year–1 at 70–80 years for thinned silver beech on fertile, low elevation sites in easdale et al. new zealand journal of forestry science (2022) 52:16 page 11 southern new zealand (valentine 1969; wardle 1984). when compared with thinned second-growth southern beech forest in patagonia, our estimated yield of 665 m3 ha–1 at 80 years for 950 stems ha–1 corresponds well with those from other southern beech forests at similar latitudes: 400–600 m3 ha–1 for a 70–80-year rotation of thinned nothofagus pumilio at latitude 45.5°s (nuñez & vera 1992) and 690 m3 ha–1 for 70–75-yearold thinned lophozonia alpina at c. 42.0°s (grosse 1989). not surprisingly, the above values sit well above a total yield of 160 m3 ha–1 for 60-year-old nothofagus betuloides stands thinned to 1000 stems ha–1 in the southernmost southern beech forests at 54.8°s (fig. 2 in martínez-pastur et al. 2010). implications for managing new zealand’s southern beech forest current legislation largely restricts harvests in southern beech forest to patch cuts not greater than 0.5 ha (new zealand ministry of agriculture and forestry 2007). we know that recruitment in such cuts is variable but often prolific and usually well exceeds the density of large trees in mature forest (wiser et al. 2007; allen et al. 2012). where recruitment is abundant, subsequent diameter growth tends to be stagnant in unthinned stands (franklin 1995) and our results demonstrate how thinning can convert more of the regenerating stands into merchantable yield within a shorter time interval. in temperate forests, above-ground woody biomass production tends to be higher (up to c. three-fold higher) in forests on soils with high nutrient availability compared with soils with low nutrient availability (vicca et al. 2012). since the alton valley has moderately fertile orthic brown soils with low to moderate base saturation (hewitt 2010), we might expect lower yields on soils that are less fertile. in addition, reductions in stand biomass productivity have been demonstrated with elevation for one southern beech species (harcombe et al. 1998) and likely translate to similar reductions in silver beech. however, we expect that, once the absolute effects of site are considered, the relative effects of density on yield found here will apply to silver beech in other sites. another consideration is managed species, with indications that responses to thinning likely differ between beech species. assessment of thinning outcomes in a mixed red-hard beech (fuscospora fusca (hook.f.) heenan & smissen; fuscospora truncata (colenso) heenan & smissen) forest at staircase creek, westland, indicated that optimum densities were lower for these species at c. 200 stems ha–1 and merchantable yield potential was substantive, with an estimated maximum 191 m3 ha−1 (stems 30-cm dbh) at 58 years (easdale et al. 2010). optimisation of yield is of course only one possible expectation from developing stands which could be managed also for other goals such as carbon sequestration (evison et al. 2012) and biodiversity (allen et al. 2012). silvicultural guidelines often suggest that thinning should be sequential. this aims to gradually liberate space and resources as trees grow without limiting crown responses to later liberations (clutter et al. 1983; smith et al. 1997). sequential thinning also aims to maintain some degree of natural competition that favours tall, clean, straight trunks (smith et al. 1997) and promotes wind stability (franklin 1995). our results suggest a single thinning suffices and its timing, between two interventions, made no essential difference to silver beech growth. the slope of the dbh curve was largely explained by final density after a second thinning with negligible improvements when accounting for density after the first thinning. noting a relatively flat peak in the yield response to density (fig. s4), and guided more by empirical than simulation results, we suggest that 600−800 stems ha–1 should maximise yield at different harvest times for a merchantable size of 30 cm dbh. with total yield being greater in dense stands (zeide 2001), lower merchantable thresholds would likely shift the optimum yield to higher densities, warranting a separate sensitivity analysis on the impacts of merchantable size. in general, the results support the view that the southern beech respond well after periods of suppression (stewart et al. 1991) when tree stems have a dbh <50 cm (wiser et al. 2005). this offers flexibility in the timing of interventions, where a single late thinning would ease the identification of crop trees (franklin 1995), promote self-pruning (smith et al. 1997) and reduce financial costs (price 1989). experience has shown that very late thinning of pole stands results in wind-throw and attack by platypus beetles (franklin 1995), so thinning should not be delayed excessively and ought to be completed by the time trees are 15 m tall and 15 cm dbh (franklin 1995). conclusions our decadal-level study of treeand stand-level responses to thinning showed the utility of remeasuring stands along a controlled and delineated density gradient. because thinning increased tree growth, but had minimal effect on tree mortality, our results alleviate concerns about the stability and productivity of thinned silver beech stands. the density which optimised yield was about two-fold greater than that previously recommended for silver beech and also gave a sevenfold gain in merchantable yield at 58 years. this density remained relatively stable through stand development and suggested that a single density will be adequate for a wide range of harvest ages, although this should take place before stand age reaches 80 years. this rotation age is much shorter than that previously estimated for maximum silver beech yield in unthinned stands. such conclusions are not only relevant to managing regeneration within coupes harvested under part 3a of the forests act 1949 but also to any areas being planted with silver beech. competing interests the authors declare that they have no competing interests. acknowledgements this trial was established and maintained by the former new zealand forest service and then supported by the foundation for research, science and technology (now ministry of business, innovation and employment) [c09x0308], the ministry of agriculture and forestry (now ministry for primary industries) [sus 8905], and the authors. we are grateful to lindsay and dixon limited for support and access to the study site, gordon baker for assistance with remeasurement, to renske terhürne for data assembly, to guy forrester, jenny hurst and dean anderson for statistical advice, to cissy pan for graphics assistance and to christine bezar and leah kearns for editorial support. comments from susan wiser, horacio bown and anonymous referees helped to improve previous versions of this work. digital records for the thinning trial are available from new zealand’s national vegetation survey databank (https://nvs. landcareresearch.co.nz). authors' contributions tae conceptualised this study, remeasured and interpreted the experiment, curated data, analysed data, and wrote the manuscript. rba conceptualised this study, remeasured and interpreted the experiment, wrote the manuscript, and obtained funding. daf conceptualised this study, established the experiment, remeasured and interpreted the experiment, and reviewed the manuscript. leb remeasured and interpreted the experiment, curated data, and reviewed the manuscript. dh remeasured and interpreted the experiment, curated data, and reviewed the manuscript. references allen, r.b., hurst, j.h., wiser, s.k., & easdale, t.a. (2012). developing management systems for the production of beech timber. new zealand journal of forestry, 57(2), 38–44. baker, g., & benecke, u. (2001). growth response to silvicultural tending of red and silver beech regeneration. new zealand journal of forestry, 46(3), 30–32. brienen, r.j., zuidema, p.a., & during, h.j. (2006). autocorrelated growth of tropical forest trees: unravelling patterns and quantifying consequences. forest ecology and management, 237(1-3), 179–190. https://doi.org/10.1016/j. foreco.2006.09.042 burnham, k.p., & anderson, d.r. (2002). model selection and multimodel inference: a practical information– theoretical approach. new york: springer. burrows, l.e., & allen, r.b. (1991). silver beech (nothofagus menziesii (hook. f.) oerst.) seedfall patterns in the takitimu range, south island, new zealand. new zealand journal of botany, 29(4), 361-365. https://doi.org/10.1080/002882 5x.1991.10415489 clutter, j.l., fortson, j.c., pienaar, l.v., brister, g.h., & bailey, r.l. (1983). timber management: a quantitative approach. malabar: krieger. deng, j.m., ran, j.z., wang, z.q., fan, z., wang, g., ji, m., liu, j., wang, y., liu, j., & brown, j.h. (2012). models and tests of optimal density and maximal yield for crop plants. proceedings of the national academy of science, 109(39), 15823–15828. https://doi. org/10.1073/pnas.1210955109 easdale, t.a., allen, r.b., peltzer, d.a., & hurst, j.j. (2012). size-dependent growth responses to competition and environment in nothofagus menziesii. forest ecology and management, 270, 223–231. https:// doi.org/10.1016/j.foreco.2012.01.009 easdale, t.a., burrows, l.e., henley, d., & baker, g. (2009). effects of thinning on silver beech growth. [landcare research contract report, lc0910/015], 24 p. lincoln, new zealand: landcare research. easdale, t.a., drew, k., henley, d., & baker, g. (2010). effects of thinning on growth of red beech and hard beech. [landcare research contract report, lc0910/179], 24 p. lincoln, new zealand: landcare research. ellis, j.c. (1979). tree volume equations for the major indigenous species in new zealand. [fri bulletin no. 237], rotorua: new zealand forest service. evison, d., easdale, t., mason, e., & sewell, a., (2012). economics of managing new zealand silver beech for timber and carbon. [nzares no. 1170-201693282], nelson: new zealand agricultural and resource economics society (inc.). farnden, c., & herring, l. (2002). severely repressed lodgepole pine responds to thinning and fertilization: 19-year results. forestry chronicle, 78(3), 404–414. https://doi.org/10.5558/ tfc78404-3 franklin, d.a. (1981). a silvicultural regime for dense beech regeneration. [fri what’s new in forest research 98], rotorua: forest research institute. franklin, d.a. (1995). management of beech. in: d. hammond (ed.), forestry handbook (pp. 95–97). christchurch: n.z. institute of forestry. gelman, a., & hill, j. (2007). data analysis using regression and multilevel/hierarchical models. new york: cup. https://doi.org/10.1017/cbo9780511790942 goffe, w.l., ferrier, g.d., & rogers, j. (1994). global optimization of statistical functions with simulated annealing. journal of econometrics, 60(1-2), 65–99. https://doi.org/10.1016/0304-4076(94)90038-8 grosse, h. (1989). renovales de raulí, roble, coigüe y tepa. expectativas de rendimiento. ciencia e investigacion forestal, 3(1), 37–72. https://doi. org/10.52904/0718-4646.1989.110 easdale et al. new zealand journal of forestry science (2022) 52:16 page 12 https://nvs.landcareresearch.co.nz https://nvs.landcareresearch.co.nz https://doi.org/10.1016/j.foreco.2006.09.042 https://doi.org/10.1016/j.foreco.2006.09.042 https://doi.org/10.1080/0028825x.1991.10415489 https://doi.org/10.1080/0028825x.1991.10415489 https://doi.org/10.1073/pnas.1210955109 https://doi.org/10.1073/pnas.1210955109 https://doi.org/10.1016/j.foreco.2012.01.009 https://doi.org/10.1016/j.foreco.2012.01.009 https://doi.org/10.5558/tfc78404-3 https://doi.org/10.5558/tfc78404-3 https://doi.org/10.1017/cbo9780511790942 https://doi.org/10.1016/0304-4076(94)90038-8 https://doi.org/10.52904/0718-4646.1989.110 https://doi.org/10.52904/0718-4646.1989.110 harcombe, p., allen, r.b., wardle, j.a., & platt, k.h. (1998). spatial and temporal patterns in structure, biomass, growth, and mortality in a monospecific nothofagus solandri var. cliffortioides forest in new zealand. journal of sustainable forestry, 6(3-4), 313–345. https://doi.org/10.1300/ j091v06n03_06 harrington, c.a., & reukema, d.l. (1983). initial shock and long-term stand development following thinning in a douglas-fir plantation. forest science, 29(1), 33–46. hewitt, a.e. (2010). new zealand soil classification. [landcare research science series 1, 3rd ed], lincoln: manaaki whenua press. husch, b., beers, t.w., & kershaw, j.a. (2003). forest mensuration. nj, usa: wiley. kariuki, m. (2008). modelling the impacts of various thinning intensities on tree growth and survival in a mixed species eucalypt forest in central gippsland, victoria, australia. forest ecology and management, 256(12), 2007–2017. https://doi. org/10.1016/j.foreco.2008.07.035 keating, b.a., carberry, p.s., bindraban, p.s., asseng, s., meinke, h., & dixon, j. (2010). eco-efficient agriculture: concepts, challenges, and opportunities. crop science, 50, s109–s119. https://doi.org/10.2135/cropsci2009.10.0594 langsaeter, a. (1941). om tynning i enaldret gran-og furuskog. meddelelser fra det norske skogforsøksves, 8, 131–216. lanner, r.m. (1985). on the insensitivity of height growth to spacing. forest ecology and management, 13(3-4), 143–148. https://doi.org/10.1016/03781127(85)90030-1 martínez-pastur, g.j., lencinas, m.v., peri, p.l., cellini, j.m., & moretto, a. (2010). long-term forest management research in south patagonia – argentina: lessons from the past, challenges from the present. revista chilena de historia natural, 83(1), 159–169. https://doi.org/10.4067/s0716078x2010000100009 online report: ministry for primary industries. (2018). measuring indigenous trees and logs a field guide. wellington: ministry for primary industrieshttps:// www.mpi.govt.nz/dmsdocument/45-indigenousforestry-measuring-indigenous-trees-and-logs-afield-guide.accessed 7 aug 2018 murphy, l. (2012). likelihood: methods for maximum likelihood estimation. r package version 1.5. h t t p s : / / c r a n . r p r o j e c t . o r g / w e b / p a c k a g e s / likelihood/ new zealand ministry of agriculture and forestry. (2007). standards and guidelines for the sustainable management of indigenous forests. wellington: maf. nishizono, t. (2010). effects of thinning level and site productivity on age-related changes in stand volume growth can be explained by a single rescaled growth curve. forest ecology and management, 259(12), 2276–2291. https://doi.org/10.1016/j. foreco.2010.03.002 nord-larsen, t., damgaard, c., & weiner, j. (2006). quantifying size-asymmetric growth among individual beech trees. canadian journal of forest research, 36(2), 418–425. https://doi. org/10.1139/x05-255 nuñez, p., & vera, o. (1992). evaluación de intervenciones silvícolas en un renoval mixto de lenga (nothofagus pumilio) y coihue (nothofagus dombeyi), ubicado en la reserva forestal coyhaique, xi región. in actas del seminario de manejo forestal de la lenga y aspectos ecológicos relacionados (pp. 111–125). esquel: ciefap. osawa, a., & allen, r.b. (1993). allometric theory explains self-thinning relationships of mountain beech and red pine. ecology, 74(4), 1020–1032. https://doi.org/10.2307/1940472 pretzsch, h. (2005). stand density and growth of norway spruce (picea abies (l.) karst.) and european beech (fagus sylvatica l.): evidence from longterm experimental plots. european journal of forest research, 124(3), 193–205. https://doi. org/10.1007/s10342-005-0068-4 price, c. (1989). the theory and application of forest economics. oxford: blackwell. r development core team. (2013). r: a language and environment for statistical computing. vienne: r foundation for statistical computing. http:// www.r-project.org. richardson, s.j., hurst, j.m., easdale, t.a., wiser, s.k., griffiths, a.d., & allen, r.b. (2011). diameter growth rates of beech (nothofagus) trees around new zealand. new zealand journal of forestry, 56(1), 3–11. sit, v., & poulin-costello, m. (1994). catalog of curves for curve fitting. [biometrics information, handbook no. 4], victoria: bc ministry of forests. skovsgaard, j.p., & vanclay, j.k. (2007). forest site productivity: a review of the evolution of dendrometric concepts for even-aged stands. forestry, 81(1), 13–31. https://doi.org/10.1093/ forestry/cpm041 smith, d.m., larson, b.c., kelty, m.j., & ashton, p.m.s. (1997). the practice of silviculture: applied forest ecology. new york: john wiley. stewart, g.h., rose, a.b., & veblen, t.t. (1991). forest development in canopy gaps in old-growth beech (nothofagus) forests, new zealand. journal of vegetation science, 2(5), 679–690. https://doi. org/10.2307/3236178 easdale et al. new zealand journal of forestry science (2022) 52:16 page 13 https://doi.org/10.1300/j091v06n03_06 https://doi.org/10.1300/j091v06n03_06 https://doi.org/10.1016/j.foreco.2008.07.035 https://doi.org/10.1016/j.foreco.2008.07.035 https://doi.org/10.2135/cropsci2009.10.0594 https://doi.org/10.1016/0378-1127(85)90030-1 https://doi.org/10.1016/0378-1127(85)90030-1 https://doi.org/10.4067/s0716-078x2010000100009 https://doi.org/10.4067/s0716-078x2010000100009 https://www.mpi.govt.nz/dmsdocument/45-indigenous-forestry-measuring-indigenous-trees-and-logs-a-field-guide https://www.mpi.govt.nz/dmsdocument/45-indigenous-forestry-measuring-indigenous-trees-and-logs-a-field-guide https://www.mpi.govt.nz/dmsdocument/45-indigenous-forestry-measuring-indigenous-trees-and-logs-a-field-guide https://www.mpi.govt.nz/dmsdocument/45-indigenous-forestry-measuring-indigenous-trees-and-logs-a-field-guide https://cran.r-project.org/web/packages/likelihood/ https://cran.r-project.org/web/packages/likelihood/ https://doi.org/10.1016/j.foreco.2010.03.002 https://doi.org/10.1016/j.foreco.2010.03.002 https://doi.org/10.1139/x05-255 https://doi.org/10.1139/x05-255 https://doi.org/10.2307/1940472 https://doi.org/10.1007/s10342-005-0068-4 https://doi.org/10.1007/s10342-005-0068-4 http://www.r-project.org http://www.r-project.org https://doi.org/10.1093/forestry/cpm041 https://doi.org/10.1093/forestry/cpm041 https://doi.org/10.2307/3236178 https://doi.org/10.2307/3236178 trewavas, a.j. (2001). the population/biodiversity paradox. agricultural efficiency to save wilderness. plant physiology, 125(1), 174–179. https://doi. org/10.1104/pp.125.1.174 valentine, j.m. (1969). some observations on the management of silver beech. new zealand journal of forestry, 14(2), 219–226. vicca, s., luyssaert, s., penuelas, j., campioli, m., chapin, f.s., ciais, p., heinemeyer, a., hogberg, p., kutsch, w.l., law, b.e., malhi, y., papale, d., piao, s.l., reichstein, m., schulze, e.d., & janssens, i.a. (2012). fertile forests produce biomass more efficiently. ecology letters, 15(6), 520–526. https://doi. org/10.1111/j.1461-0248.2012.01775.x wardle, j.a. (1984). the new zealand beeches. wellington: new zealand forest service. weiner, j. (2003). ecology – the science of agriculture in the 21st century. journal of agricultural science, 141(3-4), 371–377. https://doi.org/10.1017/ s0021859603003605 weiner, j., & freckleton, r.p. (2010). constant final yield. annual review of ecology, evolution and systematics, 41, 173–192. https://doi.org/10.1146/annurevecolsys-102209-144642 weiner, j., stoll, p., muller-landau, h., & jasentuliyana, a. (2001). the effects of density, spatial pattern, and competitive symmetry on size variation in simulated plant populations. the american naturalist, 158(4), 438–450. https://doi. org/10.1086/321988 williams, r.w.m., & chavasse, c.g.r. (1951). the silviculture of silver beech in southland. new zealand journal of forestry, 6(3), 219–235. wiser, s.k., allen, r.b., benecke, u., baker, g., & peltzer, d.a. (2005). tree growth and mortality after small-group harvesting in new zealand oldgrowth nothofagus forests. canadian journal of forest research, 35(10), 2323–2331. https://doi. org/10.1139/x05-158 wiser, s.k., baker, g., & benecke, u. (2007). regeneration of red and silver beech: how important is the size of harvested area? new zealand journal of forestry, 52(2), 31–35. xue, l., & hagihara, a. (1998). growth analysis of the self-thinning stands of pinus densiflora sieb. et zucc. ecological research, 13(2), 183–191. https:// doi.org/10.1046/j.1440-1703.1998.00256.x zeide, b. (2001). thinning and growth: a full turnaround. journal of forestry, 99(1), 20–25. zeide, b. (2008). the science of forestry. journal of sustainable forestry, 27(4), 345–473. https://doi. org/10.1080/10549810802339225 easdale et al. new zealand journal of forestry science (2022) 52:16 page 14 https://doi.org/10.1104/pp.125.1.174 https://doi.org/10.1104/pp.125.1.174 https://doi.org/10.1111/j.1461-0248.2012.01775.x https://doi.org/10.1111/j.1461-0248.2012.01775.x https://doi.org/10.1017/s0021859603003605 https://doi.org/10.1017/s0021859603003605 https://doi.org/10.1146/annurev-ecolsys-102209-144642 https://doi.org/10.1146/annurev-ecolsys-102209-144642 https://doi.org/10.1086/321988 https://doi.org/10.1086/321988 https://doi.org/10.1139/x05-158 https://doi.org/10.1139/x05-158 https://doi.org/10.1046/j.1440-1703.1998.00256.x https://doi.org/10.1046/j.1440-1703.1998.00256.x https://doi.org/10.1080/10549810802339225 https://doi.org/10.1080/10549810802339225 supplementary information this section presents a graphical examination of mortality rates (fig. s1), a growth simulation example (fig. s2), empirically calibrated simulation results (fig. s3 and s4) and simulation outputs with alternative settings for growth variability (fig. s5). figures are presented in the same order as topics covered in the main text. figure s1: empirical and modelled probabilities of mortality of crop trees as a function of density at different time intervals after the second thinning (up to 29 years). modelled probabilities (equation 2 in table 1, main text) were plotted for the midpoint of each interval (2.5, 10, 17 and 24 years after second thinning respectively). 0−5 years after second thinning ●●●●●● ●●●●●● ●●●● ●● ●● ●●●● ●● ●● 200 500 1000 2000 5000 10000 0. 00 0 0. 00 4 0. 00 8 density at second thinning (stems ha−1) a nn ua l p ro ba bi lit y of m or ta lit y ●● empirical modelled 5−15 years after second thinning ●● ●● ●● ●● ●●●● ●●●● ●● ●●●● ●● ●● ●● 200 500 1000 2000 5000 10000 0. 00 0 0. 00 4 0. 00 8 density at second thinning (stems ha−1) a nn ua l p ro ba bi lit y of m or ta lit y 15−19 years after second thinning ●● ●● ●● ●● ●● ●● ●●●● ●● ●● ●●●● ●● ●● 200 500 1000 2000 5000 10000 0. 00 0 0. 00 4 0. 00 8 density at second thinning (stems ha−1) a nn ua l p ro ba bi lit y of m or ta lit y 19−29 years after second thinning ●● ●● ●● ●●●●●● ●●●● ●●●● ●●●● ●● ●● 200 500 1000 2000 5000 10000 0. 00 0 0. 00 4 0. 00 8 density at second thinning (stems ha−1) a nn ua l p ro ba bi lit y of m or ta lit y easdale et al. new zealand journal of forestry science (2022) 52:16 page 15 0 20 40 60 80 0 20 40 60 80 100 stems ha−1 stand age (years) d b h (c m ) 0 6 12 0 20 40 60 80 n 0 20 40 60 80 0 20 40 60 80 1000 stems ha−1 stand age (years) 0 15 30 0 20 40 60 80 n 0 20 40 60 80 0 20 40 60 80 10000 stems ha−1 stand age (years) 0 20 0 20 40 60 80 n figure s2: simulated diameter (dbh) growth curves for three different densities and their corresponding dbh distributions at 80 years. for comparability, only 100 trees are shown in each case. the dotted line presents the minimum merchantable dbh size. note the greater variability of growth at low density. y s ta nd a ge (y ea rs ) 15 30 45 60 75 9 0 105 120 135 150 1 65 180 195 12 0 10 0 80 60 40 100 200 500 1000 2000 5000 10000 density at 29 years (stems ha−1) recorded yield simulations 200 500 1000 2000 5000 0 50 10 0 15 0 20 0 density at 29 years (stems ha−1) m er ch an ta bl e yi el d ( m 3 ha −1 ) 40 years 50 years 60 years 70 years 80 years 90 years 550 750 600 850 650 1000 650 1200 700 1200 700 1200 figure s3: simulation results, presented as a contour plot of merchantable yield (m3 ha–1) versus density and stand age (mean of 100 simulations). both contour lines and grey saturation of background follow the ‘topography’ of merchantable yield, with darker grey denoting higher yields. the dotted line presents the optimal densities for different stand ages. figure s4: simulation results presented as decadal changes in merchantable yield (m3 ha–1) across a gradient of stem densities (mean of 100 simulations). bold lines and associated figures indicate yields and density ranges spanning >98% of the peak yield at each age. easdale et al. new zealand journal of forestry science (2022) 52:16 page 16 figure s5: contour plot of merchantable yield for simulations that assume constant variability in tree growth across densities (a; mean of 100 simulations) and the corresponding contrast between recorded yield (points) and simulated yield (lines) at 48 years (b; open circles and thin lines) and at 58 years (b; filled circles and thick lines). note that predictions greatly exceed observations at high density. contour plot of merchantable timber yield for simulations that do not incorporate variability in tree growth (c) and corresponding contrast between recorded and simulated yield at 48 and 58 years (d). note that predictions have a sharp boundary and equate to zero when predicted diameters are smaller than the merchantable threshold (30 cm). s ta nd a ge (y ea rs ) 40 60 8 0 10 0 1 20 140 1 60 1 80 20 0 2 20 240 12 0 10 0 80 60 40 100 200 500 1000 2000 5000 10000 density at 29 years (stems ha−1) (a) ● ● ● ● ● ● ● ● ● ● ● ● ● 100 200 500 1000 2000 5000 10000 0 50 10 0 15 0 density at 29 years (stems ha−1) m er ch an ta bl e yi el d ( m 3 ha −1 ) ●● ●● ●● ●● ●● ●● ●● ●● ●● ●● ●●●● ●● (b) 48 yrs 58 yrs ● ●● r ec or d. s im ul . s ta nd a ge (y ea rs ) 20 40 6 0 8 0 1 00 1 20 1 40 1 60 1 80 2 00 12 0 10 0 80 60 40 100 200 500 1000 2000 5000 10000 density at 29 years (stems ha−1) (c) ● ● ● ● ● ● ● ● ● ● ● ● ● 100 200 500 1000 2000 5000 10000 0 50 10 0 15 0 density at 29 years (stems ha−1) m er ch an ta bl e yi el d ( m 3 ha −1 ) ●● ●● ●● ●● ●● ●● ●● ●● ●● ●● ●●●● ●● (d) 48 yrs 58 yrs ● ●● r ec or d. s im ul . easdale et al. new zealand journal of forestry science (2022) 52:16 page 17 opportunities and limitations of exotic pinus radiata as a facilitative nurse for new zealand indigenous forest restoration adam s. forbes1*, david a. norton1, fiona e. carswell2 1school of forestry, university of canterbury, christchurch 8140, new zealand 2manaaki whenua – landcare research, po box 69040, lincoln, new zealand *corresponding author: adam@forbesecology.co.nz (received for publication 24 march 2019; accepted in revised form 16 june 2019) abstract background: we investigated the long-term potential of non-harvest pinus radiata plantations for the facilitation and restoration of a natural forest community dominated by indigenous woody species. we investigated the relationship between indigenous regeneration and light levels and the hypothesis that proximity to indigenous seed sources is critical. we studied nine pinus radiata stands of different ages located within kinleith forest, which is a large (ca. 66 000 ha) commercial exotic plantation forest located in new zealand’s central north island. methods: we constructed a chronosequence of p. radiata plantation stands aged 2–89 years to represent long-term natural forest regeneration following plantation establishment. we surveyed structural, compositional and contextual aspects of this secondary succession and compared these results with an old-growth indigenous forest reference site located within the study area. results: the exotic p. radiata canopy facilitated a regeneration trajectory characterised by shade-tolerant indigenous forest species. we found that the structure and composition of p. radiata understories were strongly influenced by stand age and proximity to indigenous forest. stand age was important from the perspective of creating shaded conditions for the establishment of shade-tolerant woody forest species. our results suggest that proximal indigenous forest was required for the consistent natural establishment of larger-fruited, bird-dispersed mature forest canopy species in p. radiata plantations. conclusions: our results showed that, even at ecologically isolated sites, the microclimate conditions created by plantation pinus radiata stands supported a suite of readily-dispersed indigenous forest plants. based on these results we suggest that non-harvest p. radiata stands provide an important opportunity for the restoration of indigenous forest communities in new zealand’s production landscapes. where restoration of forest composition similar to old-growth is the restoration objective, however, interventions might be necessary to direct and accelerate the secondary forest succession. further replicated study is required into the relationship between native forest proximity and understorey regeneration patterns. new zealand journal of forestry science forbes et al. new zealand journal of forestry science (2019) 49:6 https://doi.org/10.33494/nzjfs492019x45x e-issn: 1179-5395 published on-line: 6 july 2019 © the author(s). 2019 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access keywords: biodiversity management; chronosequence survey; forest restoration; non-harvest plantation forest; pinus radiata; production landscape; seed dispersal; shade tolerance. forbes et al. new zealand journal of forestry science (2019) 49:6 page 2 introduction human habitation has caused vast reductions in the extent of natural forest cover across most of the world (crowther et al. 2015). in new zealand, for example, an estimated 71% (14 million ha) of natural forest cover has been cleared (ewers et al. 2006). this loss of forest habitat has diminished many aspects of forestrelated biodiversity (gaston et al. 2003; brockerhoff et al. 2008a; gardner et al. 2009) and the principal responses have been the formulation of strategies to reduce further forest loss, to ensure that forests are managed sustainably, and to conserve forest biodiversity (e.g. the bonn challenge; iucn, 2011). as of 2010, an estimated 6.6% (265 million ha) of the global forest area was planted forests, and new zealand ranked ninth among the top-ten countries in absolute plantation area (forest stewardship council [fsc] 2012). as increasing areas are afforested for timber production, methods to preserve or enhance indigenous biodiversity in these planted landscapes are also required (brockerhoff et al. 2001; carnus et al. 2006). existing research has examined the role of commercial plantations in facilitating the natural regeneration of indigenous forest flora (lamb et al. 2005; brockerhoff et al. 2008a). however, while commercial plantations might contribute ecologically by increasing landscape connectivity, buffering indigenous remnants, or providing sometimes scarce forest habitats (brockerhoff et al., 2008a), the long-term potential for the recruitment of mature forest canopy species is usually truncated by the disturbance associated with the plantation harvest and subsequent site preparation operations (allen et al. 1995; chapman & chapman 1996). an estimated 90% of plantation forest in new zealand comprises pinus radiata (pinaceae; ministry of primary industries [mpi] 2019) and these commercial plantations operate on a clearfell rotation of 25–30 years (mpi 2019), meaning most of the indigenous regeneration is lost between harvest rotations. alternatively, for various environmental (e.g. promoting water quality, soil conservation, or carbon sequestration), social (e.g. a shift in community aspirations), or economic (e.g. a high degree of technical difficulty or risk associated with harvesting) reasons, some p. radiata plantations are unlikely to be harvested. these “non-harvest” stands present opportunities for the restoration of indigenous forest species, with associated benefits to indigenous biodiversity (norton & forbes 2013). several european studies have assessed the role of p. radiata plantations as sites for the restoration of temperate forest species. onaindia et al. (2013) assessed the potential for p. radiata plantations to be used as a passive restoration tool for the natural establishment of indigenous mixed-quercus spp. forest in spain. they found that with increasing plantation age, p. radiata plantations became compositionally more similar to natural quercus forest. diversity indices revealed, however, that p. radiata forest maintained a greater number of both rare and dominant species compared to natural quercus stands, and several species indicative of old-growth forest were not found in the oldest p. radiata stands (aged 40 years; onaindia et al. 2013). an earlier study emphasised the role of adjacent indigenous forest remnants in providing a source of propagules to enrich regeneration patterns in p. radiata plantations (onaindia & mitxelena 2009). within pinus spp. plantations, two key predictors of understorey regeneration patterns are recognised, being: (1) stand age (keenan et al. 1997; ogden et al. 1997; onaindia & mitxelena 2009); and (2) the proximity to indigenous forest seed sources (zanne & chapman 2001; onaindia & mitxelena 2009; zamora et al. 2010). age-related changes in the p. radiata plantation canopy structure alter understorey illumination (porté et al. 2004). these temporal changes in understorey light levels make stand age an important predictor of understorey regeneration patterns (ogden et al. 1997; brockerhoff et al. 2003), and indicators of forest development such as woody species richness have been found to be positively correlated with p. radiata plantation age (keenan et al. 1997; ogden et al. 1997; onaindia & mitxelena 2009). the proximity of indigenous forest seed sources to pinus spp. plantations is also of importance to patterns in understorey regeneration (zanne & chapman 2001; onaindia & mitxelena 2009; zamora et al. 2010). this is particularly relevant to exotic pinus spp. plantations in new zealand, where the mainland forest flora is predominantly bird dispersed (ca. 70%; clout & hay 1989). most of this dispersal, however, operates over short distances (e.g. typically within several hundred metres from the parent tree; williams 2006; wotton & kelly 2012). it then follows that exotic conifer plantations, which do not produce nectar or fruit to attract key seed dispersing bird species, would have a limited ability to attract dispersers (clout & gaze 1984). this may impose limitations on the probabilities of the seed of indigenous species reaching and establishing within ecologically isolated pinus spp. plantations. here we investigated the long-term potential of nonharvest p. radiata plantations in recruiting indigenous forest flora and developing a forest community dominated by indigenous woody species, characteristic of mature natural forest. we hypothesised that the establishment of indigenous species is dependent on suitable light levels in the plantation understorey, and also on the ability of species to reach the site via dispersal from nearby indigenous forest. corresponding to modifications of the understorey light environment from the rapidly closing p. radiata canopy, we predicted an initial invasion of light-demanding early-successional woody species, followed by an accrual of shade-tolerant indigenous angiosperm forest canopy species. we also predicted that mature p. radiata stands will provide suitable establishment sites for angiosperm canopy species with high dispersal capability, but that many angiosperm species that are less dispersal capable will be limited in occurrence by the isolation of the p. radiata stand from an indigenous forest seed source, regardless of p. radiata stand age. we also expected that the relatively light-demanding conifer species of the podocarpaceae family, which normally fill prominent canopy and emergent tiers in new zealand’s natural forests, would not be present in older p. radiata stands due to either dispersal limitation or the heavy shade cast by a combination of both the p. radiata canopy and dense understorey vegetation. to represent natural forest regeneration over a period of nine decades following plantation establishment, we surveyed structural and compositional aspects over a chronosequence of p. radiata plantation stands aged 2–89 years and compared these results with an old-growth indigenous forest reference site located within the study area. methods study area the study was undertaken in kinleith forest, which is a large (ca. 66 000 ha) commercial exotic plantation forest located in new zealand’s central north island (38°23’28”s 175°57’40”e). the climate is cool and moist. annual average air temperature (1931–1990 average from tokoroa in the middle of the study area; national institute of water and atmospheric research [niwa] 2015) is 12°c. annual average rainfall (1931– 2003 average) is 1485 mm. soils of the study area are pumice and were formed from tephra parent materials (table 1; landcare research [lcr] 2015). these soils are characteristically coarse textured and free draining. topography is variable and elevations range from 350 to 550 m above mean sea level (table 1). mixed conifer-angiosperm forest is characteristic of the natural forests of the study area (leathwick & mitchell 1992). these forests typically comprise the angiosperm mature forest canopy species: beilschmiedia tawa (lauraceae), hedycarya arborea (monimiaceae), knightia excelsa (proteaceae), and weinmannia racemosa (cunoniaceae); and podocarpaceae conifers: dacrydium cupressinum, podocarpus totara, prumnopitys ferruginea, and prumnopitys taxifolia. however, the extent of these forests within the study area has been reduced to a point where only small forest remnants exist. larger tracts of indigenous forest occur > 15 km to the south (pureora forest park) and the north (kaimai-mamaku forest park) of the study area. stand selection for the purposes of a space-for-time substitution, we selected nine pinus radiata plantation stands ranging in age from 2–89 years since establishment (table 1). pinus radiata stand selection aimed to vary only stand age and kept the following factors as uniform as possible: underlying soil type, topography, aspect, slope, consistency of silviculture interventions, stand size and shape sufficient to avoid edge effects, and proximity to other chronosequence stands (table 1). field survey in each of the nine stands, and in the indigenous forest reference site, four 10 × 10 m understorey vegetation plots were randomly positioned (random gps coordinates were derived using numbered grids superimposed on satellite images), then located using a handheld gps (garmin gpsmap 64s) and then surveyed. plots were separated by a minimum of 50 m, and to avoid edge effects, plots were not located within 30 m of the plantation edge. woody understorey plant cover, height in tiers, density and tree diameter (>2 cm diameter at 1.35 m above ground level; dbh) were assessed within understorey plots using the recce method (hurst & allen 2007). cover-abundance for each understorey species was estimated using the scale: 1 = <1%; 2 = 1–5%; 3 = 6–25%; 4 = 26–50%; 5 = 51–75%; and 6 = 76–100%. understorey species were recorded when they had live foliage present within the height tiers: <0.3; >0.3–2; >2–5; >5–12; >12–25; >25 m. woody epiphytes were tallied separately. all saplings (<2 cm diameter at breast height (dbh) & >1.35 m tall) and seedlings (<1.35 m tall) occurring within each 10 × 10 m understorey plot were identified to species level and tallied. seedlings were tallied according to the height classes: <15, 16–45, 46–75, 76–105, 106– 135 cm. all p. radiata trees occurring within a circular 0.03 ha plot centred on the 10 × 10 m understorey plot were identified and their dbh and height recorded. tree heights were measured using a vertex iii hypsometer. at each plot centre, we assessed topographic exposure using the meso-scale topographic index (mcnab 1993), which required measurement of eight equidistant slope to horizon measurements from each plot centre. also measured at each plot were aspect, slope, physiography, canopy structure, and light transmission. canopy structure and light transmission were estimated using hemispherical (fisheye) photographs, taken from plot centres at 1.35 m above ground level, using a pentax k200 dslr camera fitted with a sigma 4.5 mm circular hemispherical lens. the proximity of p. radiata stands to indigenous forest was mapped using a combination of aerial photography and field inspections to verify vegetation extent and composition. plant nomenclature and dispersal mode follows the new zealand plant conservation website (www.nzpcn.org.nz, accessed september 1, 2015). indigenous forest proximity the mean proximity index (px; mcgarigal & marks 1995) was used to provide a measure of proximity to indigenous forest for each plot surveyed within the p. radiata chronosequence. the original px is the sum of the ratio of indigenous forest patch size to distance from a focal patch to each indigenous forest patch (squared) within a search radius. however, we adapted the index to suit our plot-based study. rather than calculating the edge-to-edge distance between each indigenous patch and the focal p. radiata stand, we calculated the distance from the indigenous forest edge to the p. radiata plot centre. therefore, we expressed px as the sum of