2007 NORTHEASTERN NATURALIST 14(4):571–588 Ecology and Habitat Selection of a Woodland Caribou Population in West-central Manitoba, Canada Juha M. Metsaranta1,2,* and Frank F. Mallory3 Abstract - This study examines the ecology of Rangifer tarandus caribou (woodland caribou) in the Naosap range in west-central Manitoba, Canada. This population is considered to be of high conservation concern because of potential resourcedevelopment impacts; therefore, baseline data are required to guide and evaluate the management of this species in this area. Radio-telemetry data were collected every two weeks from February 1998 to April 2001 and used in combination with forestinventory data to evaluate habitat selection, site fi delity, movement, and grouping patterns. In both summer and winter, selected habitats were mature upland spruce and pine forests, as well as treed muskeg. Hardwood forests were least selected at all scales. Mature coniferous forest was preferred over immature coniferous forests in a pair-wise comparison in winter, but not in summer. Home-range sizes were within expected ranges of variation. Animals used distinct areas in summer and winter, showing broad fi delity to seasonal ranges. However, small shifts in the core areas were observed, particularly in winter. Movement rates and grouping behavior were typical of other caribou. Habitats used in winter were common in the study area, but the ability of the animals to disperse to alternate winter areas is not known. Management efforts could focus on protecting known calving and winter-use areas, and regenerating coniferous forests after logging, which is consistent with regional forest-management objectives. Introduction This study investigated the ecology of Rangifer tarandus caribou Gmelin (woodland caribou) in an area known as the Naosap caribou range in west-central Manitoba, Canada (Fig. 1). This population is presently considered to be of high conservation concern because of potential resource- development impacts (Manitoba Conservation 2005). The purpose of this study is to describe the ecological characteristics of this species in this region because baseline data are required to guide and evaluate the management of this species in this area. These data are currently lacking, as research in the province has either historically been focused on populations located elsewhere (Brown et al. 2000, Darby and Pruitt 1984, Schaeffer and Pruitt 1991, Stardom 1975) or needs to be updated from older studies in the region (Benoit 1996, Shoesmith and Storey 1977), and thus needs to be collected in order to examine the effectiveness of mitigation plans (Tolko Industries 1999). The study has three specific 1Department of Renewable Resources, University of Alberta, 751 General Services Building, Edmonton, AB, Canada T6G 2H1. 2Current address - Canadian Forest Servie, Pacifi c Forestry Centre, 506 West Burnside Road, Victori, BC, Canada, V8Z 1M5. 3Department of Biology, Laurentian University, 935 Ramsey Lake Road, Sudbury, ON, Canada P3E 2C6. *Corresponding author - jmetsara@pfc.nrcan.gc.ca 572 Northeastern Naturalist Vol. 14, No. 4 objectives. The first is to determine and describe habitat-selection patterns of this population at two spatial scales. The second is to describe the ecology and general behavior of this population by assessing site fidelity and describing home-range sizes, annual movement cycles, and grouping behavior. The third is to discuss the implications of these observations to caribou persistence in the study area. For caribou, predator avoidance is the most important limiting factor at larger spatial scales, and forage availability is the most important at smaller spatial scales (Bergerud et al. 1990, Rettie and Messier 2000). Disturbances can change both of these factors. Fire reduces lichen abundance and increases accumulations of snow and deadfall, reducing forage availability (Schaeffer and Pruitt 1991) or impeding movement (Metsaranta et al. 2003). Forest-management practices may favor deciduous species, increasing forage availability for other ungulates (Carleton and MacClennan 1994, Strong and Gates 2006) or creating habitat unsuitable for caribou (Rettie and Messier 2000). This can also increase populations of Alces alces L. (moose), and consequently Canis lupus L. (wolves) (Bergerud and Elliot 1986, Seip 1992), which is detrimental to woodland caribou through changes to the moose-wolf-caribou predator-prey dynamic. In spring and summer, female woodland caribou with calves are solitary and