ALCES VOL. 48, 2012 CUMBERLAND - AERIAL SURVEY fOR MOOSE 67 POTVIN DOUBLE-COUNT AERIAL SURVEYS IN NEW BRUNSWICK: ARE RESULTS RELIABLE FOR MOOSE? Roderick E. Cumberland New Brunswick Department of Natural Resources, Fish and Wildlife Branch, P.O. Box 6000, Fredericton, New Brunswick, Canada E3B 5H1. ABSTRACT: following the rapid decline of deer (Odocoileus virginianus) across northern New Brunswick in the late 1980s, the New Brunswick Department of Natural Resources began to utilize a double-count helicopter survey to estimate deer numbers. Although the survey was designed for deer, moose (Alces alces) sightings were also recorded; however, no analysis was conducted on the accuracy or usefulness of these data to estimate moose numbers. The survey design was a modification of the Potvin double-count survey method for deer which accounts for most caveats to aerial surveys. This double-count (mark-recapture) technique allows calculation of bias for both observers, for single and groups of moose, and individual flights. Moose population estimates calculated from 79 flights ranged from 0.17-3.49 moose/km2 and were similar to a variety of estimates throughout North America. Population estimates from 2004-2009 correlated well with corresponding 2009 population indices for moose based on number of moose seen by deer hunters (Corr. = 0.725, P <0.001). The Potvin estimates in Wildlife Management Zone 2 were highly correlated (0.82-0.93, P <0.05) with other indices based on road kill moose, moose sightings, and harvest success rates; moose sightings and hunter success were also correlated in several other zones. This analysis indicates that Potvin surveys produce reliable population density estimates of moose in boreal/Acadian forests, given that the sighting probability is >0.4 and flights occur before mid-February when moose may occupy denser canopy cover. ALCES VOL. 48: 67-77 (2012) Key words: Aerial survey, Alces alces, deer, density estimate, moose, New Brunswick. Sustainable and effective management of large ungulates requires a reliable estimate of population size by management unit. Gauging the success of annual management actions and harvest prescriptions hinges upon the ability to obtain cost-effective density esti- mates with precision that allows detection of annual changes. following a rapid decline of deer (Odocoileus virginianus) numbers across northern New Brunswick, the New Brunswick Department of Natural Resources (NBDNR) initiated the use of a double-count aerial survey in 1996 that was developed in Quebec to estimate deer densities (Rice and Harder 1977, Potvin et al. 1992, Hardy 1994, Potvin and Breton 1995). Although the sur- vey was designed and modified specifically for the observation of deer in boreal forests, observers also recorded sightings of moose (Alces alces). However, no further analysis was conducted on the reliability of this survey or how precise or accurate it was to estimate moose numbers. Various survey methods have been used to determine ungulate densities including line transects, distance sampling, spotlight surveys, and variable plot surveys (Kie 1988, Rabe et al. 2002). Other methods include track counts, pellet group surveys, and measure of browse abundance that provide indirect counts or indices (Kendall et al. 1992, Witmer 2005), and although cost-effective, are not without their challenges and inaccuracies (Storm et al. 1992, fritzen et al. 1995, focardi et al. 2002, freddy et al. 2004, Collier et al. 2007). Some of these methods have low power to detect population change of 20-50% (Strayer 1999) or produce estimates with large confidence AERIAL SURVEY fOR MOOSE - CUMBERLAND ALCES VOL. 48, 2012 68 limits (Jordan et al. 1993). Although more expensive, aerial surveys can provide quick, reliable estimates of ungulate density (freddy et. al. 2004); however, aerial surveys have their own set of bias and problems. Early in the evolution of aerial surveys, Caughley (1974) and Caughley et al. (1976) outlined deficiencies that included width of transect surveyed, cruising speed, altitude, height above ground, observer experience, and habitat cover type. While quadrat flights had the advantages of counting a higher percentage of animals and removing the need for strip width, transects were more robust and allowed for ease in calculating observer bias and were more economical (Caughley 1977). Nearly all of these biases