indigenous forest patch area (m2) within the search radius, divided by the sum of all indigenous edges to p. radiata plot (squared (m2)) distances, for all indigenous forest patches whose edges were within 1 km (linear) of the focal plot centre. we chose 1-km lineal distance as the search radius as the distance further than which common new zealand indigenous forest birds are unlikely to reliably disperse fleshy-fruited propagules (wotton & mcalpine 2015). sites with high px value indicate higher levels of indigenous cover proximal to the survey plot. forbes et al. new zealand journal of forestry science (2019) 49:6 page 3 kruskal tests were applied in this regard to percentage canopy openness, percentage par transmission, woody indigenous seedling densities, sapling densities, and indigenous tree basal area. where kruskal tests returned statistically significant results at α=0.05, pairwise posthoc tests were carried out using nemenyi tests. any association between indigenous tree basal area within p. radiata stands and p. radiata stand age was assessed using the nonparametric spearman’s rank-order correlation. species turnover (beta diversity) was assessed from woody species presence-absence data using jaccard similarity that was calculated using the simba package (jurasinski & retzer 2012). the comparison of jaccard similarity between paired stand ages provided an assessment of changes in the level of continuity in species composition over time. the possible spatial distribution of species across a pair of quadrats is expressed as follows: a´ representing the total number of species common between both plots, b´ representing the total number of species that occur in the neighbouring plot but are absent from the focal plot, and c´ representing the total number of species that occur in the focal plot but not in the neighbouring one. species turnover as expressed by a´, b´, and c´ was visualised in similarity space using a ternary plot in accordance with the approach advocated by koleff et al. (2003). nonmetric dimensional scaling (nmds) was used to make inferences regarding compositional shifts, both across the p. radiata chronosequence and in relation to the indigenous forest reference site, using the metamds forbes et al. new zealand journal of forestry science (2019) 49:6 page 4 table 1. forest stand details of the kinleith forest pinus radiata plantation chronosequence, central north island, new zealand statistical analysis all statistical analyses were undertaken using r (r development core team team [r] 2015). vegetation plots were treated as individual units in the analyses. the effect of pinus stand age on p. radiata basal area and indigenous tree fern basal area were assessed using generalised linear regression, applying the gaussian family, and logarithm (for p. radiata) and identity (for tree ferns) links, with natural log conversion of the explanatory variable (p. radiata stand age). all regression models were tested for linearity, normality of the residuals, and homoscedasticity using diagnostic plots. canopy structure and light transmission data were extracted from hemispherical photographs using the imaging software gap light analyzer (gla, version 2; frazer et al. 1999). for gla modelling, the growing season was defined as 1 september–31 march, the default solar constant used was 1367 wm–2, and the default cloudiness index, spectral fraction, and beam fraction were all set at 0.5. the relationships between both percentage canopy openness and percentage photosynthetically active radiation (par), and p. radiata stand age, were estimated using nonparametric generalised additive models (gam), through use of the mgcv package (wood 2011). these gams were applied using beta family and logit link function. differences among p. radiata stands of varying age, and between p. radiata stands and the indigenous reference site, were assessed using nonparametric kruskal-wallis rank sum tests for the following variables. year planted age silvicultural treatment current rotation soil type^ slope aspect meso-scale topography elevation coordinates 2013 2 p = 1; r = 1 3 tph + nah 3-18°, s 6.3 350 38°24’25”s 175°54’33”e 2009 5 p = 1; r = 1 3 tph + nah 0-22°, e 12.0 350 38°23’53”s 175°55’45”e 1999 15 t = 7 2 tph + nah 22-28°, e 11.4 450 38°22’38”s 175°56’15”e 1989 25 t = 6, 8 2 tph + nah 13-33°, n 9.8 350 38°24’06”s 175°55’14”e 1979 35 t = 7 2 oih 17-30°, n 19.2 450 38°22’48”s 176°00’40”e 1970 44 – 1 w 25-35°, n 17.4 350 38°06’47”s 175°56’09”e 1954 60 t = 23 2 tph 4-25°, s 12.8 350 38°23’57”s 175°58’22”e 1927 86 t = 16 1 tph 0-25°, e 9.1 450 38°24’53”s 176°03’12”e 1925 89 – 1 tpd 0-6°, e 4.1 350 38°16’40”s 175°50’18”e n/a nf n/a n/a na + oi 5-24°, s 6.5 550 38°27’43”s 176°02’58”e note. silvicultural treatments are: “p” = herbicide land preparation, “r” = herbicide release spray, “t” = thinning, “–” = no data available; the numeral(s) indicate the plantation age (yrs) at the time of silvicultural treatment(s). elevations given are in 100 m vertical bands. aspects given are the stand mean represented as the nearest cardinal direction. ^soil abbreviations are: tph = taupo hill soils, tpd = taupo deep sand, nah = ngakuru hill soils, na = ngakuru loam, oi = oranui sand, oih = oranui hill soils, w = waiohotu (silty loam). “nf” = natural forest reference site. coordinates are shown in the wgs84 coordinate system. function of the vegan package (oksanen et al. 2008). for nmds ordination analysis, species importance values were calculated following allen et al. (1995). the following weights were allocated to the recce cover classes (cover class = weight): 1 = 1.0; 2 = 2.0; 3 = 3.0; 4 = 4.0; 5 = 5.0; and 6 = 6.0. epiphytes were given a nominal weighting of 0.5. differences in ordination space among the different aged p. radiata stands and the indigenous forest reference site were tested using permutational multivariate analysis of variance with the adnois function in vegan. the contribution of within-site variability was assessed using the function betadisper in vegan. the nmds ordination was further examined through similarity percentage analysis (simper; clarke 1993) to distinguish which species discriminate between stand ages and between the two forest types. results pinus radiata basal area and canopy height chronosequence stand summary statistics are given in appendix s1. stand age was a significant predictor of p. radiata basal area across the chronosequence (f1,7 = 74.466, p<0.001; appendix s2). predicted p. radiata basal area increased from 9.7±4 m2 ha–1 at year 2, to 49.6±5 m2 ha–1 at year 30, and to 95.3±7 m2 ha–1 at year 90. pinus radiata canopy height increased rapidly during the initial 25 years, and more gradually thereafter, reaching an eventual mean canopy top height of 53.8±1.6 m (n= 32) in 86and 89-year-old stands (appendix s1). canopy openness and total par transmission percentage canopy openness differed significantly among p. radiata stand ages (χ2(9) = 34.013, p<0.001). the 2and 5-year-old stands had significantly greater canopy openness compared to the older stands of the chronosequence. predicted mean percentage canopy openness decreased from initially completely open canopy, to 33±4% 15 years after establishment, and to 12±3% at 20 years, beyond which canopy openness plateaued at about 8% for the remainder of the chronosequence. the amount of total par transmitted to the forest understorey also differed significantly across the chronosequence (χ2(9) = 32.103, p<0.001). over the first 15 years, total par transmission was reduced from about 100% to 30±6% (fig. 1). the degree of light transmission to the forest understorey plateaued 25 years after plantation establishment; beyond 25 years, only about 25% of total available light reached the forest understorey. no significant differences in either canopy openness (all p>0.14) or total par transmission (all p >0.23) occurred among stands aged 15 years or older. nor were there significant differences in either canopy openness (all p>0.22) or total par transmission (all p >0.14) between stands aged 15 years or older and the indigenous forest reference site. forest understorey vegetation structure woody seedling density indigenous woody seedling densities differed significantly among p. radiata stand ages, and also differed from the seedling densities in the indigenous reference site (χ2(9) = 27.348, p=0.002; appendix s3). the highest indigenous woody seedling density within pinus stands was found in the 44-year-old stand (3 050±581 stems ha-1), where a significantly greater stem density was achieved compared to either 2-year-old (25±25 stems ha-1; p=0.021) or 86-year-old pinus stands (50±50 stems ha-1; p=0.025). indigenous woody seedling densities in all stands other than those aged 2 and 86 years were not significantly different to the seedling density in the indigenous forest reference site (3 750±1475 stems ha-1; all p>0.35). exotic seedlings were found in low densities in the 5-year-old (50±29 stems ha-1), 15-year-old (75±49 stems ha-1), and 25-year-old (125±95 stems ha-1) stands and were absent from all other stands of the survey. sapling density the density of indigenous saplings differed significantly across the chronosequence (χ2(9) = 26.58, p=0.002). the contributing differences were between the 2-yearold stand (where saplings were absent) and both the 35-year-old stand (1525±394 stems ha–1; p=0.05; n=61) and the indigenous forest reference site (2525±812 stems ha–1; p=0.023; n=101; data not shown). exotic saplings were present in only the 5-, 15-, and 25-year-old stands; and were most numerous in the 5-year-old stand (625±239 stems ha–1; n=25) compared to either the 15year (175 stems ha–1; n=7) or 25-year stands (n=1). indigenous trees and tree ferns there was no significant difference in indigenous tree basal area among the p. radiata stands (χ2(7) = forbes et al. new zealand journal of forestry science (2019) 49:6 page 5 figure 1: fitted generalised additive model of the percentage total photosynthetically active radiation (par) transmission assessed using hemispherical photographs taken at 1.35 m above ground level in a chronosequence of nine pinus radiata plantation stands age. dashed lines indicate the 95% ci. mean value (±1se) from the old-growth natural forest (“nf”) reference site shown for comparative purposes. 13.17, p=0.068), nor was there a significant association between indigenous tree basal area and stand age (rs(6) = 0.619, p=0.115; fig. 2a). adult tree ferns entered the chronosequence between years 15 and 25. stand age was not a significant predictor of tree fern basal area across the chronosequence (f1,4 = 6.170, p=0.068, fig. 2b). over the chronosequence, the tree-fern community was dominated by dicksonia squarrosa (dicksoniaceae; range = 400±70– 1125±433 stems ha–1), with fewer numbers of cyathea medullaris (cyatheaceae; range = 0–475±95 stems ha–1) and of c. dealbata (cyatheaceae; 0–125±75 stems ha–1), and only a small number of d. fibrosa (dicksoniaceae; n = 2). in the older (i.e. 44–89 years old) p. radiata stands, mean tree fern heights were d. squarrosa = 3.8±0.2 m (n =119), c. medullaris = 6.1±0.5 m (n=40), and c. dealbata = 3.4±0.8 m (n=8). species richness stand age was a significant predictor of indigenous woody species richness (s) in plantations across the chronosequence (f1,6 = 9.745, p=0.021; appendix s4). meso-scale topography was not a significant predictor of s. early in the chronosequence, predicted s increased rapidly until approximately 20-years old, and more gradually thereafter (fig. 3). we found evidence based on patterns of seedling density, species turnover, and compositional data to suggest a positive relationship between s and the close proximity to indigenous forest seed source, but our lack of stand-level replication means we cannot determine a statistically significant cause-and-effect relationship. there is a need for further research on the role of proximity as a predictor of s. the 44-year-old stand had both relatively high s and the highest proximity to seed source, suggesting a positive effect on s from proximity. exotic woody species richness was only a minor feature of the chronosequence, being limited to only 5-year-old (s=1.5±0.3), 15-year-old (s =1.25±0.3), 25-year-old (s=0.5±0.3), and 44-year-old (s=0.5±0.3) stands. forbes et al. new zealand journal of forestry science (2019) 49:6 page 6 figure 2: basal area of indigenous (a) trees and (b) tree ferns across a chronosequence of nine pinus radiata plantation stands aged 2–89 years, kinleith forest, central north island, new zealand. for comparative purposes, indigenous tree basal area at the old-growth natural forest was 164±33 m2 ha–1 (not shown in fig. 2a) and the tree fern basal area from the old-growth natural forest (“nf”) is shown in fig. 2b. error bars = ±1se; dashed lines indicate the 95% ci. a b figure 3: predicted indigenous woody species richness (s) as a function of stand age and mesoscale topographic exposure, across a chronosequence of nine pinus radiata plantation stands aged 2–89 years, kinleith forest, central north island, new zealand. for comparative purposes, the s from an oldgrowth natural forest (“nf”) reference site is shown. error bars = ±1se; dashed lines indicate the 95% ci. species turnover and composition patterns in woody species turnover were related to stand age and proximity to indigenous forest. the 2-year-old stands were species poor and shared few species with the older stands. these youngest stands featured lightdemanding colonisers and were grouped in jaccard similarity space near the highest extent of axis b´ and the lowest extent of axis a´ (appendix s5), signalling a high degree of species gain and low continuity in this young phase of the chronosequence. fiveand 15-yearold stands still showed <50% similarity with stands of greater age. stands >25-years of age showed the greatest between-stand continuity (i.e. sharing similarity values of 50% or greater). the 44-year-old stand was a notable exception. it showed only about 30% similarity with stands of greater age, and about 55% of species were discontinuous between the 44-year-old stand and either of the three older stands. this result was driven by the high rate of occurrence of indigenous forest species in the 44-year-old stand. species composition the pinus chronosequence comprised a total of 29 species of trees, tree ferns, and shrubs in the understorey (appendix s6). the shade-tolerant mature forest canopy and emergent species, namely: b. tawa, h. arborea, k. excelsa, litsea calicaris (lauraceae), and w. racemosa, were present in some mature p. radiata stands, particularly in the 44-year-old stand where indigenous forest was in close proximity (fig. 4; appendix s7). including the planted pines, four exotic tree and shrub species were present. the indigenous forest reference forbes et al. new zealand journal of forestry science (2019) 49:6 page 7 site featured eight additional woody species that were not found in the p. radiata chronosequence. however, fewer plots were measured in the indigenous forest (four plots), which is likely to underestimate species richness when compared to the stands where sampling was more extensive (32 plots). species composition differed significantly among (f9,29 = 8.921, p=0.001, r2 = 0.73) but not within (f9,29 = 0.687, p=0.715) the pinus and indigenous forest stands (fig. 4). changes in species composition over the first 15 years of the chronosequence were distinguished by the gain and subsequent loss of the light-demanding colonising species including coriaria arborea (coriariaceae), aristotelia serrata (elaeocarpaceae), and buddleja davidii (scrophulariaceae; exotic; table 2; fig. 4). between 15 and 25 years, increased abundance of the tree fern d. squarrosa was the most distinguishing floristic change, along with increases in the abundance of the indigenous tree species schefflera digitata (araliaceae) and coprosma robusta (rubiaceae). further accumulation of indigenous forest tree and tree fern species was apparent over the period 25–35 years. most notable from the simper analysis were the gains in abundance of melicytus ramiflorus, cyathea medullaris, and brachyglottis repanda. hedycarya arborea was the dominant tree in the indigenous forest reference site and was a distinguishing feature of the 44-year-old pinus stand. the composition of the oldest p. radiata stand was most clearly distinguished from the indigenous forest reference site by the absence of h. arborea and the high abundance of both d. squarrosa and pseudopanax arboreus. the 35and 44-year-old pinus radiata stands were closest in the ordination space to the indigenous forest reference site (fig. 4). discussion forest regeneration processes as predicted, we found that the structure and composition of pinus radiata understories were strongly influenced by stand age, and we found some, albeit limited, evidence of the importance of proximity to indigenous forest. over the first 15-year period of the forestry cycle, conditions were most suitable for the light-demanding colonising species such as coriaria arborea, a. serrata, and the exotic buddleja davidii. then as p. radiata canopy cover and shade increased, these species were replaced by generalist forest tree species with greater shade tolerance such as s. digitata, melicytus ramiflorus, and brachyglottis repanda. although these are relatively early-successional species, they do have the ability to form a forest canopy. although not the focus of this study, the successional direction observed is consistent with the acquisition of traits associated with latersuccessional species. through the chronosequence there was a shift towards species with larger seeds, greater stature, and greater longevity (weiher et al. 1999). stand age was a significant predictor of s. where indigenous forest was in close proximity, we found both a greater density of indigenous woody seedlings and a greater abundance of mature forest canopy species of closer compositional similarity to indigenous forest. figure 4: nonmetric dimensional scaling (nmds) ordination of species composition across chronosequence of nine pinus radiata plantation stands aged 2–89 years, kinleith forest, central north island, new zealand. the composition of an old-growth natural forest reference site is also included, shown by triangles. six-letter species codes are described in appendix s6. table 2: confusion matrix forbes et al. new zealand journal of forestry science (2019) 49:6 page 8 sp ec ie s st an d ag e 2– 5 6– 15 16 –2 5 26 –3 5 36 –4 4 45 –6 0 61 –8 6 87 –8 9 90 –n f co nt r. cu m . co nt r. cu m . co nt r. cu m . co nt r. cu m . co nt r. cu m . co nt r. cu m . co nt r. cu m . co nt r. cu m . co nt r. cu m . co r ar b 26 ±7 29 16 ±4 22 a r is er 22 ±1 4 52 14 ±8 40 b u d da v 15 ±1 3 69 9± 6 52 d ic sq u 16 ±7 23 7± 4 25 6± 4 36 5± 3 29 11 ±3 30 sc h di g 9± 1 36 co pr ob 8± 3 49 m el ra m 10 ±4 17 6± 2 13 cy a m ed 7± 4 30 10 ±3 15 8± 6 27 5± 3 40 b r a re p 6± 5 41 5± 4 25 11 ±5 16 7± 4 16 h ed ar b 5± 2 35 6± 2 35 13 ±5 16 ps ea rb 8± 3 40 ta b le 2 . si m ila ri ty p er ce nt ag e (s im pe r ) an al ys is o f s pe ci es c om po si ti on fr om a c hr on os eq ue nc e of n in e pl an ta ti on s ta nd s ra ng in g in a ge 2 –8 9 ye ar s lo ca te d in k in le it h fo re st , c en tr al n or th is la nd , n ew z ea la nd . t he th re e sp ec ie s di sc ri m in at in g m os t b et w ee n co ns ec ut iv e st an d ag es ( % c on tr ib ut io n = “c on tr .”) , a nd th e cu m ul at iv e pe rc en ta ge d is cr im in at io n (“ cu m .”) a re g iv en . o ld -g ro w th n at ur al fo re st ( “n f” ) re fe re nc e si te in cl ud ed . n ot e. n um er al s in b ol d si gn ify a g ai n in a bu nd an ce , n on -b ol d si gn ifi es a lo ss in a bu nd an ce , u nd er lin ed s ig ni fie s no c ha ng e in a bu nd an ce . s pe ci es c od es a re g iv en in a pp en di x s6 this result was most apparent in the 44-year-old stand where areas of indigenous forest were as close as 260 m from the sample plots and both species richness and woody seedling density were the highest of any of the plantation ages. the 44-year-old stand had a northern aspect and silty loam soils, and these attributes might also have assisted the establishment and growth of woody indigenous forest species due to warm and sunny microclimate and higher soil quality. in the absence of replicated stand ages, we are unable to generalise our findings from the 44-year-old stand. further research is required to investigate the effects of differing proximity and composition of indigenous forest seed sources on understorey regeneration patterns in p. radiata plantations. in addition to the generalist shade-tolerant forest tree species, larger-fruited, bird-dispersed species such as beilschmiedia tawa, h. arborea, and litsea calicaris were found in high abundances in the 44-year-old stand. yet, these species were either in low numbers or were absent from stands of greater age. this finding supports our expectations that stand age is important from the perspective of creating shaded conditions for the establishment of shade-tolerant species. however, we suggest that proximal indigenous forest is required for the consistent natural establishment of larger-fruited, bird-dispersed mature forest canopy species in p. radiata plantations. our results regarding the importance of stand age and the proximity to indigenous forest are consistent with existing national (allen et al. 1995; brockerhoff et al. 2003) and international literature. in the wet tropics of northern queensland, australia, pinus caribaea plantations aged 5–31 years showed a significant increase in tree species richness with age (r2 = 0.788), and 10–50-year-old p. caribaea plantations supported a total of 45 species of indigenous trees and shrubs, with the proportion of later-successional species increasing with age (keenan et al. 1997). in the temperate ethiopian highlands, seed dispersal from adjacent areas of natural forest was identified as being important to enable incorporation of later-successional species into the understories of pinus patula plantations (senbeta et al. 2002). the importance of nearby seed sources for understorey regeneration of new zealand’s ca. 240 woody plant species occurring in mainland forests, about 70% are bird dispersed (clout & hay 1989), and of the 21 tree and shrub species surveyed from the p. radiata chronosequence, 16 (76%) were dispersed by birds. aside from less frequent long-distance dispersal, effective bird and wind dispersal operates over typically short distances in new zealand. wotton and kelly (2012) found that mean dispersal distance of b. tawa by new zealand pigeon (hemiphaga novaeseelandiae) was 95±171 m. only 21% of seeds were dispersed >100 m, and <1% of seeds were dispersed >1 000 m. we found numerous seedlings of b. tawa in the 44-year-old p. radiata understorey located close to indigenous seed sources, whereas b. tawa was scarce or absent from understories of more isolated stands. new zealand pigeon is the only extant native bird species capable of dispersing the large fruit produced by b. tawa (clout & hay 1989). given the presence of b. tawa, and considering the species dependence on new zealand pigeon for seed dispersal, we can conclude that the dispersal of b. tawa was by frugivory from adjacent natural forest, and therefore that bird dispersal was operational at the 44-year-old stand, where species richness and seedling densities were greatest. studies of the ubiquitous european blackbird (turdus merula) have also shown that most seeds are dispersed by this species within only 50 m, rarely up to 1 km or more (williams 2006). nectar-feeders and frugivores are major components of new zealand’s avifauna, and given the pinus genus does not provide nectar or fruit resources, important insectivorous and frugivorous dispersers are only sometimes present in p. radiata plantations (clout & gaze 1984). therefore, the presence of mature indigenous vegetation in the surrounding landscape matrix is important for both the presence of indigenous dispersal vectors and the availability of indigenous forest propagules. this importance is reflected in our results. factors potentially affecting podocarp regeneration we predicted that the relatively light-demanding (in seedling and sapling stages) podocarp species would not be present in mature p. radiata plantations due to either heavy shading or ecological isolation from indigenous forest seed sources. even where mature indigenous forest was proximal, we found no conifer species, and while this makes it difficult to confirm the reasons for their absence, possible explanations are found in the existing literature. regeneration of new zealand’s long-lived podocarps is disturbance related, insofar as a competitive release is required for the successful establishment and growth of seedlings (mckelvey 1963; carswell et al. 2012). therefore, it is plausible that if seeds were dispersed to pinus radiata plantations, these conifer species would benefit from the disturbance of both the homogenous plantation canopy and any dense understorey growth to increase understorey light levels and stimulate podocarp seedling growth. we note however, both prumnopitys ferruginea and prumnopitys taxifolia have been found to inhabit multiple mature pinus radiata plantations across new zealand (brockerhoff et al. 2003). podocarp dispersal is dependent on frugivory (beveridge 1964), so in our study the absence of podocarps in the p. radiata chronosequence might also be attributed to the fragmented nature of the surrounding indigenous forest landscape, and to a general scarcity of podocarp seed sources in proximity to mature p. radiata stands. existing studies have identified two reasons for compositional differences between mature p. radiata stands and natural forest: insufficient stand age required to provide conditions suitable for regeneration requirements (onaindia et al. 2013), and ecological isolation (onaindia et al. 2009). interventions to create more heterogeneous light conditions in p. radiata forbes et al. new zealand journal of forestry science (2019) 49:6 page 9 plantations might assist with the recruitment of missing mature forest tree species (onaindia et al. 2013). there are grounds to expect canopy interventions (creation of canopy gaps or thinning) or understorey manipulations would benefit podocarp establishment and seedling growth in mature p. radiata plantations (tulod et al. 2018). our results strongly suggest that birds were dispersing indigenous forest seeds into the 44-year-old stand. however, sufficient stand age, proximal indigenous forest sources, and suitable light levels such that might be created by canopy gaps are all probable prerequisites for the reliable establishment of new zealand’s longlived podocarps in p. radiata plantation forests. the scarcity of mature forest canopy species colonising mature p. radiata plantations, and in particular isolated p. radiata sites, has implications for the composition of future forests at these sites. we consistently found generalist and small tree species colonising mature p. radiata understories, such as b. repanda, pseudopanax arboreus, pittosporum tenuifolium, s. digitata, myrsine australis, and melicytus ramiflorus. the pinus radiata understories also featured a high density of tree ferns. where mature forest seed sources were proximal, we found the mature forest canopy species b. tawa, h. arborea, k. excelsa, and litsea calicaris. however, even in those instances, we found no podocarps. this absence of podocarps in vegetation communities of similar ages to our pinus stands is consistent with the findings from studies of podocarp regeneration in indigenous broadleaved forest. where a podocarp seed source is available, a nurse crop of broadleaved angiosperms, or leptospermum scoparium or kunzea spp. (myrtaceae), is normally required for podocarps to establish themselves. in these natural successions the development of nursery conditions suitable for podocarp establishment is known to take 30–60 years or longer (mckelvey, 1955; cameron, 1960). these results suggest that, at ecologically isolated p. radiata stands, the future forest composition would be dominated by shorter-statured species with relatively short life expectancy, and that these forests would be distinguished from nearby natural tall forests by the absence of characteristic, very-long lived, oldgrowth forest canopy species, such as b. tawa and d. cupressinum. where mature forest canopy species were present in higher densities due to the close proximity of natural forest, the presence of species such as b. tawa, h. arborea, and k. excelsa provide an indication of convergence towards natural forest composition. however, the absence of podocarps means that a major compositional component of intact natural forests appears to be missing. where attainment of natural forest composition is the restoration objective, these results suggest that management interventions to address competition for light and seed dispersal limitation might be required to direct the regeneration processes operating in ecologically isolated mature p. radiata stands, in order for the stands to develop towards states more representative of natural forest. considerations when managing non-harvest stands as restoration sites even-aged monoculture plantations are usually initially low in structural diversity and this may limit understorey plant species richness (gamfeldt et al. 2013), recruitment of future canopy tree species on the site (royo & carson 2006), and the potential of wider forest biodiversity values (lindenmayer et al. 2006). heterogeneity in canopy cover through formation of canopy gaps, or variation in the vertical structure of the forest may take a long time to develop naturally (lust et al. 1998; kuuluvainen et al. 2002). in particular, we note that interventions mimicking the effects of disturbance, such as the creation of artificial canopy gaps, are likely to benefit the establishment and growth of podocarps. for these reasons, we suggest that early interventions such as the creation of small-scale artificial canopy gaps (by felling or stem poisoning) to increase canopy heterogeneity may provide important opportunities to maximise the biological diversity of exotic p. radiata plantations. in addition to the competitive shading effects of the plantation canopy, after several decades, additional competitive effects may result if dense understorey growth develops (royo & carson 2006), such as heavily shading herbaceous or fern dominated understories (de la cretaz & kelty 2002). this result might mean that even if p. radiata canopy openness increases with age, secondary competitive effects from a dense understorey will still limit forest floor regeneration processes. in the understories of mature spanish p. radiata plantations, species from the genera rubus and of the family poaceae grew rapidly, demonstrating fast lateral spread, achieving dominance and, as a result, were highly competitive on forest floor regeneration processes. management interventions were recommended to address these competitive effects on forest regeneration (onaindia et al. 2013). we found that adult tree ferns entered the chronosequence between 15 and 25 years and continued to increase in dominance over the next 50 years or more, reaching densities of 1125±433 stems ha–1. ogden et al. (1997) also found the understorey composition in older p. radiata stands was dominated by tree ferns (tree fern stem densities of 2000–3000 stems ha-1). in such conditions, the same competitive shading effects that have been reported from new zealand’s indigenous forests (coomes et al. 2005; gaxiola et al. 2008; brock et al. 2018) may apply, limiting regeneration processes in these exotic plantation forests. where the natural dispersal of old-growth forest canopy species is limited, their active introduction might be required. within kaingaroa forest, forbes et al. (2015) found that underplanting of podocarps under a degraded pinus ponderosa canopy accelerated forest succession by establishing the structural dominance of long-lived mature forest canopy species within only 50 years. similar interventions involving the underplanting of non-harvest plantation understories to incorporate late-successional species have also been successful in sri lankan (ashton et al. 1997), spanish (rodriguezforbes et al. new zealand journal of forestry science (2019) 49:6 page 10 calcerrada et al. 2008), and german (noack 2011) p. radiata plantations. implications for biodiversity conservation in new zealand our results show that, even at ecologically isolated sites, the microclimate conditions created by plantation pinus radiata stands can support a suite of readilydispersed indigenous forest plants. the structural and compositional aspects of indigenous forest regeneration in older p. radiata stands are broadly comparable to new zealand’s mid-successional natural forest communities (allen et al. 1995; ogden et al. 1997); and are superior in comparison to exotic pastoral landscapes. in such systems, indigenous forest species are typically restricted in distribution to degraded remnant habitat patches, or to scattered or solitary remnants, or have become locally extinct (norton & miller 2000; brockerhoff et al. 2008b). our results emphasise the importance of maintaining indigenous cover in production landscapes (craig et al. 2000) and show how the benefits of doing so can extend beyond the boundaries of those indigenous communities, by boosting biodiversity values in adjacent exotic planted forests (carnus et al. 2006). non-harvest p. radiata stands provide an important opportunity for the restoration of indigenous forest communities in new zealand’s production landscapes. however, where restoration of mature forest composition is the restoration objective, interventions might be necessary to direct and accelerate the secondary forest succession. consent for publication all authors consent for the publication of this manuscript and its content. additional files additional file 1: appendices 1–7. competing interests the authors declare that they have no competing interests. funding the financial assistance of an environment waikato environment initiative fund was gratefully received. acknowledgements we thank hancock forest management (hfm) for permitting access to kinleith forest, and both robin black and brendan morgan (both from hfm) for assistance with survey logistics and plantation forest inventory data. many thanks to vicki klein and penny andersen who assisted with the field survey. we thank the three anonymous reviewers and the coordinating editor, dr eckehard brockerhoff, for their comments on the draft of this manuscript. authors’ contributions af helped conceive the study, and designed the chronosequence, collected and analysed the data. led the manuscript preparation and responses to reviewers. dn and fc helped conceive the study and contributed to manuscript preparation. references allen, r.b., platt, k.h., & coker, r.e.j. (1995). understorey species composition patterns in a pinus radiata plantation on the central north island volcanic plateau, new zealand. new zealand journal of forestry science, 25(3), 301–317. ashton, p.m.s., gamage, s., gunatilleke, i.a.u.n., & gunatilleke, c.v.s. (1997). restoration of a sri lankan rainforest: using caribbean pine pinus caribaea as a nurse for establishing latesuccessional tree species. journal of applied ecology, 34(4), 915–925. beveridge, a.e. (1964). dispersal and destruction of seed in central north island podocarp forests. proceedings of the new zealand ecological society, 11, 48–55. brock, j.m., perry, g.l., lee, w.g., schwendenmann, l., & burns, b.r. (2018). pioneer tree ferns influence community assembly in northern new zealand forests. new zealand journal of ecology, 42(1), 1830. brockerhoff, e.g., ecroyd, c.e., & langer, e.r. (2001). biodiversity in new zealand plantation forests: policy trends, incentives, and the state of our knowledge. new zealand journal of forestry, 46, 31–37. brockerhoff, e. g., ecroyd, c. e., leckie, a. c., & kimberley, m. o. (2003). diversity and succession of adventive and indigenous vascular understorey plants in pinus radiata plantation forests in new zealand. forest ecology and management, 185(3), 307–326. brockerhoff, e.g., jactel, h., parrotta, j.a., quine, c.p., & sayer, j. (2008a). plantation forests and biodiversity: oxymoron or opportunity? biodiversity and conservation, 17(5), 925–951. brockerhoff, e.g., shaw, w.b., hock, b., kimberley, m., paul, t., quinn, j., & pawson, s. (2008b). re-examination of recent loss of indigenous cover in new zealand and the relative contributions of different land uses. new zealand journal of ecology, 32, 115–126. cameron, r.j. (1960). natural regeneration of podocarps in the forests of the whirinaki river valley. new zealand journal of forestry, 8(2), 337–354. carnus, j.