F:\ALCES\Vol_38\Pagemaker\3818. ALCES VOL. 38, 2002 COURTOIS ET AL. - HABITAT SELECTION IN CUT AREAS 177 HABITAT SELECTION BY MOOSE (ALCES ALCES) IN CLEAR-CUT LANDSCAPES Réhaume Courtois1, Christian Dussault2, François Potvin1, and Gaétan Daigle3 1 Société de la faune et des parcs du Québec, Direction de la recherche sur la faune, 675 Boulevard René-Lévesque Est, 11e étage, boîte 92, Québec, PQ, Canada G1R 5V7; 2 Université Laval, Département de Biologie, Cité Universitaire, Québec, PQ, Canada G1K 7P4; 3 Université Laval, Service de consultation statistique, Cité Universitaire, Québec, PQ, Canada G1K 7P4 ABSTRACT: Habitat selection by moose was studied over 4 years in two large sectors subject to intensive forest harvesting using a two-scale approach. At the coarser scale, i.e. location of the home range within the landscape, habitat selection did not appear to be influenced by the presence of clear-cuts. In one sector, moose preferred mature mixed stands, young coniferous, and mature coniferous stands. In the second sector, the highest preference was noted for cut areas and mature deciduous stands. Moose home ranges were located in areas with higher edge and interspersion among habitat patches. Home range size for females was positively related to the proportion of cuts, but movements were not. Habitat selection was more pronounced at the finer scale (animal locations within home range) and did not differ between sectors. Mixed stands were preferred in all seasons. Mature conifer stands were preferred in summer and in early winter while young conifer stands were preferred in late winter. Clear-cuts were avoided except in early winter. Moose were located in areas closer to edge between food and cover stands than were random locations, especially in late winter. A marked decrease in movements also was noted in late winter. This study shows differences in habitat selection pattern between the coarser and finer scales. For example, clear-cuts did not seem to markedly influence home range location at a coarser scale, and adaptations to minimize their impact seemed to operate at a finer scale. Coarser scale habitat selection was probably linked to a trade-off between predator avoidance and browse availability, whereas seasonal changes suggest behavioural adaptations of moose to maximize energy gain and counteract predation and other adverse environmental conditions at the finer scale. ALCES VOL. 38: 177-192 (2002) Key words: Alces alces, clear-cuts, cover, food, forest management, habitat selection, mortality RESUMÉ: La sélection d’habitat par l’orignal a été étudiée durant quatre ans dans deux grands secteurs comportant d’importantes coupes forestières. Deux échelles d’analyse ont été retenues. À l’échelle brute, correspondant à l’emplacement du domaine vital dans le paysage, la sélection d’habitat ne paraissait pas influencée par la présence de coupes totales. Dans l’un des secteurs, l’orignal préférait les peuplements mélangés matures ainsi que les résineux jeunes et mûrs. Dans le deuxième secteur, la préférence la plus grande était notée pour les coupes forestières, les feuillus matures et les mélangés matures. Les domaines vitaux de l’orignal étaient localisés dans des sites comportant plus de bordure et plus d’entremêlement entre les parcelles d’habitat. Chez les femelles, la superficie du domaine vital était positivement liée à l’importance des coupes, mais les déplacements ne l’étaient pas. La sélection d’habitat était plus prononcée à l’échelle plus fine (localisation des animaux à l’intérieur du domaine vital) et elle ne différait pas entre les deux secteurs. Les peuplements mélangés étaient préférés en toutes saisons. Les résineux matures l’étaient en été et en début d’hiver alors que les jeunes résineux étaient préférés en fin d’hiver. Les coupes étaient évitées, sauf en début d’hiver. Les localisations d’orignaux étaient situées plus près des bordures entre des peuplements d’alimentation et de couvert que ne l’étaient des localisations aléatoires. Une diminution marquée des déplacements était aussi notée en fin d’hiver. Cette étude montre que les préférences d’habitat diffèrent entre les échelles brute et fine. Ainsi, les coupes forestières ne HABITAT SELECTION IN CUT AREAS - COURTOIS ET AL. ALCES VOL. 38, 2002 178 Johnson (1980) suggested that habitat selection, i.e. the decision to choose a spe- cific habitat, is a hierarchical process, with decisions being made at different spatial and temporal scales. For example, an ani- mal can choose to establish its home range (HR) in an area dominated by dense conifer stands while preferring rare deciduous stands within its HR to fulfil nutritional require- ments. At a coarser scale, animals are expected to select habitats that reduce main limiting factors such as predation (Rettie and Messier 2000). When animals have successfully overcome the more important limiting factors at the coarser scale, selec- tion could diverge at a finer scale to meet more specific needs, for example adequate shelter during the calving period, food in- take in summer, and protection against un- favourable environmental conditions in late winter. Moose (Alces alces) thrive in young boreal forests originating from perturbations (Timmermann and McNicol 1988). The highest densities are recorded in mixed stands (Brassard et al. 1974, Joyal 1987, Crête 1988) and in areas disturbed by forest fires, insect outbreaks, forest harvesting, or windfall (Krefting 1974, Peek et al. 1976, Timmermann and McNicol 1988, Loranger et al. 1991). Such stands provide abundant deciduous shrubs that constitute the main food source for moose (Crête 1989). The benefits of forest harvesting for moose, however, are restricted to the 15-40 year period following disturbance (Cowan et al. 1950, Crête 1977, Franzmann and Schwartz 1985). Over the short-term, clear- cutting leads to landscapes dominated by large openings separated by small strips of uncut forest. Such recently clear-cut land- scapes seem much less favourable to moose than older cuts since previous studies have indicated low moose densities in these ar- eas (Eason et al. 1981, Girard and Joyal 1984, Eason 1989). Increased moose mor- tality caused by hunting and predation fol- lowing forest harvesting were suggested to explain these low densities (Eason et al. 1981, Girard and Joyal 1984, Dalton 1989, Eason 1989) and as shown by Rempel et al. (1997). Consequently, cutovers seem less favourable for moose survival. Whether or not moose avoid these areas in order to increase their survival and at which scale such an avoidance could