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 die- back 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 within- subclass 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 cross- referenced 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 cross- referenced 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 multi- trait 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 die- back 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 select- parent 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 die- with 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 Diplodia- associated 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 genotype- environment 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 tA- F = 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 h2- h2- h~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 A- 1\. (}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 h2- h2 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 h2- h2 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 -- ... h2- f 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 h2- f 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# h2- f 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 # h2- f 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 ------------------------------------------------------------------------------------------------------------------------------- h2- c 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 h2- c -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 h2- f 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- "'1 ~I ~ I • v