-m., parrotta,j., brockerhoff, e.g., arbez, m., jactel, h., kremer, a., lamb, d., o`hara, k., walters, b. (2006). planted forests and biodiversity. journal of forestry, 104(2), 65–77. carswell, f.e., doherty, j.e., allen, r.b., brignall-theyer, m.e., richardson, s.j., & wiser, s.k. (2012). quantification of the effects of aboveground and belowground competition on growth of seedlings forbes et al. new zealand journal of forestry science (2019) 49:6 page 11 in a conifer–angiosperm forest. forest ecology and management, 269, 188–196. chapman, c.a., & chapman, l.j. (1996). exotic tree plantations and the regeneration of natural forests in kibale national park, uganda. biological conservation, 76(3), 253–257. clarke, k.r. (1993). non-parametric multivariate analyses of changes in community structure. austral ecology, 18(1), 117–143. clout, m.n., & gaze, p.d. (1984). effects of plantation forestry on birds in new zealand. the journal of applied ecology, 21(3), 795–815. clout, m.n., & hay, j.r. (1989). the importance of birds as browsers, pollinators and seed dispersers in new zealand forests. new zealand journal of ecology, 12(supplement), 27–33. coomes, d.a., allen, r.b., bentley, w.a., burrows, l.e., canham, c.d., fagan, l., forsyth, d. m., gaxiolaalcantar, a. u., parfitt, r. l., ruscoe, w. a., wardle, d. a., wilson, d. j., wright, e.f. (2005). the hare, the tortoise and the crocodile: the ecology of angiosperm dominance, conifer persistence and fern filtering. journal of ecology, 93(5), 918–935. craig, j., anderson, s., clout, m., creese, b., mitchell, n., ogden, j., … ussher, g. (2000). conservation issues in new zealand. annual review of ecology and systematics, 31(45), 61–78. crowther, t.w., glick, h.b., covey, k.r., bettigole, c., maynard, d.s., thomas, s.m., smith, j. r., hintler, g., duguid, m. c., amatulli, g., tuanmu, m. -n., jetz, w., salas, c., stam, c., piotto, d., tavani, r., green, s., bruce, g., williams, s. j., wiser, s. k., huber, m. o., hengeveld, g. m., nabuurs, g. j., tikhonova, e., borchardt, p., li, c-f., powrie, l. w., fischer, m., hemp, a., homeier, j., cho, p., vibrans, a. c., umunay, p. m., piao, s. l., rowe, c. w., ashton, m. s., crane, p. r., bradford, m.a. (2015). mapping tree density at a global scale. nature, 525(7568), 201. de la cretaz, a.l., & kelty, m.j. (2002). development of tree regeneration in fern-dominated forest understories after reduction of deer browsing. restoration ecology, 10(2), 416–426. ewers, r.m., kliskey, a.d., walker, s., rutledge, d., harding, j.s., & didham, r.k. (2006). past and future trajectories of forest loss in new zealand. biological conservation, 133(3), 312–325. forbes, a.s., norton, d.a., & carswell, f.e. (2015). underplanting degraded exotic pinus with indigenous conifers assists forest restoration. ecological management & restoration, 16(1), 41– 49. forest stewardship council. (2012). strategic review on the future of forest plantations. helsinki, finland: forest stewardship council. frazer, g.w., canham, c.d., & lertzman, k.p. (1999). gap light analyzer (gla) (version 2) [imaging software]. retrived from http://gap-light-analyzer. software.informer.com/2.0/ gamfeldt, l., snäll, t., bagchi, r., jonsson, m., gustafsson, l., kjellander, p., ruiz-jaen, m.c., fröberg, m., stendahl, j., philipson, c.d., mikusiński, g., andersson, e., westerlund, b., andrén, h., moberg, f., moen, j., bengtsson, j. (2013). higher levels of multiple ecosystem services are found in forests with more tree species. nature communications, 4, 1340. gardner, t.a., barlow, j., chazdon, r., ewers, r.m., harvey, c.a., peres, c.a., sodhi, n.s. (2009). prospects for tropical forest biodiversity in a human-modified world. ecology letters, 12(6), 561–582. gaston, k.j., blackburn, t.m., & goldewijk, k.k. (2003). habitat conversion and global avian biodiversity loss. proceedings of the royal society of london b/ biological sciences, 270(1521), 1293–1300. gaxiola, a., burrows, l.e., & coomes, d.a. (2008). tree fern trunks facilitate seedling regeneration in a productive lowland temperate rain forest. oecologia, 155(2), 325–335. hurst, j.m., & allen, r.b. (2007). the recce method for describing new zealand vegetation: expanded manual [unpublished contract report: lc0708/029]. retrieved from https:// nvs . l a n d c a re re s e a rc h . c o . n z / c o n te n t / re c c e _ expandedmanual.pdf iucn. (2011). the bonn challenge. retrieved from h t t p : / / w w w. b o n n c h a l l e n g e . o r g / c o n t e n t / challenge. jurasinski, g., & retzer, v. (2012). a collection of functions for similarity analysis of vegetation data [computer software]. r package version 0.5-5. retrieved from http://www.r-project.org/ keenan, r., lamb, d., woldring, o., irvine, t., & jensen, r. (1997). restoration of plant biodiversity beneath tropical tree plantations in northern australia. forest ecology and management, 99(1–2), 117– 131. koleff, p., gaston, k.j., & lennon, j.j. (2003). measuring beta diversity for presence-absence data. journal of animal ecology, 72(3), 367–382. kuuluvainen, t., aapala, k., ahlroth, p., kuusinen, m., lindholm, t., sallantaus, t., siitonen, j., tukia, h. (2002). principles of ecological restoration of boreal forested ecosystems: finland as an example. silva fennica, 36, 409–422. lamb, d., erskine, p. d., & parrotta, j.a. (2005). restoration of degraded tropical forest landscapes. science, 310(5754), 1628–32. forbes et al. new zealand journal of forestry science (2019) 49:6 page 12 landcare research. (2015). s-map online [online database]. retrieved from http://smap. landcareresearch.co.nz/home leathwick, j.r., & mitchell, n.d. (1992). forest pattern, climate and vulcanism in central north island, new zealand. journal of vegetation science, 3(5), 603–616. lindenmayer, d.b., franklin, j.f., & fischer, j. (2006). general management principles and a checklist of strategies to guide forest biodiversity conservation. biological conservation, 131(2), 433–445. lust, n., muys, b., & nachtergale, l. (1998). increase of biodiversity in homogeneous scots pine stands by an ecologically diversified management. biodiversity and conservation, 7(2), 249–260. mcgarigal, k., & marks, b.j. (1995). spatial pattern analysis program for quantifying landscape structure (pnw-gtr-351). retrieved from ftp:// ftp.wsl.ch/pub/kienast/v701-0553-00_landecol/ class_5/fak_fragstats_2.pdf mckelvey, p.j. (1955). a note on the forest edge at te whaiti. new zealand journal of forestry, 7, 77–80. mckelvey, p. (1963). the synecology of the west taupo indigenous forest. wellington, new zealand: government printer. mcnab, w.h. (1993). a topographic index to quantify the effect of mesoscale landform on site productivity. canadian journal of forest research, 23(6), 1100– 1107. ministry for primary industries. (2019). national exotic forest description. retrieved from https://www. teururakau.govt.nz/news-and-resources/opendata-and-forecasting/forestry/new-zealandsforests/ national instutite of water and atmospheric research. (2015). new zealand national climate database [electronic database]. retrieved from http://cliflo. niwa.co.nz (accessed september 2015). newton, a., & healey, j.r. (1989). establishment of clethra occidentalis on stems of the tree fern cyathea pubescens in jamaican montane rain forest. journal of tropical ecology, 5(4), 441–445. noack, m. (2011). growth and nutrition of quercus petraea underplanted in artificial pine stands under conversion in the northeastern german lowlands. forest systems, 20(3), 423. norton, d.a., & forbes, a. (2013). can exotic pine trees assist in restoration? applied vegetation science, 16, 169–170. norton, b.d.a., & miller, c.j. (2000). some issues and options for the conservation of native biodiversity in rural new zealand. ecological management and restoration, 1(1), 26–34. ogden, j., braggins, j., stretton, k., & anderson, s. (1997). plant species richness under pinus radiata stands on the central north island volcanic plateau, new zealand. new zealand journal of ecology, 21(1), 17–29. oksanen, j., kindt, r., legendre, p., o’hara, b., simpson, g.l., solymos, m., & wagner, h. (2008). the vegan package [computer software]. community ecology package. retrieved from http://cran.r-project.org/ onaindia, m., & mitxelena, a. (2009). potential use of pine plantations to restore native forests in a highly fragmented river basin. annals of forest science, 66(3), 1–11. onaindia, m., ametzaga-arregi, i., san sebastián, m., mitxelena, a., rodríguez-loinaz, g., peña, l., & alday, j. g. (2013). can understorey native woodland plant species regenerate under exotic pine plantations using natural succession? forest ecology and management, 308, 136–144. porté, a., huard, f., & dreyfus, p. (2004). microclimate beneath pine plantation, semi-mature pine plantation and mixed broadleaved-pine forest. agricultural and forest meteorology, 126(1), 175– 182. r development core team (2015). r: a language and environment for statistical computing. vienna, austria: r foundation for statistical computing. retrieved from http://www.r-project.org/ rodriguez-calcerrada, j., mutke, s., alonso, j., gil, l., pardos, j.a., & aranda, i. (2008). influence of overstory density on understory light, soil moisture, and survival of two underplanted oak species in a mediterranean montane scots pine forest. investigación agraria: sistemas y recursos forestales, 17(1), 31–38. royo, a.a., & carson, w.p. (2006). on the formation of dense understory layers in forests worldwide: consequences and implications for forest dynamics, biodiversity, and succession. canadian journal of forest research, 36(6), 1345–1362. senbeta, f., teketay, d., & näslund, b.-å. (2002). native woody species regeneration in exotic tree plantations at munessa-shashemene forest, southern ethiopia. new forests, 24(2), 131–145. tulod, a.m., norton, d.a., & sealey, c. (2018). canopy manipulation as a tool for restoring mature forest conifers under an early-successional angiosperm canopy. restoration ecology. advance online publication. doi.org/10.1111/rec.12707 wardle, p., & macrae, a.h. (1966). biological flora of new zealand. new zealand journal of botany, 41(1), 114–131. forbes et al. new zealand journal of forestry science (2019) 49:6 page 13 weiher, e., werf, a., thompson, k., roderick, m., garnier, e., & eriksson, o. (1999). challenging theophrastus: a common core list of plant traits for functional ecology. journal of vegetation science, 10(5), 609–620. williams, p.a. (2006). the role of blackbirds (turdus merula) in weed invasion in new zealand. new zealand journal of ecology, 30(2), 285–291. wood, s.n. (2011). fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. journal of the royal statistical society: series b (statistical methodology), 73(1), 3–36. wotton, d.m., & kelly, d. (2012). do larger frugivores move seeds further? body size, seed dispersal distance, and a case study of a large, sedentary pigeon. journal of biogeography, 39(11), 1973– 1983. wotton, d.m., & mcalpine, k.g. (2015). seed dispersal of fleshy-fruited environmental weeds in new zealand. new zealand journal of ecology, 39(2), 155–169. zamora, r., hódar, j.a., matías, l., & mendoza, i. (2010). positive adjacency effects mediated by seed disperser birds in pine plantations. ecological applications, 20(4), 1053–1060. zanne, a.e., & chapman, c.a. (2001). expediting reforestation in tropical grasslands: distance and isolation from seed sources in plantations. ecological applications, 11(6), 1610–1621. forbes et al. new zealand journal of forestry science (2019) 49:6 page 14 new zealand journal of forestry science (2019) 49:6 https://doi.org/10.33494/nzjfs492019x45x additional file 1. opportunities and limitations of exotic pinus radiata as a facilitative nurse for new zealand indigenous forest restoration. adam s. forbes, david a. norton, fiona e. carswell2 appendix s1. chronosequence stand summary statistics. kinleith forest pinus radiata plantation chronosequence, central north island, new zealand. forest stand statistics refer to p. radiata only year planted age (years) n density (stems ha–1) basal area (m2 ha–1) mean diameter (cm) canopy height (m) px 2013 2 62 542±86 0.13±0.03 2.2 ±0.05 1.7 ±0.02 0.117 2009 5 67 558±34 8.7±0.6 14±0.4 9.2±0.2 0.001 1999 15 48 400±45 46±3 37±1 26±1 0.000 1989 25 24 200±36 43±5 52±2 38±1 0.041 1979 35 37 308±28 65±8 50±2 36±2 0.867 1970 44 39 325±34 55±3 45±2 31±1 54.774 1954 60 29 167±24 71±11 70±5 46±3 0.006 1927 86 15 125±25 82±23 89±6 55±3 0.066 1925 89 17 142± 44 107±30 95±6 52±2 0.000 note. n = number of trees sampled. px = proximity index: a landscape scale index representing the degree of ecological isolation of survey plots within pinus stands from adjacent indigenous forest cover. new zealand journal of forestry science (2019) 49:6 https://doi.org/10.33494/nzjfs492019x45x appendix s2. generalised linear regression of pinus radiata basal area from nine p. radiata plantation stands aged 2–89 years, located in kinleith forest, central north island, new zealand. dashed lines indicate the 95% ci. stand age (years) b a s a l a re a ( m 2 h a 1 ) 0 3 0 6 0 9 0 1 2 0 0 20 40 60 80 100 new zealand journal of forestry science (2019) 49:6 https://doi.org/10.33494/nzjfs492019x45x appendix s3. indigenous woody seedling densities across a chronosequence of nine pinus radiata plantation stands aged 2–89 years, kinleith forest, central north island, new zealand. the notation * indicates significant differences in mean values at the significance level p < 0.05. error bars = ±1se. mean seedling density from an old-growth natural forest (“nf”) reference site shown for comparative purposes. stand age (years) s e e d li n g d e n s it y ( s te m s h a 1 ) 0 1 0 0 0 2 0 0 0 3 0 0 0 4 0 0 0 5 0 0 0 6 0 0 0 0 20 40 60 80 100 nf * * * * new zealand journal of forestry science (2019) 49:6 https://doi.org/10.33494/nzjfs492019x45x appendix s4. analysis of deviance table from a generalised linear model fitted to estimate the effect of pinus radiata plantation age and meso-scale topographic exposure on the number of indigenous trees and shrubs found in the p. radiata understories. data collected from a chronosequence of nine plantation stands ranging in age 2–89 years. plantation stands located in kinleith forest, central north island, new zealand. analysis of deviance table: woody indigenous species richness ss df f p log age 20.610 1 9.745 0.021 * meso-scale topography 8.124 1 3.841 0.097 † appendix s5. ternary plot of pairwise age comparisons of jaccard similarity from woody species presence/absence data collected across a chronosequence of pinus plantations aged 2–89 years old, located in kinleith forest, central north island, new zealand. pinus stand ages shown as numerals. 80 2 0 8 0 60 4 0 6 0 40 6 0 4 0 20 8 0 2 0 2-5 to 2-89 5-15 5-25 5-355-44 5-60 15-87 5-89 15-25 15-35 15-44 15-60 15-86 15-89 25-35 25-44 25-60 25-86 25-89 35-44 35-6035-86 35-89 44-60 44-8644-89 60-86 60-89 86-89 spp. shared (a´) spp. loss (c´) spp. gain (b´) new zealand journal of forestry science (2019) 49:6 https://doi.org/10.33494/nzjfs492019x45x appendix s6. woody species found in kinleith forest pinus radiata plantation chronosequence (including indigenous reference site), central north island, new zealand species code family growth form aristotelia serrata ariser elaeocarpaceae tree beilschmiedia tawa beitaw lauraceae tree brachyglottis repanda brarep compositae tree buddleja davidii* buddav scrophulariaceae shrub coprosma grandifolia copgra rubiaceae tree coprosma robusta coprob rubiaceae tree coriaria arborea corarb coriariaceae tree cordyline banksii corban asparagaceae tree cotoneaster sp.* cotsp. rosaceae shrub cyathea dealbata cyadea cyatheaceae tree fern cyathea medullaris cyamed cyatheaceae tree fern dacrydium cupressinum^ daccup podocarpaceae tree dicksonia squarrosa dicsqu dicksoniaceae tree fern dicksonia fibrosa dicfib dicksoniaceae tree fern cyathea smithii dicsmi cyatheaceae tree fern fuchsia excorticata fucexc onagraceae tree gaultheria antipoda gauant ericaceae shrub geniostoma ligustrifolium genlig loganiaceae shrub hedycarya arborea hedarb monimiaceae tree knightia excelsa kniexc proteaceae tree leucopogon fasciculatus leufas ericaceae shrub leycesteria formosa* leyfor caprifoliaceae shrub litsea calicaris litcal lauraceae tree melicytus ramiflorus melram violaceae tree myrsine australis^ myraus primulaceae tree pinus radiata* pinrad pinaceae tree pittosporum eugenioides^ piteug pittosporaceae tree pittosporum tenuifolium pitten pittosporaceae tree podocarpus totara^ podtot podocarpaceae tree pomaderris amoena pomamo rhamnaceae shrub prumnopitys ferruginea^ prufer podocarpaceae tree prumnopitys taxifolia^ prutax podocarpaceae tree pseudopanax arboreus psearb araliaceae tree pseudopanax crassifolius^ psecra araliaceae tree pseudowintera colorata^ psecol winteraceae tree schefflera digitata schdig araliaceae tree weinmannia racemosa weirac cunoniaceae tree note. * = exotic flora. ^ = species found only in the mature natural forest reference site. new zealand journal of forestry science (2019) 49:6 https://doi.org/10.33494/nzjfs492019x45x appendix s7. chronosequence vegetation plot matrix, species importance values (iv). kinleith forest pinus radiata plantation chronosequence, central north island, new zealand. site a r is e r b e it a w b r a re p b u d d a v c o p a u s c o p ro b c o r a rb c o r b a n c o t sp . c y a d e a c y a m e d d a c cu p d ic fi b d ic sm i d ic sq u f u c e x c g a u a n t g e n li g h e d a rb k n ie x c l e u fa s l e y fo r l it ca l m e l ra m m y r a u s p it e u g p it te n p o d to t p o m a m o p r u fe r p r u ta x p s e a rb p s e co l p s e cr a s c h d ig w e ir a c 89a 0 0 5.5 0 0 0 0 0 0 8 3 0 0 0 11 0 0 2.5 0 0 0 0 0 6 0 0 0 0 0 0 0 7.5 0 0 0 1 89b 0 0 1 0 0 0 0 0 0 3 4 0 0 0 14 0 0 5 0 0 0 0 0 2 0 0 0 0 0 0 0 6 0 0 3 0 89c 0 0 1 0 0 0 0 0 0 0 2 0 0 0 12 0 0 4 0 0 0 0 0 8 0 0 0 0 0 0 0 7 0 0 3 0 89d 0 0 2 0 0 0 0 0 0 2 3 0 0 0 12 0 0 2 0 0 0 0 0 10 0 0 0 0 0 0 0 12 0 0 4 0 86a 0 0 11 0 8 0 0 0 0 1 0 0 0 0 8 0 0 0 0 0 0 0 0 7 0 0 0 0 0 0 0 2.5 0 0 0 0 86b 0 0 2.5 0 0 0 0 0 0 4 7 0 0 0 9 0 0 3 0 0 0 0 0 3 0 0 0 0 0 0 0 6 0 0 0 6 86c 0 0 10 0 4 0 0 0 0 8 4 0 0 0 10 0 0 2 0 0 0 0 0 2 0 0 0 0 0 0 0 5 0 0 0 0 86d 1 0 5.5 0 0 0 0 0 0 0 10 0 0 0 5 0 0 4.5 0 0 0 0 0 2.5 0 0 0 0 0 0 0 3.5 0 0 0 0 60a 4 0 0 0 6 2 0 0 0 0 3 0 0 0 0 0 0 4 0 0 0 0 0 6 0 0 8 0 0 0 0 0 0 0 0 0 60b 6 0 0 0 8 4 0 0 0 0 0 0 0 0 11 3 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 60c 2 0 0 0 1 4 0 0 0 0 0 0 4 0 13 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 60d 0 0 0 0 1 6 0 0 0 0 0 0 0 0 8 3 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 44a 0 1 0 0 2 0 0 0 0 9 9 0 0 2 5 0 0 3 3 0 1 0 1 3 0 0 0 0 0 0 0 0 0 0 0 0 44b 0 0 0 0 1 0 0 0 0 0 7 0 0 0 8 0 0 3 2 2 0 0 0 3 0 0 0 0 0 0 0 0 0 0 2 0 44c 0 1 0 0 2 2 0 0 0 0 7 0 0 0 6 0 1 8 6 4 2 0 2 2 0 0 0 0 0 0 0 0 0 0 1 0 44d 0 2 0 0 2 2 0 2 0 4 7 0 0 0 9 0 0 9 7 0 1 0 1 2 0 0 2 0 0 0 0 0 0 0 2 0 35a 0 0 0 0 11 0 0 0 0 3 0 0 2 0 8 0 0 7 0 0 0 0 0 9 0 0 0 0 0 0 0 0 0 0 9 0 35b 0 0 4 0 3 0 0 0 0 0 7 0 0 0 7 0 0 4 0 0 0 0 0 7 0 0 0 0 0 0 0 0 0 0 5 0 35c 0 0 9 0 0 0 0 0 0 4 7 0 0 0 10 0 0 7 0 0 0 0 0 6 0 0 0 0 0 0 0 0 0 0 1 0 35d 2 0 5 0 0 6 0 0 0 4 7 0 0 0 4 0 0 6 0 0 0 0 0 10 0 0 0 0 0 0 0 0 0 0 3 0 25a 5 0 0 0 0 5 0 0 0 3 0 0 0 0 8 0 0 4 0 0 0 2 0 5 0 0 0 0 0 0 0 0 0 0 0 0 25b 1 0 0 0 0 3.5 0 0 0 0 0 0 0 0 9 0 0 4 0 0 0 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0 25c 0 0 0 0 0 2 0 0 0 3 0 0 4 0 9 0 0 1 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 6 0 25d 0 0 0 0 0 3 0 0 0 0 3 0 0 0 9 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10 0 15a 2 0 0 0 0 0 0 0 0 0 0 0 0 0 5 0 0 1 0 0 0 6 0 2 0 0 2 0 0 0 0 0 0 0 0 0 15b 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 4 0 1 0 0 1 0 0 0 0 0 0 0 0 0 15c 3 0 0 0 0 0 0 0 1 0 0 0 0 0 3 0 0 1 0 0 0 6 0 0 0 0 2 0 0 0 0 0 0 0 0 0 15d 0 0 0 0 1 9 0 0 0 0 0 0 0 0 0 0 0 1 0 0 4 2 0 0 0 0 3 0 0 0 0 0 0 0 0 0 5a 7 0 0 2 0 2 6 0 0 0 0 0 0 0 3 0 0 0 0 0 1 0 0 0 0 0 6 0 0 0 0 0 0 0 0 0 5b 13 0 0 3 0 2 7 0 0 0 0 0 0 0 6 0 0 0 0 0 2 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5c 8 0 0 3 0 3 9 0 0 0 0 0 0 0 0 0 0 0 0 0 0 6 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5d 0 0 0 6 0 1 6 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2a 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2b 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 new zealand journal of forestry science (2019) 49:6 https://doi.org/10.33494/nzjfs492019x45x 2c 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2d 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 nfa 0 1 0 0 7 0 0 0 0 0 0 0 0 2 2 0 0 4 10 7 0 0 0 8 1 0 0 0 0 2 0 0 2 1 2 7 nfb 0 3 0 0 5 0 0 0 0 1 0 6 0 0 0 0 0 1 16 1 0 0 0 2 0 0 0 1 0 0 0 0 1 0 1 0 nfc 0 0 4 0 9.5 2 0 0 0 7 0 11 0 0 0 0 0 11 12 9 0 0 0 5 3 2 0 1.5 0 0 0 0 0 2 0 11 nfd 0 0 0 0 12 0 0 0 0 5 2 7 0 0 1 0 0 2 13 15 0 0 0 9 6 0 0 0 0 0 2 0 2 0 2 0 note. site = number-letter represents stand age and plot replicate reference (a-d). nf = native forest reference site. species six-letter codes are consistent with appendix s6. trade-offs between environmental and economic factors in conversion from exotic pine production to natural regeneration on erosion prone land suzanne m. lambie1*, shaun awatere1, adam daigneault2, miko u.f. kirschbaum3, michael marden4, tarek soliman5, raphael i. spiekermann3 and patrick j. walsh6 1 manaaki whenua – landcare research, private bag 3127, hamilton 3240, new zealand 2 school of forest resources, university of maine, 5755 nutting hall, 04469, usa. 3 manaaki whenua – landcare research, private bag 11052, palmerston north 4442, new zealand 4 research associate with manaaki whenua landcare research, 48 hillview terrace, mangapapa, gisborne 4010, new zealand. 5 manaaki whenua – landcare research, private bag 92170, auckland 1142, new zealand 6 us environmental protection agency, 1200 pennsylvania ave, nw, washington, dc 20460, usa. *corresponding author: lambies@landcareresearch.co.nz (received for publication 2 june 2021; accepted in revised form 29 november 2021) abstract background: some of new zealand’s exotic pine (pinus radiata d.don) forests were planted for erosion mitigation but cultural, legislative, environmental, and profitability limitations in some parts of the landscape have led to reassessment of their suitability. there is limited information to support landowner decisions on the viability of natural regeneration of native forest post-pine-harvest. methods: we evaluated scenarios of post-harvest natural regeneration, compared to remaining in pine production, using erosion susceptibility determined from historical occurrence of landslides, gullies and earthflows, biophysical growth modelling of mānuka–kānuka (leptospermum scoparium-kunzea ericoides (a.rich) joy thomps.) shrubland using the process-based cenw model, and cost-benefit analyses using nzfarm with two land use change scenarios, at two levels of erosion mitigation ± honey profits. results: in our study area, the gisborne region (north island of new zealand), ~27% of the land has moderate–very high susceptibility to landslides, 14–22% a high probability of contributing material to waterways, and 19% moderate–very high gully erosion susceptibility. pines grow 10 times faster than naturally regenerating mānuka–kānuka shrubland, but mānuka–kānuka is used for honey not wood production. natural regeneration resulted in losses of $150–250 ha-1 yr-1 compared to the current profitability of pine production. honey production offset some reduction in pine revenue, but not fully. thus, the viability of shifting from pines to native forest is highly dependent on landowner impetus and value for nonmarket ecosystem services (such as cultural and biodiversity values) provided by native forest. conclusions: a mosaic of land uses within a property may sufficiently offset income losses with other benefits, whereby highly erosion-prone land is shifted from rotational pine forest production to permanent native forest cover with honey production where possible. at the regional scale in gisborne, the conversion of the most highly susceptible land under production forestry (315–556 ha) to natural regeneration has the potential for wider benefits for soil conservation reducing erosion by 1–2.5 t yr–1 of sediment facilitating achievement of cleaner water aspirations and habitat provision. new zealand journal of forestry science lambie et al. new zealand journal of forestry science (2021) 51:14 https://doi.org/10.33494/nzjfs512021x163x e-issn: 1179-5395 published on-line: 20/12/2021 © the author(s). 2021 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access keywords: biophysical tree modelling; cost-benefit modelling; erosion susceptibility; kānuka; mānuka; pinus radiata. http://creativecommons.org/licenses/by/4.0/), lambie et al. new zealand journal of forestry science (2021) 51:14 page 2 introduction some owners of exotic pine (pinus radiata d.don) forests in new zealand aspire to convert parts of their holdings to native forest using natural regeneration after pine harvest. the viability of converting from profitable pine production to natural regeneration is unclear and the lack of information on costs and opportunities associated with adopting this conversion strategy inhibits informed decision-making by landowners. much of new zealand was cleared of native forest during polynesian and european colonisation (mcglone 1983; guild & dudfield 2010); however, some cleared areas were highly susceptible to erosion (mcglone 1989; rhodes 2001; guild & dudfield 2010). non-native conifers were planted by the nz government in the 1960s as a fast-growing solution to mitigate erosion. government-owned forests in the gisborne region were privatised in the 1980s, shifting these forests from conservation plantings to rotational pine production (rhodes 2001). nz’s pine forest industry covers >1.7 million hectares (6.3% of total area; 155,600 ha in the gisborne region) and contributes ~$3.55 billion to the new zealand economy (forest owners association & ministry for primary industries 2019). the forestry industry has positive and negative societal and environmental impacts. for example, there has been strong employment growth (nixon et al. 2017) and enhanced value through carbon credits (akpa et al. 2016; evison 2008; jindal et al. 2008) but sediment and debris deposition during storm events damages downstream environments (grant & wolff 1991; landcare research & scion 2017; arnold 2018). there are currently additional environmental and safety regulatory pressures on production forestry (e.g., national environmental standard for production forestry, national policy statement for freshwater management, national policy statement for indigenous biodiversity, health and safety at work act 2015) that may prove untenable for future wood production in some existing areas of pine forest (clark 2017; richards 2017). some of these areas may be more suitable for retirement and natural regeneration post-pine-harvest. a second pathway to permanent carbon sink and natural regeneration is via unharvested pine stands (climate change commission 2021). this scenario has not been addressed in our work as there is insufficient data on the extent of this practice (forbes & norton 2021) in nz, and the perceived negative impacts of pine seedlings and tree fall (gibson 2021) require further investigation. mānuka–kānuka shrubland, composed of mānuka (leptospermum scoparium j.r.forest & g.forst) and kānuka (kunzea ericoides (a.rich) joy thomps.) is often the first stage of natural regeneration in the gisborne region (williams 1983; newsome 1987; wilson 1994; funk et al. 2009; overdyck & clarkson 2012) providing permanent erosion mitigation and possible revenue from mānuka honey. mānuka–kānuka shrubland accounts for ~70% of regenerating forests in new zealand (ministry for primary industries 2017b) and native forests have greater potential for erosion mitigation than production pines as clear-cut forest management creates a window of vulnerability for sediment and debris transportation during storm events after harvest (phillips et al. 2012; lambie et al. 2018). this is supported by aburto et al. (2021) who showed native nothofagus forest had greater erosion mitigation than pinus radiata production forests in chile. the scale of work undertaken by aburto et al. (2021) is comparable with bergin et al. (1993; 1995) and marden and rowan (1993) in new zealand. while native forest will likely have greater erosion mitigation and cleaner water benefits compared to pine production, retirement of pine estates will decrease income from wood production (hall et al. 2017) and other avenues of income may need to be sourced such as honey, biodiversity, or carbon. to facilitate informed decision-making by landowners and regulatory authorities, we assessed the viability of converting erosion-prone land currently planted in pines to natural regeneration using high-resolution erosion susceptibility modelling, biophysical modelling of mānuka–kānuka shrubland, and an ecosystem services cost-benefit model. the cost-benefit model included scenarios with varying levels of erosion susceptibility and with and without honey production to assess economic viability of a range of scenarios. methods study area our study area was the gisborne region (835,500 ha) of te ika-a-māui, north island of aotearoa, nz (fig. 1), which has the greatest proportion of highly erosionprone land in new zealand (ministry for primary industries 2017c). the gisborne region has two main lithological terrains (fig. 1): cretaceous terrain (29% of the region) which is highly indurated and affected by earthflows and rotational slumping, and tertiary terrain (61% of region), which is less deformed but affected by landslides (black 1980). both terrains are susceptible to gully formation (pearce et al. 1987; derose et al. 1998; fuller & marden 2011). landslide susceptibility and hillslope connectivity erosion susceptibility assessments for shallow landslides, gully erosion, and earthflows were undertaken at a 1:25,000. erosion susceptibility under existing pine plantation forest was assessed. although different areas of forest were of different ages, we assumed a uniform vegetation cover as the inventory of landslide scars was mapped on pastoral land. therefore, we were unable to investigate variation in land cover type, and the assessment was limited to lithology and topographic factors. selection of conditioning factors was based on an understanding of the geomorphic process being assessed. lithology, slope, and aspect were selected as erosion-conditioning factors, and all have direct physical process relevance for slope stability. the conditioning factors were assessed using bivariate statistics to determine weights and combined in a landslide-index (juang et al. 1992; ruff & czurda 2008; sciarra et al. 2017) to characterise landslide susceptibility, with rainfall being the cause of landslides. conditions under which past landslides occurred were used to predict occurrence of future landslides where similar site conditions prevail (varnes 1984; soeters & van westen 1996; aleotti & chowdhury 1999). spatial distribution of landslides on tertiary-aged terrain was collated from betts et al. (2017) who mapped 3,164 landslide scars on similar terrain in the manawatu region, and veld and de graaf (1990) who compiled an inventory of 576 landslide scars on arai matawai and emerald hills stations ~20 km southwest of gisborne city following cyclone bola (1988). for the cretaceous terrain, a landslide inventory from the eastern ruahine ranges was used where we assumed that the greywacke bedrock is representative of the cretaceous terrain found within the gisborne district. the influence of conditioning factors in susceptibility indices was weighted as per betts et al. (2017), where slope classes were defined and weights determined by calculating ‘densities’, or the prior probability of failure within each slope class, based on the location of mapped landslides: (1) where pprior = p{s} is the conditional probability of having a landslide s and npixj (slide) are the number of pixels with landslides in slope class, and npixj (total) are the total number of pixels in slope class j. the prior probabilities are normalised to create weights wj within the range of 0–1: (2) scar-slope relationships were established using a national 15-m digital elevation model (dem) derived from contour data applied separately for tertiary and cretaceous terrains. the prior probability of landslide for the tertiary terrain was calculated as the mean of the lambie et al. new zealand journal of forestry science (2021) 51:14 page 3 table 1: description of the study sites figure 1: geological terrain in the gisborne region and exotic forest cover based on the land cover database of new zealand (landcare research 2015). scar-slope relationships, and for the cretaceous terrain, a single scar-slope relationship was used. slope aspect was calculated using the 15-m dem to create nine aspect classes, and the relationship with aspect was assessed on a pixel basis. the conditional probability for aspect was calculated in the same way as it was for slope. finally, the two conditioning factors (slope gradient for each lithology type and aspect) were multiplied to calculate the landslide susceptibility index, which was then classified into five categories none, low, moderate, high, and very high using equal weights. hillslope connectivity connectivity between hillslopes and waterways assessed if a landslide would deliver sediment to a waterway (dymond et al. 2006). the 15-m dem was used to estimate the likely flow-path of material generated by a landslide, its flow direction, and potential intervening accumulation zones, to determine whether sediment and/or slash could potentially enter a stream network. if the flow path encountered any significant flat land (consecutive pixels below four degrees of slope), the source pixel was tagged as ‘non-connected’, as we assumed material to be deposited on the flat terrain before reaching the stream. otherwise, the pixel was tagged as ‘connected’. connectivity was determined for parcels of land with moderate–very high landslide susceptibility as these land parcels were most likely to be retired from pine forest production and reverted to natural forest that affords longer-term erosion control. gully and earthflow susceptibility gully inventories mapped from aerial photography flown in 1957 (1:15,000) and 1997 (1:26,000) were combined (marden et al. 2012) to identify actively eroding gullies, and the units (with an average size of 110 ha) within the new zealand land resource inventory (nzlri) (landcare research 2010) that would be most likely to develop gullies. the combined mapped gully layers were then intersected with the nzlri units. a matrix was used to create gully susceptibility classes (1–5) based on the number of existing gullies (frequency), the potential for future gullies to develop within each nzlri unit, and the proportion (% of area) of each unit affected by current and past gullying (magnitude). of the 2,097 nzlri units in the gisborne region identified in the 1970s as having ‘present gully erosion’, 126 units showed no evidence of gullies when mapped in 1957 and 1997 (marden et al. 2012). on the assumption that gullies were present in these units in the 1970s, the gully severity ranking assigned to these units at the time of the original mapping of lri units was adopted. earthflow susceptibility assessment draws on the nzlri (landcare research 2010), which includes an erosion severity ranking for earthflows as mapped in 1990 and assigned to classifications between ‘none’ to ‘very high’. biophysical modelling provision of erosion mitigation by forests is strongly linked to its growth characteristics, which are influenced by climatic, soil and tree species factors. lambie et al. new zealand journal of forestry science (2021) 51:14 page 4 natural regeneration in new zealand often begins with mixed mānuka (leptospermum scoparium and kānuka (kunzea ericoides var. ericoides) stands (stephens et al. 2005). although natural regeneration can begin with broadleaf/podocarp species (e.g., cameron 1960), there is no published information on the growth and structure of early growth podocarp forest. we, therefore, used mānuka–kānuka shrubland as a representative of natural regeneration as this is the most likely pathway in the gisborne region (funk et al. 2009; ministry for primary industries 2017b). growth simulations of mānuka for the gisborne region were assessed using a comprehensive dataset of tree growth parameters (e.g., height and diameter at breast height) from across new zealand from payton et al. (2010), the new zealand national vegetation survey databank (nvs) and other unpublished datasets. carbon (c) accumulation in mānuka–kānuka shrubland was estimated using model simulations with the physiological model cenw version 5.0 (kirschbaum 1999a). cenw has been used extensively to predict the growth of pine stands (e.g., kirschbaum 1999a, b; kirschbaum & watt 2011; kirschbaum et al. 2012). the model and its source code are available at: http:// www.kirschbaum.id.au/welcome_page.htm, with a list of relevant equations available at http://www.kirschbaum. id.au/cenw_equations.pdf. for the present work, it was parameterised against the observations from payton et al. (2010) to simulate the growth of mānuka based on external environmental drivers (temperature and rainfall), stand-internal factors (stand density), and weed competition. in cenw, weeds are modelled as a separate entity that competes with the main species of interest for nutrients, water, and radiation. normally, the tree canopy eventually overtops weeds and out shades them, with the time course depending on the initial biomass of weeds and overstorey species and the parameters that define the competitive properties of the weed layer. here, we assumed a maximal height of the weed layer of 0.5 m and initial weed biomass as 1500 kgdm ha-1 and mānuka–kānuka stands as 200 kgdm ha-1. these parameters correspond to primary competition by a well-established grass understory. cenw can be run over periods of many decades with an underlying daily simulation time step. it simulates stand properties and dynamics, such as leaf-area development, stand height, basal area development, litter-fall and exchange of both water and carbon dioxide based on daily inputs of minimum and maximum temperature, solar radiation, rainfall, and vapour pressure (kirschbaum 1999a). it also requires estimates of site fertility, soil water-holding capacity, and silt and sand fractions as a measure of soil texture. we used 20 years of daily weather input data from the virtual climate station network (vcsn; national institute of water and atmospheric research ltd, appendix 1). daily vcsn data were estimated on a 0.05° latitude/longitude grid (tait et al. 2006; tait 2008; tait & liley 2009) as described by kirschbaum and watt (2011). soil waterholding capacity and the percentage of silt plus clay were obtained from the national soils database (landcare research 2020). characteristic climate variables for the gisborne region are presented in appendix figure a1. predictions of different measures of growth were fitted to independent observations from 69 stands located at 52 distinct locations throughout new zealand (payton et al. 2010), with growth measures mostly consisting of total stand biomass at respective ages inferred from ring counts. we also used the observed distribution of mānuka–kānuka throughout new zealand to derive functions to constrain the environmental performance of mānuka–kānuka under more extreme climatic conditions than the sites sampled by payton et al. (2010). some parameter estimates were based on the earlier modelling work of whitehead et al. (2004) and whitehead and walcroft (2005) and the observations of burrows et al.1 and burrows2. other specific information on stand allometric properties was sourced from scott et al. (2000) and leaf nitrogen concentrations from ross et al. (2009). the range of parameter values was further constrained to remain within physiologically plausible bounds to retain the physiological integrity of the simulations. the modelling procedure was similar to that described by kirschbaum and watt (2011) for pine growth. cost-benefit analysis a cost-benefit analysis using a combination of a spatially explicit agri-environmental economic land-use model (nzfarm) (daigneault et al. 2018) and other nonmarket valuation methods was used to monetise changes in land use based on categories of erosion susceptibility. landslide susceptibility categories 4 (moderate susceptibility with high waterway connectivity), 6 (high susceptibility with high waterway connectivity), 8 (very high susceptibility with high waterway connectivity), and land highly prone to gullying (as per spiekermann and marden 2018) were the focus of model scenarios. the model scenarios had two levels of landsliding susceptibility with more categories included in scenario 2 and consistent gully susceptibility across all scenarios. the erosion susceptibility scenarios were tested with and without honey production as a potential offset to losses in wood profits. scenario 1, landslide classes 6+8 (high and very high landslide susceptibility with high waterway connectivity) and high gully category converted from pines to native forest with honey production. scenario 2, landslide classes 4+6+8 (moderate, high, and very high landslide susceptibility and high waterway connectivity) and high gully category converted from pines to native forest with honey production. scenario 3, same as scenario 1, without honey production and scenario 4, same as scenario 2, without honey production. model scenarios were compared to the ‘baseline’ scenario, where all identified land remains in pine plantation forestry. the baseline was established using a land use map of the gisborne region (agribase; asure quality 2020) and the new zealand land cover database (landcare research 2015). the model includes assessments of production impacts including profit from land converted from pine forestry, profit from land converted to mānuka honey or related production, value of the land, and planting/native afforestation costs and environmental impacts including carbon cycling and water quality. the analysis was conducted over 62 years to reflect two cycles of 30-year pine rotation (hockey & page unpublished3; ministry for forestry 1994). since the benefits and costs occur over different time periods, the net present value was determined at two discount rates (4 and 6%; te tai ōhanga 2018) to calculate overall impacts: (3) where t is each year (up to 62), and r is the discount rate. treasury new zealand recommended the use 4 and 6% rates to convert future values to present values as money in the future is worth less than the same amount in the present (te tai ōhanga 2018). the range of discount rates also reflects the different objectives of landowners. for instance, the lower end of the range might reflect the objective of achieving intergenerational equity as some landowners (e.g., māori) might not be driven solely by profits but by other societal goals such as intergenerational resource sustainability as well as maintaining cultural and spiritual benefits for future generations (goulder & williams 2012; harmsworth and awatere 2013). net present value of economic returns for the pine forestry sector were calculated as follows: (4) where p is the log price (te uru rākau 2020), vol is the log volume harvested in each year (from cenw; kirschbaum & watt 2011), and cost is the fixed and variable costs of production. fixed and variable costs including logging, cartage, roading, and growing costs were derived from the literature (olssen et al. 2012). all future values of the production function were then discounted to get total present values. we finally derived the equal annual lambie et al. new zealand journal of forestry science (2021) 51:14 page 5 1 burrows, l.e. (2010). estimate of biomass and leaf area index (lai) for kānuka at avoca station. [unpublished report], lincoln, new zealand: landcare research. 2 burrows, l., carswell, f., brignall-theyer, m., mckenzie, s. (2009). changes in biomass carbon during kānuka shrubland succession to forest. [landcare research internal report lc0809/084], lincoln, new zealand: landcare research. 