occur is unknown. In this study, we used radio-telemetry to evaluate habitat selection by moose in an area subject to intensive forest harvesting during the last 10-15 years. We used a two- scale approach to test (1) if moose avoid landscapes with clear-cuts, and (2) if they avoid clear-cuts and edges within their HR. At a coarser scale (HR within the land- scape), we expected moose to select habi- tats that are believed to decrease their vulnerability to predation and hunting (Eason et al. 1981, Girard and Joyal 1984, Dalton semblaient pas influencer grandement l’emplacement des domaines vitaux à l’échelle brute, et les adaptations visant à diminuer leur impact semblaient opérer à fine échelle. À l’échelle brute, la sélection d’habitat semblait résulter d’un compromis entre l’évitement des prédateurs et la recherche de nourriture alors qu’à l’échelle fine, les changements saisonniers suggéraient des adaptations comportementales visant à maximiser les gains énergétiques et à contrer les effets de la prédation et des conditions environnementales adverses. ALCES VOL. 38: 177-192 (2002) Mots clés: Alces alces, aménagement forestier, coupes totales, couvert, mortalité, nourriture, sélection d’habitat ALCES VOL. 38, 2002 COURTOIS ET AL. - HABITAT SELECTION IN CUT AREAS 179 1989, Eason 1989). At that scale moose s h o u l d e x h i b i t l o w p r e f e r e n c e (disproportional use, Hall et al. 1997) for areas dominated by clear-cuts, and should increase their HR size and movements in cut areas as suitable habitats are more scattered. At a finer scale (within HR), we expected that habitat use would be oriented toward maximizing energy gains and would vary seasonally depending on the severity of environmental conditions. Habitat re- quirements are more easily fulfilled during the growing season and in early winter when high quality food is abundant and snow (< 60 cm) does not restrict movement. However, during the growing season, fe- males must protect their calves from preda- tion by wolves (Canis lupus) and black bears (Ursus americanus) (Ballard 1992). Consequently, in summer and early winter, moose should then look for stands offering both ample food supply to maximize nutri- tional intake (Schwartz 1992), and cover to keep predation risk at a moderate level (Edenius 1992), such a trade-off being more pronounced in females than males. Moose are sensitive to heat stress in late winter and their movements are hampered by snow depth and hardness (Timmermann and McNicol 1988). During this period, moose should reduce their daily movements, prefer cover to food supply, and be located closer to edges between food and cover than in other seasons (Peek et al. 1992). METHODS Study Area The study was conducted in a 3,200 km2 area located in northwestern Québec, Canada (Fig. 1). The study area is a typical boreal forest dominated by conifer stands that were intensively harvested (22% of the Fig. 1. Location of the study area in northwestern Québec (78° 40’ W, 47° 50’ N). Dots indicate sites of capture of marked animals. HABITAT SELECTION IN CUT AREAS - COURTOIS ET AL. ALCES VOL. 38, 2002 180 area) between mid-1980 and 1994. At the end of the study, 31% conifer, 13% mixed, and 8% deciduous stands covered the area. Other habitat classes available include stands currently unproductive for forestry (12%, mostly alder, Alnus rugosa), conifer stands a f f e c t e d b y t h e s p r u c e b u d w o r m (Choristoneura fumiferana) outbreak (5%) during the 1980s, and water bodies or lakes (9%). Dominant trees are black spruce (Picea mariana), balsam fir (Abies balsamea), jack pine (Pinus banksiana), paper birch (Betula papyrifera), and trem- bling aspen (Populus tremuloides). Clear- cutting without protecting advanced regen- eration was practiced in the study area 7 to 11 years (94% of all cuts) before the study. In areas cut between 1992 and 1994, feller- bunchers and skidders circulated in trails spaced 10-15 m apart to protect advanced regeneration. The terrain is gently rolling with hills rarely exceeding 350 m above sea level. Temperature (mean ± SE (n years)), as measured at the nearby Belleterre mete- orological station, was -16.2°C ± 0.6 (29) in January, and 17.3°C ± 0.2 (29) in July. Annual precipitation was 1013 mm ± 23 (24) including 291 mm ± 12 (26) that fell as snow. Maximum snow depth (66 cm ± 7 (17)) occurs in February and did not exceed 90 cm during the study. Wolf and black bear are found in the study area at approxi- mately 1 and 14 individuals / 100 km2 re- spectively (Lamontagne et al. 1999, Larivière et al. 2000). Capture and Telemetry Sixty-five moose (1991: 29; 1992: 16; and 1993: 20) were equipped with radio transmitters. Track networks in the snow were identified by aerial surveys, numbered and randomly selected to determine the collaring sequence. Males (14), females (37), and calves (14) were collared accord- ing to their proportion in the population of the hunting zone in which the study was conducted (Paré and Courtois 1990). All animals present in a chosen track network were collared provided they corresponded to the desired category. Moose were im- mobilized with succinylcholine chloride (Van Ballenberghe 1989, Delvaux et al. 1999), at a dosage of 5-6 mg / 100 kg of visually estimated body weight. Moose were equipped with LMRT-4 VHF transmitters (Lotek Engineering, Newmarket, Ontario) and ear tagged. To estimate habitat use and movements, moose were located from an aircraft (helicopter: 96%; fixed wing: 4%) in 3 seasons: summer and fall (15 April to 31 October), early winter (mid-December to late January: snow depth < 60 cm), and late winter (mid-February-end of March: snow depth > 60 cm). Each year, marked animals were located 3-4 times per season. Transmitters included motionless sen- sors that allowed the identification of dead individuals. For dead animals, the cause of death was determined by inspection of the carcass and its surrounding area. Predation was presumed when the presence of wolves was obvious around the carcass. Animals killed by hunters were reported to manda- tory registration stations. Home Range Estimates The minimum convex polygon (Eddy 1977) was retained to estimate HR size due to its widespread use in the literature (Aebischer et al. 1993). Post-capture outliers (distance > 10 km in < 3 days post- capture) were discarded to eliminate their effect on HR size, leaving 2,160 valid loca- tions (1991: 297 locations; 1992: 626; 1993: 732; and 1994: 505). Most animals were followed for 2-3 years, yielding 32.6 loca- tions per animal (SE = 2.1; n = 65 moose). The