result in fewer sightings and less precise estimates. Estimates obtained from aerial surveys improved as these problems were addressed (Peterson and Page 1993). The use of heli- copters addressed aircraft speed (Peterson and Page 1993) and improved the accuracy of moose estimates by 78%, on average, over fixed winged aircraft (Gosse et al. 2002). Beasom et al. (1981) tested sightability within the survey strip and found that observers saw 34-73% fewer deer in the outer 50 m of a 100 m survey strip compared to the first 50 m of the strip. DeYoung et al. (1989) found that they could improve density estimates with little or no loss in precision by reducing their strip width from 200 to 40 m. In Wyoming, 42% of moose missed occurred >50 m from the transect path (Anderson and Lindzey 1996), evidence that the search area not exceed this width. Aerial surveys are typically flown over open habitat types because dense overstories combined with oblique angles of view hide ungulates from observers (Anderson and Lidn- zey 1996). However, floyd et al. (1978) found that modifying search methodology greatly improved observability of deer in coniferous forests, such as altering the viewing angle (Potvin et al. 1992, Gauthier and Cumberland 2000). In addition, background snow greatly improved sightability to as high as 78% in an oak-hickory forest (Beringer et al. 1998). Because not all animals are observed in aerial surveys, several methods are used to correct for this bias, including calculating cor- rection factors by habitat type (Anderson and Lindzey 1996) or incorporating some form of mark-recapture using 2 observers, which has proved the simplest and most efficient means (Pollock and Kendall 1987). A double-count procedure is essentially a mark-recapture tech- nique that improves estimates by accounting for missed animals (Magnusson et. al. 1978, Eberhardt and Simmons 1987, Borchers et al. 1998). DeYoung et al. (1989) used cor- rection factors to improve their population under-estimate (42.3% less than their Bailey’s estimate) to within 6% of the estimated popu- lation size. further improvement to precision is possible by calculating visibility bias for both observers, and for both single and groups of deer (Samuel and Pollock 1981, Rivest et al. 1995). With such modifications, double- count aerial surveys that address these issues arguably hold promise for estimating ungulate density in coniferous-deciduous forests. Potvin deer estimates are considered reliable in New Brunswick. Deer densities estimated in WMZ 22 from 2000-2009 were strongly correlated to estimates derived from population reconstruction with the Provincial deer model. Estimated rates of increase and decrease for Potvin estimates, the deer model, and actual harvest changes in WMZ 22 were strongly correlated throughout the 2000s (NBDNR unpublished data). The objective of this study was to evaluate the reliability of Potvin estimates of moose population density in New Brunswick by using >10 years of aerial survey data from multiple Wildlife Manage- ment Zones (WMZ) of variable moose density across New Brunswick. ALCES VOL. 48, 2012 CUMBERLAND - AERIAL SURVEY fOR MOOSE 69 METHODS Study Area New Brunswick is located on the east coast of Canada at 46.00° N and 66.54° W. Over half of it borders the Atlantic Ocean along the Chaleur Bay, Northumberland Strait, and the Bay of fundy. The majority (>80%) of the landbase is forested, and elevation varies from lowlands at sea level with little relief along the eastern coast, to highlands of the Appalachian Mountains in the north to elevations of 764 m. Woodlands are characterized by Acadian for- est, with spruce (Picea spp.), balsam fir (Abies balsamea), hemlock (Tsuga canadensis), and white cedar (Thuja occidentalis) as climax species on softwood sites, and sugar maple (Acer saccharum), beech (Fagus grandifolia), and yellow birch (Betula alleghaniensis) characterizing mature hardwood sites. Early successional forests are predominated by poplars (Populus spp.), birches (Betula spp.), willows (Salix spp.), and cherry (Prunus spp.). New Brunswick is divided into 27 WMZs (fig. 1) to facilitate localized and site- specific management through seasonal and zone-specific quotas for antlerless deer and all sexes of moose. Field Techniques Our survey method (here- after referred to as Potvin) was modified slightly (Gau- thier and Cumberland 2000) from the