dispersed or spaced out in predator-free habitats like islands and shorelines (Bergerud et al. 1990), or in the absence of these features, at low densities over large areas (Bergerud 1996, Stuart-Smith et al. 1997). Caribou aggregate in fall, and are found in small groups in winter (Brown et al. 2000, Fuller and Keith 1981, Rettie and Messier 2001, Stuart-Smith et al. 1997). Home ranges are larger in fall and winter than in spring and summer (e.g., Mahoney and Virgil 2003, Rettie and Messier 2001, Stuart-Smith et al. 1997). In some populations, individuals use non-overlapping seasonal ranges (e.g., Bergerud et al. 1990, Cumming and Beange 1987), and in others seasonal ranges overlap signifi cantly (Ouellet et al. 1996, Poole et al. 2000, Figure 1. Location of the study area in west-central Manitoba, Canada. 2007 J.M. Metsaranta and F.F. Mallory 573 Stuart-Smith et al. 1997). Caribou show inter-year fi delity to calving sites, and make shifts in the areas that they use each winter (e.g., Schaeffer et al. 2000, Wittmer et al. 2006), though these areas can be broadly similar (e.g., Cumming and Beange 1987). Study Area The study area is in west-central Manitoba, Canada, northeast of the towns of Flin Flon and The Pas (Fig. 1). It is intersected by the boundary of the Churchill River upland ecoregion of the boreal shield ecozone to the north and the mid-boreal lowland ecoregion of the boreal plains ecozone to the south. The boreal shield consists of interspersed uplands and lowlands with bedrock outcrops, lakes, and low topographic relief. In contrast, the boreal plains are topographically level to gently rolling, consisting of lacustrine or organic parent materials. Tree species include Picea mariana (Mill.) Britt. (black spruce), Picea glauca (Moench) Voss (white spruce), Pinus banksiana Lamb. (jack pine), Larix laricina (DuRoi) (tamarack), Populus tremuloides Michx. (trembling aspen), and Betula papyrifera Marsh. (white birch) The climate is continental. Mean daily temperatures range from 17.7 ºC in July to -21.4 ºC in January. Mean annual rainfall and snowfall range from 323.3 mm and 170.2 cm, respectively, in The Pas to 345.3 mm and 143.9 cm, respectively, in Flin Flon. Snow is present from mid-November to early April, with maximum depths in January and February. Highway and rail transportation corridors, forestry road development, hydro transmission lines, and habitat disturbance from logging or forest fi res all potentially affect this population. Forest management began in the early 1970s. Fires occur naturally and are currently suppressed. A large part of the range burned during a major fi re year in 1989 (Hirsch 1991). Methods Data collection Radio-telemetry. Radio-telemetry data were collected from February of 1998 to April of 2001. Animals were captured by net gunning and outfi tted with standard VHF radio collars (Lotek Wireless Inc, Newmarket, ON, Canada). Between 14 and 25 female woodland caribou were located every two weeks using standard aerial radio-tracking methods. Positions were recorded by marking the location of the animal on aerial-photo mosaics and by recording the latitude and longitude of the position using a GPS receiver. Actual sightings occurred in 32% of cases, with sightings being more common in winter (52%) than in summer (13%). A database of locations was created using ArcView GIS (ESRI, Redlands, CA). Most locations were manually digitized by comparing features on the aerial photo mosaics with those on a Forest Resource Inventory (FRI) database displayed on screen. Remaining locations were generated from the GPS co-ordinates recorded. A total of 1358 locations were obtained. 574 Northeastern Naturalist Vol. 14, No. 4 Habitat data. Habitat data were obtained from the Manitoba FRI. A high degree of correspondence between FRI’s and habitat characteristics important to woodland caribou has previously been shown (Rettie et al. 1997). Here, combinations of FRI components were aggregated into 12 habitat types, based upon vegetative associations in the study area (Table 1). Areas disturbed by fi re or logging were classifi ed as immature hardwood or immature conifer, depending on the regeneration present. The FRI components used to classify stands into each of the habitat types are described in Metsaranta (2002). The FRI data were updated from aerial photography obtained in either 1982–83 or 1987–88. Individual stands were updated to account for events occurring before the end of 1998. Statistical analyses Habitat selection. Habitat selection