plot of multi-annual HR size against the number of locations per animal indicated an inflection point at 20 locations per animal, as noted by Courtois et al. (1998a). For moose ALCES VOL. 38, 2002 COURTOIS ET AL. - HABITAT SELECTION IN CUT AREAS 181 collared at 0.5-year-old, only locations col- lected after they were 18 months old were used in analyses to avoid the influence of dispersing animals (Labonté et al. 1998) and to prevent pseudo-replication with the data of their mothers, which were also monitored. Consequently, analyses were conducted only for the 47 moose (11 males and 36 females > 1.5 year-old) having > 20 locations. Multi-annual HR sizes were cal- culated in ArcView GIS 3.1 (ESRI 1996) using the Animal Movement extension de- veloped by Hooge and Eichenlaub (1997). Habitat Composition We used the term “habitat classes” to designate forest stands and unproductive areas identified on forest maps. Habitat composition was obtained from digitized 1:20,000 forest maps produced by interpre- tation of 1:15,000 aerial photographs (MER 1984). The minimum size of stands on these maps is usually 8 ha. This information was used to produce a 10 m × 10 m raster map imported into ArcView 3.1 and managed with the Spatial Analyst extension (ESRI 1996). To maximize the statistical power of comparisons (Alldredge and Ratti 1992) and to avoid having missing values in the availability data sets (Aebischer et al. 1993), habitats were grouped into eight classes according to their age (young: 20-60 years; mature: >60 years) and composition (Y_DEC, M_DEC, Y_MIX, M_MIX, Y_CON, M_CON, re- spectively, young and mature deciduous, mixed and coniferous stands; CUTS: clear- cuts < 11 years; OTHER: water (mainly) and unproductive stands). All cuts were grouped because those harvested between 1992 and 1994 were not available to all animals. Such grouping of habitat classes provides good correspondence between the forest maps and ground survey validations (Dussault et al. 2001). It also reflects the basic require- ments of moose in terms of food (browse, leaves, and aquatics) and cover (heat stress, snow) (Thompson and Stewart 1998). CUTS can potentially provide summer and winter food as well as Y_DEC and M_DEC which could also offer summer cover against heat; OTHER was expected to give access to aquat- ics; Y_MIX and M_MIX could offer both food and cover all year long while Y_CON and M_CON were supposed to provide summer and winter cover. One habitat map per year (1 April to 31 March) was produced to take into account the chronology of cutting op- erations. Habitat composition of a given area (i.e., study area, HR) was computed as the percentage of that area covered by each habitat class. Habitat Structure Habitat structure at the coaser scale was estimated with 7 Fragstats landscape pattern indices (McGarigal and Marks 1995) computed with the ArcView extension Patch Analyst (Elkie et al. 1999). We selected two indices to measure edge (the length of the interface) between habitat patches (ED: edge density, perimeter of all habitat patches per unit area [m/ha]; CWED: contrast-weighted edge density [m/ha], edge between decidu- ous or mixed vs. coniferous stands [weights = 1; other edges = 0]). One metric (CAD: core area density) was selected to estimate the density of core areas per 100 ha. Core area was defined as the interior of the patch exclusive of a band 100 m wide immediately inside the patch boundary. The size of the buffer was based on observations indicating that moose overused the 100 m zone on each side of food-cover edges (Dussault 2002). Two indices measured the diversity of the landscape (SDI: Shannon’s diversity index, amount of information per habitat patch [without units]; IJI: interspersion and juxtaposition index, measure the extent to which patch types are equally adjacent to each other [percent]). Two other indices were selected to quantify the shape and size of stands (MSI: mean shape index, average HABITAT SELECTION IN CUT AREAS - COURTOIS ET AL. ALCES VOL. 38, 2002 182 perimeter-to-area ratio [without units]; TCAI: total core area index, relative importance (%) of the core areas). Finally, an index of edge preference was calculated as the mean distance (m) between moose locations and the nearest edge between deciduous or mixed vs. coniferous stand. Data Analysis Habitat components were ranked in order of preference, rather than classified as preferred or avoided, because prefer- ence and avoidance depend markedly upon the array of components assumed to be available (Johnson 1980). In all analyses, the level of rejection of the null hypothesis was set at α= 0.05. Conclusions drawn from habitat selec- tion studies are critically dependent upon t h e d e l i n e a t i o n o f a v a i l a b l e h a b i t a t (Aebischer et al. 1993, McClean et al. 1998, Wilson et al. 1998). To ensure that habitats deemed available were really accessible to each animal, we delimited two sectors (Rapide-Sept, 1,065 km2, in the south east- ern part of the study area; Rouyn-Noranda, 1,184 km2, in the north western) using 100% convex polygons joining all the locations of moose frequenting each sector (Fig. 1). Each of these sectors were 13-15 times the size of the mean HR size, which seems sufficient to provide a precise estimate of availability outside a given HR. At the same time such an area remains potentially ac- cessible to moose. For example, dispersal of yearlings is usually 10-30 km and some animals can travel up to 100 km before settling down (Labonté et al. 1998). Pref- erence analyses (percent used/percent available) were conducted using the animal as the sample unit (Aebischer et al. 1993). Zero values in percent use were replaced by a small value (0.01%) about one order of magnitude smaller than the minimum en- countered in the data set. Analyses were conducted at two scales to identify: (1) preferences revealed by the multiannual HR location within the land- scape (coarser scale); and (2) seasonal preferences within the HR (finer scale). At the coarser scale, habitat within the HR was defined as use while availability corre- sponded to habitat composition of the sector where each animal was collared. To test the influence of landscape pattern indices, each moose HR was paired with 30 areas of identical size and shape (random HR) ran- domly positioned (x and y coordinates of the centroid) and oriented (0-360 degrees of rotation from the observed orientation) in the sector frequented by that animal (Wilson et al. 1998, Potvin et al. 2001, Dussault 2002). The use of random HR facilitated the