double-count aerial survey used to estimate deer numbers by Potvin et al. (1992). We flew a 4-seat, Bell 206 Jet Ranger helicopter with a GPS (Global Position- ing System) navigation unit. The aircraft was equipped with side bubble windows on the left (port) side that al- lowed for a wide field of view to reduce parallax, and greatly reduced the effect of dense conifer cover on sightability of moose by reducing parallax angles <45° in the survey strip. The survey strip itself was limited to a 60 m swath and a flight altimeter was installed to maintain consistent height above undulating topography. flight speeds did not exceed 60 knots and aircraft height above the canopy was kept constant at 60 m. We found that slow speeds and low altitude of the survey aircraft usually initiated some form of movement of deer and moose below the helicopter and greatly assisted with ob- servability. The flight crew consisted of the pilot, a navigator-recorder, and 2 observers seated on the left side of the helicopter. Two observers were employed for the double-count estimator; observability bias was calculated for both observers, for single and groups of deer, and for each individual flight. Every effort was made to ensure that observers were experienced and remained the same within fig. 1. Distribution of Wildlife Management Zones (WMZ) in New Brunswick; WMZs 1-3, 6, and 10 were used to compare with moose population estimates in adjacent Maine. AERIAL SURVEY fOR MOOSE - CUMBERLAND ALCES VOL. 48, 2012 70 each annual survey. To approximate a true mark-recapture study, observers must be isolated from audio to prevent overhearing conversation that might bias sightings. The helicopter was equipped with a modified intercom system designed to allow the navigator-recorder to hear the observers while controlling their access to audio with a toggle switch. Although snow is not necessary for the survey, the color contrast created between deer, moose, and snow greatly assists observers and increases sightability. Therefore, we only conducted aerial surveys following snowfall that completely covered the ground. Because both deer and moose were counted, surveys were flown as long as snow occurred, but this was altered as we experienced behavioral change of both deer (yarding) and moose (restricted movement). Survey Block Selection Ballard et al. (1997) conducted flights of the Potvin double-count survey and suggested that the survey block size should be >200 km2 with transect lines spaced 1 km apart and of equal length to address spatial distribution of deer, and to meet statistical requirements. Therefore, we selected large survey blocks with gross habitat characteristics that matched the WMZ as a whole. Although the spatial distribution of habitat types could not be evalu- ated, proportions of habitat types were sum- marized for each WMZ using a GIS database. To make selection of survey blocks easier, GIS map tiles were used as components to define survey blocks and their boundaries. Because GIS tiles were 43 km2, combinations of 6 tiles (approximately 258 km2) in a matrix were selected in each zone (n = 20-40 per WMZ) as potential survey blocks; each was summa- rized by the 9 habitat types. Survey blocks which most closely matched the total zone habitat characteristics based on simple linear regression analysis were selected (adjusted r² values >85%) and subsequently evaluated for topographic characteristics. The block with topographical characteristics with the fewest obstructions (i.e., large lakes, dense housing, suburban areas) and that was most consistent with the habitat in the WMZ as a whole was selected. Transect lines were flown systemati- cally beginning at the edge of the block and spaced 1 km apart and oriented either N-S or E-W. Transects were approximately 40 km long to encompass as much of the variation in habitat as possible to ensure variation be- tween transects was minimized and precision maximized (Caughley et al. 1976). Moose in New Brunswick are currently managed by tracking various indices of popula- tion change through time. Indices across and within WMZs include the number of moose seen per hour by moose hunters, the number of moose seen by deer hunters, annual road kill in each WMZ, and success rate of moose hunters. Density estimates were compared to the relative index of moose seen per deer hunter in each zone to determine if these relative values