3 hockey, m., page, m. (1983). site index data for the gisborne-east coast region. collated in 1981 and partially revised in 1983. wellington, new zealand: unpublished internal new zealand forest service report. equivalent (eae) from the estimated net present value to compare the annual economic returns from pine forestry with other land uses. economic returns from mānuka honey produced from naturally regenerating mānuka–kānuka shrublands were based on daigneault et al. (2015). environmental factors (ei) including nutrient leaching, erosion, and carbon sequestration were monetised following estimation on a per hectare basis (γenv), as impacted by soil type, land cover, and land use. by aggregating the per hectare values of these parameters across the land area under pine (x), we could estimate the total environmental factor outputs from pine forestry: (5) surficial and stream bank erosion were simulated using the nzeem model (dymond et al. 2010), and nutrient values obtained (e.g., parfitt et al. 1997; lilburne et al. 2010). ghg emissions were sourced from mpi carbon look-up tables (ministry for primary industries 2017a). to reflect the impact of land conversion from pine production to native forest on the environmental outputs, equation 5 was modified to: (6) where z is the area of the land converted to native forest. the parameter γenv specifies the environmental impacts of pine forest after accounting for land conversion, while μenv describes the impact of native forest on the environmental factors. several non-market techniques have been developed to place a price on changes in water quality, including hedonic pricing (boyle et al. 1999), recreation demand (massey et al. 2006), and stated preferences (moore et al. 2018). of those techniques, estimates from stated preference studies capture the widest range of people and values, both use and non-use, in their application depending on the scope of the study. we therefore focus on stated preference values and used willingness to pay (wtp) for improved water quality. to monetise water quality improvements associated with each scenario, a benefit transfer for people’s wtp for reductions in nutrient leaching was used, which is common technique for determining the value of an ecosystem service (e.g., aguilar et al. 2018; huber and finger 2020; tian et al. 2020). nutrient leaching values were sourced from takatsuka et al. (2009), where a choice experiment was used to estimate monetary values of improvements in water quality as well as other ecosystem services. based on the values estimated by takatsuka et al. (2009), a linear transfer function was estimated (although a nonlinear function produced comparable results). estimates are corrected for differences in income and nominal dollars between the original and current study. to value carbon sequestration, lookup tables were first used to calculate changes in carbon for each scenario (ministry for primary industries 2017b) using a natural logarithm regression model for 50–62 years (r2 = 99.99). carbon sequestration values were then assumed to be $25 tco2e–1, based on prices from recent transactions of the new zealand emission trading scheme (ets). the change in erosion was monetised using a mid-range estimate of $3 t–1 sediment based on two independent estimates of $1 t–1 (dymond et al. 2012) and $6.50 t–1, where the upper value includes only avoided flood damage and water treatment costs (barry et al. 2014). annual honey profits in mānuka–kānuka shrublands vary largely depending on honey quality ranging between $98 ha–1 for low unique mānuka factor (umf) to $1000 ha–1 for high umf (walsh et al. 2017). these profit estimates were based on a production per hive of 30 kg and 35 kg and a price per kg of $26.30 and $40 for the $98 ha–1 and $1000 ha–1, respectively (burke 2015; wetere 2015). the variation in profits was due to differences in mānuka productivity, and capital and operating costs. umf is a rating system designed to ensure purity, quality, and authenticity of mānuka honey, and the higher the umf rating the higher the monetary value of the honey. while no spatially explicit data is available on where high umf honey has been produced, recent research predicted that soil quality, rainfall, climate, and genotype are critical variables in determining the umf level in a particular area. to identify areas suitable for high umf honey, we therefore used temperature and precipitation prediction equations (watt et al. 2012) to define the probability of occurrence of mānuka–kānuka shrubland as per walsh et al. (2017). the probabilities were summed and assigned to nzfarm polygons based on an area-weighted average to create an index describing the relative likelihood of occurrence of mānuka– kānuka. based on conversations with local farmers and associations, the top 10% of polygons was assumed to be areas suitable for high umf honey and therefore allocated a profit of $1000 per hectare, while the rest of the polygons were assumed to produce low umf and were allocated a profit of $98 per hectare. much of the value of agricultural land is tied to the profits associated with it, so we assume that the opportunity cost of removing land from production is reflected in the change in the npv of future profits. nzfarm outputs change in net revenue represent the central opportunity costs of the change from pine production to native forest with and without honey production. we were able to monetise several notable impacts associated with changes in land use. however, it is important to note that there are a number of other impacts that we were not able to quantify or monetise (table 1). biodiversity was quantified as per walsh et al. (2019) and presented as ‘restored significance’ as per mason et al. (2012) and carswell et al. (2015), but were not monetised and therefore not included in the economic analysis. further, the impacts of weed control and competition were not able to be included in the analysis due to limited support information. lambie et al. new zealand journal of forestry science (2021) 51:14 page 6 results erosion susceptibility incidences of landslides increased markedly on slopes >16° in tertiary terrain and often delivered sediment and forest slash into the nearest watercourse. in contrast, on the cretaceous terrain, shallow landslides are predominantly restricted to the steep flanks of the raukumara range. slope aspect was an important factor for landslide susceptibility in our study area, as found by others (e.g., yalcin & bulut 2007; galli et al. 2008; ruff & czurda 2008; van westen et al. 2008). we found a disproportionate number of landslides on north to north-east facing slopes, and landslide density on north facing slopes was 50% higher than average. in the gisborne region, 73.2% (545,700 ha) had very low–low landslide susceptibility and 21.1% (157,600 ha) had moderate–very high susceptibility. moderate– very high areas had landslides that were likely to deliver sediment and other material into nearby waterways. of the two terrain types, tertiary underlies the largest proportion of the gisborne region and landslide susceptibility was very low–low on 67.4% of hill country areas, and moderate–very high on the remaining 32.6% of hill country, of which approximately 165,000 ha (22%) of hill country slopes are directly connected to waterways (fig. 2). overall, 125,000 ha (24.7%) of hill country slopes are moderate–very highly susceptible and have potential for landslides and/or anthropogenic disturbances resulting in sediment, and any associated woody debris, entering a water course. in comparison, cretaceous terrain has slopes that were generally less steep and landslide susceptibility was very low–low on 85.5% of hill country areas, and moderate–very high on 14.5% (34,800 ha) of remaining hill country areas, of which 13.7% (32,700 ha) of hill country slopes were directly connected to waterways (fig. 2). land under pine forests has higher levels of landslide susceptibility than the average for the whole region (fig. 2) as they were established on erosion prone land to mitigate erosion. a larger proportion of the current forest estate is on cretaceous than tertiary terrain (landcare research 2015). however, land occupied by pine forests lambie et al. new zealand journal of forestry science (2021) 51:14 page 7 main impacts monetised quantified production-related impacts profit from land converted from exotic forestry x profit from land converted to honey or related production x value of land x planting/native afforestation costs x cultural impacts cultural medicines cultural harvests aesthetic/landscape changes environmental impacts biodiversity x carbon x water quality x water quantity recreation threatened or endangered species health and community impacts injuries changes in unemployment changes in population welfare impacts of native forest proximity table 1: summary of main impacts of land conversion scenarios on tertiary terrain has a much higher proportion of land susceptible to landslides (35.9% in classes moderate– very high) than the pine forests on cretaceous terrain (26.8%). within the gisborne region, 19% of the region is moderate–very high for gully erosion susceptibility (fig. 3a). 22.5% of hill country is classified as susceptible to earthflow erosion, with just 8.8% classed as moderate– very high susceptibility (fig. 3b) and is associated with areas dominated by mudstone or crushed argillite. tree growth to better understand the growth potential of mānuka– kānuka stands, we parameterised the cenw model against the observed data of payton et al. (2010) (fig. 4). stands showed some moderate growth potential over the first 20 years, reaching stand biomass of about 50 tc ha-1. growth rates then slow, with peak stand biomass typically reached by about 50 years. stands then tend to degenerate over further time or are invaded by taller trees that then dominate. the observed growth lambie et al. new zealand journal of forestry science (2021) 51:14 page 8 figure 2: landslide susceptibility, as a percentage of: (a) major catchments and gisborne region; and (b) for the area of tertiary and cretaceous terrains, including pine forest areas only. for landslide susceptibility, the percentage of hill country with moderate to very high susceptibility and potential to deliver sediment (connectivity) to waterways is also shown. figure 3: erosion susceptibility for: (a) gully; and (b) earthflow erosion, as a percentage of major catchments and gisborne region. patterns could be captured well by the cenw simulations (fig. 4b), with a calculated model efficiency of 0.65. growth potential of stands is strongly affected by initial stand density, which interacts with weed competition. once seedlings over-top the competing grass layer, a period of rapid growth can commence. stand growth rate tends to be reduced through self-thinning that reduces the number of living trees with associated loss of stand biomass. growth rates are therefore typically only a fraction of those achieved by well-managed p. radiata stands (cf. kirschbaum et al. 2011). analysis of growth rates (payton et al. 2010) and the patterns of temperature and rainfall in relation to the natural distribution of mānuka–kānuka stands across the country were used to estimate tree growth in response to environmental drivers. we had insufficient information to make a distinction between the growth of mānuka and kānuka and assumed the same environmental limitations for mānuka stands and mānuka–kānuka shrublands. mānuka–kānuka stands require mean annual temperatures of more than ~5°c (appendix fig. a2a), and growth increases rapidly with increasing temperature to about ~12°c (fig. a2a). similarly, mānuka–kānuka stands require annual rainfall > ~300 mm yr–1 (appendix fig. a2b), and growth increases steeply for optimal performance at ~800–1000 mm yr–1, with no further growth response at higher rainfall. however, it is not currently known whether extremely high rainfall beyond 2,000 mm yr–1 adversely affects the growth and persistence of stands. for the gisborne region, simulated growth of mānuka–kānuka was highest in the north-east, where it could reach up to 1.2 tc ha–1 yr–1, and marginally lower in the coastal regions on both the north and southeast (fig. 5a). growth was slightly lower along the higherelevation inland ridge where cooler temperatures limited growth to 0.7–0.8 tc ha–1 yr–1. low values were also obtained for a few southern coastal sites where low precipitation combined with soils with low soil waterholding capacity to cause water-stress limitations. pines are capable of much faster biomass and wood accumulation than naturally regenerating shrubland. stem diameters in naturally regenerating stands are much smaller than for pines because of the much higher stand density of naturally regenerating stands, so that their biomass must be spread over a larger number of stems (fig. 5). pine growth was modelled to be highest in the regions with moderately high rainfall, with biomass growth of up to 12 tc ha–1 yr–1 (fig. 5b). this corresponds to wood production of 350–450 tdm ha–1 over a 30-year rotation (data not shown). in contrast, growth in the highrainfall raukumara range ridge running from north-east to south-west was slightly lower with biomass growth of only 7–8 tc ha–1 yr–1. unlike mānuka (stephens et al. 2005), pines generally do not grow well in regions with excessive rainfall (kirschbaum & watt 2011). cost-benefit analysis the estimated area of pine forests for conversion in each scenario was 315 and 556 ha, or 0.33 and 0.59% of the total planted pine area in the study area. in the first two scenarios, which included honey production in native forest areas, 1–2 ha was classified as high-umf production (table 2) and were estimated to have high economic returns, while the remainder of the land was assumed to produce lower-value honey with poorer economic returns. seven and 15 ha were estimated to incur additional expenses for successful native regeneration (table 2), including purchasing plants and planting costs. the remaining land within the scenarios was predicted to have sufficient landscape and geographic factors to permit natural regeneration (walsh et al. 2019) and was likely to be mānuka–kānuka shrubland initially (bray et al. 1999; funk et al. 2009; ministry for primary industries 2017b). there was lower reduced carbon storage and erosion under all four scenarios compared to the baseline (table 2). there were also some changes in biodiversity and water quality but given the size of lambie et al. new zealand journal of forestry science (2021) 51:14 page 9 figure 4: observed and modelled growth potential of mānuka-kānuka stands of different ages, showing observed (symbols) and modelled (solid curve) biomass (a) and observed biomass plotted against modelled data (b). observed data are from payton et al. (2010) and simulations from cenw. lambie et al. new zealand journal of forestry science (2021) 51:14 page 10 those changes and the lack of usable nonmarket values, they were difficult to monetise. restored significance, as an estimate of biodiversity improvements, was 324±20 ppb for all scenarios. overall, the monetised npv of converting from pine in each of our scenarios is negative (table 3). converting pine forestry land reduces profits from production and the land value, and the two combined can represent significant decreases. although revenue can be obtained from other enterprises based on natural afforestation, such as mānuka honey or oil, these are not as profitable as growing pines. however, there were some environmental benefits that could not be monetised and therefore were omitted from the npv calculations. highly erosion-prone land is usually on steep slopes and the analysis of profits and land values may be overestimated (table 3). across the four scenarios, losses in npv ranged between $3 and $8 million over the 62-year period or ~$150-$250 ha–1 yr–1 and lost profits from forestry represented the largest costs of conversion. decreases in carbon sequestration were also notable, as native forest stores less carbon across 62 years than pines. other social benefits, such as environmental and cultural factors, are more difficult to quantify. yao et al. (2014) found that there is a willingness of the new zealand public to pay for enhancement of forest ecosystems for biodiversity provision, but further work is needed to adequately provide values for biodiversity and cultural parameters, which may further offset wood derived profits under our scenarios. several other categories could be neither quantified nor monetised. in terms or overall impacts, these omitted categories would likely decrease the overall negative impacts of our scenarios. these omitted values may provide compelling reasons figure 5: simulated growth over 20 years of: (a) total biomass for naturally regenerating mānuka–kānuka stands (100,000 stems ha–1); and over 30 years of: (b) total biomass for pine (p. radiata) in the gisborne region. note the 10-fold difference in scales between mānuka–kānuka stands and pine. component scenario 1 scenario 2 scenario 3 scenario 4 ls 6+8, +gully + honey ls 4+6+8, +gully + honey ls 6+8, +gully ls 4+6+8, +gully total area converted (ha) 315 556 315 556 high umf area (ha) 1.04 2.18 0.00 0.00 active afforestation area (ha) 7.56 14.46 7.56 14.46 stored carbon δ (tc) –13,347 –23,600 –13,347 –23,600 erosion δ (t yr-1) –1,077 –1,968 –1,351 –2,469 table 2: model inputs for cost-benefit valuation for four scenarios, two scenarios with high (6) to very high (8) landslide susceptibility and two with moderate (4), high (6), and very high (8) landslide susceptibility, all scenarios included high gully prone land. scenarios 1 and 3 are the same except for the inclusion of honey in scenario 1. scenarios 2 and 4 are the same except for the inclusion of honey in scenario 2. ls is the landslide susceptibility category, and δ indicates the change from the baseline scenario. to promote native afforestation. cultural impacts, for instance may justify the loss in forestry profit is some cases. it is also difficult to calculate potential changes in employment resulting from our modelled land use conversions. however, since the scenarios we analysed affect only small land areas, the employment impact should be minor. larger conversions would likely result in proportionately larger employment changes with wider regional economic impacts. discussion both plantation pine and natural regeneration have longterm benefits for catchments and communities. however, clear-felling of pine stands can lead to introduction of sediment and woody debris to waterways during storm events (marden et al. 2006; imaizumi et al. 2008; payn et al. 2015; issaka & ashraf 2017; spiekermann & marden 2018). erosion susceptibility is inherently variable within catchments and geological terrains suggesting that land cover should ideally be as diverse as the landforms to create a mosaic of land uses across a property or catchment. economic impacts of conversion from pine production to native forest if assessing the viability of converting pines to natural regeneration on a purely economic basis, pines are considerably more profitable than natural regeneration, even when other market benefits of natural regeneration are included. the costs of conversion would be borne by pine forest owners, in terms of lost profit and changes in land value. conversely, many of the benefits from forests accrue to the general population, such as improvements in local air and downstream water quality, carbon sequestration, and cultural impacts (e.g., nowak et al. 2012). mānuka–kānuka shrubland is often the first phase of natural regeneration (overdyck & clarkson 2012), and under suitable circumstances, honey profits may be used to offset some of the losses from retiring land under pines. scenario 2 may represent the ideal situation whereby the most highly vulnerable parts of the landscape (susceptible to both landslides and gullying) are retired from production forestry and allowed instead to develop a permanent tree cover, with the associated benefit of honey production. many historical pine plantations were established in very difficult, steep terrain, often far from urban centres, ports, and mills increasing operational and infrastructural costs (raymond 2012). these impacts on profits were not included in our analysis and strategies are being developed to improve harvesting efficiency on steep sites, it is likely that profits for these slopes may be underestimated (raymond 2012, 2014; amishev et al. 2013). on particularly difficult slopes, trees may remain unharvested, especially during periods of low log prices (review panel 2014) and therefore act as a nursery crop of natural regeneration and any profits for these sites will be unrealised. component change in npv from baseline (nz$) scenario 1 scenario 2 scenario 3 scenario 4 ls 6+8, +gully + honey ls 4+6+8, +gully + honey ls 6+8, +gully ls 4+6+8, +gully carbon seq (4%) -1,427,491 -2,524,004 -1,427,491 -2,524,004 carbon seq (6%) -1,373,127 -2,427,881 -1,373,127 -2,427,881 reduced erosion (4%) 76,638 140,035 96,113 175,621 reduced erosion (6%) 54,504 99,591 68,354 124,899 native afforest (4%) -54,088 -103,375 -54,088 -103,375 native afforest (6%) -54,088 -103,375 -54,088 -103,375 net revenue (4%) -2,106,090 -3,745,326 -2,886,646 -5,133,173 net revenue (6%) -1,497,820 -2,663,621 -2,052,940 -3,650,638 water quality (4%) 9,871 17,454 9,871 17,454 water quality (6%) 6,721 11,884 6,721 11,884 total npv δ (4%) -3,501,160 -6,215,216 -4,262,241 -7,567,477 total npv δ (6%) -2,863,810 -5,083,402 -3,405,080 -6,045,111 table 3: baseline npv (nz $) across 62 years and change in npv for four scenarios (4% and 6% discount rates), two scenarios with high (6) to very high (8) landslide susceptibility and two with moderate (4), high (6), and very high (8) landslide susceptibility, all scenarios included high gully prone land. scenarios 1 and 3 are the same except for the inclusion of honey in scenario 1. scenarios 2 and 4 are the same except for the inclusion of honey in scenario 2. where ls is the landslide susceptibility category, ‘seq’ represents sequestration, ‘afforest’ represents afforestation, and δ indicates the change from the baseline scenario. lambie et al. new zealand journal of forestry science (2021) 51:14 page 11 weeds can reduce the growth and survival of target species and is an important factor for long-term erosion mitigation and monetary profits. weeds can initially compete strongly with mānuka–kānuka seedlings for light, water, and nutrients, but once mānuka trees are taller than competing weeds, they gain largely unrestricted access to solar radiation and will outshade low-stature weeds once the canopy is sufficiently large and dense. our growth simulations were based on assuming a well-established grass layer as the principal weed competitor. because of its high initial biomass, grass acts as an effective competitor in the early establishment phase, but because of its low maximum height, mānuka–kānuka seedlings could soon out-shade the grass layer and become the dominant plant type. if shrubs, such as gorse or broom, were the principal competitive weed species, their competitive inhibition of the growth of mānuka–kānuka seedlings could have persisted for longer or could even have prevented the establishment of the shrubland altogether. these impacts of weeds have been well characterised for pines (e.g., richardson et al. 1996; watt et al. 2007), but we are not aware of systematic studies of weed effects on the establishment of mānuka stands. pine seedlings are another substantial weed problem occurring post-harvest and during the natural regeneration stage. pine seedlings can have a substantial impact on native species establishment and survival (marlborough district council et al. 2016). further, the costs of pine seedling control can be substantial ($150– 500 ha–1) and affect the species composition of naturally regenerating forests as different native species differ in their sensitivity to wilding control with herbicide applications (marlborough district council et al. 2016; lambie & marden 2020). drivers for conversion of pine production to native forest land-management decisions are also subject to social influences as well as the economic factors (miller et al. 2007; bhandari et al. 2015). enhancing biodiversity values is a driver for shifts from pine to native forests (marlborough district council et al. 2016; lambie & marden 2020). pine forests have good biodiversity values, providing habitat for many plants and animals including threatened species (brockerhoff et al. 2001; bremer et al. 2010; pawson et al. 2010; michelsen et al. 2014; berndt & brockerhoff 2019). native forests typically contain a greater species richness compared with pine forests, as native forest has a greater range of food sources that are unimpacted by harvest and forest structural diversity providing a greater range of habitat; however, native forest is highly temporally and spatially variable (clout & gaze 1984; díaz et al. 2005; pawson et al. 2010). as a result, some landowners who highly value these attributes may be willing to reduce their profits from pine revenues. biodiversity offsetting is also a potential pathway for recognising the value associated with native forests (department of conservation 2014). however, a monetary or currency value associated with biodiversity offsetting in new zealand remains highly complex and a valuation framework is required (department of conservation 2014). biodiversity offsetting is generally supported through resource management undertaken by councils, under the umbrella of the resource management act. substantial work has been undertaken to establish guidance on biodiversity offsetting (department of conservation 2014; maseyk et al. 2018) and it is possible that this mechanism will be further supported under the resource management act reforms and in the future (ministry for the environment 2021). there are also pathways where financial value can be assigned to projects that enhance biodiversity. for example, green bonds, green funds, sustainability linked loans and biodiversity credits (chartres 2021). biodiversity credits are the most likely pathway by which monetary value associated with increased biodiversity and systems for biodiversity credits are being put in place internationally (e.g., porras and steel 2020; nsw department of planning, industry and environment 2021). biodiversity credits can be linked to biodiversity offsetting (nsw department of planning, industry and environment 2021) but require significant development of a framework in which to undertake credit trading and recognition in new zealand. shifts from pine to native forest cover can also be driven by cultural values (hēnare 2014). former stateowned forest assets are returning to māori under treaty of waitangi claims and could result in 41% of post-harvest pine forests being māori-owned (miller et al. 2007). māori connect with indigenous forests on a spiritual level associated with whakapapa (genealogy) and kaitiakitanga (guardianship), which provide additional non-economic factors for decisions on land use (miller et al. 2007) that may be attractive to māori landowners. in the gisborne region much of the most erosion-prone land is on māori-owned land (miller et al. 2007), which may provide an added incentive for increasing indigenous forests. carbon credits may also drive natural regeneration. however, changing from a fast-growing tree crop such as pine trees to potentially slower growing native trees will reduce carbon stocks and carbon sequestration and ultimately result in less carbon credits compared to pine trees (kimberley et al. 2014) and is more profitable when converting from pasture to forests. further, carbon accrual in regenerating forests is calculated generically from look-up tables which are particularly lacking with respect to native species, and do not contain regional or species-specific information (ministry for primary industries 2017b). carver and kerr (2017) suggest that updating the tables to include more native tree specific information will facilitate inclusion of native forests in the ets and therefore greater recognition of these forests for carbon income. conclusions we suggest a mosaic of land use within a property (or catchment) may be the best overall option, where moderate to highly susceptible land is left to naturally lambie et al. new zealand journal of forestry science (2021) 51:14 page 12 regenerate post-pine-harvest, while pine production is maintained on land less susceptible to erosion. for this multi-land use approach to be considered viable and adopted successfully, landowners will need to understand the erosion susceptibility of their land to create an impetus to shift from pine trees to native forest. this would benefit from support from local and central government agencies with the overall aim to meet cleaner water aspirations. many of the current mechanisms to encourage erosion mitigation are not applicable to those wanting to shift from pines to native forests and only support planting in previously unforested areas (e.g., one billion trees programme, erosion control funding programme). further, the relatively lower amount of carbon credits that native forests accrue relative to pine plantations may also discourage shifting to native species. overall, these results illustrate the economic value of the pine forestry industry, and the consequent challenges with transitioning to other land uses despite the greater environmental benefits associated with reducing erosion. list of abbreviations umf: unique mānuka factor; npv: net present value. competing interests the authors declare that they have no competing interests. authors’ contributions sl: conceptualisation, funding acquisition, methodology, project administration, led manuscript development, visualisation, supervision. sa: conceptualisation, funding acquisition, methodology, manuscript development. ad: methodology, manuscript development, visualisation. mk, mm, ts, rs, pw: conceptualisation, funding acquisition, methodology, formal analysis, manuscript development. acknowledgements the authors gratefully acknowledge the contribution of pia pohatu and tui warmenhoven (he oranga mo nga uri tuku iho trust), thomas robertson, and participants of the workshops for this project. further, we thank sarah richardson, larry burrows and robbie holdaway for provision of the data utilised in the biophysical modelling and the anonymous reviewers for their valuable contributions. funding this work was funded by the ministry for primary industries through the sustainable land management and climate change fund (slmacc 405415) and supported by manaaki whenua landcare research with strategic science investment funding. references aburto, f., cartes, e., mardones, o., rubilar, r. (2021). hillslope soil erosion and mobility in pine plantations and native deciduous forest in the coastal range of south-central chile. land degradation and development, 32, 453-466. https://doi.org/10.1002/ldr.3700 aguilar, f.x., obeng, e.a., cai, x. (2018). water quality improvments elicit consistent willingness-topay for the enhancement of forested watershed ecosystem services. ecosystem services, 30(part a), 158-171. https://doi.org/10.1016/j. ecoser.2018.02.012 akpa, s.i.c., odeh, i.o.a., bishop, t.f.a., hartemink, a.e., amapu, i.y. (2016). total soil organic carbon and carbon sequestration potential in nigeria. geoderma, 271, 202-215. https://doi. org/10.1016/j.geoderma.2016.02.021 aleotti, p., chowdhury, r. (1999). landslide hazard assessment: summary review and new perspectives. bulletin of engineering geology and the environment, 58, 21-44. https://doi. org/10.1007/s100640050066 amishev, d., basher, l., phillips, c., hill, s., marden, m., bloomberg, m., moore, j. (2013). new forest management approaches to steep hills. [mpi technical paper no. 2014/39], 109 p. wellington, new zealand: minstry for primary industries. arnold, n. (2018). when the rain came for tolaga bay. new zealand geographer, 152, 10-15. asure quality. (2020). agribase2020. https://www. asurequality.com/services/agribase/ (accessed 31 october 2020). barry, l.e., yao, r.t., harrison, d.r., paragahawewa, u.h., pannell, d.j. (2014). enhancing ecosystem services through afforestation: how policy can help. land use policy, 39, 135-145. https://doi.org/10.1016/j. landusepol.2014.03.012 berndt, l. a., brockerhoff, e. g. (2019). effects of land cover type on carabid beetles (coleoptera: carabidae) of the canterbury foothills, new zealand. new zealand journal of forestry science, 49. https://doi.org/10.33494/nzjfs492019x54x bergin d.o., kimberley, m.o., marden, m. (1993). how soon does regenerating scrub control erosion? new zealand journal of forestry, 38, 38-40. bergin, d.o., kimberley, m.o., marden, m. (1995). protective value of regenerating tea tree stands on erosion-prone hill country, east coast, north island, new zealand. new zealand journal of forestry science, 25, 3-19. betts, h., basher, l., dymond, j., herzig, a., marden, m., phillips, c. (2017). development of a landslide component for a sediment budget model. environmental modelling and software, 92, 28-39. https://doi.org/10.1016/j.envsoft.2017.02.003 lambie et al. new zealand journal of forestry science (2021) 51:14 page 13 https://doi.org/10.1002/ldr.3700 https://doi.org/10.1016/j.ecoser.2018.02.012 https://doi.org/10.1016/j.ecoser.2018.02.012 https://doi.org/10.1016/j.geoderma.2016.02.021 https://doi.org/10.1016/j.geoderma.2016.02.021 https://doi.org/10.1007/s100640050066 https://doi.org/10.1007/s100640050066 https://www.asurequality.com/services/agribase/ https://www.asurequality.com/services/agribase/ https://doi.org/10.1016/j.landusepol.2014.03.012 https://doi.org/10.1016/j.landusepol.2014.03.012 https://doi.org/10.33494/nzjfs492019x54x https://doi.org/10.1016/j.envsoft.2017.02.003 bhandari, k.p., aryal, j., darnsawasdi, r. (2015). a geospatial approach to assessing soil erosion in a watershed by integrating socio-economic determinants and the rusle model. natural hazards, 75, 321-342. https://doi.org/10.1007/ s11069-014-1321-2 black, r.d. (1980). upper cretaceous and tertiary geology of mangatu state forest, raukumara peninsula, new zealand. new zealand journal of geology and geophysics, 23, 293-312. https://doi. org/10.1080/00288306.1980.10424141 boyle, k.j., poor, p.j., taylor, l.o. (1999). estimating the demand for protecting freshwater lakes from eutrophication. american journal of agricultural economics, 81, 1118-1122. https://doi. org/10.2307/1244094 bray, j.r., burke, w.d., struick, g.j. (1999). propagule dispersion and forest regeneration in leptospermum scoparium (mānuka) and l. ericoides (kānuka) forests following fire in golden bay, new zealand. new zealand natural sciences, 24, 35-52. bremer, l.l., farley, k.a. (2010). does plantation forestry restore biodiversity or create green deserts? a synthesis of the effects of land-use transitions on plant species richness. biodiversity and conservation, 19(14), 3893-3915. https://doi. org/10.1007/s10531-010-9936-4 brockerhoff, e.g., ecroyd, c.e., langer, e.r. (2001). biodiversity in new zealand plantation forests: policy, incentives, and the state of our knowledge. new zealand journal of forestry, 31-37. burke, j. (2015). high performance manuka plantations. presented at rotorua land use opportunities symposium, 16-17 june 2015. cameron, r.j. (1960). natural regeneration of podocarps in the forests of the whirinaki river valley. new zealand journal of forestry, 8, 337-354. carswell, f.e., mason, n.w.h., overton, j.m., price, r., burrows, l.e., allen, r.b. (2015). restricting new forests to conservation lands severely constrains carbon and biodiversity gains in new zealand. biological conservation, 18, 206-218. https://doi. org/10.1016/j.biocon.2014.11.002 carver t., kerr s. (2017). facilitating carbon offsets from native forests. wellington, nz: motu economic and public policy research. https://doi.org/10.29310/ wp.2017.01 chartres, a. (2021). valuing biodiversity. accessed 30 august 2021. https://pureadvantage.org/ financing-biodiversity/ clark p. (2017). clarky’s comment february. accessed 11 march 2021. https://nz.pfolsen.com/marketinfo-news/wood-matters/2017/february/clarkys-comment-february/.clout, m.n., gaze, p.d. (1984). effects of plantation forestry on birds in new zealand. journal of applied ecology, 21, 795815. https://doi.org/10.2307/2405048 daigneault, a., greenhalgh, s., samarasinghe, o. (2018). economic impacts of multiple agro-environmental policies on new zealand land use. environmental and resources economics, 69, 763-785. https://doi. org/10.1007/s10640-016-0103-6 daigneault, a., wright, w., samarasinghe, o. (2015). economic analysis of land use opportunities in maniapoto rohe. [landcare research contract report lc2415], 60 p. te kuiti, new zealand: maniapoto māori trust board. department of conservation (2014). guidance on biodiversity offsetting in new zealand. wellington, nz: department of conservation. derose, r.c., gomez, b., marden, m., trustrum, n.a. (1998). gully erosion in mangatu forest, new zealand, estimated from digital elevation models. earth surface processes and landforms, 23, 1045-1053. https://doi.org/10.1002/ (sici)1096-9837(1998110)23:11<1045::aidesp920>3.0.co;2-tdíaz, i.a., armesto, j.j., reid, s., sieving, k.e., willson, m.f. (2005). linking forest structure and composition: avian diversity in successional forests of chiloé island, chile. biological conservation, 123(1), 91-101. https:// doi.org/10.1016/j.biocon.2004.10.011 dymond, j.r., ausseil, a.-g.e., ekanayake, j.c., kirschbaum, m.u.f. (2012). tradeoffs between soil, water, and carbon a national scale analysis from new zealand. journal of environmental management, 95, 124-131. https://doi.org/10.1016/j. jenvman.2011.09.019 dymond, j.r., ausseil, a.g., shepherd, j.d., buettner, l. (2006). validation of a region-wide model of landslide susceptibility in the manawatu-wanganui region of new zealand. geomorphology, 74, 70-79. https://doi.org/10.1016/j.geomorph.2005.08.005 dymond, j.r., betts, h.d., schierlitz, c.s. (2010). an erosion model for evaluating regional landuse scenarios. environmental modelling and software, 25, 289-298. https://doi.org/10.1016/j. envsoft.2009.09.011 evison, d. (2008). the impact of carbon credits on new zealand radiata pine forestry profitability. new zealand journal of forestry, 53, 42. forbes, a., norton, d. (2021) transitioning exotic plantations to native forest: a report on the state of knowledge. [mpi technical report no: 2021/22], 37 p. wellington, new zealand: ministry for primary industries. forest owners association, ministry for primary industries. (2019). fact and figures 2018/2019: new zealand plantation forest industry. 63p. wellington, new zealand: ministry for primary industries. fuller, i.c., marden, m. (2011). slope-channel coupling in steepland terrain: a field-based conceptual model from the tarndale gully and fan, waipaoa catchment, new zealand. geomorphology, lambie et al. new zealand journal of forestry science (2021) 51:14 page 14 https://doi.org/10.1007/s11069-014-1321-2 https://doi.org/10.1007/s11069-014-1321-2 https://doi.org/10.1080/00288306.1980.10424141 https://doi.org/10.1080/00288306.1980.10424141 https://doi.org/10.2307/1244094 https://doi.org/10.2307/1244094 https://doi.org/10.1007/s10531-010-9936-4 https://doi.org/10.1007/s10531-010-9936-4 https://doi.org/10.1016/j.biocon.2014.11.002 https://doi.org/10.1016/j.biocon.2014.11.002 https://doi.org/10.29310/wp.2017.01 https://doi.org/10.29310/wp.2017.01 https://pureadvantage.org/financing-biodiversity/ https://pureadvantage.org/financing-biodiversity/ https://nz.pfolsen.com/market-info-news/wood-matters/2017/february/clarky-s-comment-february/ https://nz.pfolsen.com/market-info-news/wood-matters/2017/february/clarky-s-comment-february/ https://nz.pfolsen.com/market-info-news/wood-matters/2017/february/clarky-s-comment-february/ https://doi.org/10.2307/2405048 https://doi.org/10.1007/s10640-016-0103-6 https://doi.org/10.1007/s10640-016-0103-6 https://doi.org/10.1016/j.biocon.2004.10.011 https://doi.org/10.1016/j.biocon.2004.10.011 https://doi.org/10.1016/j.jenvman.2011.09.019 https://doi.org/10.1016/j.jenvman.2011.09.019 https://doi.org/10.1016/j.geomorph.2005.08.005 https://doi.org/10.1016/j.envsoft.2009.09.011 https://doi.org/10.1016/j.envsoft.2009.09.011 128, 105-115. https://doi.org/10.1016/j. geomorph.2010.12.018 funk, j., field, c., kerr s., trotter, c. (2009). modeling the impact of carbon farming on a new zealand landscape. phd thesis, chapter ii. stanford university: palo alto, ca. galli, m., ardizzone, f., cardinali, m., guzzetti, f., reichenbach, p. (2008). comparing landslide inventory maps. geomorphology, 94, 268-289. https://doi.org/10.1016/j.geomorph.2006.09.023 gibson, e. (2021). 