comparison of landscape pattern indices because it eliminates the influence of size and shape on landscape pattern indices of HR under comparison (McGarigal and Marks 1995). Preference at the coarser scale was studied using the 1994 habitat map. However, this will overestimate moose preference towards cuts since some stands that were frequented before cutting will appear to be cut in the analysis. Any conclusion that cutovers are avoided will therefore be conservative. At the finer scale, habitat composition in a 100-m buffer zone around each moose location was compared to that of the total HR of that animal. The buffer minimized the influence of spatial imprecision of habi- tat map and telemetry locations (Rettie and McLoughlin 1999). The size of the buffer was relatively small because locations were precise (20-100 m; Potvin 1998) due to the use of a helicopter that permitted seeing most animals (69%) during telemetry loca- tions. Habitat use was estimated using the annual forest map corresponding to each location. Data collected during a given season (summer, early winter, and late win- ter) and year were pooled per animal to avoid temporal pseudo-replication (thus giv- ALCES VOL. 38, 2002 COURTOIS ET AL. - HABITAT SELECTION IN CUT AREAS 183 ing one record per animal per season of each year). Edge preference (distance [m] from an edge) was evaluated by comparing moose locations with paired random loca- tions (one per moose location) within the HR of each animal. Paired locations were retained to assure that habitat characteris- tics at the random locations were accessi- ble to the animal under consideration. Preference analyses (Aebischer et al. 1993) were preceeded by multivariate analy- ses of variance (Proc GLM, SAS Institute 1989) to evaluate the effect of independent factors (year, sector, sex, season, and their interactions) on habitat preference for all habitat classes taken simultaneously. In these analyses, percentage use and avail- ability were transformed into log-ratios us- ing OTHER habitats as the denominator to render habitat variables independent (Aebischer et al. 1993). Multivariate nor- mality of the residuals was assessed with Mardia skewness and kurtosis (program multnorm.sas, available from SAS Insti- tute). Potential outliers were identified with the program OUTLIER.SAS (M. Friendly, http:/ / w w w . m a t h . y o r k u . c a / S C S / s s s g / outlier.html). Predicted preference indices were estimated by a posteriori univariate analysis (Proc MIXED, SAS Institute 1989) followed by a bootstrap procedure (avail- able at gaetan.daigle@mat.ulaval.ca). The bootstrap technique with 500 samples was used in order to approximate the exact distribution of the estimation procedure, conditional on the observed data (Efron and Tibshirani 1993). Bootstrap technique al- lows estimating the statistic of interest with many subsamples drawn from the original data with replacement. The associated distribution of the estimates is an approxi- mation of its true distribution, and thus can be used in statistical analyses. Univariate analyses (Proc GLM) were used to test the effect of independent fac- tors on HR size, female productivity, and distance between successive locations. We used the same factors as in the multivariate analyses but time interval between two successive locations was added in analyses involving distance. The normality and homoscedasticity of the residuals were evalu- ated respectively with the Shapiro-Wilk’s test and the plot of the residuals against predicted values. Paired t-tests were used to compare (1) landscape pattern indices between moose HR and the mean value of the 30 paired random HRs, and (2) the distance from an edge between moose and random locations. This last analysis was followed by an univariate analysis to assess the influence of independent factors on that variable. To test the influence of clear-cuts on moose mortality, animals were partitioned into 2 groups according to the proportion of cuts within their HR (few: < 35%; many: > 35%; based on the mean proportion found in HR). Proc LIFETEST (SAS Institute 1989) was then used to compare equality of survival curves over these two groups using the non-parametric log-rank test. RESULTS Habitat Preference at the Coarser Scale At the coarser scale, two outliers were removed to respect the multinormality of the residuals. No transformation of the data permitted reaching multinormality with these two outliers. We detected a sector effect in habitat preference (F[ 7,35 ] = 9.25, P <0.0001) but no sex (F[ 7,35 ] = 0.31, P = 0.9461) or sector*sex effect (F[ 7,35 ] = 1.25, P = 0.3009). Consequently, preferences were estimated by sector. Important differences were noted between the two sectors (Fig. 2). In the Rapide-Sept sector, the highest preference was noted for mature mixed stands followed by mature coniferous and young coniferous. The rank order for all habitat classes was: M_MIX > M_CON = Y_CON > M_DEC = CUTS = Y_MIX = OTHER > Y_DEC. In HABITAT SELECTION IN CUT AREAS - COURTOIS ET AL. ALCES VOL. 38, 2002 184 the Rouyn-Noranda sector, the most pre- ferred habitat classes were clear-cuts fol- lowed by mature deciduous, and all other habitat classes (CUTS > M_DEC > M_MIX = OTHER = M_CON = Y_CON = Y_DEC = Y_MIX). Six of the seven habitat pattern indices differed between moose and random HR (Table 1). Moose HR had higher total edge and number of core areas per surface area, higher interspersion and diversity (IJI and SDI), and lower perimeter-to-area ratio, and core areas size than random HR within the landscape. The difference between moose and random HR was particularly important for the number of core areas (7.1 vs. 5.1 core areas/ha) and total edge (81 vs. 69 m/ ha). Other significant differences were low (<9%). Habitat Preference at the Finer Scale Habitat preference within the HR var- ied among seasons (F[ 14,566 ] = 2.24, P < 0.0059) without any sector (F[ 7,37 ] = 1.80, P < 0.1156), sex (F[ 7,37 ] = 0.50, P < 0.8266) or year (F[ 21,813 ] = 1.43, P < 0.0960) influence, and no significant interactions. Preference analyses indicated that mature mixed and coniferous stands, and mature deciduous stands, were the most preferred habitat classes in summer and fall (M_MIX = M_CON > M_DEC > Y_CON = OTHER > CUTS = Y_DEC = Y_MIX). In early winter, moose preferred mature mixed stands, and cuts and mature coniferous (M_MIX > CUTS = M_CON > M_DEC > Y_CON = OTHER = Y_DEC = Y_MIX). In late winter, only M_MIX and Y_CON were pre- ferred over other habitat classes (M_MIX = Y_CON > M_CON = Y_MIX = Y_DEC = OTHER = M_DEC = CUTS). Globally, the highest pref- erences were noted for mature mixed in all seasons (Fig. 3). The major seasonal changes were an important increase in the preference for CUTS in early winter, and Y_CON in late winter. Young and mature deciduous, and OTHER habitats were more preferred in summer-fall than in the two other seasons. Moose were located closer to edge between deciduous or mixed stands and coniferous stands (food-cover edge) than random locations within the HR (mean dif- ference = 52 ± 16 m; n = 297; t = 4.63, P < 0.001; Fig. 4). Changes were noted among seasons (F[ 2,296 ] = 9.39, P = 0.0001) but not between sex (F[ 1,296 ] = 0.55, P = 0.4615), sectors (F[ 1,296 ] = 1.91, P = 0.1736), year Fig. 2. Habitat preference (mean ± SE) by moose at the coarse scale in the two sectors of the study area in northwestern Québec, 1991-1994. Preference indices (% used/% available) were standard- ized to sum to unity. 