were comparable. Changes in population and indices were determined by the slope of a regression line fitted to each index, and by measuring r by regressing log e of the estimated population size (Caughley and Birch 1971). These rates were compared between estimated population change by the aerial survey and other indices of change. In- dex values were correlated to Potvin estimates over time to determine if changes in moose indices were reflected by related changes in the Potvin estimates. Our estimated moose densities were also compared to estimates from surrounding jurisdictions and others in the literature. RESULTS The NBDNR flew 154 individual flights over various habitat types and in nearly all WMZs since 1996. Thirty-five (22.5%) were flown as test flights in February and March 1996 and 1997 to determine appropriate block size, alignment, and location to calibrate the survey and to form sight images for deer and ALCES VOL. 48, 2012 CUMBERLAND - AERIAL SURVEY fOR MOOSE 71 moose; therefore, these test flights were not included in the analysis. Since 1997, several flights were flown in February or later to es- timate deer numbers, and several others had sighting probabilities <0.40. These additional 14 flights were also excluded from the analy- sis because moose typically move into more coniferous cover in mid-late winter thereby reducing sightability; simulations indicate a loss of robustness when sighting probability is <0.45 (Magnusson et al. 1978, Potvin et al. 1992). Of the remaining 105 flights, 26 flights were excluded from analysis because they did not produce a population estimate due to too few moose. Our 1996-1997 test flights sug- gested that densities <0.5 per km² resulted in >80% of survey lines without sightings. As expected, central and southern WMZs with lower moose densities and harvests accounted for 17 of these 26 flights. Population estimates (unpublished data, NBDNR) from the remaining 79 flights indi- cated high moose density in northern WMZs where deer populations are substantially lower, and lower moose density in southern WMZs. Significant within-zone variability was only evident in 4 (5%) flights. Estimated moose densities varied from 3.49 moose/km² in northern zones (Table 1) to as low as 0.17 moose/km² in southern zones (Table 2). Aver- age sightability for front and rear observers was 0.85 and 0.83, respectively. Fewer than 8 WMZs were flown yearly due to financial and weather limitations; therefore, few zones have continuous data. Because Pot- vin estimates were never obtained in all zones in one year, the most recent Potvin estimate in each from 2004-2009 was compared with the 2009 index of moose seen by deer hunt- ers. These two values correlated well (corr 0.725, P <0.001), and better than most other indices with each other except the 2 moose sighting indices that were correlated (corr. 0.80, P <0.001). WMZ 2 was flown most consistently (1996-1998, 2000, 2003-2004, 2006, and 2008-2009) and provided the best data set for temporal comparisons of the Potvin esti- mates. The observed rate of increase (r) for Potvin estimates and all other moose indices in WMZ 2 were positive and of similar mag- nitude (Table 3). Potvin estimates were highly correlated (0.82-0.93; all P <0.05) with all other popula- WMZ Year # Moose Density ± SD 1 2008 24 1.39 ± 0.42 2 2009 52 3.49 ± 1.14 3 2006 16 1.18 ± 0.32 4 2008 10 0.62 ± 0.18 5 2008 42 2.66 ± 0.77 6 2009 27 1.70 ± 0.83 7 2007 20 1.28 ± 0.44 8 2008 13 0.82 ± 0.23 10 2009 11 0.69 ± 0.36 12 2009 9 0.54 ± NA. Table 1. Potvin estimates of moose population density (moose/km²) in northern WMZs (higher density WMZs) in New Brunswick, 2006-2009; WMZs 1-3, 6, and 10 border Maine. A WMZ was measured in November or December (see Methods). WMZ Year Moose Density ± SD 13 2004 18 1.07 ± 0.32 14 2006 17 1.13 ± 0.34 15 2004 29 1.79 ± 0.44 16 2007 7 0.48 ± 0.15 17 2003 9 0.66 ± NA 18 2007 21 1.37 ± NA 20 2007 11 0.68 ± 0.29 21 2009 43 1.29 ± 0.23 22 2009 3 0.17 ± NA 23 1999 4 0.25 ± 0.13 25 2009 10 0.72 ± 0.26 Table 2. Potvin estimates of moose population density (moose/km²) in southern WMZs (lower moose density WMZs) in New Brunswick, 2003- 2009. A WMZ was measured in December or January (see Methods). AERIAL SURVEY fOR MOOSE - CUMBERLAND ALCES VOL. 48, 2012 72 tion indices in WMZ 2 (i.e., road kill, moose sightings, and harvest success rate; fig. 2). Potvin estimates were also highly