was analyzed at two seasonal periods (winter and summer) and two spatial scales (second- and third-order selection Johnson 1980). Seasons were defi ned as October 16th to April 15th (winter) and April 16th to October 15th (summer), based on movement patterns and the presence of snow cover. Years began on April 16th of one year and ended on April 15th of the following year. Three annual periods (1998, 1999, and 2000) were available for analysis. Habitat-selection patterns could be assessed for 23 animals in summer and 22 in winter. At the second-order (study-area) scale, availability was defi ned as the outermost boundary of a combination of the 100% minimum convex polygon (MCP) and 95% isopleth of the fi xed-kernel home range for all locations. Use was defi ned by the seasonal 50% isopleth of the fi xed-kernel home range for individual animals. At the third-order (home-range) scale, availability was defi ned as the outer boundary of the 100% MCP home range for all telemetry locations for an individual animal. To account for habitat-dependant bias on location precision (Rettie and McLoughlin 1999), use was defi ned as a Table 1. Habitat types available to woodland caribou in the Naosap range in west-central Manitoba Habitat type Total area (km2) Proportion of study area (%) UCP (upland conifer pine) 420.7 9.3 UCS (upland conifer spruce/fi r) 498.8 11.0 OUC (open upland conifer) 54.5 1.2 IC (immature conifer) 667.6 14.7 OW (open wetland) 274.3 6.0 UH (upland hardwood) 966.8 2.1 OUH (open upland hardwood) 100.0 0.2 IH (immature hardwood) 542.2 1.2 LC (lowland conifer) 303.4 6.7 W (water) 760.1 16.7 TM (treed muskeg) 1385.8 30.5 NV (non-vegetated) 15.8 0.04 Total 4542.2 100.0 2007 J.M. Metsaranta and F.F. Mallory 575 circular buffer of radius 350 m in summer and 710 m in winter. These were the mean daily movement rates of all animals in each season. Habitat-selection patterns were determined using the compositional analysis of log-ranks method (Aebischer et al. 1993). Habitats without any use were replaced by 0.001, a value smaller than any other value in the dataset (Aebischer et al. 1993). Third-order habitat selection was fi rst examined by year to see if annual differences existed in habitat-type rankings. Using a Wilcoxon sign-rank test, no signifi cant difference was found (see Results), so data were pooled among years for all further analyses. Site fi delity. Fixed-kernel home-range estimates were calculated at four probability isopleths (95%, 75%, 50%, and 25%), and four combinations of years and seasons (summer 1999 and 2000, and winter 1999 and 2000). Home ranges could be calculated for n = 21 animals in both years. The number of data points used to calculate the home ranges was low (mean = 12, range 9 to 17), but these calculations were focused on assessing site fi delity, and not accurate and precise home-range size estimation. To determine fi delity, the intersection of the resulting vector polygons for each season or for each year at each probability isopleth was calculated in ArcView. From this spatial intersection, the Dice similarity coeffi cient (Dice 1945) was calculated as: 2 (Pa∩Pb) 2 (Pa∩Pb) + Pa∉ Pb + Pb ∉ Pa Where Pa ∩ Pb is the area covered by both Pa and Pb, Pa ∉ Pb is the area of Pa not contained within Pb, and Pb ∉ Pa is the area of Pb not contained within Pa. Here, the Dice coefficient is used as a spatial overlap index to test for site fidelity. Zidjenbos et al. (1994) derive the Dice similarity coefficient Cohen’s kappa coefficient of agreement, and Landis and Koch (1977) proposed that kappa coefficient values in the range 0.41–0.60 represent moderate similarity, those between 0.61–0.80 represent substantial similarity, and those greater than 0.81 represent almost perfect similarity. Values less than 0.4 represent essentially no similarity. As a result, this study considered values greater than 0.4 to be evidence of site fidelity. The intersection of the seasonal polygons for different years was a test for interyear site fidelity, while the intersection of seasonal polygons within a year was a test for intra-year site fidelity. To provide additional context on the degree of home-range shift on an inter- and intra-annual basis, the distance between home-range centroids (mean x- and y-locations) both within and between years was also calculated. Home-range estimation. The Animal Movement extension for ArcView (Hooge and Eichenlaub 1999) was used for all home-range calculations. Fixed-kernel estimates used least-squares cross validation (Seaman and Spd = 576 Northeastern Naturalist Vol. 14, No. 4 Powell 1996). The fi nal reported seasonal home-range sizes use data pooled for the entire three-year study period. Home ranges were calculated for 23 animals in summer (mean of 28 locations [range = 20–44]), and 22 animals in winter (mean of 27 locations [range = 21–43]), using both the 100% MCP and the 95%-isopleth of the fi xed-kernel estimator. Movement rates and group sizes. Movement rates were calculated as the mean daily distance traveled between successive relocations, considering only locations from 12 to 18 days apart, and were assigned to the month in which the mean date between successive locations fell. Each time a radio-collared caribou was sighted during a telemetry flight, the number of animals associated with it was recorded. In a small number of cases when other groups of caribou were incidentally sighted, their numbers were also recorded. Although not formally assessed, there did not appear to be any obvious evidence that the size or composition of these groups differed. These data were used to describe the distribution of group sizes observed by month. Results Third-order habitat selection. Habitat use was not random at the thirdorder scale in both summer and winter, and in each of the three study years (Λ = 0.012 to 0.055, 11 df, p < 0.01). However, habitat rankings were not signifi cantly different from year to year (Wilcoxon signed rank test: Zw = 0.00 to -0.48, n = 5 to 10, p > 0.63). As a result habitat-use data were pooled to look for habitat-selection patterns at all scales. The pooled data also indicated that habitat use differed significantly from random in both summer and winter (Λ = 0.043 and 0.048, respectively, 11 df, p < 0.01; Table 2). The highest-ranking habitats in summer were: treed muskeg, water, upland conifer-pine, and upland coniferspruce. The highest ranking habitats in winter were: upland conifer-pine, treed muskeg, and upland conifer-spruce. Hardwood habitats were the lowest-ranking subset of habitats in both seasons. Upland conifer-spruce and upland conifer-pine habitats were preferred over immature conifer habitats in a pair-wise comparison in winter, but were not preferred during summer (Table 2). Second-order habitat selection. At the second-order scale, habitat use again differed signifi cantly from random in both summer and winter (Λ = 0.082 and 0.044, respectively, 11 df, p < 0.01; Table 3). The highest-ranking habitats in summer were: water, treed muskeg, and upland conifer-spruce. The highest-ranking habitats in winter were: upland conifer-spruce, upland conifer-pine, treed muskeg, and open wetland. The lowest-ranking habitats in both seasons were all hardwood habitat types. Again, in a pair-wise comparison, upland conifer-spruce and upland conifer-pine habitats were signifi cantly preferred over immature conifer habitat in winter, but were not preferred in summer (Table 3). 2007 J.M. Metsaranta and F.F. Mallory 577 Site fi delity. The distribution of Dice coeffi cient values for each probability isopleth for intra-year fi delity in 1999 and 2000 are plotted in Figure 2. One animal had a Dice coeffi cient value less than 0.4 at the 95% isopleth in both years. All other animals had Dice coeffi cient values less than 0.4 at all of the other fi xed-kernel isopleths. The distribution of Dice coeffi cient values for each probability isopleth for inter-year fi delity to summer and winter ranges are plotted in Figure 3. Most animals (81%) had Dice coeffi cient values greater than 0.4 between 1999 and 2000 at the 95% isopleth. At the 75%, 50%, and 25% isopleths, the number of animals that have Dice coeffi cient values greater than 0.4 decreases with each successive isopleth. In summer, 33% of animals have values greater than 0.4 for the 25% isopleth, and 14% of animals do so in winter. The mean distance between winter and summer home-range centroids in 1999 was 22.3 km (range = 5.9 to 46.5 km, SD = 12.5 km); in 2000 it was 22.1 km (range = 3.4 to 54.7 km, SD = 13.4 km). The mean distance between winter home-range centroids between years was 9.5 km (range = 2.5 to 25.4 km, SD = 5.4 km) and in summer it was 3.3 km (range = 1.0 to 7.6 km, SD = 1.8 km). Table 2. Compositional analysis matrix of (1) summer and (2) winter third-order habitat selection (buffered telemetry locations within 100%-MCP home range). Each mean log difference is replaced by a sign (++ or --) indicating signifi cant differences. Habitat types are ranked in the order of their importance, with