0.00 0.05 0.10 0.15 0.20 0.25 Y_DEC M_DEC Y_MIX M_MIX Y_CON M_CON CUTS OTHER P re fe re n c e i n d e x Rapide-Sept Rouyn-Noranda ALCES VOL. 38, 2002 COURTOIS ET AL. - HABITAT SELECTION IN CUT AREAS 185 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Y_DEC M_DEC Y_MIX M_MIX Y_CON M_CON CUTS OTHER P re fe re n c e i n d e x Summer-fall Early winter Late winter (F[ 3,296 ] = 2.31, P = 0.0767), or any interac- tions among factors. The difference was greater in late winter (80.9 m) than in the two other seasons (30.0 and 33.5 m). Moose were closer to an edge in late winter (254.2 ± 22.9 m, n = 118 moose-season) than in summer-fall (279.4 ± 14.5, 120) or in early winter (282.1 ± 17.9, 124). Effects of Clear-Cuts on Space use, Mortality and Productivity Mean multi-annual HR size was related to sex (87.7 km2 ± 20.3 and 73.7 ± 10.9 for males and females respectively; F[ 1,44 ] = 4.68, P = 0.037) and the interaction between sex and the proportion of cuts within the HR (F[ 1,44 ] = 9.23, P = 0.004); HR size in- creased with the proportion of cuts for females (r = 0.44, P = 0.009, n = 34) (Fig. 5) but not for males. However, the proportion of cuts within HR, or sex, sector, and inter- actions among these variables did not influ- ence the mean distance travelled between two successive locations (F[ 4,44 ] = 1.38, P = 0.2573). Movements depended on season (F[ 2,346 ] = 32.59, P < 0.001), interaction Table 1. Mean value ± SE of seven landscape pattern indices in moose (n = 47) and random (n = 30 per moose) home ranges in northwestern Québec, 1991-1994. Paired Landscape pattern indices Moose HR Random HR Student’s t P ED: total edge (m/ha) 80.6 ± 1.5 68.8 ± 0.8 7.40 <0.0001 CWED: food-cover edge (m/ha) 16.3 ± 0.7 16.5 ± 0.5 -2.27 0.7874 CAD: n core areas beyond 100 m/100 ha 7.1 ± 0.3 5.1 ± 0.1 7.17 <0.0001 MSI: perimeter to area ratio (without units) 1.84± 0.01 1.92 ± 0.01 -6.19 <0.0001 TCAI: importance (%) of core areas beyond 100 m 88.1 ± 0.2 89.6 ± 0.1 -6.80 <0.0001 JI: interspersion of habitat patches (%) 80.0 ± 1.2 75.5 ± 0.7 3.81 <0.0004 SDI: Shannon’s diversity index (without units) 1.75 ± 0.03 1.61 ± 0.02 4.94 <0.0001 Fig. 3. Seasonal changes in habitat preference (mean ± SE) within the home range by moose in northwestern Québec, 1991-1994. Preference indices (% used/% available) were standardized to sum to unity. HABITAT SELECTION IN CUT AREAS - COURTOIS ET AL. ALCES VOL. 38, 2002 186 between year and season (F[ 6,346 ] = 2.48, P= 0.024), sector*year (F[ 2,346 ] = 5.06, P= 0.0069), and time interval between loca- tions (F[ 1,146 ] = 3.99, P= 0.047). Mean distance travelled by moose was lower in late winter during all the years of the study (Fig. 6). Twenty-three collared moose died dur- ing the study from hunting (17), predation (4), and other causes (2). Survival rate of adults was similar for both sexes (χ2 = 1.84, df = 1, P = 0.1746; males: 83.2% ± 5.8 [12,364 collar-days]; and females: 86.9% ± 3.3 [33,340 collar-days]). Similarly, no sig- nificant difference was found between sur- vival for moose with many or few cuts in their HR (χ2 = 0.50, df = 1, P = 0.4812). Female productivity depended on the sector (F[ 1,34 ] = 5.29, P = 0.0284) but not on the proportion of cuts within the HR (F[ 1,34 ] = 0.56, P = 0.4615), and no interaction was detected. DISCUSSION Johnson (1980) suggested a four-order scale to explain habitat selection by an animal: the geographic range of the species, the HR, the specific habitats within the HR, and the micro-habitat (e.g., feeding sites). Further, Rettie and Messier (2000) sug- gested that differences in selection pattern among the scales could reflect the impor- tance of limiting factors with animals es- caping the most important limiting factors at the coarser scales. Habitat Selection at the Coarser and Finer Scales Habitat selection patterns of moose dif- fered between the coarser and finer scales in an area subject to intensive forest har- vesting, suggesting that moose resource selection at the level of HR and within HR may reside in different domains (sensu Wiens 1989). However, our results do not support the hypothesis that moose avoid clear-cuts at the coarser scale. In the Rapide-Sept sector, moose preferred forested habitats over clear-cuts, but the reverse trend was noted in the Rouyn-Noranda sector. Fe- males increased HR size, but not move- ments, in the presence of cuts. Differences in habitat preference be- Fig. 4. Distance (mean ± SE) between moose locations or random locations and the proxi- mal limit of the nearest edge between decidu- ous or mixed stands vs. coniferous stands in northwestern Québec, 1991-1994. Fig. 6. Seasonal changes in distance (mean ± SE) travelled by moose between two successive locations in northwestern Québec, 1991-1994. 