correlated to moose sightings (corr. 0.90, P <0.001) and hunter success (corr. 0.79, P <0.011) in WMZ 8. Although fewer data points were available for Potvin estimates in WMZ 4 and 5, they were strongly correlated with moose sightings (corr. 0.84, P <0.001) and hunter success (corr. 0.75, P <0.001) in WMZ 4, and to a lesser extent in WMZ 5 (corr. 0.61, P <0.082 and corr. 0.46, P <0.21, respectively). DISCUSSION Thirteen years of deer and moose popula- tion estimates have been done with the Potvin double-count survey in New Brunswick. The survey was designed to address limitations of aerial surveys by flying at slow speed (60 kph) and at low altitude (60 m) to increase the likelihood of deer and moose movement that, in turn, increases the likelihood of sightability. Specifically, bubble windows reduced paral- lax and vastly improved the ability to sight ungulates under closed coniferous canopies. The flight altimeter allowed the pilot to main- tain survey height and reduce fluctuations in survey width that can affect population estimates. Use of a survey transect width of 60 m also reduced parallax, providing easier observation of a narrower strip, including the portion most likely to have observable ani- mals. Using 2 observers in the double-count allowed use of the mark-recapture technique that vastly improves estimates, and calcula- tion of observer bias for each flight as well as singles and groups of animals from which the population estimate is corrected. Behavioral and ecological differences of ungulate species may introduce bias and affect the accuracy and use of aerial surveys. for example, McCorquodale (2001) found that the spatial distribution of male and female elk during winter was a potential source of bias in helicopter surveys because females were 9 times more likely to be observed than males. However, Anderson and Lindzey (1996) successfully observed 59% of moose groups over a wide range of habitat types with strip widths of 150-250 m (3-5 x larger than here). Given the size and color of moose, and that Potvin estimates were found reliable for deer under similar protocols and conditions, it was presumed that moose had equal or higher sightability than deer, and that bias was negligible. Because moose shift from deciduous to conifer habitats as snow depth increases (Coady 1974, Peek et al. 1976), some caution against flying in late winter (February-April) when moose tend to occupy denser vegetation (Lynch 1975, Karns 1982, Gasaway et al. 1985, Crete et al. 1986, Anderson and Lindzey 1996, Ballard et al. 1997). The most optimal time to survey moose is when they occupy more open habitat types (Anderson and Lindzey 1996); thus, most aerial surveys are flown late fall and into early winter when snow covers the ground and moose occupy relatively open habitat types (Coady 1974, Lynch 1975, Peek et al. 1976, Karns 1982, Gasaway et al. 1985, Crete et al. 1986, Anderson and Lindzey 1996, Quayle et al. 2001). Quayle et al. (2001) found that % vegetative cover and snow cover, as well as temperature, affected sightability of moose in British Columbia. Our estimates of moose density are comparable with those in adjacent Maine (L. Kantar, Maine Department of Inland fisheries and Wildlife, pers. comm.) where the range of density estimates along the New Brunswick Metric Estimated r r² adj. doubling time Potvin 0.176 84.9 3.9 Years Harvest 0.226 70.8 3.1 Years Road kill 0.276 77.3 2.5 Years Success 0.358 82.0 1.9 Years Sightings by MH 0.364 65.9 1.9 Years Table 3. The estimated rate of increase (r) and the corresponding doubling time for the population (Caughley and Birch 1971) for moose population indices in WMZ 2, New Brunswick. ALCES VOL. 48, 2012 CUMBERLAND - AERIAL SURVEY fOR MOOSE 73 border is 1.24-3.05 moose/ km²; density in bordering WMZs 1, 2, and 6 were 1.39- 3.49 (Table 1). In nearby New Hampshire, the density estimate was 1.19 moose/ km² using the fLIR method (Bontaites et al. 2000). Population density of heav- ily harvested populations or those in poor habitat are typically <1.0 moose/km², but in less-exploited situa- tions, density can approach 3.0/km², and in extreme cases be as high as 5-10/km². Collectively, these other estimates indicate that the Potvin estimates reported in this study (0.17-3.49 moose/ km²) are reasonable. Due to cost and weather conditions, <10 of 25 WMZs were flown in any given year. Further, consecutive flights