73 million trees ‘ not nearly enojugh’, says company using pine to nurture native forest. stuff, accessed 26 august 2021. https://www.stuff.co.nz/environment/climatenews/124826669/73-million-trees-not-nearlyenough-says-company-using-pine-to-nurturenative-forest goulder, l. h., & williams iii, r. c. (2012). the choice of discount rate for climate change policy evaluation. climate change economics, 3, 1250024. https:// doi.org/10.1142/s2010007812500248 grant, g. e.; wolff, a. l. (1991). long-term patterns of sediment transport after timber harvest, western cascade mountains, oregon, usa. in: n.e. peters, d.e. walling (eds.), sediment and stream water quality in a changing environment: trends and explanation, [proceedings of the vienna iahs symposium, vienna, austria. iahs publication no. 203], (pp 31-40). oxfordshire, united kingdom: international association of hydrological sciences. guild, d., dudfield, m. (2010). a history of fire in the forest and rural landscape in new zealand part 2, post 1830 influences, and implications for future fire management. new zealand journal of forestry, 54, 31-38. hall, d., lindsay, s., judd, s. (2017). permanent forest bonds: a pioneering environmental impact bond for aotearoa new zealand. [working paper 17/01], 45 p. wellington, new zealand: institute for governance and policy studies. harmsworth, g. r., & awatere, s. (2013). indigenous māori knowledge and perspectives of ecosystems. in: j. dymond (ed.), ecosystem services in new zealand-conditions and trends (pp 274-286). lincoln, new zealand: manaaki whenua press. hēnare, m. (2014). iwi signal end to pine plantations. http://www.uabsknowledge.ac.nz/en/researcha n d c o m m e n t / r e s e a r c h a n d a n a l y s i s / i w i signal-end-to-pine-plantations.html (accessed 3 february 2020). huber, r., finger r. (2020). a meta-analysis of the willingness to pay for cultural services from grasslands in europe. journal of agricultural economics, 71, 357-383. https://doi. org/10.1111/1477-9552.12361 imaizumi, f., sidle, r.c., kamei, r. (2008). effects of forest harvesting on the occurence of landslides and debris flows in steep terrain of central japan. earth surfaces proceses and landforms, 33, 827840. https://doi.org/10.1002/esp.1574 climate change comission (2021). ināia tonu nei: a low emissions future for aotearoa 2021. wellington, nz: climate change comission. issaka, s., ashraf, m.a. (2017). impact of soil erosion and degradation on water quality: a review. geololgy, ecology, and landscapes, 1, 1-11. https://doi.org/ 10.1080/24749508.2017.1301053 kimberley, m., bergin, d., beets, p. (2014). carbon sequestration by planted native trees and shrubs. hamilton, nz: tane’s tree trust. jindal, r., swallow, b., kerr, j. (2008). forestry-based carbon sequestration projects in africa: potential benefits and challenges. natural resources forum, 32(2), 116-130. https://doi.org/10.1111/j.14778947.2008.00176.x juang, c.h., lee, d.h., sheu, c. (1992). mapping slope failure potential using fuzzy sets. journal of geotechical engineering,118, 475494. https://doi.org/10.1061/(asce)07339410(1992)118:3(475) kirschbaum, m.u.f. (1999a). cenw, a forest growth model with linked carbon, energy, nutrient and water cycles. ecological modelling, 118, 17-59. https://doi.org/10.1016/s0304-3800(99)000204 kirschbaum, m.u.f. (1999b). modelling forest growth and carbon storage with increasing co2 and temperature. tellus, 51b, 871-888. https://doi. org/10.1034/j.1600-0889.1999.t01-4-00002.x kirschbaum, m.u.f., watt, m.s. (2011). use of a processbased model to describe spatial variation in pinus radiata productivity in new zealand. forest ecology and management, 262, 1008-1019. https://doi. org/10.1016/j.foreco.2011.05.036 kirschbaum, m.u.f., watt, m.s., tait, a., ausseil, a.-g.e. (2012). future wood productivity of pinus radiata in new zealand under expected climatic changes. global change biology, 18, 1342-1356. https://doi. org/10.1111/j.1365-2486.2011.02625.x lambie, s.m., marden, m. (2020). transitioning from exotic to native forest through natural regeneration: benefits and risks. [landcare research contract report lc3676], 19 p. napier, new zealand: hawkes bay regional council. lambie, s.m., marden, m., kirschbaum, m.u.f., soliman, t., walsh, p. (2018). best options for land use following radiata harvest in the gisborne district under climate change: literature review. [mpi technical paper no: 2018/46]. 76 p, wellington, new zealand: ministry for primary industries. landcare research. (2015). lcdbv4.1 land cover database version 4.1, mainland new zealand. https://lris.scinfo.org.nz/layer/48423-lcdb-v41land-cover-database-version-41-mainland-newlambie et al. new zealand journal of forestry science (2021) 51:14 page 15 https://doi.org/10.1016/j.geomorph.2010.12.018 https://doi.org/10.1016/j.geomorph.2010.12.018 https://doi.org/10.1016/j.geomorph.2006.09.023 https://www.stuff.co.nz/environment/climate-news/124826669/73-million-trees-not-nearly-enough-says-company-using-pine-to-nurture-native-forest https://www.stuff.co.nz/environment/climate-news/124826669/73-million-trees-not-nearly-enough-says-company-using-pine-to-nurture-native-forest https://www.stuff.co.nz/environment/climate-news/124826669/73-million-trees-not-nearly-enough-says-company-using-pine-to-nurture-native-forest https://www.stuff.co.nz/environment/climate-news/124826669/73-million-trees-not-nearly-enough-says-company-using-pine-to-nurture-native-forest https://doi.org/10.1142/s2010007812500248 https://doi.org/10.1142/s2010007812500248 http://www.uabsknowledge.ac.nz/en/research-and-comment/research-and-analysis/iwi-signal-end-to-pine-plantations.html http://www.uabsknowledge.ac.nz/en/research-and-comment/research-and-analysis/iwi-signal-end-to-pine-plantations.html http://www.uabsknowledge.ac.nz/en/research-and-comment/research-and-analysis/iwi-signal-end-to-pine-plantations.html https://doi.org/10.1111/1477-9552.12361 https://doi.org/10.1111/1477-9552.12361 https://doi.org/10.1002/esp.1574 https://doi.org/10.1080/24749508.2017.1301053 https://doi.org/10.1080/24749508.2017.1301053 https://doi.org/10.1111/j.1477-8947.2008.00176.x https://doi.org/10.1111/j.1477-8947.2008.00176.x https://doi.org/10.1061/(asce)0733-9410(1992)118:3(475) https://doi.org/10.1061/(asce)0733-9410(1992)118:3(475) https://doi.org/10.1016/s0304-3800(99)00020-4 https://doi.org/10.1016/s0304-3800(99)00020-4 https://doi.org/10.1034/j.1600-0889.1999.t01-4-00002.x https://doi.org/10.1034/j.1600-0889.1999.t01-4-00002.x https://doi.org/10.1016/j.foreco.2011.05.036 https://doi.org/10.1016/j.foreco.2011.05.036 https://doi.org/10.1111/j.1365-2486.2011.02625.x https://doi.org/10.1111/j.1365-2486.2011.02625.x https://lris.scinfo.org.nz/layer/48423-lcdb-v41-land-cover-database-version-41-mainland-new-zealand/ https://lris.scinfo.org.nz/layer/48423-lcdb-v41-land-cover-database-version-41-mainland-new-zealand/ zealand/ (accessed 24 january 2020). landcare research. (2020). national soils database. h t t p s : / / s o i l s . l a n d c a r e r e s e a r c h . c o . n z / s o i l data/national-soils-data-repository-and-thenational-soils-database/the-nsdr-improvementprogramme/ (accessed 23 jaunary 2020). landcare research (2010). nzlri north island, edition 2 (all attributes). https://lris.scinfo.org.nz/ layer/48134-nzlri-north-island-edition-2-allattributes/ (accessed 24 january 2020). landcare research, scion (2017). debris flows. https:// w w w. n z fo a . o r g . n z / re s o u rc e s / f i l e l i b ra r i e s resources/environment/factsheets/582-debris/ file (accessed 6 december 2017). lilburne, l., webb, t., ford, r., bidwell, v. (2010). estimating nitrate-nitrogen leaching rates under rural land uses in canterbury. [environment canterbury technical report r10/127], 37p. kaikoura, new zealand: canterbury regional council. marden, m., arnold, g., seymour, a., hambling, r. (2012). history and distribution of steeplands gullies in response to land use change, east coast region, north island, new zealand. geomorphology, 153-154, 81-90. https://doi.org/10.1016/j. geomorph.2012.02.011 marden, m., rowan, d. (1993). protective value of vegetation on tertiary terrain before and during cyclone bola, east coast, north island, new zealand. new zealand journal of forestry science, 23, 255-63. marden, m., rowan, d., phillips, c. (2006). sediment sources and delivery following plantation harvesting in a weathered volcanic terrain, coromandel peninsula, north island, new zealand. soil research, 44, 219-232. https://doi. org/10.1071/sr05092 marlborough district council, department of conservation, marlborough sounds restoration trust. (2016). guidelines for converting pine plantations to native vegetation in the marlborough sounds. 11 p. marlborough, new zealand: marlborough district council. maseyk, f., ussher, g., kessels, g., christensen, m., brown m. (2018). biodiversity offsetting under the resource management act: a guidance document. new zealand: biodiversity working group. massey, d.m., newbold, s.c., gentner, b. (2006). valuing water quality changes using a bioeconomic model of a coastal recreational fishery. journal of environmental economics and management, 52, 482-500. https://doi.org/10.1016/j. jeem.2006.02.001 mason, n.w.h., ausseil, a.-g.e., dymond, j.r., overton, j.m., price, r., carswell, f.e. (2012). will use of non biodiversity objectives to select areas for ecological restoration always compromise biodiversity gains? biological conservation, 155, 157-168. https://doi. org/10.1016/j.biocon.2012.05.019 mcglone, m.s. (1983). polynesian deforestation of new zealand: a preliminary synthesis. archaeology in oceania, 18, 11-25. https://doi.org/10.1002/ arco.1983.18.1.11 mcglone, m.s. (1989). the polynesian settlement of new zealand in relation to environmental and biotic changes. new zealournal journal of ecology, 12, 115-129. michelsen, o., mcdevitt, j.e., coelho, c.r.v. (2014) a comparison of three methods to assess land use impacts on biodiversity in a case study of forestry plantations in new zealand. the international journal of life cycle assessment, 19, 1214-1225. https://doi.org/10.1007/s11367-014-0742-1 miller, r., dickinson, y., reid, a. (2007). māori connections to forestry in new zealand. in: s. feary (ed.) forestry for indigenous peoples: learning from experiences in forest industries. (pp. 13-22). canberra, australia: australian national university. ministry for forestry. (1994). a guide to the east coast forestry project. wellington, new zealand: ministry for forestry. ministry for primary industries. (2017a). carbon look-up tables for forestry in the emissions trading scheme. wellington, new zealand: ministry for primary industries. ministry for primary industries. (2017b). a guide to the carbon look-up tables for forestry in the emission trading scheme. 44 p. wellington, new zealand: ministery for primary industries. ministry for primary industries (2017c). erosion control funding programme. http://www.mpi.govt.nz/ funding-and-programmes/forestry/erosioncontrol-funding-programme (accessed 23 january 2020). ministry for the environment 2021. overview of the resource management reforms. https:// environment.govt.nz/what-government-is-doing/ key-initiatives/resource-management-systemre fo r m / ove r v i e w / # o b j e c t ive s o f r m re fo r m (accessed 26 august 2021). moore, c., guigne, d., dockin, c., maguire, k.b., simo, n.b. (2018). valuing ecological improvements in the chesapeake bay and the importance of ancillary benefits. journal of benefit-cost analysis, 9, 1-26. https://doi.org/10.1017/bca.2017.9 newsome, p.f.j. (1987). the vegetation cover of new zealand. wellington, nz: water and soil directorate, ministry of works and development. nixon, c., amperle, d., pambudi, d., clough, p. (2017). plantation forestry statistics: contribution of forestry to new zealand. 83 p. wellington, new zealand: new zealand institute for economic research. nowak, d.j., hirabayashi, s., bodine, a., greenfield, e. (2014). tree and forest effects on air quality and lambie et al. new zealand journal of forestry science (2021) 51:14 page 16 https://lris.scinfo.org.nz/layer/48423-lcdb-v41-land-cover-database-version-41-mainland-new-zealand/ https://soils.landcareresearch.co.nz/soil-data/national-soils-data-repository-and-the-national-soils-database/the-nsdr-improvement-programme/ https://soils.landcareresearch.co.nz/soil-data/national-soils-data-repository-and-the-national-soils-database/the-nsdr-improvement-programme/ https://soils.landcareresearch.co.nz/soil-data/national-soils-data-repository-and-the-national-soils-database/the-nsdr-improvement-programme/ https://soils.landcareresearch.co.nz/soil-data/national-soils-data-repository-and-the-national-soils-database/the-nsdr-improvement-programme/ https://lris.scinfo.org.nz/layer/48134-nzlri-north-island-edition-2-all-attributes/ https://lris.scinfo.org.nz/layer/48134-nzlri-north-island-edition-2-all-attributes/ https://lris.scinfo.org.nz/layer/48134-nzlri-north-island-edition-2-all-attributes/ https://www.nzfoa.org.nz/resources/file-libraries-resources/environment/factsheets/582-debris/file https://www.nzfoa.org.nz/resources/file-libraries-resources/environment/factsheets/582-debris/file https://www.nzfoa.org.nz/resources/file-libraries-resources/environment/factsheets/582-debris/file https://www.nzfoa.org.nz/resources/file-libraries-resources/environment/factsheets/582-debris/file https://doi.org/10.1016/j.geomorph.2012.02.011 https://doi.org/10.1016/j.geomorph.2012.02.011 https://doi.org/10.1071/sr05092 https://doi.org/10.1071/sr05092 https://doi.org/10.1016/j.jeem.2006.02.001 https://doi.org/10.1016/j.jeem.2006.02.001 https://doi.org/10.1016/j.biocon.2012.05.019 https://doi.org/10.1016/j.biocon.2012.05.019 https://doi.org/10.1002/arco.1983.18.1.11 https://doi.org/10.1002/arco.1983.18.1.11 https://doi.org/10.1007/s11367-014-0742-1 http://www.mpi.govt.nz/funding-and-programmes/forestry/erosion-control-funding-programme http://www.mpi.govt.nz/funding-and-programmes/forestry/erosion-control-funding-programme http://www.mpi.govt.nz/funding-and-programmes/forestry/erosion-control-funding-programme https://doi.org/10.1017/bca.2017.9 human health in the united states. environmental pollution, 193, 119-129. https://doi.org/10.1016/j. envpol.2014.05.028 nsw department of planning, industry and environment. (2021). what are biodiversity credits? accessed 30 august 2021. https://www.environment.nsw. gov.au/topics/animals-and-plants/biodiversityoffsets-scheme/about-the-biodiversity-offsetsscheme/what-are-biodiversity-credits olssen, a., zhang, w., evison, d., kerr, s. (2012). a forest-profit expectations dataset for new zealand, 1990-2008. [motu working paper 12-07], 25p. wellington, new zealand: motu economic and public policy research. https://doi.org/10.29310/ wp.2012.07 overdyck, e., clarkson, b.d. (2012). seed rain and soil seed banks limit native regeneration within urban forest restoration plantings in hamilton city, new zealand. new zealand journal of ecology, 36, 177190. parfitt, r.l., percival, h.j., dahlgren, r.a., hill, l.f. (1997). soil and solution chemistry under pasture and radiata pine in new zealand. plant and soil, 191, 279290. https://doi.org/10.1023/a:1004266000509 pawson, s.m., ecroyd, c.e., seaton, r., shaw, w.b., brockerhoff, e.g. (2010). new zealand’s exotic plantation forest as habitat for threatened indigenous species. new zealand journal of ecology, 3, 342-355. payn, t., phillips, c., basher, l., baillie, b., garrett, l., harrison, d., heaphy, m., marden, m. (2015). improving management of post-harvest risks in steepland plantations. new zealand journal of forestry, 60, 3-6. payton, i.j., barringer, j., lambie, s., lynn, i., forrester, g., pinkney, t. (2010). carbon sequestration rates for post-1989-compliant indigenous forests. [landcare research contract report lc0809/107], 35 p. lincoln, new zealand: landcare research. pearce, a.j., o’loughlin, c.l., jackson, r.j., zhang, x.b. (1987). reforestation: on site effects on hydrology and erosion, eastern raukumara range, new zealand. forest hydrology and watershed management, 167. phillips, c., marden, m., basher, l. (2012). plantation forest harvesting and landscape response what we know and what we need. new zealand journal of forestry, 56, 4-12. porras, i., steele, p. (2020). making the market work for nature: how biodiversity credits can protect biodiversity and reduce poverty. [iied issue paper], london: iied. raymond, k. (2012). innovation to increase profitabiilty of steep terrain harvesting in new zealand. new zealand journal of forestry, 57, 19-23. raymond, k. (2014, november). new harvest technology for a safer future: challenges on harvesting steepland. new zealand tree grower, 3-6. review panel. (2014). independent forestry safety review: an agenda for change in the forestry sector. 12 p. wellington, new zealand: new zealand forest owners association. rhodes, d. (2001). rehabilitation of deforested steep slopes on the east coast of new zealand’s north island. unasylva, 207, article 4. richards k. (2017). national environmental standard for plantation forests. accessed 11 march 2021. https://nz.pfolsen.com/market-info-news/woodmatters/2017/october/national-environmentalstandard-for-plantation-forests/ richardson, b., vanner, a., ray, j., davenhill, n., coker, g. (1996). mechanisms of pinus radiata growth suppression by some common weed species. new zealand journal of forestry science, 26, 421-432. ross, d.j., scott, n.a., lambie, s.m., trotter, c.m., rodda, n.j., townsend, j.a. (2009). nitrogen and carbon cycling in a new zealand pumice soil under a manuka (leptospermum scoparium) and kanuka (kunzea ericoides) shrubland. soil research, 47, 725-736. https://doi.org/10.1071/sr08261 ruff, m., czurda, k. (2008). landslide susceptibility analysis with a heuristic approach in the eastern alps (vorarlberg, austria). geomorphology, 94, 314-324. https://doi.org/10.1016/j. geomorph.2006.10.032 sciarra, m., coco, l., urbano, t. (2017). assessment and validation of gis-based landslide susceptibility maps: a case study from feltrino stream basin (central italy). bulletin of engineering geology and the environment, 76, 437-456. https://doi. org/10.1007/s10064-016-0954-7 scott, n.a., white, j.d., townsend, j.a., whitehead, d., leathwick, j.r., hall, g.m.j., marden, m., rogers, g.n.d., watson, a.j., whaley, p.t. (2000). carbon and nitrogen distribution and accumulation in a new zealand scrubland ecosystem. canadian journal of forest research, 30, 1246-1255. https:// doi.org/10.1139/x00-048 soeters, r., van westen, c.j. (1996). slope stability: recognition, analysis and zonation. in: a.k. turner, r.l. shuster (eds.), landslides: investigation and mitigation, [transportation research board special report 247], (pp. 129-177). washington, usa, transportation research board. spiekermann, r., marden, m. (2018). slmacc 405415: best options for land use following radiata harvest in the gisborne district under climate change: spatial analysis of erosion susceptibility in plantation forests, east coast region. [landcare research contract report lc3202], wellington, new zealand: ministry for primary industries. stephens, j.m.c., molan, p.c., clarkson, b.d. (2005). a review of leptospermum scoparium (myrtaceae) in new zealand. new zealand journal of botany, lambie et al. new zealand journal of forestry science (2021) 51:14 page 17 https://doi.org/10.1016/j.envpol.2014.05.028 https://doi.org/10.1016/j.envpol.2014.05.028 https://www.environment.nsw.gov.au/topics/animals-and-plants/biodiversity-offsets-scheme/about-the-biodiversity-offsets-scheme/what-are-biodiversity-credits https://www.environment.nsw.gov.au/topics/animals-and-plants/biodiversity-offsets-scheme/about-the-biodiversity-offsets-scheme/what-are-biodiversity-credits https://www.environment.nsw.gov.au/topics/animals-and-plants/biodiversity-offsets-scheme/about-the-biodiversity-offsets-scheme/what-are-biodiversity-credits https://www.environment.nsw.gov.au/topics/animals-and-plants/biodiversity-offsets-scheme/about-the-biodiversity-offsets-scheme/what-are-biodiversity-credits https://doi.org/10.29310/wp.2012.07 https://doi.org/10.29310/wp.2012.07 https://doi.org/10.1023/a:1004266000509 https://nz.pfolsen.com/market-info-news/wood-matters/2017/october/national-environmental-standard-for-plantation-forests/ https://nz.pfolsen.com/market-info-news/wood-matters/2017/october/national-environmental-standard-for-plantation-forests/ https://nz.pfolsen.com/market-info-news/wood-matters/2017/october/national-environmental-standard-for-plantation-forests/ https://doi.org/10.1071/sr08261 https://doi.org/10.1016/j.geomorph.2006.10.032 https://doi.org/10.1016/j.geomorph.2006.10.032 https://doi.org/10.1007/s10064-016-0954-7 https://doi.org/10.1007/s10064-016-0954-7 https://doi.org/10.1139/x00-048 https://doi.org/10.1139/x00-048 43, 431-449. https://doi.org/10.1080/0028825 x.2005.9512966takatsuka, y., cullen, r., wilson, m., wratten, s. (2009). using stated preference techniques to value four key ecosystem services on new zealand arable land. international journal of agricultural sustainability, 7(4): 279-291. https:// doi.org/10.3763/ijas.2009.0334 tait, a. (2008). future projections of growing degree days and frost in new zealand and some implications for grape growing. weather climate, 28, 17-36. https://doi.org/10.2307/26169696 tait, a., henderson, r., turner, r., zheng, x. (2006). thinplate smoothing spline interpolation of daily rainfall for new zealand using a climatological rainfall surface. international journal of climatology, 26, 2097-2115. https://doi.org/10.1002/joc.1350 tait, a., liley, b. (2009). interpolation of daily solar radiation for new zealand using a satellite-derived cloud cover surface. weather and climate, 29, 7088. https://doi.org/10.2307/26169706 te tai ōhanga. (2018). discount rates. https://treasury. govt.nz/information-and-services/state-sectorleadership/guidance/financial-reporting-policiesand-guidance/discount-rates (last accessed 25 june 2020). te uru rākau. (2020). historic indicative new zealand radiata pine log prices. indicative prices of radiata pine logs by quarter since 1992. https://www. teururakau.govt.nz/news-and-resources/opendata-and-forecasting/forestry/wood-productmarkets/historic-indicative-new-zealand-radiatapine-log-prices/ (last accessed 24 june 2020). tian, y., wu, h., zhang, g., wang, l., zheng, d., li, s. (2020). perceptions of ecosystem services, dissercies, and willingness-to-pay for urban green space conservation. journal of environmental management, 260, 110140. https://doi. org/10.1016/j.jenvman.2020.110140 van westen, c.j., castellanos, e., kuriakose, s.l. (2008). spatial data for landslide susceptibility, hazard, and vulnerability assessment: an overview. engineering geology, 102, 112-131. https://doi. org/10.1016/j.enggeo.2008.03.010 varnes, j.d. (1984). iaeg comission on landslides and other mass movements, landslide hazard zonation: a review of principles and practice. pp 63. paris, france: unesco press. veld, g.j., de graaf, f. (1990). erosion damage as a result of cyclone bola: an assessment on arai matawai and emerald hills properties east coast region, north island, new zealand. gisborne, 41 p. new zealand: forest research institute and the ministry of forestry. walsh, p.j., soliman, t., greenhalgh, s., mason, n.w.h., palmer, d. (2017). valuing the benefits of permenant forests. [ministry for primary industries technical paper no: 2017/68], wellington, new zealand: ministry for primary industries. walsh, p., soliman, t., robertson, t. (2019). a cost-benefit analysis of transitions from exotic to native forestry in gisborne. [landcare research ccontract report lc3553] lincoln, new zealand: landcare research. watt, m.s., kimberley, m.o., coker, g., richardson, b., estcourt, g. (2007). modelling the influence of weed competition on growth of young pinus radiata. development and parametization of a hybrid model across an environmental gradient. canadian journal of forest research, 37, 607-616. https://doi.org/10.1139/x06-254 watt, m.s., kirschbaum, m.u.f., meason, d., jovner, a., pearce, h.g., moore, j.r., nicholas, i., bulman, l., rolando, c., palmer, d.j., harrison, d., hock, b.k., tait, a., ausseil, a-g.e. (2012). future forest systems. 180 p. [scion client report]. rotorua, new zealand: scion. wetere, w. (2015). what’s our potential? maniapoto and manuka honey. presentation at ngā aho rangahau o maniapoto threads of research symposium, te kuiti, july 2015. whitehead, d., walcroft, a.s. (2005). forest and shrubland canopy carbon uptake in relation to foliage nitrogen concentration and leaf area index: a modelling analysis. annals of forest science, 62, 525-535. https://doi.org/10.1051/ forest:2005045 whitehead, d., walcroft, a.s., scott, n.a., townsend, j.a., trotter, c.m., rogers, g., rogers, g.n.d. (2004). characteristics of photosynthesis and stomatal conductance in the shrubland species mānuka (leptospermum scoparium) and kānuka (kunzea ericoides) for the estimation of annual canopy carbon uptake. tree physiology, 24, 795-804. https://doi.org/10.1093/treephys/24.7.795 williams, p.a. (1983). secondary vegetation succession on the port hills, banks peninsula, canterbury, new zealand. new zealand journal of ecology, 21, 237-247. https://doi.org/10.1080/002882 5x.1983.10428556 wilson, h.d. (1994). regeneration of native forest on hinewai reserve, banks peninsula. new zealand journal of ecology, 32, 373-383. https://doi.org/1 0.1080/0028825x.1994.10410480 yao, r.t., scarpa, r., turner, j.a., barnard, t.d., rose, j.m., palma, j.h.n., harrison, d.r. (2014). valuing biodiversity enhancement in new zealand’s planted forests: socioeconomic and spatial determinants of willingness-to-pay. ecological economics, 98, 90-101. https://doi.org/10.1016/j. ecolecon.2013.12.009 yalcin, a., bulut, f. (2007). landslide susceptibility mapping using gis and digital photogrammetric techniques: a case study from ardesen (neturkey). natural hazards, 41, 201-226. https:// doi.org/10.1007/s11069-006-9030-0 lambie et al. new zealand journal of forestry science (2021) 51:14 page 18 https://doi.org/10.3763/ijas.2009.0334 https://doi.org/10.3763/ijas.2009.0334 https://doi.org/10.2307/26169696 https://doi.org/10.1002/joc.1350 https://doi.org/10.2307/26169706 https://treasury.govt.nz/information-and-services/state-sector-leadership/guidance/financial-reporting-policies-and-guidance/discount-rates https://treasury.govt.nz/information-and-services/state-sector-leadership/guidance/financial-reporting-policies-and-guidance/discount-rates https://treasury.govt.nz/information-and-services/state-sector-leadership/guidance/financial-reporting-policies-and-guidance/discount-rates https://treasury.govt.nz/information-and-services/state-sector-leadership/guidance/financial-reporting-policies-and-guidance/discount-rates https://www.teururakau.govt.nz/news-and-resources/open-data-and-forecasting/forestry/wood-product-markets/historic-indicative-new-zealand-radiata-pine-log-prices/ https://www.teururakau.govt.nz/news-and-resources/open-data-and-forecasting/forestry/wood-product-markets/historic-indicative-new-zealand-radiata-pine-log-prices/ https://www.teururakau.govt.nz/news-and-resources/open-data-and-forecasting/forestry/wood-product-markets/historic-indicative-new-zealand-radiata-pine-log-prices/ https://www.teururakau.govt.nz/news-and-resources/open-data-and-forecasting/forestry/wood-product-markets/historic-indicative-new-zealand-radiata-pine-log-prices/ https://www.teururakau.govt.nz/news-and-resources/open-data-and-forecasting/forestry/wood-product-markets/historic-indicative-new-zealand-radiata-pine-log-prices/ https://doi.org/10.1016/j.jenvman.2020.110140 https://doi.org/10.1016/j.jenvman.2020.110140 https://doi.org/10.1016/j.enggeo.2008.03.010 https://doi.org/10.1016/j.enggeo.2008.03.010 https://doi.org/10.1139/x06-254 https://doi.org/10.1051/forest:2005045 https://doi.org/10.1051/forest:2005045 https://doi.org/10.1093/treephys/24.7.795 https://doi.org/10.1080/0028825x.1983.10428556 https://doi.org/10.1080/0028825x.1983.10428556 https://doi.org/10.1080/0028825x.1994.10410480 https://doi.org/10.1080/0028825x.1994.10410480 https://doi.org/10.1016/j.ecolecon.2013.12.009 https://doi.org/10.1016/j.ecolecon.2013.12.009 https://doi.org/10.1007/s11069-006-9030-0 https://doi.org/10.1007/s11069-006-9030-0 lambie et al. new zealand journal of forestry science (2021) 51:14 page 19 appendix gisborne climate solar radiation across the gisborne region is fairly uniform, with just slightly higher values at the coast and slightly lower values inland at higher elevations (fig. a1a). temperature shows greater variation, with mean annual temperatures above 15°c at the north-eastern coastal tip and dropping to below 10°c in the higher inland locations, especially towards the south-west (fig. a1b, c, d). there is rainfall of 2,500 to over 3,000 mm yr–1 in the upland regions, but less than 1,500 mm yr–1 in most of the coastal regions, and rainfall of only around 1,000 mm yr–1 inland from gisborne city (fig. s1e). in terms of temperature extremes, the area near gisborne city has experienced the greatest temperature extremes of recorded maxima above 35°c, while the northern side of the peninsula has reached only 26–28°c. absolute minima have generally been lowest in the highland region with values below –7°c, whereas the coastal regions generally remained milder with minima mostly not falling below –3°c, except for a large part of the east coast where temperatures have fallen to below –5°c. figure a1: key environmental variables and soil characteristics of the gisborne region, radiation (a); mean temperature (b); recorded maximum (c); and minimum temperatures over a 20-year period (d); rainfall (e); and soil water-holding capacity (f ). climatic data refer to the 1980–1999 period, with panels giving either the averages recorded over that period (a, b, e) or the absolute extremes recorded over that period (c, d). all data are shown at 0.05-degree resolution. figure a2: the relative growth response of mānuka–kānuka stands for different mean annual temperatures (a); and mean annual rainfall (b). these relationships have been developed based on analysis of the data of payton et al. (2010) and the observed distribution of mānuka–kānuka stands across new zealand. science, policy, and sustainable indigenous forestry in new zealand matt s. mcglone1, peter j. bellingham1,2 and sarah j. richardson1* 1manaaki whenua – landcare research, po box 69040, lincoln 7640, new zealand 2school of biological sciences, university of auckland, auckland 1142, new zealand *corresponding author: richardsons@landcareresearch.co.nz (received for publication 29 august 2021; accepted in revised form 2 march 2022) abstract background: over 80% of new zealand’s indigenous forests are in public ownership with logging prohibited, and logging of private indigenous forests is restricted to sustainable harvesting only. such limitations are highly unusual globally and were imposed only in the last few decades of the 20th century. previously, the national goal had been indigenous wood production in perpetuity. here we review the role of forestry science in this outcome, and in particular in relation to the policies and practices adopted by the new zealand forest service. methods: literature review. results: as early as 1900, it was recognised that economically viable management of indigenous forests for timber production was marginal at best. nevertheless, the forest service, from its formation in 1919 to its abolition in 1987, advocated sustainable commercial management of indigenous forests. however, it failed to bring any significant areas under such management nor prevented conversion of substantial tracts of old-growth forest to exotic plantations or agriculture. indigenous forest logging would have continued until commercial exhaustion of tall conifer species if a confluence of factors (urbanization, political upheaval, rise of an assertive conservation movement, and declining economic contribution) had not weakened the influence of provincial logging advocacy. forestry research played a minor role in this saga as it focused on the technical issues of indigenous silviculture (e.g., coupe vs group vs single-tree harvesting methods) while the main drivers of change were economic, social, and cultural. conclusions: commercially valuable indigenous forests were protected only when the political cost of continuing logging was greater than that of halting it. however, it is an open question if the current policy settings will remain. changes in governance (including increased māori participation), land use change, planted indigenous forests and formation of exoticindigenous forest communities will affect public attitudes as regards their use. if indigenous forestry science is to be of more consequence than in the past, new zealand will need clear forestry goals and policies to deal with these changed circumstances, and the will to implement them. new zealand journal of forestry science mcglone et al. new zealand journal of forestry science (2022) 52:8 https://doi.org/10.33494/nzjfs522022x182x e-issn: 1179-5395 published on-line: 25/03/2022 © the author(s). 2022 open access this article is distributed under the terms of the creative commons attribution 4.0 international license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the creative commons license, and indicate if changes were made. research article open access (allen et al. 2013), aside from occasional salvage logging (watson 2017), and strictly regulated on private land. for a well-wooded country, new zealand thus has an unusual, perhaps unique, forestry industry (benecke 1996). this is a recent development. for most of the period between 1850 and 1960, indigenous timbers supplied most or a large proportion of local demand and, in the earlier years, were an export mainstay (roche 1990). from the 1870s onwards, forestry professionals consistently argued for management of the indigenous introduction ‘no issue in forestry evokes such strong emotions as logging…’ (singh 2001). the present day indigenous forests of new zealand cover 30% (8.0 million ha) of its land area but yield only 0.08% of its timber products (ministry for primary industries 2015). exotic forest plantations (7.9% of the land area) account for virtually all the wood production. indigenous forestry is prohibited on state-owned land keywords: indigenous forest; logging; sustainability; conservation; silviculture mailto:richardsons@landcareresearch.co.nz http://creativecommons.org/licenses/by/4.0/), mcglone et al. new zealand journal of forestry science (2022) 52:8 page 2 forests to ensure sustained production (star 2002) and progressively larger areas eventually amounting to over 80% of the total were brought under state control. from the 1920s onwards, areas suitable for sustained harvest were designated as state forests, inventories of the indigenous forest resource conducted, and its silviculture and protection from pests and weeds researched. nevertheless, by the early years of the 21st century, the last state-administered forests were withdrawn from timber production and strict controls imposed on logging in privately-owned tracts (allen 2004, allen et al. 2013). this outcome of near-total reservation of indigenous forest is often presented as a triumphalist narrative of conservationists prevailing in the face of the machinations of a forest service bent on destruction of the native forests (bensemann 2018). it is more complex than that, however, with economics, state restructuring, changing electoral politics and urbanisation also playing key roles. however, the failure of the hundred-year effort made by generations of scientists, foresters, managers, and policy makers to achieve large-scale and sustainable production of affordable timber from indigenous forests is arguably the critical element. it is this effort that we review here, as it has important lessons about the limitations of science in resolving dilemmas in which biology, economics, and public opinion collide. there is an extensive literature on the history of new zealand forestry, and we found the following particularly informative. michael roche included detailed and insightful chapters on the history of indigenous forest utilisation and the conservation debates in his book on this topic (roche 1990). we have also relied on alan reid’s comprehensive coverage of the history and impact of logging restrictions in new zealand’s indigenous forests (reid 2001). for insights into the policies and activities of the state forest service (1919–1949) and its successor, the new zealand forest service (1949–1987), the annual reports of these organisations to parliament are invaluable1. until the retirement of the directorgeneral, pat entrican, these remarkable documents gave frank, outspoken assessments of the progress of forestry and did not hesitate to warn and admonish successive government administrations, even to the extent of outright opposition to government policy. they also launched uninhibited critiques of sawmillers, the timber industry as a whole, local and visiting experts and, in particular, conservationists (‘preservationists’ in their language). we refer to the state forest service annual reports to parliament as sfsar and the new zealand forest service annual reports as nzfsar. there were intense internal debates over indigenous forestry within the forest service, with a wide range of opinions expressed, and some staff did their utmost to restrict the area logged and to minimise the damage done during logging. however, as our focus is on the policies as promulgated and their consequences, we leave an account of these internal debates to others. two recent books provide such accounts. elizabeth orr (2017) gives an insider’s view (she was the daughter of pat entrican, the director-general of forests, 1939– 1961) of the controversies surrounding state forestry. the drama of conservation: the history of pureora forest, new zealand (king et al. 2015) provides a scientific and social history of a forest pivotal in the late 20th century debate on indigenous logging and presents the views and reminiscences of scientists involved. in this review we examine the issues surrounding exploitation of indigenous forests but not those concerning indigenous plantation forestry, a related but separate matter (see bergin & kimberley 2003, steward et al. 2014). we first summarise the ecology of the indigenous forests in relation to the commercial timber species before summarising the history of their exploitation, the attempts to achieve sustainable harvesting and its ultimate abandonment. we then examine the role of scientific investigation in this history and then pose the question as to whether sustainable indigenous forestry was ever an economically, ecologically, or socially credible goal for new zealand. finally, we discuss whether the imposition of a forest policy that restricts the exploitation of indigenous forests will persist, given changes in governance and the expansion of planned and naturalising hybrid exotic– indigenous forest mixtures. ecology of new zealand indigenous forests composition and structure new zealand was almost entirely forest covered below tree line before human settlement in the thirteenth century. although growing under an oceanic climate similar to that of north-western north america or western europe, new zealand forests are markedly different to either (mcglone et al. 2016) and observers have often commented on their ‘tropical’ appearance (dawson & sneddon 1969). the forests are nearly entirely evergreen and, in the lowlands and many montane locations, structurally complex, with abundant tall lianas and monopodial trees (palms, tree ferns), epiphytic trees and shrubs, strangler lianas, fern-rich ground layers and few understory forbs (mcglone et al. 2016). most lowland and some northern montane forests are characterised by emergent conifers (araucariaceae, podocarpaceae and cupressaceae) over a canopy layer of broad-leaved angiosperms. these forests in general are species-rich for their latitude and the new zealand flora as a whole has over 250 almost entirely endemic species of trees and lianas (mcglone et al. 2010). cooler uplands in the north and the lowlands of the far south have extensive nanophyll nothofagaceae (collectively referred to as “southern 1 annual reports of the director of forestry (state forest service) (1921-1949) were presented to parliament from 1921 to 1949 as appendices to the journals of the house of representatives, section c (crown lands) 03. annual reports of the director of forestry (from 1961 director-general of forests) to parliament were continued under the new zealand forest service until its dissolution in 1987. the state forest service reports are digitally available in the national library collection (https://paperspast.natlib.govt.nz/parliamentary/ajhr). the new zealand forest service annual reports are held as physical copies in larger libraries. https://paperspast.natlib.govt.nz/parliamentary/ajhr beech” and consisting of two genera, lophozonia and fuscospora) forests, which are sometimes monodominant (wardle 1984). deciduous tree species are few, and tend to occupy niches characterised by cool winters and fertile soils (mcglone et al. 2004). ectomycorrhizal trees include only five genera, leptospermum and kunzea (myrtaceae), pomaderris (rhamnaceae), and fuscospora and lophozonia (nothofagaceae) (orlovich & cairney 2004). new zealand trees are, for the most part, slowgrowing (bee et al. 2007) especially in comparison with the introduced plantation trees (mainly ectomycorrhizal pinaceae – chiefly from europe and north america – but including ectomycorrhizal eucalyptus spp.) and a number of species (e.g., acer pseudoplatanus, ligustrum spp., salix spp.) some of which are plantation wildings (pinus spp., pseudotsuga menziesii) (table 1). why they mcglone et al. new zealand journal of forestry science (2022) 52:8 page 3 table 1: description of the study sites species mean growth rate (range), cm yr-1 references native conifers kauri (agathis australis) 0.23 (0.11–0.61) palmer & ogden (1983), mean across all natural populations cited in steward & beveridge (2010) rimu (dacrydium cupressinum) 0.25 (0.08–1.00) katz (1980), smale & kimberley (1986), stewart & white (1995) and references therein, mean across all natural populations cited in norton et al. (1988) tōtara (podocarpus totara) 0.51 (0.29–1.05) katz (1980), ebbett (1998), willems (1999), bergin & kimberley (2003) kahikatea (dacrycarpus dacrydioides) 0.62 (0.19–0.96) katz (1980), burns