0 50 100 150 200 250 300 350 400 450 Summer-fall Early winter Late winter D is ta n c e f ro m a f o o d -c o v e r e d g e ( m ) Moose Random R² = 0.19 0 50 100 150 200 250 0 10 20 30 40 50 60 % of HR cut H R s iz e ( k m ²) 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 1991 1992 1993 1994 D is ta n c e t ra v e ll e d ( k m ) Summer-fall Early winter Late winter Fig. 5. Influence of the proportion of cuts within the home range (HR) on female moose HR size in northwestern Québec, 1991-1994. ALCES VOL. 38, 2002 COURTOIS ET AL. - HABITAT SELECTION IN CUT AREAS 187 tween sectors could not be related to differ- ent availability since both sectors comprised similar proportions of clear-cuts (24.4 vs. 23.6%) of similar age (≤ 11-year-old). The use of the 1994 habitat map and multi- annual HR could have over-represented the proportion of clear-cuts within moose HR, but the same situation prevailed for both sectors. Moreover, annual HRs were ex- amined and the general trend was a propen- sity to stay in the same area despite the presence of clear-cuts. Only 3 of the 47 animals seemed to gradually shift HR fol- lowing the progression of the cuts. Preference at the finer scale was more pronounced than at the landscape scale. In both sectors, a marked preference was noted for mature over young stands, princi- pally mixed and coniferous ones. In early winter, however, there was an increase in preference for cuts. Mixed stands and regenerated clear-cuts could provide sig- nificant browse availability (Courtois et al. 1998b). During the summer-fall period, there was an increased preference for young and mature deciduous, and for OTHER habitat class. Deciduous trees can provide food and thermal cover while unproductive areas and water give access to aquatics. Preference for cover slightly increased in late winter. According to Thompson and Euler (1987), in cut areas, moose preferably choose young cutovers in early winter be- fore moving to older ones, and finally leave clear-cuts to use undisturbed areas in late winter. During that period, between mid- March and mid-April, they also reduce their movements (Courtois and Crête 1988) prob- ably because snow accumulation > 65 cm and crust impede movements (Peek et al. 1976). Also, moose are in poorer physical condition in late winter and can be heat stressed when temperatures reach -5 to 0 ° C ( R e n e c k e r a n d H u d s o n 1 9 8 6 , Timmermann and McNicol 1988). Many authors have concluded that moose seek coniferous shelter in late winter (Desmeules 1964, Brusnyk 1983, Thompson and Euler 1987), whereas Crête (1988) doubted its necessity when deciduous browse is abun- dant. Our results showed that moose re- duced their movements during that period, used cover in a greater proportion, and clear-cuts in a lesser proportion, than in summer or in early winter, but they still preferred mixed stands. Consequently, we suggest that dense cover is not a major component of late winter habitat, at least in regions where snow depth is usually < 90 cm as in our study area (Paré and Courtois 1990). Similar findings were made in an agro-forested site and in black spruce clear- cuts near our study area (Girard and Joyal 1984, Joyal and Bourque 1986) and also on the North Shore of the St. Lawrence River (Courtois et al. 1993). Moose decrease movements in deep snow but do not neces- sarily confine themselves under dense co- niferous cover stands. For example, they can take advantage of small conifer patches within mixed stands. Forbes and Theberge (1993) also noted differences in habitat preference at differ- ent scales in Algonquin Park in Ontario. At a regional scale (>1,000 km2), moose pre- ferred sites with at least 33% of their area disturbed by spruce budworm epidemics or cuts (mainly uniform shelterwood and se- lection cutting). At a finer scale (<100 km2), they preferred conifer stands, mostly hemlock (Tsuga canadensis), presumably because their study area was dominated by deciduous stands providing low cover. In our case, mature mixed stands were the most preferred habitat class both at the coarser and the finer scales in all seasons. Moreover, moose searched for diversified landscapes with high edge, interspersion and number of cores areas per surface area, and a lower size of core areas relative to habitat patch size. Consequently, prefer- ence at the coarser scale could result from HABITAT SELECTION IN CUT AREAS - COURTOIS ET AL. ALCES VOL. 38, 2002 188 a trade-off between predator avoidance and access to ample food supply (Dussault 2002). At the finer scale, habitat selection seemed to be oriented towards maximizing energy gains and counterbalancing adverse effects of the environment (Dussault 2002). Presumed Relationship Between Habi- tat Selection and Limiting Factors We expected that moose would avoid clear-cut areas at the coarser scale be- cause they are considered more vulnerable to hunting and predation in these environ- ments (Eason et al. 1981, Girard and Joyal 1984, Dalton 1989, Eason 1989). However, clear-cuts were not avoided in either sector, and mortality and productivity were not related to the abundance of clear-cuts. Establishment of HR within the land- scape may be more tightly related to socio- logical behaviour, namely philopatry for ju- venile females and dispersal for males (Cederlund et al. 1987, Labonté et al. 1998) than habitat constraints. However, adopt- ing philopatry as opposed to a more flexible strategy overcoming the influence of the factors affecting survival and reproduction could not be evolutionarily stable. Habitat selection probably has some portion of ge- netic programming. Our results suggest that avoidance of clear-cuts acts at the finer scale rather than at the coarser scale. It is also possible that moose may have tried to reduce the impact of several limiting factors at both scales. For example, it would not necessarily be a good strategy for a moose to choose a home range with very adverse environmental conditions even if there is no predator in that area. Our approach of coarser scale habitat selection presupposes a top-down driven process due to limiting factors such as pre- dation (Rettie and Messier 2000). An alter- native hypothesis may be a bottom-up proc- ess where the establishment of the home range is the result of daily decisions in locating food patches and encounters with conspecifics and potential predators, like wolves and hunters, along with other vari- ables such as daily and seasonal weather patterns. Our data could support this sec- ond interpretation since analyses demon- strate equivocal results for clear-cut avoid- ance at the coarser scale, and the fact that moose are unlikely to explore large land- scapes before selecting a home range. Behavioural changes that could dimin- ish the impact of clear-cuts seemed to oper- ate mainly at the finer scale, i.e., within the HR. Many studies have shown that moose, and particularly females with calves, avoid open habitats (Peek et al. 1987, Courtois and Crête 1988, Dalton 1989, Eason 1989, Hundertmark et al. 1990, Dussault 2002). In late winter, females with calves rarely go 60 m beyond the forest cover, whereas other moose browse up to 80 m from the forest fringe (Thompson and Euler 1987). Remaining close to cover could reduce pre- dation risk. At the finer scale, cuts were one of the less preferred habitat classes, except in early winter. Moose preferred the forested part of their HR which prob- ably contributed