within WMZs were sporadic due to the need for specific data from other zones in successive years. While this limits the continuous nature of the data set, sufficient data existed in several zones to make comparisons to other moose population indices in New Brunswick. I com- pared the 2006-2009 Potvin estimates to the number of moose observed by deer hunters because this index was more robust than the observation rate by moose hunters. Moose observations by 3,500 moose hunters dur- ing the 3-day season was low in some WMZs, versus the number of moose observed 0 0.5 1 1.5 2 2.5 3 3.5 4 0 5 10 15 20 25 30 35 40 45 50 Po tv in d en si ty (m oo se /k m 2 ) Number of road killed moose y = 0.0679x + 0.1764 r2 adj. = 0.774, P <0.05 0 0.5 1 1.5 2 2.5 3 3.5 4 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 Po tv in d en si ty (m oo se /k m 2 ) Moose hunter success rate y = 8.1442x - 4.7078 r2 adj. = 0.766, P <0.05 0 0.5 1 1.5 2 2.5 3 3.5 4 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Po tv in d en si ty (m oo se /k m 2 ) # moose sighted/hour hunted y = 10.358x - 0.0384 r2 adj. = 0.859, P <0.05 fig. 2. Relationships between Potvin estimates (moose/km²) and 3 indices of moose population density in WMZ 2 in New Brunswick, 1996-2009: a) road kill, b) moose hunter success rate, and c) moose sightings per hour by moose hunters. Each index was positively correlated (P <0.05) with the Potvin estimates. AERIAL SURVEY fOR MOOSE - CUMBERLAND ALCES VOL. 48, 2012 74 by 50,000 deer hunters over 4 weeks. The Potvin estimates correlated well with the relative number of moose observed by deer hunters across WMZs. Although the absolute change in the Pot- vin estimates for WMZ 2 increased over the course of our study from 1077 to 8731, the estimated rate of increase was less than that calculated for the other indices of population change (i.e., roadkill, hunter sightings, hunter success, and harvest; Table 3). The Potvin estimates were highly correlated to changes in hunter success and moose sightings in northern zones where a sufficient number of flights allowed for this analysis. Although the Potvin estimates correlated well with the other indices of moose density, 9 flights did not produce a density estimate in areas where the moose population was pre- sumably high enough to provide an estimate. Further, 4 of the 79 flights where estimates were obtained had a fluctuation >50% from the previous or consecutive estimate. Pre- liminary analysis shows little evidence that this was attributed to weather or temperature anomalies. The question remains whether such estimates should be excluded from an annual population analysis, or be treated as a low population and managed accordingly; current analysis has not formed the basis for management decisions. Rivest et al. (1995) analyzed visibility bias of the Potvin survey and found that double- count estimates were superior to those derived from independent sightability studies such as Gasaway and other single survey types. Potvin estimators also have the advantage of reflecting the prevailing conditions during the actual survey. Potvin and Breton (2005) tested the double-count method simultaneously with infrared surveys and found that the thermal infrared technique provided more variable results; accuracy was 54-89% and was mostly influenced by forest canopy. Conversely, 4 of 6 Potvin surveys yielded reliable results and the density estimates were within 64-83% of the assumed densities based on population reconstruction. Provided sighting probabilities are >0.45, the double count method provides valid estimates of deer density for management purposes; however, if sighting probabilities are <0.40 it might underestimate density (Potvin et al. 2004). Moose in New Brunswick are currently managed by tracking various indices of popula- tion change through time, although none have been tested or validated with actual population sizes or rates of change. The Potvin method could be a valuable tool in producing more reliable moose population estimates in boreal/ Acadian forests provided sighting probability is >0.40 and flights occur prior to February. Because unreported harvest of moose by first Nation’s people in New Brunswick creates a void in age data which limits the ability to implement population reconstruction (e.g., Paloheimo and fraser model [1981]), early winter Potvin estimates might provide both a population estimate