et al. (1999), duncan (1991), smale (1984) exotic conifers radiata pine (pinus radiata) 2.9 (2.8–3.0) richardson et al. (1999), watt et al. (2004) douglas-fir (pseudotsuga menziesii) 1.2 (1.1–1.5) lausberg et al. (1995) native angiosperms tawhai raunui/red beech (fuscospora fusca) 0.28 (0.08–1.49) june & ogden (1978), ogden (1978), wardle (1984), runkle et al. (1997), wiser et al. (2005), richardson et al. (2011) tawhai/silver beech (lophozonia menziesii) 0.23 (0.08–1.14) wardle (1980), wardle (1984), runkle et al. (1997), wiser et al. (2005), richardson et al. (2011) tawa (beilschmiedia tawa) 0.22 (0.04–0.55) ogden & west (1981), smale & kimberley (1986), smale et al. (2014) exotic angiosperms sugar maple (acer saccharum) 0.38 godman (1957) red maple (acer rubrum) 0.43 (0.26–0.57) walters & yawney (1990), zhang et al. (2015) white oak (quercus alba) 0.47 rogers (1990) red oak (quercus rubra) 0.65 (0.5–1.0) sander (1990) tulip tree (liriodendron tulipifera) 0.75 (0.5–1.0) mcdonald & urban (2004) table 1: representative mean annual diameter growth rates for new zealand native timber tree species and exotic tree species used in commercial plantations. are so slow growing is not clear as new zealand climates are mild and, while new zealand soils in general are not particularly fertile (hewitt et al. 2021), introduced weed and plantation tree species often grow at several times the rate of their indigenous competitors. there are apparent functional gaps as well (lee 1998). the new zealand tree flora lacks a fast-growing stress-tolerant pioneer equivalent to the northern hemisphere firs and pines. indigenous trees on droughty, frost-prone sites grow much more slowly and achieve lower heights than invasive northern hemisphere conifers. there are no close local equivalents to the clonal, rapidly spreading salix spp. that invade new zealand’s riparian zones and wetlands (dansereau 1964). fire-resistant or firepromoting canopy trees are notably absent and fireadapted seral trees, shrubs and ferns are few in species but now abundant in the fire-prone landscapes created by human settlement (e.g., leptospermum scoparium, kunzea ericoides, pittosporum tenuifolium, cordyline australis, discaria toumatou, pteridium esculentum; perry et al. 2014). regeneration slow growth rates (table 1) and the multi-layered structure of conifer–broadleaved forests mitigate against a rapid replacement of the overstorey dominants after disturbance. to simplify a complex and spatially highly variable situation, tall emergents, which are often conifers (agathis australis, dacrycarpus dacrydioides, dacrydium cupressinum, podocarpus spp., prumnopitys spp.) but sometimes angiosperms (metrosideros robusta, laurelia novae-zelandiae, knightia excelsa), form a usually discontinuous layer above a canopy of tall broad-leaved angiosperms, under which or in canopy gaps are smaller understory trees and tree ferns, and beneath that shrubs and small trees and a fern-rich ground layer. while disturbance often results in most of the components of this forest re-establishing within a relatively short period, shade cast by thickets of shrubs, small trees and tree ferns may completely exclude recruitment of canopy and emergent species for long periods, or severely reduce the growth rates of their seedlings and saplings (lusk et al. 2015). heavily stocked conifer forest tends to prevail either on deep, recent soils after disturbance, or on waterlogged, podzolised soils or drought-prone ridges where broadleaved trees cannot thrive. these dense conifer stands tend to have little or no conifer regeneration. in conifer–angiosperm stands, conifers tend to occupy more stressed (waterlogged or dry) sites than most angiosperm trees (coomes & bellingham 2011), and often occur as isolated individuals or small clusters (ogden & stewart 1995, mcglone et al. 2017). a key insight from ogden & stewart (1995) is that conifer– angiosperm forests in new zealand are essentially two-component forests in which one component, the long-lived conifers, occupies a superior, often emergent stratum, while the other component, angiosperms and tree ferns, although capable of outcompeting conifers in the juvenile and sapling stages, are eventually overtopped or open up through senescence, permitting a conifer mcglone et al. new zealand journal of forestry science (2022) 52:8 page 4 cohort to re-establish. it is typical for 50 years or more to pass before substantial regeneration of the previous dominants is apparent. kauri (agathis australis), treasured for its extraordinarily versatile timber, may take 25 years to attain 1.4 m in height, even on optimal sites, and reach full productivity only after 200 years (steward et al. 2014). on stressed sites such as ridges or frostand drought-prone valley floors, tall pioneers such as kānuka (kunzea ericoides) may dominate for 80 or more years, with recruitment of the previous dominants occurring only after significant dieback and opening up of the dense kānuka canopy (bellingham et al. 2010, richardson et al. 2014). long-lived tree fern glades have a similar effect (coomes et al. 2005). oscillations in dominance between conifers and angiosperms can differ markedly according to location and history. examples are the postulated 2000-year linear sequence from dense podocarps to scattered podocarps over a broadleaved canopy sparked by the devastation and bared landscape created by the massive taupō volcanic eruption c. ad 230 in the central north island (mckelvey 1963); the ‘catastrophe’ cycle of 700 to 1000 years caused by widespread fire or cyclonic damage and resulting in succession through a prolonged kānuka stand phase, followed by dense podocarps (cameron 1954); and the regeneration cycle lasting 500–700 years when large conifer trees or small groups fall and the small gap made follows a tree fern–broadleaved– podocarp succession (beveridge 1973). beech forests are quite different and have much simpler regeneration patterns, which have been well described and quantified (wardle 1984, ogden et al. 1996). these ectomycorrhizal trees often grow in species-poor, sometimes monodominant, stands, often with sparse understories. episodic seeding events provide large numbers of slow-growing but long-lived seedlings which act as a permanent store of advance regeneration. stand-clearing events (aside from fire) are therefore usually followed by rapid regeneration (stewart et al. 1991, coomes & allen 2007). however, a major limitation for the southern beeches is their poorly dispersed, small but heavy wind-blown seeds and requirement for ectomycorrhizal infection (baylis 1980, dickie et al. 2012, forsyth et al. 2015). beech forests therefore often have sharp boundaries with competing forest types or fail to spread into cleared adjacent ground. on more climatically stressed sites in mountainous areas, beech regenerate freely on bared ground, though taking some years to suppress a dense fern or herb ground cover should the latter colonise the site first. in the case of blow-down where the soil remains intact, advanced seedling banks of long-lived stunted individuals quickly respond to increased light and reduction of root competition to recapture the gap. under progressively more benign climatic regimes, the nearly monospecific stands that characterise harsher sites share canopy dominance with podocarps and a range of broadleaved trees, and the understories have a greater proportion of ferns and tree ferns. under these circumstances, beech regeneration is suppressed by broader-leaved angiosperm trees such as kāmahi (pterophylla racemosa, until recently weinmannia racemosa) and quintinia acutifolia and tree ferns (lusk & smith 1998). the course of indigenous forest exploitation in 1840, when new zealand became a british colony, much of the forest had been cleared through burning by māori, and the eastern half of the south island and around a third of the north island was largely in grassland, fernland or shrubland. exploitation of the timber in kauri forests had been in progress since the 1820s as well as clearance for agriculture (cameron 1961, roche 1990). even though māori rights to forests were guaranteed by the treaty of waitangi (1840), colonial settlement saw a rapid shift towards state ownership and private freehold which accelerated after the new zealand wars of the 1860s and subsequent confiscations and dubiously legal alienation of māori land. freehold or leased indigenous forest tracts during the peak periods of milling provided two-thirds of the timber output, although making up less than one-third of the forested area (roche 1990). forestry exports were important during the early colonial era, mainly highly-valued kauri timber, but rapid depletion of the most accessible coastal stands and expansion of wool and gold receipts led to them falling below 3% of total export value until 1880, after which they increased as improved transport enabled greater access, peaking in the 1890s. exhaustion of the most sought-after timbers through over-cutting and the global depression of the 1930s reduced forestry exports to a new low but they increased in the 1950s as exotic plantations came on stream (fig.1). however, compared with the overwhelming dominance of pastoral agricultural exports after the 1850s, forestry export earnings always took a distant second place. this, plus the administrative arrangement whereby forested crown lands were managed by the lands department whose chief concern was farm development, accounts for the scant regard for protecting forests. kauri was severely depleted by the early 1920s (orwin 2019), and kahikatea (dacrycarpus dacrydioides), favoured for construction of boxes for dairy products both in new zealand and australia and often growing on first-rate agricultural soils, was reduced to remnants by 1950. rimu (dacrydium cupressinum) became the mainstay of the local timber industry but had no particular merits to attract much international interest. from the earliest days of the colony, new zealand imported timber from north america, australia and the baltic states, and later in the 20th century from numerous other sources, including hardwoods from tropical countries. in part, this was for specialist uses not well provided for by indigenous or exotic plantations in new zealand but also because internationally sourced timber was usually of higher quality and often cheaper than locally sourced offerings. this too tended to devalue indigenous forests. mass exotic plantings on agriculturally marginal land during the great depression of the late 1920s and early 1930s were designed to head off projected timber shortages resulting from rapid depletion of the indigenous forest. by the 1950s these plantations had begun to mature at the same time as a resurgent post-world war ii economy increased the demand for construction timber (fig. 2). in the course of 10 years, locally grown exotic timber rose from less than 10% of new zealand production to be approximately equivalent with indigenous timber and the long-predicted ‘timber famine’ was averted. many of the early pine plantings had been poorly sited and the forests badly managed, mcglone et al. new zealand journal of forestry science (2022) 52:8 page 5 figure 1: contribution by forestry to new zealand’s merchandise export revenue, 1853 to 2015. grey boxes and text show two periods during which exports were typically greater than 5% of new zealand’s export revenue, the earlier period being indigenous exports, and the latter being exotic pine exports. data from briggs (2003) and new zealand forest owner’s association 2003-2018 (https:// www.nzfoa.org.nz/resources/publications/ facts-and-figures; accessed march 2020). figure 2: volumes of rough sawn timber extracted from indigenous forests in new zealand, 19282000. three historical events are marked: gd = great depression (1929-1933); ww2 = world war 2 (1939-1945); new policy for indigenous state forests passed in 1975. data from roche (1990), devoe & olson (2001), griffiths (2002), and griffiths (2016). https://www.nzfoa.org.nz/resources/publications/facts-and-figures https://www.nzfoa.org.nz/resources/publications/facts-and-figures https://www.nzfoa.org.nz/resources/publications/facts-and-figures and what to do with an inferior product became a pressing issue. with the government-sponsored ‘kawerau scheme’ and the private kinleith venture, large pulp and paper mills began to absorb the surplus from the early 1950s on, and commodity log exports surged as radiata pine (pinus radiata) quality improved because of major investment in its genetics, silviculture, and post-harvest treatment (fig. 3). meanwhile, production of indigenous timber (now largely rimu) began a steep decline accelerating from 1975, and by the turn of the 21st century, it was no longer a significant contributor to the new zealand economy (fig. 3). factors behind this dramatic shift are discussed in the following section. acts, policies, controversies, and indigenous forests ‘to lock up 40,000 acres, however, as a plant museum or “tree cemetery” would be regarded by the service and a large body of its supporters as fantastically wasteful of land and natural resources.’ commentary on the waipoua forest reserve proposal. (nzfsar 1948: p22). in the following sections, we outline the significant acts, policies, controversies and events that determined the use of indigenous forests in new zealand. key dates and events are summarised in table 2. pre-1920 the forest flora was well documented by the late 19th century (kirk 1889) and although little systematic quantification of forest resources had been carried out, the impact of unrestrained logging and clearance of lowland forests was clear. the government geologist, james hector presented estimates for 1873 showing a 40% decline since 1830 in forest cover of which an astonishing 15% had occurred in the previous 5 years (appendix to the journals of the house of representatives, 1874. h-5 papers relating to state forests, their conservation, planting, management, etc. part 1: pp 35-36). dismay at the wasteful clearance (less than 10% of the timber was harvested; nzfsar 1959: p 17) and inefficient sawmill practices, concern regarding the enormous conflagrations that were an inevitable consequence of contemporary logging and land clearance practices (arnold 1994), and widespread apprehension that unconstrained and unchecked forest clearance would lead to flooding and local climate change (beattie 2003) resulted in the passing of the forests act 1874. a chief forest conservator, captain inches campbell-walker, who had considerable forestry experience in india and familiarity with german forestry, was appointed in 1875 and he presented a detailed parliamentary report on the state of the forests (campbell-walker 1877). a strong advocate for scientific state forestry, he was supported by local foresters with continental european experience (campbell-walker 1876; lecoy 1879). his vision, kept alive by government agencies until the beginning of the 21st century, was for reservation under state control of large areas of indigenous forest that could be managed as a sustainable source of timber. the forests act and the formation of the state forestry branch of government was vigorously opposed in parliament and by sawmillers and rural landholders (roche 1990). the prevailing opinion was that all suitable forested land should be converted to pasture or crops. the royal commission on forestry (1913) reflected this consensus, stating: “…as a broad principle that no forest land, except it be required for the special purposes of a climatic or a scenic reserve and which is suitable for farm land, should be permitted to remain under forest if it can be occupied and resided upon in reasonably limited areas. should the area under consideration contain milling-timber the question will arise whether it be more profitable to mill before settlement or to fell, burn, and mcglone et al. new zealand journal of forestry science (2022) 52:8 page 6 figure 3: timber harvesting totals from new zealand, 1951-2018: (a) removals of logs versus pulp and chips from old growth indigenous forests; (b) roundwood removals from old growth indigenous forests and planted exotic forests. data from nz ministry for primary industries (https://www.mpi.govt.nz/forestry/newzealand-forests-forest-industry/forestry/ wood-processing/; accessed march 2020). https://www.mpi.govt.nz/forestry/new-zealand-forests-forest-industry/forestry/wood-processing/ https://www.mpi.govt.nz/forestry/new-zealand-forests-forest-industry/forestry/wood-processing/ https://www.mpi.govt.nz/forestry/new-zealand-forests-forest-industry/forestry/wood-processing/ mcglone et al. new zealand journal of forestry science (2022) 52:8 page 7 act, policy, event date provisions or consequences treaty of waitangi/tiriti o waitangi 1840 new zealand colony established; accelerated emigration of british settlers; beginning of alienation of māori forests forests act 1873 regulated sales of native timber; provision made for state forest; minister of the crown as commissioner of state forests state forest act 1885 chief conservator of the state forests (branch of crown lands department) established. controls imposed on logging in high elevation forests royal commission on the timber and timberbuilding industries 1909 reported that indigenous forests would be depleted within 40 years royal commission on forestry 1913 recommended which forests to be retained for soil protection, water conservation, and scenic reserves and which for agriculture or logging. recommended state forest service director of forests and forests act 1919-1922 implementation of royal commission: appointment step towards independent service, formation of state forest service, and acquisition of forests forests act 1949 the forest service to direct forest policy, regulations and commercial operations. clarified extent and purpose of state forests but not preservation of indigenous forests. south island beech scheme 1971 new zealand forest service plans to log for timber and pulp c. 340,000 ha of indigenous forest in the south island. only partly implemented; sparked a severe conservation backlash. new forest service native forest policy 1975-1977 indigenous clear-felling abandoned in favour of selection logging and reservation of large area maruia declaration 1977 influential parliamentary petition calling for the end to indigenous logging west coast accord 1986 agreement between the west coast united council, environmental, industry groups, conservation groups, and local communities on south island west coast forests strategy. some indigenous sawmilling permitted. disestablishment of new zealand forest service 1987 conservation forests assigned to the newly created department of conservation. timberlands west coast ltd 1990 a state-owned enterprise created to manage production forestry, including sustainably-managing indigenous west coast forests. resource management act 1991 a fundamental reshaping of environmental law to encourage sustainable management of natural and physical resources. amendment to forests act 1949 1993 mandated cessation of unsustainable indigenous logging. certain māori owned forests excluded. end of logging west coast, south island 2002 government’s directive to cease all indigenous logging on crown owned land forests amendment act 2004 further amendment of the 1949 forests act. prohibits felling of indigenous timber on state land and export of indigenous forest produce. requires sustainable forest management plans for felling indigenous forest on private land. no second cut permitted until volume of timber is equivalent to that at commencement of first permit. west coast wind-blown timber (conservation lands) act 2014 2014 allows the removal of timber from west coast south island conservation forests damaged by cyclone ita. table 2: acts and policies influencing the use or conservation of indigenous forests in new zealand. grass. obviously the answer is purely one of finance, and each case must be dealt with on its merits, the main factors being the enhanced value of the timber if reserved for a stated period, its distance from the centre of demand, and the expense of the milling…” and regarding beech forests: “…these forests are the only ones amongst those indigenous to new zealand which may regenerate rapidly enough to warrant their permanent retention”. the commission gave detailed financial justifications for why planting of podocarps would be “…an utter absurdity.” dr leonard cockayne, a vocal advocate for forest reserves (cockayne 1927), was a member of the committee; of the other four, two were farmers, one a builder and one a woodware manufacturer (goulding 2013). supporting this view of forest as an obstacle to agriculture was the assumption that indigenous forest management was a pointless exercise. the rapid retreat of indigenous forest and the spectacular growth of exotic conifers and woody weeds suggested to many that the timber trees of new zealand were inferior to those of the northern hemisphere (campbell-walker 1876). observations of dense forests with old stagheaded podocarp trees lacking pole or seedling cohorts versus vigorous but economically valueless angiosperm regeneration in cutover tracts of ‘wasteland’ confirmed this opinion. campbell-walker was vilified in the press and his appointment was not renewed. some progress was made under the conservators: reports on indigenous forests were made, 800,000 ha set aside as forest reserves and regulations introduced to reduce misuse. however, the long depression of the 1880s, popular opposition and changes in government led to the position of chief conservator of forests being disestablished in 1887. in any case, provisions of the forests act had been poorly funded and not enforced. logging and clearance proceeded as before. formation of the state forest service and 1945 policy among other far-reaching recommendations (roche 2013), the 1913 royal commission suggested establishment of a state forest service and, after a delay because of the world war, a canadian forester, l. macintosh ellis, was appointed director of forestry in 1919 and head of the newly formed state forest service. he was well suited for the job as he had academic training and practical forestry experience in canada and had just completed military service in france as a forester. within a year of taking up his position, he provided a detailed report for parliament (ellis 1920) setting out the rapidly depleting state of the indigenous forests of new zealand, pointing out the failures of the government to regulate logging and presenting a comprehensive vision of a sustainable, multiple-use future for the indigenous forests. he proposed a well-funded forest service that would have custody of most of the forested land, both for production and preservation, control logging, manage forests, and carry out scientific investigations into all aspects of silviculture. much of his ambitious vision was to come to pass. the state forest service (renamed the new zealand forest service in 1949) managed most of the forested land; the logging industry was controlled to some extent; and scientific investigations proceeded. just as importantly, some of it did not. by as early as 1925 the planting of what were to become immense plantations of exotic conifers had begun and this eventually reduced the need to conserve indigenous forests. the forest service was not given control of all state forests; scenic and scientific reserves and national parks were placed under the residual department of lands (later department of lands & survey); and the department of internal affairs given responsibility for indigenous avifauna. this narrowly focussed the forest service on timber production and maintenance of landscape protection forests. the forest service’s plan was that forests on poorer soils or in suboptimal climates would be reserved for sustainable timber yield or protection forestry while on better soils and under milder climates, clear-felling and conversion to agriculture or plantations would be the norm. although this plan was widely supported, conservationist assertiveness grew over time. before 1900, conservation concerns centred on wasteful exploitation of forests and the fear that adverse climate change, flooding and soil erosion would follow their removal (beattie 2003). by the 1920s, conservationists’ concerns had shifted to an emphasis on wildlife, scenery and amenity value (star, 2002). early logging excesses were for the most part supported or largely unnoticed by the public and tolerated by the forest service. however, public disquiet increased, and the lack of forest service investment in research and management of indigenous forest relative to exotic plantation noted. the forest service response was that this focus on exotics was the best approach: “… it has only been interim concentration on exotic forestry which has enabled the forest service, since its inception 30 years ago, to save already 150,000 acres of the best indigenous forest from milling.” (nzfsar 1951: p 5). 1945–1975: state control and continuing exploitation during world war ii, the new zealand government developed a highly centralised control over the economy that was maintained for four decades (easton 2020). successive governments expanded the public service and were willing to sponsor or undertake commercial activities in order to improve the nation’s productivity. in 1949, sweeping powers were given the renamed and expanding new zealand forest service. the forest service was a major player in the supply of both indigenous and exotic timber, virtually the only provider of forestry research, the enforcer of forestry regulations and the prime source of policy advice. it was thus deeply engaged with politicians, the public, rural interests and private forestry companies. after 1945, a national priority was to address the shortfall in housing and other construction, a legacy of the 1930s depression and world war ii, and the demand for timber surged (roche 1990). therefore, even though significant areas of indigenous forest were reserved and all high-elevation forests strictly protected, it was mcglone et al. new zealand journal of forestry science (2022) 52:8 page 8 in the context of accelerated logging of the rest of the indigenous forests. the burgeoning supply of exotic conifer timber after 1950 did not reduce the demand for indigenous timber as radiata pine was regarded as unsuitable for many uses and builders preferred the indigenous timbers they were familiar with. with the decline of the kauri industry in the north – although some exploitation continued until the 1980s and the reservation of the largest remaining kauri forest (waipoua) in 1952 (a severe blow to the forest service’s ambitions for sustainable harvest), three large areas of relatively unexploited indigenous forest timber remained: the forests of the central north island (largely podocarps, in particular rimu); the west coast of the south island (podocarps and beech); and the far south of the south island (mostly beech). post-war governments encouraged indigenous logging in these areas and low stumpage fees, guaranteed sawmiller profit margins, and price control of sawn timber was maintained until 1979 (bassett 1987). indigenous timber was therefore cheap and was used mostly for basic construction rather than for the more specialised uses for which its higher grades were best suited. the government continued to side with the sawmillers, going to the extent of endorsing a committee report in 1953 which stated that “…the preservation of indigenous timber supplies is of no consequence..” over the strenuous objections of the forest service which was attempting to slow the cutting rate to a sustainable level (halkett 1991). government policy remained in essence the ‘best land use’ policy promoted by the 1913 royal commission, which encouraged conversion of indigenous forest to exotic plantation or farmland should this result in higher productivity. cutover indigenous forest that was not converted was left in a derelict state. the forest service promoted pine, and called for restraint in indigenous logging, raising the spectre of disease or climatic events affecting exotic plantations as a justification for continuing to reserve indigenous forests for future logging (conway 1977). as regeneration of merchantable timber in cutover tracts was sporadic and sometimes non-existent on any reasonable timescale, a continuing supply of indigenous timber could not be maintained in a given region without strict controls. however, strongly pro-development governments and intense pressure from local sawmillers meant that the forest service had difficulty in ensuring adherence to their forest management plans. the over-cutting that went on was indistinguishable from clear-felling in many cases, and damage from logging machinery to the forest environment considerable. especially in north island forests, logging often targeted scattered emergent podocarps in a matrix of angiosperm trees, notably tawa (beilschmiedia tawa), resulting in logged forests that were substantially altered in structure and composition. local sawmillers became adept at circumventing controls. the 1950 forest service annual report discussed the difficulties in preventing them from circumventing plans for the sustained and equitable allocation of cutting rights. among the ‘innumerable’ tactics the forest service faced was sawmillers erecting mills alongside unallocated state forest and then requesting access on compassionate grounds. the forest service had eventually to admit its inability to control the cut “whether on state forest, māori forest or that owned by any private interest” and that it would have to rely on exhaustion of the supplies available to the mills (nzfsar 1960: p 23). after world war ii, national opinion with regard to indigenous forests shifted steadily from a focus on economic benefits and employment to more biocentric issues. in the 1960s there was increasing emphasis on scenery, birdlife and larger reserves (salmon 1960) and from the 1970s a strongly preservationist agitation began for retaining natural old-growth forest. protest intensified with continued encroachment of logging into old-growth podocarp forest (tihoi, pureora and whirinaki) in the central north island and kauri forest in northland (warawara) and fuelled the rise of conservation campaigns in the late 1960s and early 1970s (young 2004, orwin 2019). as national-level support for indigenous logging weakened, silvicultural techniques – as practised, not as promoted – came under scrutiny and were found wanting. conservation organisations rightly pointed out that indigenous logging operations were still completely focussed on maximum extraction of merchantable timber, made only token efforts towards ensuring regeneration, and left the forests in an unnatural state. meanwhile, the rise of an international market for wood chips made previously neglected indigenous angiosperm species economically viable, in particular north island tawa and south island beech. logging indigenous tawa forests for chipping and pulp began in the north island in 1970 and a large-scale south island scheme based on beech was proposed in 1971. by 1989, 1000 ha of indigenous forest was being cleared for chipping every year (wilson 1994). these wood chip and pulp initiatives inflamed opinion because they were less selective and used much more of the forest biomass. conservationists by now were well organised, focused and adept at using the media (young 2004, bensemann 2018). we can surmise that there was also a moral dimension in play as the chips were almost entirely for export and for paper, demolishing the rhetoric of wise use of indigenous forest for specialist local needs. 1975–present: reduction in indigenous forestry economic arguments for continuing exploitation of indigenous forests lost credibility as the availability of exotic timber grew, the indigenous forest cut declined and cheap imports of hardwood timber and products with similar or superior qualities became readily available. as well, indigenous forestry made only a vestigial contribution to the export balance sheet and thus its national importance waned at a time when conservationist protest was intensifying. the official forest service policy was amended in 1975 and a candid admission made that, despite years of promoting sustainable indigenous forestry “…the practice over the last quarter century throughout much of the country has been to extract all the merchantable trees mcglone et al. new zealand journal of forestry science (2022) 52:8 page 9 mcglone et al. new zealand journal of forestry science (2022) 52:8 page 10 when logging an area. where they comprised the bulk of the crop this resulted in what is termed clear felling. in mixed forests of hardwoods and softwoods it was the latter which by and large were removed, and while a forest cover of sorts remained it bore little or no resemblance to the original stand” (nzfsar 1976). the new policy had as its key elements that: (1) clear-felling should only proceed if a “clear need” is evident; (2) indigenous forest should be cleared only when a study of the social, environmental and economic factors show national welfare would be enhanced; (3) a decision on whether timber production is of greater importance than other conflicting values should be deferred until a commitment one way or the other is necessary; (4) logging of indigenous forests is not precluded but should be carried out in a way that leaves “open the options of maintaining an indigenous forest structure with a wide range of values or clearing for other uses at some unspecified future time” (conway 1977). the new policy slowed exploitation of indigenous state forests. cutting in privately-owned forests (which provided ca. 66% of the total) fell at the same rate, even though not directly affected by the policy, probably because the diminishing state forest supply disrupted the entire indigenous timber infrastructure (fig. 2). nevertheless, the forest service promotion of multiple use of state forests became increasingly controversial because of the suspicion that it was a cover for a continuation of forest mining. oversight of logging was poor and there were apparently no good-faith intentions or resources to manage regeneration. the 1969 forestry development conference recommended a second major expansion of the exotic forestry estate and cutover native state forest was seen as an obvious land bank for this. moreover, indigenous conversions continued: from 1970 to the mid-1980s large areas of secondary forest (often rich in regenerating conifers) were destroyed to make way for radiata pine plantations, and even old-growth forests (on the mamakū plateau, in northland, in north westland) were clear-felled and burnt to make way for plantations (fleming 1969). while the forest service had made progress towards its ideal forest model in that some attempt at best-practice logging techniques was made and scientific reserves set aside, it was now lagging well behind changing public opinion. leading conservation movements called for a ban on logging in public forests, sustainable management of indigenous forests in private hands, and a ban on indigenous exports (gillman 2008). a bitter, complex debate ensured (thompson 1987, roche 1990, king et al. 2015). in 1984, in the aftermath of a national financial crisis, a neoliberal government with a focus on efficiency and economic return was elected (tilling 1992). the new labour government was less concerned than its predecessors with rural issues and richard prebble, a key minister in the administration, pointed out that in 70 years of existence the forest service had failed to make a profit from its forestry operations (prebble 1996). the government was also ideologically antagonistic to the sprawling forest service with its multiple, conflicting agendas, opaque decision-making, entanglements with private forestry companies, and propensity to antagonise conservationists. disestablishment of the forest service came in 1987. protection and conservation forests were allocated to the newly formed department of conservation and production forestry privatised or incorporated into state-owned enterprises. the end to state-sponsored indigenous forestry came with the demise of timberlands, a state-owned company set up to manage crown-owned forest tracts in westland for timber. with financial support from central government, timberlands invested in the development and trialling of small-scale (<0.2 ha) coupe and group harvesting methods in beech forests (wiser et al. 2005, wiser et al. 2007), and single-tree harvesting methods for rimu (richards 1994, james & norton 2002). these methods were designed to minimise canopy disturbance and thus maintain the character of natural forests. after several decades of monitoring across replicated sites, there was evidence for adequate regeneration by beech in small coupes (wiser et al. 2005, wiser et al. 2007, allen et al. 2012). comparable studies are not available from rimu forests, but light-demanding rimu seedlings (norton et al. 1998) are unlikely to regenerate after single-tree harvesting because of rapid canopy closure by adjacent trees (james & norton 2002, allen et al. 2013). in spite of promising results from beech forests, there were analyses (mason 2000) that suggested the beech timber extraction necessary to keep the timberlands enterprise afloat could not be balanced by the speed with which natural regeneration replaced harvested canopy trees. when it was revealed that the company had been undertaking a secret lobbying campaign to bolster its position, including denigration of the prime minister (hager & burton 1999), its fate was sealed. a new policy was announced in the amendment to the forest act 1949 in 1993 where part 3a (section 7) promotes the sustainable management of indigenous forest land defined as: “…the management of an area of indigenous forest land in a way that maintains the ability of the forest growing on that land to continue to provide a full range of products and amenities in perpetuity while retaining the forest’s natural values.” since the forests amendment act 2004, indigenous forest milling has taken place only on privately-owned land and is subject to a requirement for sustainable forest management. nevertheless, calls for a new, holistic forest policy in which the distinction between indigenous and plantation forestry is dissolved, and for sustainable management of the department of conservation forests for timber, are made from time to time (levack 2006). māori-owned forest lands māori were the original owners of new zealand forests and their rights to forest resources were guaranteed in the treaty of waitangi, but subsequent confiscation, seizure under the public works act and other legal and illegal stratagems greatly reduced their holdings. māori were mostly willing participants in the exploitation and conversion of indigenous forests, undertaking major clearances for agriculture on their own behalf in the mcglone et al. new zealand journal of forestry science (2022) 52:8 page 11 early 19th century (cameron 1961), and providers of a considerable amount of forest resource and forestry labour in the 20th century. peak logging on māori land came after 1950 when increased prices spurred a dramatic rise in logging from less than 71,000 m3 in 1947 to about 190,000 m3 in 1958 (nzfsar 1959: p29). māori continue to be a major supplier of indigenous timber in the sustainable logging industry. currently around 600,000 ha of indigenous forest are owned or under control of māori (150,000 ha covenanted), that is around 29% of all privately-owned indigenous forest (wilson & memon 2005, holt & bennett 2014). the policy history we have presented, which resulted in the ultimate victory of those who viewed continuing exploitation for timber on publicly-owned forest land as impermissible and logging on private land acceptable only on sustainable grounds, is largely a settler-pākehā narrative. it has been argued that the prevailing biocentric view of forests derives from european scientific notions of how indigenous forests should be managed and that for māori “…cultural, economic and social factors combine to create forest management pathways that are complex and that may not necessarily be compatible with western (i.e. pakeha) notions of ‘ideal’ biodiversity preservation” (wilson & memon 2010). geoff park (2000) went further in suggesting that for māori ‘preservation was also subordination’. research into sustainable indigenous forests ‘the pure botanist has little if any conception of silviculture, but relies on an academic knowledge of the natural growth of individual plants and plant societies. he has no appreciation of the concept of forestry as an art and as an applied science, and yet has purported to advise the public on the future of waipoua’ (sfsar 1948: p23). until late in the 20th century, advanced training of new zealand’s foresters and ecologists largely took place in the united states, britain, and australia (mckelvey 1999) and the international literature was dominated by