to minimizing the influence of clear-cuts on mortality. It is not known whether the impact of clear-cuts depends on moose density. At a higher density and particularly near to carrying capacity, the impact of habitat modifications may have been higher. In such situations, more ani- mals would have been forced to frequent marginal habitats and would have been ex- posed to higher mortality risks (Sinclair and Arcese 1995). MANAGEMENT IMPLICATIONS Our work adds more evidence that moose and forest management can co-ex- ist. However, recent clear-cuts are avoided except in early winter. The type of stands and the magnitude of forest harvesting de- termine the impact of logging (Joyal 1987). ALCES VOL. 38, 2002 COURTOIS ET AL. - HABITAT SELECTION IN CUT AREAS 189 Limited - diameter cuttings carried out in shade-tolerant hardwood stands are usually favourable to moose (Crête 1977). Open- ings created by such forest harvesting tech- niques, in which usually 40% of the basal area of the deciduous trees and 75% of the conifers are removed, rapidly stimulate the growth of browse. In shade-intolerant hard- wood, > 75% of the basal area of deciduous and conifers is usually removed. In conifer- ous stands where all commercial trees (> 9 cm) are harvested; the influence of cuts is more pronounced and depends on the time lag between cutting and regeneration of browse species and re-establishment of cover. An acceptable threshold for the propor- tion of cuts in the landscape can be esti- mated from the examination of the HR composition. In this study, moose HR com- prised on average 30-40% of mixed, de- ciduous and spruce budworm-affected stands, 30-35% of coniferous stands, 20- 25% of clear-cut areas, and 10-15% of other habitats. In northeastern Minnesota, Peek et al. (1976) suggested creating land- scapes comprising 40-50% of cutovers and only 5-15% of conifers in order to favour moose. The spatial organization of the land- scape has been studied on some occasions. Considering the area of track networks seen during aerial surveys, residual blocks < 20-50 ha should be unattractive to moose (Courtois et al. 1998b). Eason (1989) sug- gested 70 ha as the minimum size for an attractive block of forest, with observed moose densities increasing with the size of residual blocks. Courtois et al. (1998b) and Potvin et al. (1999) suggested alternating 50-100 ha cut and leave blocks. Based on landscape pattern indices, this study shows that small cuts regularly interspersed with residual stands should be beneficial. ACKNOWLEDGEMENTS We would like to thank Michel Crête from Société de la Faune et des Parcs du Québec (FAPAQ), Jean-Pierre Ouellet and Luc Sirois from Université du Québec à Rimouski (UQAR), and two anonymous referees for providing helpful comments on earlier versions of this paper. Alain Caron of UQAR assisted in the preparation of habitat maps. We are particularly indebted to Aldée Beaumont, Laurier Breton, and André Gaudreault from FAPAQ for their assistance in field work. REFERENCES AEBISCHER, N.J., P.A. ROBERTSON, and R.E KENWARD. 1993. Compositional analy- sis of habitat use from animal radio- tracking data. Ecology 74:1313-1325. ALLDREDGE, J.R., and J.T. RATTI. 1992. Further comparison of some statistical techniques for analysis of resource se- lection. Journal of Wildlife Manage- ment 56:1-9. BALLARD, W.B. 1992. Bear predation on moose: a review of recent North Ameri- can studies and their management im- plication. Alces Supplement 1:162-176. BRASSARD, J.M., E. AUDY, M. CRÊTE, and P. GRENIER. 1974. Distribution and winter habitat of moose in Québec. Naturaliste Canadien 101:67-80. BRUSNYK, L.M., and F.F GILBERT. 1983. Use of shoreline timber reserves by moose. Journal of Wildlife Manage- ment 47:673-685. CEDERLUND, G., F. SANDEGREN, and K. LARSSON. 1987. Summer movements of female moose and dispersal of their offspring. Journal of Wildlife Manage- ment 51:342-352. CO U R T O I S , R., and M. CR Ê T E. 1988. Déplacements quotidiens et domaines vitaux des orignales du sud-ouest du Québec. Alces 24:78-89. , , and F. BARNARD. 1993. HABITAT SELECTION IN CUT AREAS - COURTOIS ET AL. ALCES VOL. 38, 2002 190 Productivité de l’habitat et dynamique d’une population d’orignaux du sud de la taïga québécoise.Publication 2144. Ministère du Loisir, de la Chasse et de la Pêche du Québec, Québec, Canada. , J. LABONTÉ, and J.-P. OUELLET. 1998a. Déplacements et superficie du domaine vital de l’orignal, Alces alces, dans l’est du Québec. Canadian Field- Naturalist 112:602-610. , J.-P. OUELLET, and B. GAGNÉ. 1998b. Characteristics of cutovers used by moose (Alces alces) in early winter. Alces 34:201-211. COWAN, I. MCT., W.S. HOAR, and J. HAT- TER. 1950. The effect of forest succes- sion upon the quantity and nutritive val- ues of woody plants used as food by moose. Canadian Journal of Research 28:249-271. CRÊTE, M. 1977. Importance de la coupe forestière sur l’habitat hivernal de l’orignal dans le sud-ouest du Québec. Canadian Journal of Forest Research 7:241-257. . 1988. Forestry practices in Québec and Ontario in relation to moose popu- lation dynamics. Forestry Chronicle 64:246-250. . 1989. Approximation of K carry- ing capacity for moose in eastern Québec. Canadian Journal of Zoology 67:373-380. DALTON, W.J. 1989. Use by moose (Alces alces) of clear-cut habitat where 100% or 50% of the production forest was logged. COFDRA Project 32001. On- tario Ministry of Natural Resources, Toronto, Canada. DELVAUX, H., R. COURTOIS, L. BRETON, and R. PATENAUDE. 1999. Relative effi- ciency of succinylcholine, xylazine, and carfentanil/xylazine mixtures to immo- bilize free-ranging moose. Journal of Wildlife Diseases 35:38-48. DESMEULES, P. 1964. The influence of snow on the behavior of moose. Report 3. Ministère du Tourisme, de la Chasse et de la Pêche du Québec, Québec, Canada. DU S S A U L T , C. 2002. Influence des contraintes environnementales sur la sélection de l’habitat de l’orignal (Alces alces). Ph.D. Thesis. Université Laval, Québec, Canada. , R. COURTOIS, J. HUOT, and J.-P. OUELLET. 2001. The use of forest maps for the description of wildlife habitats: limits and recommendations. Canadian Journal of Forest Research 31:1227- 1234. EASON, G. 1989. Moose response to hunting and 1-km2 block cutting. Alces 25:63- 74. , E. THOMAS, R. JERRARD, and K. OSWALD. 1981. Moose hunting closure in a recently logged area. Alces 17:111- 125. EDDY, W.F. 1977. A new convex algorithm for planer sets from ACM. Transac- tions on Mathematical Software 3:398- 403. EDENIUS, L. 1992. Moose feeding in relation to position of food plants. Alces Supple- ment 1: 132-138. EFRON, B., and R.J. TIBSHIRANI. 