and a basis for under- standing any population response associated with unreported harvest. In New Brunswick, deer and moose data were collected simultaneously and such an approach could be more cost-effective than tra- ditional, single species surveys in jurisdictions with multiple species of ungulates. However, at northern latitudes where deer tend to occupy closed-canopy mature coniferous forest in winter, the timing of favorable weather, pres- ence of snow, and yarding migration greatly narrows the window when combined surveys could occur. Although the Potvin estimates were limited to <10 WMZs flown annually and few zones had continuous annual data, the population estimates were reasonable and correlated with other population indices. An- nual flights may not be necessary to calibrate other modeling and population reconstruction efforts, and this technique may also prove advantageous in high density areas requiring local management strategies. ALCES VOL. 48, 2012 CUMBERLAND - AERIAL SURVEY fOR MOOSE 75 REFERENCES Anderson, C. r. Jr., and F. G. Lindzey. 1996. Moose sightability model developed from helicopter surveys. Wildlife Society Bul- letin 24: 247-259. BALLArd, W. B., J. A. KershAw, H. A. whit- LAw, D. L. sABine, G. J. ForBes, and S. younG. 1997. A Preliminary Evaluation of the Aerial Double Count Technique for Estimating White-tailed Deer Densities in New Brunswick. Unpublished Report. University of New Brunswick, frederic- ton, New Brunswick, Canada. BeAsom, S. L., J. C. hood, and J. R. CAin. 1981. The effect of strip width on heli- copter censusing of deer. Journal of Range Management 34:36-37. BerinGer, J., L. P. hAnsen, and O. sexton. 1998. Detection rates of white-tailed deer with a helicopter over snow. Wildlife Society Bulletin 26: 24-28. BontAites, K. M., K. A. GustAFson, and R. mAKin. 2000. A Gasaway-type moose survey in New Hampshire using infrared thermal imagery: preliminary results. Alces 36: 69-76. BorChers, D. L., W. ziCChini, and R. M. Fewster. 1998. Mark-recapture models for line transect surveys. Biometrics 54: 1207-1220. CAuGhLey, G. 1974. Interpretation of age ratios. Journal of Wildlife Management 38: 557-562. _____. 1977. Analysis of Vertebrate Popula- tions. John Wiley and Sons Ltd., New York, New York, USA. _____, and L. BirCh. 1971. Rate of increase. Journal of Wildlife Management 35: 658-663. _____, R. sinCLAir, and D. sCott-Kemmis. 1976. Experiments in aerial survey. Journal of Wildlife Management 40: 290-300. CoAdy, J. W. 1974. Influence of snow on behaviour of moose. Naturaliste Canadian 101: 417-436. CoLLier, B. A., S. S. ditChKoFF, J. B. rAGLin, and J. M. smith. 2007. Detection prob- ability and source of variation in white- tailed deer spotlight surveys. Journal of Wildlife Management 71: 277-281. Crete, M., L. rivest, H. JoLiCoeur, J. BrAs- sArd, and f. messier. 1986. Predicting and correcting helicopter counts of moose with observations made with fixed-wing aircraft in southern Quebec. Journal of Applied Ecology 23: 751-761. deyounG, C. A., f. S. Guthrey, S. L. BeAsom, S. P. CouGhLAn, and J. R. heFFeLFinGer. 1989. Improving estimates of white-tailed deer from helicopter surveys. Wildlife Society Bulletin 17: 275-279. eBerhArdt, L. L., and M. A. simmons. 1987. Caliberating population indices by double sampling. Journal of Wildlife Manage- ment 51: 665-675. FLoyd, T. J., L. D. meCh, and M. E. neLson. 1978. An improved method of census- ing deer in deciduous-coniferous forests. Journal of Wildlife Management 43: 258-261. FoCArdi, S., R. isotti, and A. tineLLi. 2002. Line transect estimates of ungulate popula- tions in a Mediterranean forest. Journal of Wildlife Management 66: 48-58. Freddy, D. J., G. C. white, M. C. KKeeLAnd, R. H. KAhn, J. W. unsworth, W. J. deverGie, V. K. GrAhAm, J. H. eLLenBerGer, and C. H. wAGner. 2004. How many mule deer are there? Challenges of credibility in Colorado. Wildlife Society Bulletin 32: 916-927. Fritzen, D. E., R. f. LABisKi, D. E. eAston, and J. C. KiLGo. 1995. Nocturnal movements of white-tailed deer: implications for re- finement of track count surveys. Wildlife Society Bulletin 23: 187-193. GAsAwAy, W. C., S. D. DuBois, and S. J. hArBo. 1985. Biases in aerial transect surveys for moose during May and June. Journal of Wildlife Management 49: 777-784. GAuthier, P., and R. E. CumBerLAnd. 