an (H) indicating that they are not signifi cantly different from the highest-ranked habitat type, and an (L) indicating that they are not signifi cantly different from the lowest-ranked habitat type. See Table 1 for explanation of habitat-type abbreviations. (1) Summer TM W UCP UCS OW IC LC OUC NV OUH IH UH Rank TM ++ ++ ++ ++ ++ ++ ++ ++ 1 (H) W ++ ++ ++ ++ ++ ++ ++ 2 (H) UCP ++ ++ ++ ++ ++ ++ 3 (H) UCS ++ ++ ++ ++ ++ 4 (H) OW -- ++ ++ ++ ++ 5 IC -- -- ++ ++ ++ 6 LC -- -- -- -- ++ 7 OUC -- -- -- ++ 8 NV -- -- -- -- -- 9 (L) OUH -- -- -- -- -- -- 10 (L) IH -- -- -- -- -- -- 11 (L) UH -- -- -- -- -- -- -- -- 12 (L) (2) Winter UCP TM UCS LC OW W OUC UH NV IC OUH IH Rank UCP ++ ++ ++ ++ ++ ++ ++ ++ ++ 1 (H) TM ++ ++ ++ ++ ++ ++ ++ ++ ++ 2 (H) UCS ++ ++ ++ ++ ++ ++ ++ 3 (H) LC -- -- ++ ++ ++ ++ ++ ++ ++ 4 OW -- -- ++ ++ ++ ++ ++ ++ ++ 5 W -- -- -- -- -- ++ ++ ++ ++ ++ 6 OUC -- -- -- -- -- ++ ++ 7 UH -- -- -- -- -- -- ++ ++ 8 NV -- -- -- -- -- -- ++ ++ 9 IC -- -- -- -- -- -- ++ ++ 10 OUH -- -- -- -- -- -- -- -- -- -- 11 (L) IH -- -- -- -- -- -- -- -- -- -- 12 (L) 578 Northeastern Naturalist Vol. 14, No. 4 Home-range Sizes. Home-range sizes were larger in winter than in summer using either estimator. Mean home-range size in winter for n = 22 animals was 856 km2 (range = 103 to 2206 km2, SD = 430 km2) using the 100%-MCP estimator and 1386 km2 (range = 126 to 3256 km2, SD = 709 km2) using the 95%-isopleth fi xed-kernel estimator. Mean home-range size in summer for n = 23 animals was 162 km2 (range = 7 to 975 km2, SD = 201 km2) using the 100%-MCP estimator, and 175 km2 (range = 10 to 670 km2, SD = 155 km2) using the 95%-isopleth fi xed-kernel estimator. Movement rates. Movement rates were lowest from May to September, corresponding to the summer calving and post-calving period (Fig. 4). Movement rates were highest in April, November, and January. The November and April peaks corresponded to periods of seasonal range-use shifts, while the January peak represented a movement from early to late winter-use areas (Fig. 4). Table 3. Compositional analysis matrix of (1) summer and (2) winter second-order habitat selection (50% adaptive-kernel home range within cumulative population range). Each mean log difference is replaced by a sign (++ or --) indicating signifi cant differences. Habitat types are ranked in the order of their importance, with an (H) indicating that they are not signifi cantly different from the highest-ranked habitat type, and an (L) indicating that they are not signifi - cantly different from the lowest-ranked habitat type. See Table 1 for explanation of habitat type abbreviations. W TM UCS OW UCP LC OUC NV IC OUH UH IH Rank (1) Summer W ++ ++ ++ ++ ++ ++ ++ ++ 1 (H) TM ++ ++ ++ ++ ++ ++ 2 (H) UCS ++ ++ ++ ++ 3 (H) OW -- ++ ++ ++ ++ 4 UCP ++ ++ ++ 5 LC -- ++ ++ ++ 6 OUC -- -- -- 7 (L) NV -- -- -- 8 (L) IC -- -- -- 9 (L) OUH -- -- -- -- -- -- 10 (L) UH -- -- -- -- -- -- 11 (L) IH -- -- -- -- -- -- 12 (L) (2) Winter UCS ++ ++ ++ ++ ++ ++ ++ ++ 1 (H) UCP ++ ++ ++ ++ ++ ++ ++ ++ 2 (H) TM ++ ++ ++ ++ ++ ++ ++ ++ 3 (H) OW ++ ++ ++ ++ ++ ++ ++ ++ 4 (H) W -- -- -- -- ++ ++ ++ ++ ++ 5 LC -- -- -- -- ++ ++ ++ ++ 6 OUC -- -- -- -- ++ ++ ++ 7 NV -- -- -- -- -- ++ ++ 8 UH -- -- -- -- -- -- ++ ++ 9 IC -- -- -- -- -- -- -- ++ 10 IH -- -- -- -- -- -- -- -- -- 11 (L) OUH -- -- -- -- -- -- -- -- -- -- 12 (L) 2007 J.M. Metsaranta and F.F. Mallory 579 Grouping behavior. Group sizes varied greatly over a year (Fig. 5). From May to September, group size was limited to two animals, which usually represented cow-calf pairs. From October to April, the average group size was 5.1 animals (SD = 3.1, n = 282 sightings). Although not formally investigated, observations of groups of animals not associated with radio-collared animals during this period appeared to be of similar Figure 2. Distribution of Dice coeffi cient values in 1999 and 2000 for intra-year home-range fi delity of n = 21 woodland caribou in the Naosap area. 580 Northeastern Naturalist Vol. 14, No. 4 size, and have similarly varying sex and age compositions to those that had radio-collared individuals (Metsaranta 2002). The largest individual aggregation observed was 20 animals, and groups of more than 10 animals were regularly sighted. Figure 3. Distribution of Dice coeffi cient values in winter and summer for inter-year home-range fi delity of n = 21 woodland caribou in the Naosap area. 2007 J.M. Metsaranta and F.F. Mallory 581 Discussion Habitat selection. Mature coniferous forests were highly ranked habitat types in