1993. An introduction to the bootstrap. Chapman and Hall, New York, New York, USA. ELKIE, P.R., R.R. REMPEL, and A.P. CARR. 1999. Patch analyst user’s manual. Publication TM-002. Northwest Science and Technology, Ontario Ministry of Natural Resources, Thunder Bay, On- tario, Canada. (ESRI) ENVIRONMENTAL SYSTEMS RESEARCH INSTITUTE INC. 1996. ArcView GIS. The geographic systems for everyone. Environmental Systems Research In- stitute Inc., Redlands, California, USA. FORBES, G.J., and J.B. THEBERGE. 1993. Multiple landscape scales and distribu- tion of moose, Alces alces, in a forest ALCES VOL. 38, 2002 COURTOIS ET AL. - HABITAT SELECTION IN CUT AREAS 191 ecotone. Canadian Field-Naturalist 107:201-207. FRANZMANN, A.W., and C.C. SCHWARTZ. 1985. Twinning rates: a possible popu- lation condition assessment. Journal of Wildlife Management 49:394-396. GIRARD, F., and R. JOYAL. 1984. L’impact des coupes à blanc mécanisées sur l’orignal dans le nord-ouest du Québec. Alces 20:40-53. HALL, L.S., P.R. KRAUSMAN, and M.L. MORRISON. 1997. The habitat concept and the plea for standard terminology. Wildlife Society Bulletin 25:173-182. HOOGE, P.N., and B. EICHENLAUB. 1997. A n i m a l m o v e m e n t e x t e n s i o n t o ArcView. Ver. 1.1. Alaska Biological Science Center, U.S. Geological Sur- vey. Anchorage, Alaska. HUNDERTMARK, K.J., W.L. EBERHARDT, and R.E. BALL. 1990. Winter habitat use by moose in southeastern Alaska: implica- tions for forest management. Alces 26:108-114. JOHNSON, D.H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65-71. JOYAL, R. 1987. Moose habitat investiga- tions in Québec and management impli- cations. Swedish Wildlife Research Supplement 1: 139-152. , and C. BOURQUE. 1986. Varia- tions, selon la progression de l’hiver, dans le choix de l’habitat et du régime a l i m e n t a i r e c h e z t r o i s g r o u p e s d’orignaux (Alces alces) en milieu agro- forestier. Canadian Journal of Zoology 64:1475-1481. KREFTING, L.W. 1974. Moose distribution and habitat selection in north central North America. Naturaliste Canadien 101:81-100. LABONTÉ, J., J.-P. OUELLET, R. COURTOIS, and F. BÉLISLE. 1998. Moose dispersal and its influence in the maintenance of harvested populations. Journal of Wild- life Management 62:225-235. LAMONTAGNE, G., H. JOLICOEUR, and R. LAFOND. 1999. Plan de gestion de l’ours noir, 1998-2002 Publication 3960. Société de la faune et des parcs du Québec, Québec, Canada. LARIVIÈRE, S., H. J OLICOEUR, and M. CRÊTE. 2000. Status and conservation of the gray wolf (Canis lupus) in wildlife re- serves of Québec. Biological Conser- vation 94:143-151. LORANGER, A.J., T.N. BAILEY, and W.W. LARNED. 1991. Effects of forest suc- cession after fire in moose wintering habitats on the Kenai Peninsula, Alaska. Alces 27:100-109. MCCLEAN, S.A., M.A. RUMBLE, R.M. KING, and W.L. BAKER. 1998. Evaluation of resource selection methods with differ- ent definitions of availability. Journal of Wildlife Management 62:793-801. MCGARIGAL, K., and B.J. MARKS. 1995. Fragstats. Spatial pattern analysis pro- gram for quantifying landscape struc- ture. Oregon State University, Corvalis, Oregon, USA. (MER) MINISTÈRE DE L’ÉNERGIE ET DES R E S S O U R C E S . 1 9 8 4 . N o r m e s d’inventaire forestier. Ministère de l’Énergie et des Ressources du Québec, Québec, Canada. PARÉ, M., and R. COURTOIS. 1990. Inventaire aérien de l’orignal dans les zones de chasse 12 en janvier 1988 et dans la zone de chasse 13 en janvier 1989.Pub- l i c a t i o n 1 7 6 4 . M i n i s t è r e d e l’Environnement et de la Faune du Québec, Québec, Canada. PEEK, J.M., R.J. MACKIE, and G.L. DUSEK. 1992. Over-winter survival strategies of North American Cervidae. Alces Supplement 1:156-161. , D.J. PIERCE, D.C. GRAHAM, and D.L. DAVIS. 1987. Moose habitat use and implications for forest management HABITAT SELECTION IN CUT AREAS - COURTOIS ET AL. ALCES VOL. 38, 2002 192 in Northcentral Idaho. Swedish Wildlife Research Supplement 1:195-199. , D.L. URICH, and R.J. MACKIE. 1976. Moose habitat selection and re- lationships to forest management in northeastern Minnesota. Wildlife Mono- graphs 48. POTVIN, F. 1998. La martre d’Amérique (Martes americana) et la coupe à blanc e n f o r ê t b o r é a l e : u n e a p p r o c h e télémétrique et géomatique. Ph.D. The- sis. Université Laval, Québec, Canada. , R. COURTOIS, and L. BÉLANGER. 1999. Short-term response of wildlife to clear-cutting in Québec boreal for- est: multiscale effects and management implications. Canadian Journal of For- est Research 29:1120-1127. , K. LOWELL, M.-J. FORTIN, and L. BÉLANGER. 2001. How to test habitat selection at the home range scale: a resampling random windows technique. Écoscience 8:399-406. REMPEL, R.S., P.C. ELKIE, A.R. RODGERS, and M.J. GLUCK. 1997. Timber-man- agement and natural-disturbance ef- fects on moose habitat: landscape evalu- ation. Journal of Wildlife Management 61:517-524. RENECKER, L.A., and R.J. HUDSON. 1986. Seasonal energy expenditures and ther- moregulatory responses of moose. Ca- nadian Journal of Zoology 64:322-327. RETTIE, W.J., and P.D. MCLOUGHLIN. 1999. Overcoming radiotelemetry bias in habi- tat-selection studies. Canadian Journal of Zoology 77:1175-1184. , and F. MESSIER. 2000. Hierarchi- cal habitat selection by woodland cari- bou: its relationship to limiting factors. Ecography 23:466-478. SAS INSTITUTE INC. 1989. SAS/STAT user’s guide. SAS Institute Incorpo- rated, Cary, North Carolina, USA. SCHWARTZ, C.C. 1992. Physiological and nutritional adaptations of moose to north- ern environments. Alces Supplement 1:139-155. SINCLAIR, A.R.E., and P. ARCESE. 1995. Population consequence of predation- sensitive foraging: the Serengeti wilde- beest. Ecology 76:882-891. THOMPSON, I.D., and D.L. EULER. 1987. Moose habitat in Ontario: a decade of change in perception. Swedish Wildlife Research Supplement 1:181-193. , and R.W. STEWART. 1998. Man- agement of moose habitat. Pages 377- 402 in A. W. Franzmann and C.C. Schwartz, editors. Ecology and man- agement of the North American moose. Smithsonian Institution Press, Wash- ington, D.C., USA. TIMMERMANN, H.R., and J.G. MCNICOL. 1988. Moose habitat needs. Forestry Chroni- cle 64:238-245. VAN BALLENBERGHE, V. 1989. Twenty years of moose immobilization with suc- cinylcholine chloride. Alces 25:25-30. WIENS, J.A. 1989. Spatial scaling in ecol- ogy. Functional Ecology 3:385-397. WILSON, S.F., D.M. SHACKLETON, and K.L. CAMPBELL. 1998. Making habitat-avail- ability estimates spatially explicit. Wild- life Society Bulletin 26:626-631.