management issues specific to northern hemisphere conifer and hardwood forests. as well, research carried out by ecologists and botanists was particularly important during the 1920s, in part due to the lack of forestry professionals (sveding 2019). each generation of new zealand forest scientists therefore had to address the issue of the extent to which the ecological concepts and forestry principles taught them applied locally. there had been lively scientific debates regarding the slow growth rates of indigenous timber trees beginning in the 1830s (roche 1997). the pioneer ecologist leonard cockayne noted the widespread lack of regeneration of the conifer overstorey in many undisturbed forests and suggested that the podocarp forests of northern regions were “..turning by degrees into a climax with the tawa (beilschmiedia tawa) dominant, and that the podocarps of south island forest, to the south of lat. 42°, would be eventually replaced by weinmannia racemosa.” (cockayne 1928: p. 153). he believed the valuable timber trees of the dense podocarp and kauri stands to be a temporary phase which was destined to give way to a ‘climax’ forest dominated by angiosperms, supporting his view that the evolutionarily more advanced angiosperms would eventually replace the conifers. this bolstered the widespread popular belief that all indigenous forest was likely to succumb to invading exotics due to some undefined shortfall in their nature, and that valuable conifers were destined to be replaced by angiosperm trees of little value as timber. on the other hand, cockayne saw a great future for management of the extensive beech forests (cockayne 1926, 1928). systematic scientific investigations into the extent, management and timber potential of indigenous forests began in the 1920s but little progress was made: “it is difficult to estimate to what extent the botanist’s preoccupation with naming and classification, and the forester’s interest in the economic possibilities of exotics has been responsible for the lack of experimenting. the fact remains that planned experimentation as the basis of scientific study has been notoriously absent. in the whole of the country’s indigenous forest estate there are probably not five sets of even small plots which have been studied and tended continually over a period of the 40 consecutive years of departmental existence, let alone any longer. one is forced to speculate whether the woeful lack of scientific observation is not due to some basic defect in the teaching of research in new zealand.” (nzfsar 1960: p. 19). new zealand was poorly provided with forestry professionals until the late 1940s. the newly formed state forest service was supposed to carry out most of the forest science necessary to implement ellis’s original vision. however, few forestry professionals were appointed. their number was never greater than 8 (including scientists, foresters, and engineers) until after 1945, falling as low as 3 in the 1930s (nzfsar 1950). the two forestry schools (auckland and christchurch) founded in the 1920s were poorly funded, undermined by internal dissent, and defunct by 1934 (mckelvey 1999). the resources and staff later invested in sustainable forestry research by the forest service were always minimal, never amounting to more than a few scientists and technicians. given the poor understanding of the extent of merchantable timber in the forest estate and growing public pressure for reserves, a national forest inventory was carried out (1921–1923). by the 1940s, the limitations of the national forest inventory were clear. between 1946 and 1952, the much more ambitious national forest survey was undertaken, based on fieldwork, systematically placed ground-based plots, aerial photograph interpretation, and statistical methods (masters et al. 1957). it provided estimates of the amount of merchantable indigenous timber in non-protection forests and documented regeneration and damage by introduced herbivores, in particular deer and goats. this project had profound effects on the direction of forestry science in new zealand, and not least by provision of a cadre of scientists with extensive field experience in indigenous forests. the lack of conifer regeneration in mature stands it revealed was interpreted as evidence mcglone et al. new zealand journal of forestry science (2022) 52:8 page 12 that the conifers were ill suited to contemporary climates (holloway 1954), and this was echoed in later papers (nicholls 1956, wardle 1963, mckelvey 1963), a view not supported by later research (veblen & stewart 1982, wardle 1985). however, even though contested at the time (cumberland 1962), the climatic hypothesis provided further support to popular and industry views that logging indigenous conifers was simply anticipating their inevitable demise. the forest service annual report of 1949 made this clear: “….much of our forest is in an unstable condition. present distribution and composition appear to reflect a warmer and more humid period in the past…the bulk of the podocarp forest displays symptoms of over-maturity and stagnation with the virtual absence of young growth” (p. 9). after world war ii, the central issue that forestry scientists faced was how to promote the regeneration of valuable, slow-growing conifer timber trees. the problem was exacerbated by the widespread belief among foresters that many indigenous forests left to their own devices would lack conifer regeneration and be replaced by commercially valueless hardwood species, so that they doubted the forests were even worth managing. although some early investigations had given hope of a reasonable yield from well-stocked rimu stands on the west coast of the south island (hutchinson 1928, hutchinson 1931) or kauri in northland (sando 1936), these claims proved to be illusory. even under ideal conditions and optimistic calculations the main timber tree, rimu, would take 85 years to recover to a merchantable state after clear-felling of a stand (hutchinson 1931). a frank assessment of the situation was given by a forest service forester david kennedy: “the new zealand forester has long been the target of criticism for his neglect of indigenous silviculture. he is continually being asked why he does not replant rimu after his stands have been harvested by the sawmiller. it is true that small areas could conceivably be replanted with trees raised in nurseries from the minute, irregular, and predominantly infertile seed crops that are a characteristic of rimu, the main species. likewise, it would be possible, by oft-repeated and long-continued release cuttings to ensure that such plantings were allowed to develop without undue competition from the ubiquitous second growth that follows logging. the answer is, of course, that such forms of silviculture would be hopelessly uneconomic, and could not possibly be justified as practical forestry.” (kennedy 1951). technological advances in forestry practices in the second half of the 20th century provided the impetus to have another look at the situation. increasing mechanization and replacement of bush tramways by a more intensive road network reduced the costs of timber extraction, making feasible more sophisticated alternatives to clear-felling or removal of conifers from mixed forests. renewed scientific investigations and experimental trials to support the development of sustainable indigenous forest began in the 1950s and have continued to the present. key areas for these investigations were the large remaining stands of lowland dense podocarp forest on the west coast of the south island and on the volcanic plateau of the central north island (chavasse & travers 1966, james & franklin 1977, six-dijkstra et al. 1985, smale et al. 1998, beveridge et al. 2000, carswell et al. 2007, beveridge et al. 2009). there were also investigations into the sustainable management of young kauri stands that had developed after clear-felling of old-growth stands in the 19th century (halkett 1983, barton & madgwick 1987, steward & beveridge 2010). the emphasis in the earlier years was on the optimal logging pattern to ensure regeneration. strip-felling, in which 80 m wide bands of untouched forest were left, was promoted in westland terrace rimu forest (chavasse 1954) but proved to be less successful than anticipated in achieving regeneration and preserving old growth. conventional methods of extraction by rope hauling and tramways, replaced later by crawler tractors, resulted in unacceptable damage. many forests were on soils prone to compaction, waterlogging and windthrow resulting in premature deaths and lack of regeneration of the remaining trees. in podocarp forest, regeneration was sparse in the cleared strips and wind damage excessive in the untouched strips, and the technique (first trialled in 1956) was abandoned in 1965 in favour of selection logging and small coupe approaches (halkett 1991). under selection logging, large areas of forest were thinned out by removal of trees in a range of age classes. in the 1980s, more environmentally friendly technology in the form of low ground pressure bulldozers, portable bush sawmills and helicopter log extraction increased the range of possibilities for selective logging by taking single or small groups of stems (james & norton 2002). with regard to podocarp–angiosperm forests, given the great range of initial conditions of the stands, the percentage of stems removed, damage to the forest and soils during logging operations, and subsequent deaths of remaining timber trees, conclusions varied to how sustainable such initiatives were. estimates for cutting cycles of selectively logged forest for the most valuable large diameter stems were typically long: for instance, for south island west coast rimu forests, a suggested cycle was three cuts over 225 years but even on the best sites this might have been optimistic (six-dijkstra et al. 1985). in some cases introduced deer browsing caused a complete hiatus of podocarp regeneration on otherwise suitable sites (richardson et al. 2014). analysis of a north island podocarp–tawa forest block, selectively harvested in 1961 for one third of its merchantable conifer and hardwood volume with preferential removal of unthrifty trees, revealed that after 43 years the size class structures of commercial species were maintained after harvesting, regeneration of conifers and tawa was proceeding, and stability of the forest was not affected but that nearly a century would be needed for the forest to recover 80% of its previous basal area (smale & beveridge 2007). it was noted that the least marketable timber tree, tawa, was the most suited to selective logging management and that in similar forests elsewhere, conifer regeneration after logging had been poor. the overall conclusion was that the selection system, while often regarded as that most likely to mimic natural processes, produces outcomes that resemble few natural forests, and that careful intervention is needed to maintain it. beech stands generally have more prolific and assured regeneration than podocarps and studies of their management yielded more optimistic conclusions (wardle 1984), but there were provisos. many harvestable trees had defects which precluded their use as sawlogs which meant that for many stands, selection logging was uneconomic, relative to imported timbers, and was viable only if the timber was used for pulp (johnston 1972). woody debris in beech stands is important as a carbon store, moderating energy flow, in the nutrient cycle and as a substrate for invertebrate and fungal biodiversity (allen et al. 2000). however, it also provides brood material for wood-boring insects that during outbreaks can sometimes damage even healthy trees sufficiently to lead to deterioration or death through fungal pathogens (ogden et al. 1996). stand hygiene to reduce outbreaks is necessary, but fundamentally alters the natural functioning of the forest. although the results of some of these studies could be seen as supportive of selective logging, this support extended only as far as the issue of effective replacement of the focus timber tree. two elements were generally missing: the broader forest biodiversity, and the economic or social implications. the philosophy behind sustainable forestry until the 1980s was essentially the ‘better-than-natural’ ideal (o’hara 2002) in which indigenous forests were to be managed so as to be evenaged over economic cutting compartments and producing maximum volume increments of merchantable timber. descriptors for forests in this condition were ‘healthy’ or ‘vigorous’. these were contrasted with uneven-aged, old-growth stands which were described as ‘overmature’, ‘senescent’, ‘stagnant’, ‘moribund’, ‘unthrifty’, or ‘decadent’. that such management (even if realistic in new zealand) would result in forests resembling plantations was mostly unremarked. however, by the 1990s, it was clear that intrinsic biodiversity values also had to be catered for, including all animal and plant life, invertebrates, and fungi. inclusion of these elements – a large component of which are favoured by undisturbed soils, old, large trees, dense undergrowth, dead standing and fallen wood – completely changed the argument. for maximum biodiversity, a diverse age structure is required and this includes old-growth stands previously characterised as ‘stagnant’ or ‘moribund’. from a forester’s point of view, the goalposts had now shifted. the ideal forest was no longer young, vigorous and productive but ancient and steady-state, and they no longer had either the ideological mindset or the research basis to deal with the new dispensation. while high-quality and sophisticated biophysical work has been carried out since the 1990s on how technically to address sustainable forestry (e.g., wiser et al. 2005), economic or social analyses are almost always lacking. even comprehensive accounts rarely make any but a passing reference to the costs relative to outcomes (reay & norton 1999, james & norton 2002, smale & beveridge 2007, forbes et al. 2021). economic studies tend to show that the net annualised return for biophysically sustainable indigenous forestry generally lies below zero (evison et al. 2012), a finding that echoes the views of some forest owners (hawes & memon 1998). could sustainable indigenous forestry have been implemented? “the ultimate objective is to convert the stagnant and over-mature indigenous stands so characteristic of the country’s forests today into vigorous growing and highly productive stands. only if this is achieved will any indigenous forests remain for the enjoyment of prosperity” nzfsar 1951: p. 7. the official forestry policy in new zealand from 1921 until 2004 encouraged management of indigenous forests in perpetuity. as discussed above, a number of issues prevented this from becoming a reality: • the often irregular seeding and inherently slow growth of the conifer species with the most desirable wood properties, and thus the extremely long conifer regeneration cycle; • the poor soils, terrain and climates that characterise most remaining forests; • post-logging issues including soil compaction and waterlogging, weed invasion and pest herbivores; • the lack of a market for timber from associated angiosperm trees and the weak market for beech. confronted with this unpromising set of issues – most of which were well understood by 1900 – research focused on silvicultural manipulation during harvesting and observations of the subsequent regrowth. the problems with the recalcitrant biology and the long regeneration cycle of conifers cannot be solved solely by manipulating tree extraction procedures; subsequent silvicultural investment is needed if shorter harvest cycles are required. while the costs of carrying out different harvesting strategies are essentially paid for by the timber extracted, this is not the case for subsequent silviculture. the relative lack of investigation of the effects of further silvicultural interventions (such as planting, release of seedlings, pruning, weed and herbivore control) can be attributed to their high cost and the reluctance of forest managers to invest in them. moreover, given the very long cycle involved in even the fastest-growing indigenous timber trees, the costs that can be recovered through improved value are minimal. calculations for westland rimu in 1951 suggested that the timber produced from a planted and silviculturally tended stand would need to be priced at £127 (= $8000 current) per cubic metre merely to cover the investment, more than 150 times the going rate at the time – an economic outcome referred to as “…startlingly poor” (nzfsar 1951: p. 6). while efforts at creating indigenous plantations were regularly highlighted in successive mcglone et al. new zealand journal of forestry science (2022) 52:8 page 13 forest service annual reports, they are estimated to cover only a few thousand hectares at most (forbes et al. 2021). recent analyses of the potential for indigenous plantation forestry for rimu in a north island location showed that on flat terrain it was not profitable, and on steep slopes only profitable with a low discount rate, high stumpage price and with other benefits such as avoided erosion, carbon sequestration, and biodiversity and cultural enhancement factored in (pizzirani et al. 2019). the high costs of establishing indigenous conifers in cutover native forest or exotic plantations regions is confirmed by more recent work in the central north island (forbes et al. 2021); establishing and tending podocarps until year 5 cost at least $40,000/ha (rob allen pers. comm. august 2021). the current one billion trees programme for accelerated tree-planting allocates up to $6,000/ha for planting high-quality indigenous forest (https://www.mpi.govt.nz/forestry/fundingtree-planting-research/one-billion-trees-programme/ one-billion-tree-fund/; accessed july 2021), a very substantial underestimate of the true costs of a successful project. this underestimation extends to the research needed. as late as 1997, after more than 75 years of reiteration by the forest service on a more-or-less yearly basis of the need for research into indigenous silviculture, when it became apparent that demand for permits to log freehold indigenous forests was escalating, a ministry of forestry workshop concluded: “although indigenous forest management for timber production remains one of the most contentious land-use issues in new zealand, there is little research currently funded to ensure the sustainable management of private indigenous forests.” (allen & benecke 1997). the one billion tree initiative is currently funding a limited amount of research into indigenous forest planting and natural successions. some attempts have been made to improve the market, in particular for indigenous angiosperm timber. in the early days specialist uses were trialled including matches, tool handles, deep sea fishing rods, bowling balls, telephone cross-arms and coach building to supplement the more common use in furniture, interior finishing and veneer for the better grades, and mine props, fencing, boxing and rough construction for the lower grades. while indigenous angiosperm timbers with superior wood qualities have a wide range of uses (wardle 2011; nguyen et al. 2021), they are effectively niche products and, in general, do not command sufficiently high prices to justify intensive management of the source forest, especially when it is as slow growing as, for instance, tawa (smale et al. 1986) (table 1). a recent appraisal of indigenous timber production in new zealand concluded that the key element in the sector failing to realise expectations (value of harvest in 2016 was < 40% of 2001) was not the cost of establishing sustainable forestry plans but low demand for the product (griffiths 2017). competing imports of furniture from countries with lower costs of production and the subsequent retrenchment of the local indigenous manufacturing industries were thought to be largely responsible and hence there has been agitation from the forest owners for permission to export logs and pulp, rather than just sawn timber and manufactured products. this of course demolishes the argument for continuing logging of these forests because of local, niche manufacturing demand. while arguably the official policy for over 80 years, sustainable indigenous forestry was never seriously attempted on any substantial scale. although large areas were set aside as state forest, much of it cutover, only very limited attempts were made to manage them for timber production. the state of the effectively abandoned cutover forest was well known: “everyone is familiar with the depressing areas of logged native forest, abounding in weeds and useless scrub, the refuge of noxious animals and pests […] the best economics is often to raise on the land a crop of exotic trees…” nzfsar 1962: p. 15. today private efforts continue but on a small scale, and mostly confined to owners of forest tracts who, by registering a sustainable forest management (sfm) plan under the forests amendment act 2004, are permitted to take a portion of the annual increment. thus, with a few exceptions, these sfm plans are a legal means by which owners can recover some short-term value from forest which otherwise provides little income. in practice, sustainable indigenous forestry is effectively almost extinct in new zealand, with the exception of limited beech operations in the south island (allen et al., 2012). need this have been the outcome? for sustainable forestry to have been feasible in new zealand, a number of factors would have had to have aligned. first, silvicultural and forest management techniques would have to have been developed which minimised the slow growth/poor regeneration handicap of nearly all indigenous timber species; secondly, the economic return for the timber produced through utilization of those techniques would, at the very least, have had to cover the silvicultural outgoings; thirdly, local provincial interests would need to be receptive; and fourthly, national public opinion in agreement. as we have seen, silvicultural techniques were developed that minimised damage and gave a fair chance of adequate regeneration but, although time to a second harvest could be improved, it still remained too long to be economically competitive against imports. local opinion remained adamantly against ‘locking up’ (i.e., not clear-felling as expeditiously as possible) their forests and maintained ongoing political pressure to continue the unofficial clear-felling policy. as noted in 1955 (nzfsar: p16), “pressure for the release of state forest land for agricultural development has continued unabated throughout the year”, and in 1962 the forest service admitted that it had to abandon all hope of a reduction in cutting in indigenous forests, whether state or private, despite earlier attempts at regulation (nzfsar 1962: p23). if campbell-walker’s (1877) and ellis’ (1920) vision had been put into practice, that is, a much greater area of lowland indigenous forest on productive soils reserved for sustainable wood production, would such a system be currently economically viable given the contemporary requirement to maintain near-natural states? central to this issue is the economic value of mcglone et al. new zealand journal of forestry science (2022) 52:8 page 14 https://www.mpi.govt.nz/forestry/funding-tree-planting-research/one-billion-trees-programme/one-billion-tree-fund/ https://www.mpi.govt.nz/forestry/funding-tree-planting-research/one-billion-trees-programme/one-billion-tree-fund/ https://www.mpi.govt.nz/forestry/funding-tree-planting-research/one-billion-trees-programme/one-billion-tree-fund/ the timber produced versus the costs incurred through low-impact harvesting to maintain all-aged forest tracts and ensure regeneration and the ongoing costs of suppressing competing vegetation and controlling weeds and pests. the timber price on the local market is linked to those prevailing internationally, and in the absence of tariff protection, the value of indigenous timber is depressed. the biology of the valuable conifer species (kauri, rimu and tōtara podocarpus totara) mitigates against short rotations and, while the beech species apparently can be managed within a commercial framework, their timber does not command as high a premium. the scarcity of private investment in sustainable indigenous forestry, and the fact that state efforts were confined to small experimental treatments and the short-lived timberlands schemes at the turn of the century, demonstrate the lack of a commercially viable return relative to other opportunities. we conclude that even extensive lowland forests would not yield competitively priced timber unless mined as it was in the past. some public good would have to be factored in to make logging acceptable. some balancing considerations can be suggested: provision of jobs for rural communities; non-timber products (such as honey, sphagnum moss, possum fur); recreational and tourist opportunities (hunting, bushwalking, sightseeing); the environmental value of weed and pest control; carbon sequestration; and finally, the existence value of the managed forest (yao et al. 2017). however, the problem with using these considerations as counterweights is that all but jobs and weed and pest control are provided simply by leaving the forests as they are. unless the profit realised by the timber is higher than has been the case to date, the extraction of timber barely pays for the silviculture necessary to ensure a second viable crop within a commercially viable timespan, and there is little left over for weed and pest control. that leaves the sole significant balancing consideration as jobs for rural communities, and that indeed was a major element in the equation until recent years. the implicit assumption for many years was that sustainable forest management was a desirable alternative to much more damaging clearfelling or conversion to pasture or exotic plantations, and that it merely had to show that it was technically feasible. that is why the focus has been on how to avoid damage during logging, how to ensure adequate regeneration and how to maintain a forest structure that bears some resemblance to the natural state (allen et al. 2013; wiser et al. 2005). however, with much more strict regulation and a near-total ban on conversion of indigenous forest to other uses, the focus now has to be on the intrinsic value of indigenous forests. leaving indigenous forests free of timber extraction is underpinned as a policy by the ease with which timber can be sourced elsewhere at competitive prices. however, a valid case can be made that, by importing hardwood timber products, new zealand can be supporting unsustainable logging practices elsewhere in the globe (mayer et al. 2005, allen et al. 2013). global product certification may go some way towards rectifying this situation but it seems certain that in the medium term, unless imported furniture and fittings cheaply manufactured from overseas hardwoods are excluded or subject to tariffs, prices for indigenous new zealand timber will not reflect the true costs of their sustainable production (griffiths 2017). science, policy, and practice “to the rest of the developed world such rights (including access to their native timbers) are integral to their culture and heritage. perhaps it is time that we as a culture developed a more mature perspective as well, and stopped fighting the phantom, bushclearing pioneer in every forest management plan” (perley 1998). applied sciences such as forestry can thrive only when the products of their research are taken up and used. if the economic rationale is weak, and consistent, clear policies, regulation and enforcement are lacking, research struggles to be relevant. throughout the nearly 70 years of the new zealand forest service’s existence, the economic basis for sustainable indigenous forestry was questionable, official policy often did not reflect the actual situation and, although well-intentioned regulations were promulgated, enforcement was slack. as well, the new zealand forest service itself had a dilemma that became only worse through time. we can best see this through using the concept of an ‘overton window of political possibilities’ (szałek 2013). politicians have only a limited number of policy options available to them at any one time, and are constrained by societal willingness to support them. options can be portrayed as a sequence from less government control and regulation to greater regulation, or in the case of new zealand indigenous forests from policies that prioritise social and economic factors to biocentric ones. the overton window depicts the range of acceptable options available at a given time (fig. 4). our estimation is that the overton window for new zealand indigenous forestry shifted in the course of 140 years from a range of policy options that favoured destruction of the lowland forests to a policy range that put preservation of the forests first. in the early years, the new zealand forest service was a consistent advocate for reservation of forest for timber production and reduction of waste and at the upper end of the contemporary overton window. by 1980, they were well towards the bottom of the overton window. however, these positions relate only to the policies that the new zealand forest service was advocating, that is for silvicultural management of indigenous forests, and with multiple-use forests preferred over strict reservation. as we have seen, actual forestry practice was different. varying felling techniques were mandated but oversight of private loggers was weak and silvicultural management was rarely done. there was thus only a limited amount of data that could be collected to inform best practice and expensive experimental investigations had to be undertaken to provide this. these investigations were necessarily limited. as well, despite the official policies advocated mcglone et al. new zealand journal of forestry science (2022) 52:8 page 15 regarding perpetuation of vigorous, healthy productive forests, the new zealand forest service remained highly pessimistic about whether economically viable indigenous forestry was possible. as the annual report of 1960 (p19) stated: “…overseas professional visitors have almost invariably described the forest service attitude to indigenous forestry as one of “defeatism” ”. this is reflected in the statement some years later that: “…there is no intention of planning restrictions on the milling of private and maori-owned timber so that the only method of husbanding our timber resource is to control cutting of state-owned timber” (nzfsar 1965: p7). as the overton window later shifted inexorably downwards, the focus on timber production and the relative neglect of other aspects of indigenous forests vitiated forest service policy recommendations. scientific research and forest research scientists have only a limited role in this story. research into sustainable forest practice was only sporadically carried out in state forests and largely ignored on private land until legislation early in the 20th century (halkett 1991). even then, conservation-minded forest scientists supported pastoral development over retention of forest. as late as the 1970s, the ecological society of new zealand offered only muted criticism of the ill-conceived west coast beech scheme of 1971, focussing largely on provision for reserves and stating: “…if in the future, the reservations are found to be needlessly generous in ecological, scientific, recreational, tourism and “conservation” grounds, the timber resource will still be available and be of even greater economic value than if it is imprudently cut now” (new zealand ecological society 1978). the dedicated effort by forest service and department of scientific and industrial research scientists to document and argue the case for ecological reserves (bassett & miers 1984) is, more than indigenous silviculture, the enduring scientific success of this period. scientific research therefore was operating in an unreal environment. the prevailing pessimistic attitudes towards indigenous forestry meant that takeup of harvest and silvicultural recommendations on state forest land were likely to be half-hearted and non-existent on private land. the economic basis for sustainable indigenous forestry was lacking and this probably lay behind the failure to document costs and to project future returns. however, whether intended or not, this scientific effort enabled the forest service to keep the illusion going until the end that indigenous forestry was viable on a sustainable but economically unquantified basis. ultimately, rosy visions of a time in the undetermined future when research would deliver the means for sustainable forestry was insufficient. in the early 1970s, the forest service could not point to extensive areas of indigenous forest where plans for their sustainable management could be demonstrated, nor to a thriving specialist timber industry, but instead were figure 4: an overton window analysis of new zealand forestry policy, 1880 to 2020. forest policy settings (as we have interpreted them from new zealand forest service annual reports and ministry for primary industries reports) are listed (top to bottom) from those advocating little regulation (1) to those advocating complete cessation of logging and clearance (7). the boxes to the left are our assessment of the feasible policy settings (= overton windows) given the public opinion and economic imperatives at various times in the past. the black lines within each time period indicate the position taken by the new zealand forest service and successor organisations at those times. mcglone et al. new zealand journal of forestry science (2022) 52:8 page 16 advocating indigenous log and pulp exports and clearfelling natural forests for their replacement by exotic conifer plantations. the lesson from this history is that if it is to have any hope of making an effective contribution to environmental debates, applied scientific research needs to address realistic policy goals and to partner with broader economic and social analyses (tilling 1988). while this is a bleak assessment of the influence of science on indigenous forestry policy and outcomes, the work that the forest service and other forestry scientists carried out was and remains of great value. the forest plots they established throughout new zealand forests have provided a secure basis for later monitoring and research (wiser et al. 2001; bellingham et al. 2020). through their unparalleled knowledge of the forests, forest service scientists such as john nicholls were well placed to ensure that ecologically and scientifically important reserves were created, even in the midst of the post-war indigenous logging frenzy (bassett & miers 1984). the research that the forest service conducted and the infrastructure of plots they established underpinned physiological work on the fundamentals of tree growth and carbon sequestration (beets 1980, benecke & nordmeyer 1982, hollinger et al. 1994, hall et al. 2001), climate-tree relationships (leathwick, et al. 1996, leathwick & whitehead 2001), and forest modelling (hall & hollinger 2000), advances now critical to dealing with the issues raised by climate change (holdaway et al. 2017), pest animal impacts, and novel pathogens (mccarthy et al. 2021). the future “the resulting ‘natural forestry’ approach represents a paradigm shift that is long overdue in new zealand and overseas. unfortunately recent public statements suggest that many conservationists cannot adapt to twc’s challenge, nor divorce their religious fundamentalist beliefs from scientific inference that the proposed forestry represents a commendable compromise between conservation, economics and other societal needs.” (moller 1998). “…it just isn’t appropriate to run the sort of campaign against indigenous forest managers that we ran back in the70s and 80s. indeed, it seems counterproductive...” (salmon 1998) “the beech scheme has all the makings of an economic as well as ecological fiasco” (sage 1998). the radically divergent positions (quoted above) on the timberlands beech scheme adopted by prominent conservationists at the height of the controversy show that the issue of sustainable indigenous forestry in new zealand is not settled. and indeed, it never was for māori. as michael roche points out, the original mandate of the department of conservation for preservation is ‘no longer beyond question’ as authorization by parliament for salvage logging of windthrown indigenous forest and disestablishment of a national park and its reconstitution under a joint iwi–crown management shows (roche 2017). many māori, as discussed above, may have a more use-orientated attitude to indigenous forest than the dominant pākehā paradigm of biocentric and nonextractive use conservation (lyver et al. 2017). this pākehā–urban paradigm has deep roots, and appears to have begun to spread early in the 20th century when the difficulties of sustained management of indigenous forests first became apparent. as star (2002) states: ‘… a curiously fractured society has evolved in new zealand, which seeks spiritual sustenance through its native forests while being physically remote and gaining no material sustenance from them. whether this approach was (and is) necessary or, on balance, the most constructive way forward – for new zealand or any country – should be the subject of continuing debate.’ roche argues that the current ‘new zealand forest model’ of strict separation of exotic production forest and indigenous preservation reflects mono-functional tendencies which were inscribed in cadastral patterns and land tenure systems since colonial times. this mono-functional trend may be reversed, not only by changing legislation and regulation, but also by continuing research and the nature of the forests themselves. as an example of the research, more than 200,000 ha of northland freehold land include stands of regenerating tōtara. a pioneering species, it often produces dense stands on otherwise unproductive land (young & norton 2017), and it has been shown that merchantable timber can be extracted from them with little disturbance to their stand structure (steward & quinlan 2019) and that the end product is highly acceptable and favoured by architects and furniture makers as an attractive native timber (quinlan 2011). similar research is underway on the abundant beech forests of the south island that already produce the bulk of indigenous timber milled in the country (allen et al. 2012), and support the only substantial indigenous timber operation left in the country, that of lindsay & dixon in the longwood range. invasion of exotic trees into indigenous forests and vice versa challenges the new zealand forest model. for instance, regeneration on abandoned pasture or cutover forest will often include vigorous woody exotics. exotic successions may lead to indigenous forests quite different from those following indigenous successions (sullivan et al. 2007) and mixed exotic–indigenous forest types are not uncommon. underplanting of pine plantations with native trees is another instance where ‘not quite natural’ indigenous forests will result (forbes et al. 2019). exactly where will these novel forest types fall on the current forest model spectrum? will they be available for commercial exploitation without the restrictions applying to indigenous forests? roche (2017) argues that new zealand does not currently have a comprehensive national forest policy to answer these and similar questions, but instead an assemblage of institutions with governance or oversight responsibility and a range of legislation. lack of such a policy is negatively impacting the forestry sector (mcewen 2013). as we have seen, in the absence of clear goals responsive to public opinion as well as commercial mcglone et al. new zealand journal of forestry science (2022) 52:8 page 17 imperatives, and lack of well-designed policy linked to practical actions, scientific research struggles to deliver. potential problems of the lack of a comprehensive national forest policy can be seen in the one billion trees programme (te uru rākau). like the massive exotic plantings of the 1920s and 1930s, it is primarily driven by an economic imperative although, rather than a looming timber shortage, it is a need to sequester carbon in response to the imminent disaster of climate change (bastin et al. 2019), although tree planting will not solve climate change (holl & brancalion 2020). as with the earlier exotic forest programme of the early 20th century, it relies on mass plantings (although in this case 70% indigenous) using rural labour. for indigenous plantings that are centred on establishing podocarps, the economic and ecological problems that saw such efforts fail in the past will be encountered again. te uru rākau is funding scientific research (one billion trees science plan) with the aim of preventing such failure and that intends to reactivate plots established by forest service scientists in the 1980s. it is to be hoped that this research will be generously funded, long-term, and supported by clear goals. an essential component of this research must be attention to the costs of the various interventions and provision of high-quality monitoring of progress. what needs to be avoided at all costs is the ‘smart solutions’ approach of short-term technologydriven innovation focussing on establishment. the key issues are not nursery-raising of seedlings or planting techniques, but the more fundamental concerns about what happens after, and that requires a long-term perspective. conclusion scientific research was far less useful than it might otherwise have been in providing solutions to new zealand’s sustainable indigenous forestry problems. in the 20th century researchers were assigned a goal of economically viable sustainable indigenous timber production that was known with some certainty to be unobtainable even in the 19th century and which collapsed in the 21st. the new zealand forest service for almost 70 years promoted sustainability policies for multiple-use forests which would be ‘better than natural’, which they never seriously attempted to implement and that would have been blocked by government and rural interests had they tried. nevertheless, there is a valuable science legacy consisting of forests protected from exploitation and underpinning research into all aspects of indigenous forests. there are opportunities for future applied research, but policy guidance is needed. we still need a national consensus, vision and plan: that is an understanding of what roles, including provision of timber, indigenous forests will play in new zealand landscapes, and the pathways to achieve them. competing interests the authors declare that they have no competing interests. authors’ contributions all authors conceived of the study, and participated in its direction and completion. all authors read and approved the final manuscript. acknowledgements we thank the journal editor, greg steward, and two anonymous reviewers for their insightful comments and criticisms which led to substantial changes in the manuscript, and rob allen for provision of unpublished data on reforestation costs and many years of discussion on this and related topics. research was funded by strategic science investment funding from the new zealand ministry of business, innovation and employment to manaaki whenua – landcare research. references allen, r., hurst, j., wiser, s., & easdale, t. (2012). developing management systems for the production of beech timber. new zealand journal of forestry, 57, 38–44. allen, r.b. (2004). sustainable indigenous research funding. new zealand tree grower, february 2004. allen, r.b., & benecke, u. (1997). sustainable indigenous forest management research – christchurch, april 1997. new zealand forestry, august, 7–8. allen, r.b., bellingham, p.j., holdaway, r.j., & wiser, s.k. (2013). new zealand’s indigenous forests and shrublands. in j. dymond (ed.) ecosystem services in new zealand—condition and trends, pp. 34–48. lincoln: manaaki whenua press. allen, r.b., buchanan, p.w., clinton, p.w., & cone, a.j. (2000). composition and diversity of fungi on decaying logs in a new zealand temperate beech (nothofagus) forest. canadian journal of forest research, 30(7), 1025–1033. https://doi. org/10.1139/x00-037 arnold, r. (1994). new zealand’s burning: the settlers’ world in the mid 1880’s. wellington, new zealand: victoria university press. barton, i.l., & madgwick, h.a.i. (1987). response of a kauri stand to fertiliser addition and thinning. new zealand forestry, 32(2), 19–21. bassett, c. (1987). the role of the forest service in conservation of native state forests. new zealand forestry, 32(2), 15–18. bassett, c., & miers, k. (1984). scientific reserves in state forest. journal of the royal society of new zealand, 14(1), 29–35. https://doi.org/10.1080/03036758. 1984.10421725 bastin, j.–f., finegold, y., garcia, c., mollicone, d., rezende, m., routh, d., zohner, c.m, & crowther, t.w. (2019). the global tree restoration potential. science, mcglone et al. new zealand journal of forestry science (2022) 52:8 page 18 https://doi.org/10.1139/x00-037 https://doi.org/10.1139/x00-037 https://doi.org/10.1080/03036758.1984.10421725 https://doi.org/10.1080/03036758.1984.10421725 365(6448), 76–79. https://doi.org/10.1126/ science.aax0848 baylis, g. (1980). mycorrhizas and the spread of beech. new zealand journal of ecology, 3, 151–153. beattie, j. (2003). environmental anxiety in new zealand, 1840–1941: climate change, soil erosion, sand drift, flooding and forest conservation. environment and history, 9(4), 379–392. https:// doi.org/10.3197/096734003129342881 bee, j.n., kunstler, g., & coomes, d.a. (2007). resistance and resilience of new zealand tree species to browsing. journal of ecology, 95(5), 1014–1026. https://doi.org/10.1111/j.1365– 2745.2007.01261.x beets, p.n. (1980). amount and distribution of dry matter in a mature beech/podocarp community. new zealand journal of forestry science, 10, 395–418. bellingham, p.j., richardson, s.j., gormley, a.m., allen, r.b., cook, a., crisp, p.n., forsyth, d.m., mcglone, m.s., mckay, m., macleod, c.m., van dam bates, p., & wright, e.f. (2020). implementing integrated measurements of essential biodiversity variables at a national scale. ecological solutions and evidence, 1(2), e12025. https://doi.org/10.1002/26888319.12025 bellingham, p.j., wiser, s.k., wright, a.e., cameron, e.k., & forester, l.j. (2010). disperser communities and legacies of goat grazing determine forest succession on the remote three kings islands, new zealand. biological conservation, 143(4), 926–938. https://doi.org/10.1016/j.biocon.2010.01.001 benecke, u. (1996). ecological silviculture: the application of age–old methods. new zealand forestry, 41(2), 27–33. benecke, u., & nordmeyer, a.h. (1982). carbon uptake and allocation by nothofagus solandri var cliffortioides (hook. f.) poole and pinus contorta douglas ex loudon ssp. contorta at montane and subalpine altitudes. in r.h. waring (ed.) carbon uptake and allocation in subalpine ecosystems as a key to management, pp. 9–21. corvallis, usa: forest research laboratory, oregon state university. bensemann, p. (2018). the fight for the forests. nelson: potton & burton. bergin, d.o., & kimberley, m.o. (2003). growth and yield of tōtara in planted stands. new zealand journal of forestry science, 33(2), 244–264 beveridge, a., christensen, b., smale, m., & bergin, d. (2009). ecology, management and history of the forests of the mamaku plateau, new zealand. wellington, new zealand: department of conservation. beveridge, a., holzapfel, a., & smale, m. (2000). ecology and management of pureora forest park. science & research division, department of conservation. beveridge, a.e. (1973). regeneration of podocarps in a central north island forest. new zealand journal of forestry, 18(1), 23–35. briggs, p. (2003). looking at the numbers: a view of new zealand’s economic history: wellington, new zealand institute of economic research. burns, b.r., smale, m., & merrett, m.f. (1999). dynamics of kahikatea forest remnants in middle north island: implications for threatened and local plants: science for conservation 113. wellington, department of conservation. cameron, r.j. (1954). mosaic or cyclical regeneration in north island podocarp forests. new zealand journal of forestry, 7(1), 55–64. cameron, r.j. (1961). maori impact upon the forests of new zealand. historical review (whakatane historical society), 9(3), 131–141. campbell–walker, i. (1876). state forestry: its aim and object. transactions and proceedings of the new zealand institute, 9, 187–203. campbell–walker, i. (1877). report of the conservator of state forests. appendix to the journals of the house of representatives, session1, c–03. carswell, f.e., richardson, s.j., doherty, j.e., allen, r.b., & wiser, s.k. (2007). where do conifers regenerate after selective harvest? a case study from a new zealand conifer–angiosperm forest. forest ecology and management, 253(1–3), 138–147. https://doi. org/10.1016/j.foreco.2007.07.011 chavasse, c. (1954) potentialities for indigenous and exotic forestry in westland. new zealand journal of forestry 7(1) 34–49. chavasse, c., & travers, w.w. (1966). growth habits of rimu in westland’s terrace forests and their implications for forest management: new zealand forest service. cockayne, l. (1926). monograph on the new zealand beech forests (vol. 1). wellington: wag skinner, government printer. cockayne, l. (1927). new zealand plants and their story (3rd ed.). wellington: government printer. cockayne, l. (1928). the vegetation of new zealand (2nd rev ed.). leipzig: engelmann. conway, m.j. (1977). management policy for new zealand’s indigenous state forests. wellington: new zealand forest service. coomes, d.a., & allen, r.b. (2007). mortality and tree-size distributions in natural mixed-age forests. journal of ecology, 95(1), 27–40. https://doi.org/10.1111/ mcglone et al. new zealand journal of forestry science (2022) 52:8 page 19 https://doi.org/10.3197/096734003129342881 https://doi.org/10.3197/096734003129342881 https://doi.org/10.1002/2688-8319.12025 https://doi.org/10.1002/2688-8319.12025 https://doi.org/10.1016/j.biocon.2010.01.001 https://doi.org/10.1016/j.foreco.2007.07.011 https://doi.org/10.1016/j.foreco.2007.07.011 https://doi.org/10.1111/j.1365-2745.2006.01179.x j.1365-2745.2006.01179.x coomes, d.a., & bellingham, p.j. (2011). temperate and tropical podocarps: how ecologically alike are they? in b. l. turner & l. a. cernusak (eds.), smithsonian contributions to botany: ecology of the podocarpaceae in tropical forests, (vol. 95, pp. 119– 140). washington, d.c.: smithsonian institution scholarly press. https://doi.org/10.5479/ si.0081024x.95.119 coomes, d.a., allen, r.b., bentley, w.a., burrows, l.e., canham, c.d., fagan, l., forsyth, d.m., gaxiola– alacantar, a., parfitt, r.l., ruscoe, w.a., wardle, d.a., wilson, d.j., & wright, e.f. (2005). the hare, the tortoise and the crocodile: the ecology of angiosperm dominance, conifer persistence and fern filtering. journal of ecology, 93(5), 918–935. https://doi.org/10.1111/j.13652745.2005.01012.x cumberland, k.b. (1962). ‘climate change’ or cultural interference? new zealand in moahunter times. in m. mccaskill (ed.), land and livelihood: geographical essays in honour of george j jobberns (pp. 88–142). christchurch: new zealand geographical society. dansereau, p. (1964). six problems in new zealand vegetation. bulletin of the torrey botanical club, 91(2), 114–140. https://doi. org/10.2307/2483613 dawson, j., & sneddon, b. (1969). the new zealand rain forest: a comparison with tropical rain forest. pacific science, 23, 131–147. devoe, n., & olson, s. (2001). why a strong indigenous forestry sector is in the national interest. new zealand journal of forestry, 46, 22–26. dickie, i.a., davis, m., & carswell, f.e. (2012). quantification of mycorrhizal limitation in beech spread. new zealand journal of ecology, 36(2), 210–215. duncan, r.p. (1991). competition and the coexistence of species in a mixed podocarp stand. journal of ecology, 79(4), 1073–1084. https://doi. org/10.2307/2261099 easton, b. (2020). not in narrow seas: the economic history of aotearoa new zealand. wellington: victoria university of wellington press. ebbett, r. l. (1998). the ecology of lowland totara in south island, new zealand: distribution, regeneration, and future survival in a fragmented landscape. a thesis submitted in fulfilment of the requirements for the degree of doctor of philosophy. school of forestry, university of canterbury, christchurch, new zealand ellis, l.m. (1920). forest conditions in new zealand, and the proposals for a new zealand forest policy. wellington. evison, d., easdale, t., mason, e., & sewell, a. (2012). economics of managing new zealand silver beech for timber and carbon. nelson, new zealand: nz agricultural and resource economics society annual conference, 31 august 2012. fleming, c.a. (1969). mammon on the mamaku. new zealand listener, 14 november 1969. downloaded from te ara – the encyclopedia of new zealand http://www.teara.govt.nz forbes, a.s., allen, r.b., herbert, j.w., kohiti, k., shaw, w.b., & taurua, l. (2021). determining the balance between active and passive indigenous forest restoration after exotic conifer plantation clear– fell. forest ecology and management, 479, 118621. https://doi.org/10.1016/j.foreco.2020.118621 forbes, a.s., norton, d.a., & carswell, f.e. (2019). opportunities and limitations of exotic pinus radiata as a facilitative nurse for new zealand indigenous forest restoration. new zealand journal of forestry science, 49: 6. https://doi. org/10.33494/nzjfs492019x45x forsyth, d.m., wilson, d.j., easdale, t.a., kunstler, g., canham, c.d., ruscoe, w.a., wright, e.f., murphy, l., gormley, a.m., gaxiola, a., coomes, d.a. (2015). century-scale effects of invasive deer and rodents on the dynamics of forests growing on soils of contrasting fertility. ecological monographs, 85(2), 157–180. https://doi.org/10.1890/14-0389.1 gillman, l.n. (2008). assessment of sustainable forest management in new zealand indigenous forest. new zealand geographer, 64(1), 57–67. https:// doi.org/10.1111/j.1745-7939.2008.00127.x godman, r.m. (1957). silvical characteristics of sugar maple (acer saccharum). lake states forest experiment station, forest service: us department of agriculture. goulding, c. (2013). the 1913 new zealand royal commission on forestry. new zealand journal of forestry, 58(1), 3. griffiths, a. (2002). managing nz’s indigenous forested lands for timber; an update, christchurch, new zealand: ministry of agriculture and forestry. griffiths, a. (2016). indigenous forestry on private land: present trends and future potential – an update, christchurch, new zealand: ministry of agriculture and forestry. griffiths, a. (2017). indigenous forestry on private land: present trends and future potential, christchurch, new zealand: ministry of agriculture and forestry. hager, n., & burton, b. (1999). secrets and lies. nelson: craig potton publishing. mcglone et al. new zealand journal of forestry science (2022) 52:8 page 20 https://doi.org/10.1111/j.1365-2745.2006.01179.x https://doi.org/10.1111/j.1365-2745.2005.01012.x https://doi.org/10.1111/j.1365-2745.2005.01012.x https://doi.org/10.2307/2483613 https://doi.org/10.2307/2483613 https://doi.org/10.2307/2261099 https://doi.org/10.2307/2261099 http://www.teara.govt.nz https://doi.org/10.1016/j.foreco.2020.118621 https://doi.org/10.33494/nzjfs492019x45x https://doi.org/10.33494/nzjfs492019x45x https://doi.org/10.1890/14-0389.1 https://doi.org/10.1111/j.1745-7939.2008.00127.x https://doi.org/10.1111/j.1745-7939.2008.00127.x halkett, j. (1983). a basis for the management of new zealand kauri (agathis australis (d. don) lindl.) forest. new zealand journal of forestry, 28(1), 15– 23. halkett, j. (1991). the native forests of new zealand. wellington: gp books. hall, g.m.j., & hollinger, d.y. (2000). simulating new zealand forest dynamics with a generalized temperate forest gap model. ecological applications, 10(1), 115–130. https://doi.org/10.1890/10510761(2000)010[0115:snzfdw]2.0.co;2 hall, g.m.j., wiser, s.k., allen, r.b., beets, p.n., & goulding, c.j. (2001). strategies to estimate national forest carbon stocks from inventory data: the 1990 new zealand baseline. global change biology, 7(4), 389–403. https://doi.org/10.1046/j.1365– 2486.2001.00419.x hawes, p., & memon, p.a. (1998). prospects for sustainable management of indigenous forests on private land in new zealand. journal of environmental management, 52(2), 113–130. https://doi.org/10.1006/jema.1997.0164 hewitt, a.e., balks, m.r., & lowe, d.j. (2021). the soils of aotearoa new zealand. world soils book series, springerlink. https://doi.org/10.1007/978-3030-64763-6 holdaway, r.j., easdale, t.a., carswell, f.e., richardson, s.j., peltzer, d.a., mason, n.w., brandon, a.m., coomes, d.a. (2017). nationally representative plot network reveals contrasting drivers of net biomass change in secondary and old–growth forests. ecosystems, 20(5), 944–959. https://doi. org/10.1007/s10021–016–0084–x holl, k.d., & brancalion, p.h. (2020). tree planting is not a simple solution. science, 368(6491), 580–581. https://doi.org/10.1126/science.aba8232 hollinger, d., kelliher, f., byers, j., hunt, j., mcseveny, t., & weir, p. (1994). carbon dioxide exchange between an undisturbed old-growth temperate forest and the atmosphere. ecology, 75(1), 134–150. https:// doi.org/10.2307/1939390 holloway, j.t. (1954). forests and climate in the south island of new zealand. transactions of the royal society of new zealand, 82(2), 329–410. holt, l., & bennett, p. (2014). connecting science and technical research with māori interests in forestry: ka tangi hoki ahau. new zealand journal of forestry, 58(4), 13. hutchinson, f. (1928). a hypothesis in regard to the westland rimu bush. te kura ngahere, 2(3), 3–12. hutchinson, f. (1931). an approach to the management of the rimu forests. te kura ngahere, 3, 17–24. james, i., & franklin, d. (1977). preliminary results on the effects of selection management of terrace rimu forest. new zealand journal of forestry science, 7, 349–358. james, i.l., & norton, d.a. (2002). helicopter–based natural forest management for new zealand’s rimu (dacrydium cupressinum, podocarpaceae) forests. forest ecology and management, 155, 337–346. https://doi.org/10.1016/s0378-1127(01)005709 johnston, a.d. (1972). management of west coast beech forest. new zealand journal of forestry, 17, 180– 188. june, s., & ogden, j. (1978). studies on the vegetation of mount colenso, new zealand 4. an assessment of the processes of canopy maintenance and regeneration strategy in a red beech (nothofagus fusca) forest. new zealand journal of ecology, 1, 7–15. katz, a. (1980). structure and growth of dense podocarp forest in whirinaki. new zealand forest service production forestry division indigenous forest management report no. 25. rotorua, forest research institute, new zealand forest service. kennedy, d. (1951). new zealand forestry practice: review and preview. new zealand journal of forestry, 6(3), 189–199. king, c.m., gaukrodger, d.j., & ritchie, n.a. (2015). the drama of conservation: the history of pureora forest, new zealand. berlin: springer. https://doi. org/10.1007/978-3-319-18410-4 kirk, t. (1889). the forest flora of new zealand. wellington: government printer. lausberg, m., cown, d., mcconchie, d., & skipwith, j. (1995). variation in some wood properties of pseudotsuga menziesii provenances grown in new zealand. new zealand journal of forestry science, 25(2), 133–146. leathwick, j., whitehead, d., & mcleod, m. (1996). predicting changes in the composition of new zealand’s indigenous forests in response to global warming: a modelling approach. environmental software, 11(1–3), 81–90. https://doi. org/10.1016/s0266-9838(96)00045-7 leathwick, j.r., & whitehead, d. (2001). soil and atmospheric water deficits and the distribution of new zealand’s indigenous tree species. functional ecology, 15(2), 233–242. https://doi.org/10.1046/ j.1365-2435.2001.00504.x lecoy, a. (1879). the forests question in new zealand. transactions and proceedings of the new zealand institute, 12, 3–23. mcglone et al. new zealand journal of forestry science (2022) 52:8 page 21 https://doi.org/10.1890/1051-0761(2000)010%5b0115:snzfdw%5d2.0.co;2 https://doi.org/10.1890/1051-0761(2000)010%5b0115:snzfdw%5d2.0.co;2 https://doi.org/10.1046/j.1365–2486.2001.00419.x https://doi.org/10.1046/j.1365–2486.2001.00419.x https://doi.org/10.1006/jema.1997.0164 https://doi.org/10.1007/978-3-030-64763-6 https://doi.org/10.1007/978-3-030-64763-6 https://doi.org/10.1126/science.aba8232 https://doi.org/10.2307/1939390 https://doi.org/10.2307/1939390 https://doi.org/10.1016/s0378-1127(01)00570-9 https://doi.org/10.1016/s0378-1127(01)00570-9 https://doi.org/10.1007/978-3-319-18410-4 https://doi.org/10.1007/978-3-319-18410-4 https://doi.org/10.1016/s0266-9838(96)00045-7 https://doi.org/10.1016/s0266-9838(96)00045-7 https://doi.org/10.1046/j.1365-2435.2001.00504.x https://doi.org/10.1046/j.1365-2435.2001.00504.x lee, w.g. (1998) the vegetation of new zealand – functional, spatial, and temporal gaps. in: lynch p. (ed.). ecosystems, entomology and plants: proceedings of a symposium held at lincoln university to mark the retirement of bryony macmillan, john dugdale, peter wardle and brian malloy. the royal society of new zealand miscellaneous series 48, pp. 91–101. wellington. levack, h. (2006). rectifying bad forest governance in new zealand. new zealand journal of forestry, 51(1), 34. lusk, c.h., & smith, b. (1998). life history differences and tree species coexistence in an old-growth new zealand rain forest. ecology, 79(3), 795–806. https://doi.org/10.1890/00129658(1998)079[0795:lhdats]2.0.co;2 lusk, c.h., jorgensen, m.a., & bellingham, p.j. (2015). a conifer–angiosperm divergence in the growth vs. shade tolerance trade–off underlies the dynamics of a new zealand warm–temperate rain forest. journal of ecology, 103(2), 479–488. https://doi. org/10.1111/1365-2745.12368 lyver, p.o., timoti, p., gormley, a.m., jones, c.j., richardson, s.j., tahi, b.l., & greenhalgh, s. (2017). key māori values strengthen the mapping of forest ecosystem services. ecosystem services, 27, 92–102. https:// doi.org/10.1016/j.ecoser.2017.08.009 mason, e.g. (2000). evaluation of a model beech forest growing on the west coast of the south island of new zealand. new zealand journal of forestry, 44, 26–31. masters, s.e., holloway, j.t., & mckelvey, p.j. (1957). the national forest survey of new zealand, 1955. volume. i. the indigenous forest resources of new zealand. wellington, nz: government printer. mayer, a.l., kauppi, p.e., angelstam, p.k., zhang, y., & tikka, p.m. (2005). importing timber, exporting ecological impact. science, 308(5720), 359–360. https://doi.org/10.1126/science.1109476 mccarthy, j.k., wiser, s.k., bellingham, p.j., beresford, r.m., campbell, r.e., turner, r., & richardson, s.j. (2021). using spatial models to identify refugia and guide restoration in response to an invasive plant pathogen. journal of applied ecology, 58(1), 192– 201. https://doi.org/10.1111/1365-2664.13756 mcdonald, r.i., urban, d.l. (2004). forest edges and tree growth rates in the north carolina piedmont. ecology, 85(8), 2258–2566. https://doi. org/10.1890/03-0313 mcewen, a. (2013). why we need a national forest policy. new zealand journal of forestry, 58(1), 19. mcglone, m.s., buitenwerf, r., & richardson, s.j. (2016). the formation of the oceanic temperate forests of new zealand. new zealand journal of botany, 54(2), 128–155. https://doi.org/10.1080/002882 5x.2016.1158196 mcglone, m.s., dungan, r.j., hall, g.m.j., & allen, r.b. (2004). winter leaf loss in the new zealand woody flora. new zealand journal of botany, 42(1), 1–19. https://doi.org/10.1080/002882 5x.2004.9512887 mcglone, m.s., richardson, s.j., & jordan, g.j. (2010). comparative biogeography of new zealand trees: species richness, height, leaf traits and range sizes. new zealand journal of ecology, 34(1), 137–151. mcglone, m.s., richardson, s.j., burge, o.r., perry, g.l.w., & wilmshurst, j.m. (2017). palynology and the ecology of the new zealand conifers. frontiers in earth science, 5, 94 https://doi.org/10.3389/ feart.2017.00094 mckelvey, p.j. (1963). the synecology of the west taupo indigenous forest. new zealand forest service bulletin no. 14. wellington: new zealand forest service. mckelvey, p.j. (1999). earlier professional schools of forestry in new zealand. new zealand forestry, 43, 30–34. ministry for primary industries (2015). sustainable management of new zealand’s forests: new zealand’s third country report on the montreal process criteria and indicators. wellington: ministry for primary industries. moller, h. (1998). conservation through sustainable beech forest management: breaking the historical shackles of conservationists and foresters. new zealand forestry, 43 (3), 10–11. new zealand ecological society, (1978). the future of west coast forests and forest industries: policy submission. new zealand journal of ecology, 1, 166–172. nguyen, l., bayne, k., & altaner, c. (2021). a review of kōwhai (sophora spp.) and its potential for commercial forestry. new zealand journal of forestry science, 51: 8. https://doi.org/10.33494/ nzjfs512021x157x nicholls, j. (1956). the historical ecology of the indigenous forest of the taranaki upland. new zealand journal of forestry, 7(3), 17–34. norton, d., herbert, j., & beveridge, a. (1988). the ecology of dacrydium cupressinum: a review. new zealand journal of botany, 26(1), 37–62. https:// doi.org/10.1080/0028825x.1988.10410098 mcglone et al. new zealand journal of forestry science (2022) 52:8 page 22 https://doi.org/10.1890/0012-9658(1998)079%5b0795:lhdats%5d2.0.co;2 https://doi.org/10.1890/0012-9658(1998)079%5b0795:lhdats%5d2.0.co;2 https://doi.org/10.1111/1365-2745.12368 https://doi.org/10.1111/1365-2745.12368 https://doi.org/10.1016/j.ecoser.2017.08.009 https://doi.org/10.1016/j.ecoser.2017.08.009 https://doi.org/10.1126/science.1109476 https://doi.org/10.1111/1365-2664.13756 https://doi.org/10.1890/03-0313 https://doi.org/10.1890/03-0313 https://doi.org/10.1080/0028825x.2016.1158196 https://doi.org/10.1080/0028825x.2016.1158196 https://doi.org/10.1080/0028825x.2004.9512887 https://doi.org/10.1080/0028825x.2004.9512887 https://doi.org/10.3389/feart.2017.00094 https://doi.org/10.3389/feart.2017.00094 https://doi.org/10.33494/nzjfs512021x157x https://doi.org/10.33494/nzjfs512021x157x https://doi.org/10.1080/0028825x.1988.10410098 https://doi.org/10.1080/0028825x.1988.10410098 ogden, j. (1978). on the diameter growth rates of red beech (nothofagus fusca) in different parts of new zealand. new zealand journal of ecology, 1, 16–18. ogden, j., & stewart, g.h. (1995). community dynamics of the new zealand conifers. in n.j. enright & r.s. hill (eds.), ecology of the southern conifers (pp. 81–119). melbourne: melbourne university press. ogden, j., & west, c. (1981). annual rings in beilschmiedia tawa (lauraceae). new zealand journal of botany, 19(4), 397–400. https://doi.org/10.1080/002882 5x.1981.10426397 ogden, j., stewart, g., & allen, r. (1996). ecology of new zealand nothofagus forests. in t.t. veblen, r.s. hill & j. read (eds.) the ecology and biogeography of nothofagus forests (pp. 25–82). new haven: yale university press. o’hara, k.l. (2002). the historical development of uneven-aged silviculture in north america. forestry, 75(4), 339–346. https://doi.org/10.1093/ forestry/75.4.339 orlovich, d.a., & cairney, j.g. (2004). ectomycorrhizal fungi in new zealand: current perspectives and future directions. new zealand journal of botany, 42(5), 721–738. https://doi.org/10.1080/002882 5x.2004.9512926 orr, e. (2017). keeping new zealand green: our forests and their future. wellington: steele roberts aotearoa. orwin, j. (2019). kauri: witness to a nation’s history (2nd edition). auckland: new holland publishers. palmer, j., & ogden, j. (1983). a dendrometer band study of the seasonal pattern of radial increment in kauri (agathis australis). new zealand journal of botany, 21(2), 121–125. https://doi.org/10.1080/002882 5x.1983.10428535 park, g. (2000). the ecology of the visit (nature in new zealand). landfall, 200, 23–34. perley, c. (1998). assessing timberlands’ sustainable beech management using concepts of ecosystem health and ecosystem management. new zealand forestry, 43, 3–7. perry, g.l.w., wilmshurst, j.m., & mcglone, m.s. (2014). ecology and long–term history of fire in new zealand. new zealand journal of ecology, 38(2), 157–176. pizzirani, s., monge, j.j., hall, p., steward, g.a., dowling, l., caskey, p., & mclaren, s.j. (2019). exploring forestry options with māori landowners: an economic assessment of radiata pine, rimu, and manuka. new zealand journal of forestry science, 49: 5. https://doi.org/10.33494/nzjfs492019x44x prebble, r. (1996). i’ve been thinking. auckland: seaview publishing. quinlan, p. (2011). existing uses and market development opportunities for naturally regenerating totara timber. sustainable farming fund project (l10/145) report for the northland totara working group, wellington, nz: ministry of agriculture and forestry. https://www.tanestrees. org.nz/site/assets/files/1100/existing_uses_and_ market_development_opportunities_for_naturally_ regenerating_totara_timber_full_project_report_ june_2011.pdf reay, s.d., & norton, d.a. (1999). assessing the success of restoration plantings in a temperate new zealand forest. restoration ecology, 7(3), 298–308. https:// doi.org/10.1046/j.1526-100x.1999.72023.x reid, a. (2001). impacts and effectiveness of logging bans in natural forests: new zealand. in p.b. durst, t.r. waggener, t. enters, & t.l. cheng (eds.), forests out of bounds: impacts and effectiveness of logging bans in natural forests in asia–pacific. bangkok, thailand: food and agricultural organisation of the united nations, asia–pacific forestry commission. richards, k. (1994). sustainable management of rimu forests – the timberlands west coast experience. new zealand journal of forestry, 39(2), 7–9. richardson, b., kimberley, m.o., ray, j.w., & coker, g.w. (1999). indices of interspecific plant competition for pinus radiata in the central north island of new zealand. canadian journal of forest research, 29(7), 898–905. https://doi.org/10.1139/x99-099 richardson, s.j., hurst, j.m., easdale, t.a., wiser, s.k., griffiths, a.d., & allen, r.b. (2011). diameter growth rates of beech (nothofagus) trees around new zealand. new zealand journal of forestry, 56(1), 3–11. richardson, s.j., holdaway, r.j., & carswell, f.e. (2014). evidence for arrested successional processes after fire in the waikare river catchment, te urewera. new zealand journal of ecology, 38(2), 221–229. roche, m. (1990). history of new zealand forestry. wellington: new zealand forestry corporation. roche, m. (1997). growth rates: debating pathways for new zealand forestry. in j. dargavel (ed.), australia’s everchanging forests iii (pp. 48–57). canberra: australian national university. roche, m. (2013). the royal commission on forestry 1913 viewed from 2013. new zealand journal of forestry, 58(1), 7. roche, m. (2017). forest governance and sustainability pathways in the absence of a comprehensive mcglone et al. new zealand journal of forestry science (2022) 52:8 page 23 https://doi.org/10.1080/0028825x.1981.10426397 https://doi.org/10.1080/0028825x.1981.10426397 https://doi.org/10.1093/forestry/75.4.339 https://doi.org/10.1093/forestry/75.4.339 https://doi.org/10.1080/0028825x.2004.9512926 https://doi.org/10.1080/0028825x.2004.9512926 https://doi.org/10.1080/0028825x.1983.10428535 https://doi.org/10.1080/0028825x.1983.10428535 https://doi.org/10.33494/nzjfs492019x44x https://doi.org/10.1046/j.1526-100x.1999.72023.x https://doi.org/10.1046/j.1526-100x.1999.72023.x https://doi.org/10.1139/x99-099 national forest policy–the case of new zealand. forest policy and economics, 77, 33–43. https:// doi.org/10.1016/j.forpol.2015.12.007 rogers, r. (1990). quercus alba l. white oak. silvics of north america, 2, 60–69. royal commission on forestry. (1913). appendix to the journals of the house of representatives, session i, c–12. runkle, j.r., stewart, g.h., & mcclenahen, j.r. (1997). temporal changes in height and diameter growth for two nothofagus species in new zealand. journal of vegetation science, 8(3), 437–446. https://doi. org/10.2307/3237335 sage, e. (1998). beech scheme – an ecological and economic fiasco. new zealand forestry, 43(3), 9–10. salmon, g. (1998). timberlands’ beech project deserves to succeed. new zealand forestry, 43(3), 8. salmon, j.t. (1960). heritage destroyed. the crisis in scenery preservation in new zealand. wellington: ah & aw reed. sander, i.l. (1990). quercus rubra l. northern red oak. silvics of north america, 2, 727–733. sando, c.t. (1936). notes on agathis australis. te kura ngahere, 4, 16–21. singh, r.p. (2001). foreword. in p. b. durst, t. r. waggener, t. enters, & t. l. cheng (eds.), forests out of bounds: impacts and effectiveness of logging bans in natural forests in asia–pacific (pp. 43). bangkok, thailand: food and agricultural organisation of the united nations, asia–pacific forestry commission six–dijkstra, h., mead, d., & james, i. (1985). forest architecture in terrace rimu forest of saltwater forest, south westland, and its implications for management. new zealand journal of forestry science, 15(1), 3–22. smale, m. (1984). white pine bush – an alluvial kahikatea (dacrycarpus dacrydioides) forest remnant, eastern bay of plenty, new zealand. new zealand journal of botany, 22(2), 201–206. https://doi.org/10.1080/ 0028825x.1984.10425252 smale, m., & kimberley, m. (1986). growth of naturally regenerated beilschmiedia tawa and podocarps in unlogged and selectively logged podocarp/tawa forest, pureora. new zealand journal of forestry science, 16(2), 131–141. smale, m., bathgate, j.l., & guest, r. (1986). current prospects for tawa. new zealand forestry, 31(1), 13–18. smale, m., beveridge, a., & herbert, j. (1998). selection silviculture trials in north island native forests: impacts on the residual forest and their implications for sustainable forest management. new zealand journal of forestry, 43(3), 19–29. smale, m.c., & beveridge, a.e. (2007). long–term ecological impacts of selective harvesting on a new zealand conifer–hardwood forest. new zealand journal of forestry science, 37(1), 3–22. smale, m.c., richardson, s.j., & hurst, j.m. (2014). diameter growth rates of tawa (beilschmiedia tawa) across the middle north island, new zealand– implications for sustainable forest management. new zealand journal of forestry science, 44: 20. https://doi.org/10.1186/s40490-014-0020-9 star, p. 2002. native forest and the rise of preservation in new zealand (1903–1913). environment and history, 8(3), 275–294. https://doi. org/10.3197/096734002129342675 steward, g.a., & beveridge, a.e. (2010). a review of new zealand kauri (agathis australis (d. don) lindl.): its ecology, history, growth and potential for management for timber. new zealand journal of forestry science, 40, 33–59. steward, g.a., kimberley, m.o., mason, e.g., & dungey, h.s. (2014). growth and productivity of new zealand kauri (agathis australis (d. don) lindl.) in planted forests. new zealand journal of forestry science, 44: 27. https://doi.org/10.1186/s40490-014-0027-2 steward, g.a.; quinlan, p. 2019. tōtara industry pilot project: a fresh look at a familiar northland species. new zealand tree grower november, 30–33. stewart, g.h., & white, j.c. (1995). tree ring analysis in rimu (dacrydium cupressinum): implications for studies of forest dynamics and sustained yield management. new zealand forestry, 40, 27–31. stewart, g.h., rose, a.b., & veblen, t.t. (1991). forest development in canopy gaps in old-growth beech (nothofagus) forests, new zealand. journal of vegetation science, 2(5), 679–690. https://doi. org/10.2307/3236178 sullivan, j.j., williams, p.a., & timmins, s.m. (2007). secondary forest succession differs through naturalised gorse and native kanuka near wellington and nelson. new zealand journal of ecology, 31(1), 22–38. sveding, a. (2019). providing guideline principles: botany and ecology within the state forest service of new zealand during the 1920s. international review of environmental history 5(1), 113–128. https://doi.org/10.22459/ireh.05.01.2019.07 szałek, b.z. (2013). some praxiological reflections on the so–called ‘overton window of political possibilities, mcglone et al. new zealand journal of forestry science (2022) 52:8 page 24 https://doi.org/10.1016/j.forpol.2015.12.007 https://doi.org/10.1016/j.forpol.2015.12.007 https://doi.org/10.2307/3237335 https://doi.org/10.2307/3237335 https://doi.org/10.1080/0028825x.1984.10425252 https://doi.org/10.1080/0028825x.1984.10425252 https://doi.org/10.1186/s40490-014-0020-9 https://doi.org/10.3197/096734002129342675 https://doi.org/10.3197/096734002129342675 https://doi.org/10.1186/s40490-014-0027-2 https://doi.org/10.2307/3236178 https://doi.org/10.2307/3236178 https://doi.org/10.22459/ireh.05.01.2019.07 ‘framing’ and related problems. reality of politics. estimates–comments–forecasts 4, 237–257. thompson, a.p. (1987). the national business review conservation debate. new zealand journal of forestry, 32(1), 7–11. tilling, a.j. (1988). multiple–use indigenous forestry on west coast of south island. new zealand forestry, 34(4), 13–18. tilling, a.j. (1992). indigenous forest management in new zealand: from interventionist to monetarist policies and the special case of the south island’s west coast. new zealand forestry, 38, 8–13. veblen, t.t., & stewart, g.h. (1982). on the conifer regeneration gap in new zealand – the dynamics of libocedrus bidwillii stands on south island journal of ecology, 70(2), 413–436. https://doi. org/10.2307/2259912 walters, r.s., & yawney, h.w. (1990). acer rubrum l. red maple. silvics of north america, 2, 605–613. wardle, j.a. (1984). the new zealand beeches: ecology, utilisation and management. christchurch, new zealand: new zealand forest service. wardle, j.a. (2011). wardle’s native trees of new zealand and their story. wellington, new zealand: new zealand farm forestry association. wardle, p. (1963). the regeneration gap of new zealand gymnosperms. new zealand journal of botany, 1, 301–315. https://doi.org/10.1080/002882 5x.1963.10429001 wardle, p. (1980). ecology and distribution of silver beech (nothofagus menziesii) in the paringa district, south westland, new zealand. new zealand journal of ecology, 23–36. wardle, p. (1985). environmental influences on the vegetation of new zealand. new zealand journal of botany, 23(4), 773–788. https://doi.org/10.1080/ 0028825x.1985.10434242 watson, j. (2017). the recovery of windblown timber from west coast public conservation land. new zealand journal of forestry, 62(2), 13–16. watt, m.s., kimberley, m.o., richardson, b., whitehead, d., & mason, e.g. (2004). testing a juvenile tree growth model sensitive to competition from weeds, using pinus radiata at two contrasting sites in new zealand. canadian journal of forest research, 34(10), 1985–1992. https://doi.org/10.1139/ x04-072 willems, n. (1999). forest structure and regeneration dynamics of podocarp/hardwood forest fragments, banks peninsula, new zealand. a thesis submitted in partial fulfilment of the requirements for the degree of master of applied science, lincoln university, new zealand. wilson, g.a. (1994). wood chipping of indigenous forest on private land in new zealand 1969–1993. australian geographical studies, 32(2), 256–273. h t t p s : / / d o i . o rg / 1 0 . 1 1 1 1 / j . 1 4 6 7 8 4 7 0 . 1 9 9 4 . tb00675.x wilson, g.a., & memon, p.a. (2005). indigenous forest management in 21st–century new zealand: towards a ‘postproductivist’ indigenous forest– farmland interface? environment and planning a, 37(8), 1493–1517. https://doi.org/10.1068/ a37144 wilson, g.a., & memon, p.a. (2010). the contested environmental governance of maori–owned native forests in south island, aotearoa/new zealand. land use policy, 27(4), 1197–1209. https://doi. org/10.1016/j.landusepol.2010.04.003 wiser, s.k., allen, r.b., benecke, u., baker, g., & peltzer, d. (2005). tree growth and mortality after small–group harvesting in new zealand old– growth nothofagus forests. canadian journal of forest research, 35(10), 2323–2331. https://doi. org/10.1139/x05-158 wiser, s.k., bellingham, p.j., & burrows, l.e. (2001). managing biodiversity information: development of new zealand's national vegetation survey databank. new zealand journal of ecology, 25(2), 1-17. wiser, s.k., baker, g., & benecke, u. (2007). regeneration of red and silver beech: how important is the size of harvested area? new zealand journal of forestry, 52(2), 31-35. yao, r.t., harrison, d.r., & harnet, m. (2017). the broader benefits provided by new zealand’s planted forests. new zealand journal of forestry, 61(4), 7–15. young, d. (2004). our islands, our selves: a history of conservation in new zealand. dunedin: otago university press. young, l.m., & norton, d.a. (2017). sustainable tōtara management and biodiversity conservation in northland. new zealand journal of forestry, 62(2), 23–25. zhang, y., bergeron, y., zhao, x.h., & drobyshev, i. (2015). stand history is more important than climate in controlling red maple (acer rubrum l.) growth at its northern distribution limit in western quebec, canada. journal of plant ecology, 8(4), 368–379. https://doi.org/10.1093/jpe/rtu029 mcglone et al. new zealand journal of forestry science (2022) 52:8 page 25 https://doi.org/10.1080/0028825x.1963.10429001 https://doi.org/10.1080/0028825x.1963.10429001 https://doi.org/10.1080/0028825x.1985.10434242 https://doi.org/10.1080/0028825x.1985.10434242 https://doi.org/10.1139/x04-072 https://doi.org/10.1139/x04-072 https://doi.org/10.1111/j.1467-8470.1994.tb00675.x https://doi.org/10.1111/j.1467-8470.1994.tb00675.x https://doi.org/10.1068/a37144 https://doi.org/10.1068/a37144 https://doi.org/10.1016/j.landusepol.2010.04.003 https://doi.org/10.1016/j.landusepol.2010.04.003 https://doi.org/10.1139/x05-158 https://doi.org/10.1139/x05-158 https://doi.org/10.1093/jpe/rtu029