both summer and winter, as in other studies (e.g., Bradshaw et al. 1995, Mahoney and Virgil 2003, Mosnier et al. 2003, Rettie and Messier 2000). However, in a pair-wise comparison, mature coniferous habitats were not used more than immature conifer forests. Rettie and Messier (2000) also found that, in certain populations, animals showed a selective inclusion of immature forest types, speculating that this represented historical habitatselection patterns, and that where annual shifts occurred, they tended to show avoidance of immature forest types. This is likely also the case in this study. Avoidance of immature forest types would tend to show the selective avoidance of moose and consequently higher predator populations associated with moose, since moose tend to be more prevalent in young forests. Winter habitat types selected consisted primarily of a mosaic of mature upland spruce and pine conifer forests and open or semi-open treed muskegs. The habitat types selected in winter are common throughout the study area, together accounting for about 50% of the total available habitat in the study area. In summer, the strong selective inclusion of water at both spatial scales is similar to previous studies where caribou have been noted to occur near water (on islands and peninsulas or near lakeshores) during this period as an Figure 4. Box and whisper plot of daily distance traveled in each month by woodland caribou in the Naosap area. The ends of the box represent the lower and upper quartiles, the whiskers represent the 10th and 90th percentiles, and the dots represent the 5th and 95th percentiles. The solid line in the box represents the mean, and the dashed line in the box represents the median. 582 Northeastern Naturalist Vol. 14, No. 4 anti-predator strategy (Bergerud et al. 1990). Unlike recent studies (Ferguson and Elkie 2005), caribou here did not show any selective use of frozen lakes during the winter. Animals in this population did not select mature over immature coniferous habitats, which could indicate that their habitat quality remains adequate during the summer, and that specific mitigation plans for forest harvesting near summer-use areas that maintain buffer areas and access corridors around lakes where caribou are known to use islands and peninsulas to calve appear to be allowing adult female caribou to continue using those areas. It may also indicate that moose populations may not have had enough time to increase after disturbance, and thus, predation pressure on caribou had not yet increased. Moose would tend to be found in higher numbers in hardwood forest types, which were ranked the lowest at all spatial and temporal scales. This suggests that should silvicultural practices successfully regenerate coniferous forests, this would help ensure future availability of woodland caribou habitat, and possibly maintain present populations. If silvicultural practices do not successfully regenerate coniferous habitats (e.g., Carleton and MacClennan 1994), if management Figure 5. Box and whisper plot of group sizes in each month by woodland caribou in the Naosap area. The ends of the box represent the lower and upper quartiles, the whiskers represent the 10th and 90th percentiles, and the dots represent the 5th and 95th percentiles. The solid line in the box represents the mean, and the dashed line in the box represents the median. 2007 J.M. Metsaranta and F.F. Mallory 583 practices favor increased forage for other ungulates (e.g., Strong and Gates 2006), or if post-logging successional pathways differentially favor other ungulates over caribou (e.g., Metsaranta, in press), then this situation would be a concern for maintaining caribou populations. Home-range sizes. As in many other studies (e.g., Edmonds 1988, Mahoney and Virgil 2003, Rettie and Messier 2001, Stuart-Smith et al. 1997), home-range sizes were larger in winter than in summer. In studies with the most-similar seasonal defi nitions (Edmonds 1988, Stuart-Smith et al 1997), the reported home-range sizes ranged from 147 to 650 km2 in winter and 24 to 536 km2 in summer. These are smaller than the 856-km2 winter-range size, but similar to the 162-km2 summer-home size, observed here. Intra-year site fidelity. Caribou in this region used distinct areas in summer and winter. This is consistent with some studies (e.g., Bergerud et al. 1990, Cumming and Beange 1987, Shoesmith and Storey 1977), but not others (e.g., Ouellet et al 1996, Stuart-Smith et al. 1997). In cases where animals calve and spend the