2000. A AERIAL SURVEY fOR MOOSE - CUMBERLAND ALCES VOL. 48, 2012 76 New Brunswick manual for aerial surveys of white-tailed deer populations utilizing the Potvin mark-recapture technique. Deer Technical Report No. 5. New Bruns- wick Department of Natural Resources, fredericton, New Brunswick, Canada. Gosse, J., B. mCLAren, and E. eBerhArdt. 2002. Comparison of fixed-wing and helicopter searches for moose in a mid- winter habitat-based survey. Alces 38: 47-53. hArdy, T. 1994. Methods of estimating white-tailed deer densities and population trends - a review. Senior Thesis, faculty of forestry. University of New Brunswick, fredericton, New Brunswick, Canada. JordAn, P. A., R. O. Peterson, P. CAmPBeLL, and B. mCLAren. 1993. Comparison of pellet counts and aerial counts for esti- mating density of moose at Isle Royale: a progress report. Alces 29: 267-278 KArns, P. D. 1982. Twenty-plus years of aerial moose census in Minnesota. Alces 18: 186-196. KendALL, K. C., L. H. metzGAr, D. A. PAt- terson, and B. M. steeLe. 1992. Power of sign surveys to monitor population trends. Ecological Applications 2: 422-430. Kie, J. G. 1988. Performance in wild ungu- lates: Measuring population density and condition of individuals. USDA forest Service General Technical Report PSW- 106. Pacific Southwest forest and Range Experiment Station, Berkeley, California, USA. LynCh, G. M. 1975. Best timing of moose surveys in Alberta. Alces 11: 141-153. mAGnusson, W. E., G. J. CAuGhLey, and G. C. GriGG. 1978. A double survey esti- mate of population size from incomplete counts. Journal of Wildlife Management 42: 174-176. mCCorquodALe, S. M. 2001. Sex-specific bias in helicopter surveys of elk: sight- ability and dispersion effects. Journal of Wildlife Management 65: 216-225. PALoheimo, J. E., and D. FrAser. 1981. Es- timation of harvest rate and vulnerability from age and sex data. Journal of Wildlife Management 45: 948-958. PeeK, J. M., D. L. uuriCh, and R. J. mACKie. 1976. Moose habitat selection and rela- tionships to forest management in north- eastern Minnesota. Wildlife Monographs 48: 3-65. Peterson, R. O., and R. E. PAGe. 1993. Detec- tion of moose in midwinter from fixed- wing aircraft over dense forest cover. Wildlife Society Bulletin 21: 80-86. PoLLoCK, K. f., and W. L. KendALL. 1987. Visibility bias in aerial surveys: a review of estimation procedures. Journal of Wildlife Management 51: 502-510. Potvin, f., and L. Breton. 1995. Standard for Aerial Surveys of White-Tailed Deer Populations. Quebec Department of En- vironment and Wildlife, Rene-Levesque, Quebec, Canada. _____, and _____. 2005. from the field: test- ing 2 aerial survey techniques on deer in fenced enclosures – visual double-counts and thermal infrared sensing. Wildlife Society Bulletin 33: 317-325. _____, _____, and L-P. rivest. 2004. Aerial surveys for white-tailed deer with the double-count technique in Quebec: two 5-year plans completed. Wildlife Society Bulletin 32: 1099-1107. _____, _____, _____, and A. GinGrAs. 1992. Application of a double-count aerial survey technique for white-tailed deer, Odocoileus virginianus, on Anticosti Is- land, Quebec. Canadian field-Naturalist 106: 435-442. quAyLe, J. f., A. G. mAChutChon, and D. N. Jury. 2001. Modeling moose sightability in south-central British Columbia. Alces 37: 43-54. rABe, M. J., S. S. rosenstoCK, and J. C. devos, Jr. 2002. Review of big game survey methods used by wildlife agencies of the western United States. Wildlife Society ALCES VOL. 48, 2012 CUMBERLAND - AERIAL SURVEY fOR MOOSE 77 Bulletin 30: 46-52. riCe, W. R., and J. D. hArder. 1977. Ap- plication of multiple aerial sampling to a mark-recapture census of white-tailed deer. Journal of Wildlife Management 41: 197-206. rivest, L-P., f. Potvin, H. CrePeAu, and G. dAiGLe. 1995. Statistical methods for aerial surveys using the double-count technique to correct visibility bias. Bio- metrics 51: 461-470. sAmueL, M. D., and K. H. PoLLoCK. 1981. Cor- rection of visibility bias in aerial surveys where animals occur in groups. Journal of Wildlife Management 45: 993-997. storm, G. L., D. f. CottAm, R. H. yAhner, and J. D. niChoLs. 1992. A comparison of 2 techniques for estimating deer density. Wildlife Society Bulletin 20: 197-203. strAyer, D. L. 1999. Statistical power of presence-absence data to detect popula- tion declines. Conservation Biology 13: 1034-1038. witmer, G. W. 2005. Wildlife population monitoring: some practical consider- ations. Wildlife Research 32: 259-263.