summer on islands and peninsulas of large lakes typical of the boreal shield ecozone, wintering areas are often separate and distinct mainland areas (Bergerud et al. 1990, Cumming and Beange 1987, Shoesmith and Storey 1977). In cases where animals occupy large peatland areas, typical of the boreal plains ecozone, distinct winter and summer areas are absent (Stuart-Smith et al. 1997). The animal in this study that used overlapping seasonal ranges was in the southern portion of the study area, where large continuous peatlands typical of the boreal plains are common. The rest of the animals in this study occupied distinct areas during summer and winter, and were in the northern portion of the study area, where numerous lakes and islands typical of the boreal shield are common. Inter-year site fidelity. Animals broadly use the same seasonal areas year after year (95% and 75% isopleths). Other studies have also documented strong tendencies to site fidelity in this species (Rettie and Messier 2001, Schaeffer et al. 2000, Wittmer et al. 2006). However, animals in the present study exhibited slight shifts in the core areas (50% and 25% isopleths) used in winter. Caribou broadly used the same areas in winter (similarity of 95% and 75% isopleth between years), but did not necessarily return to precisely the same locations (less similarity in the 50% and 25% isopleths between years). Cumming and Beange (1987) and Wittmer et al. (2006) made similar observations of small changes in wintering areas from year to year, even though the areas were broadly similar from year to year. Cumming (1996) speculated that winter-use areas are implicit refuges from predation, while Wittmer et al. (2006) speculated that changes in winter-use areas occur in response to forage availability. In either case, if alternate habitats are available, then disruption of winter habitats may not be detrimental, as long as forests regenerate to suitable conditions and caribou are able to disperse to alternate areas. It is not known if alternate 584 Northeastern Naturalist Vol. 14, No. 4 wintering areas are present on this range, and they may not be detectable if animals themselves do not disperse there (Cumming 1996). Movement rates. Individuals traveled an average of 0.3 km day-1 in summer, and an average of 0.8 km day-1 in winter. These can only be considered rough estimates of movement rates since the straight-line distance between two locations taken two weeks apart may not be strongly correlated with the actual distances moved. Regardless, they are consistent with observations elsewhere (e.g., Benoit 1996, Fuller and Keith 1981, Stuart-Smith et al. 1997). Peak rates occurred in April, November, and January. The April peak is consistent with previous observations in the region made in the late 1970s (Benoit 1996). However, these previous observations also show two smaller peaks of movement in October and late December (Benoit 1996), not in November and January. This could be a change in the timing of seasonal migrations in response to changing snow-depth patterns (e.g., Darby and Pruitt 1984, Stardom 1975). As these peaks occur earlier now than they did 25 years ago, this indicates that snow accumulation have recently occurred later in the year than in the past. Weather records at The Pas airport support this contention. Peak snow depths have been declining since records began in 1944, and the winter of 1999–00 had the lowest and 2000–01 had the sixth lowest recorded peak snow accumulation since that time. Grouping behavior. From May to September, cow-calf pairs were the primary group observed. Group size peaked in November at a mean of 6.3 animals. From October to April, the mean group size was 5.1 (SD 3.1), which is consistent with woodland caribou observed elsewhere (e.g., Brown et al. 2000, Darby and Pruitt 1984, Rettie and Messier 1998, Stuart-Smith et al. 1997). In a previous study of this region, Shoesmith and Storey (1977) noted group sizes ranging from 2 to 14 animals, with a peak average of 6 in December. Thus, grouping behavior of the present study is concordant with past observations of this population. Conclusions This study found that the population of caribou living in the Naosap range in west central Manitoba has ecological characteristics that are similar to other populations of woodland caribou across Canada. Hardwood forest types were the lowest-ranked habitat types at all scales examined, which is a concern if silvicultural practices are not successful at regenerating conifer forests. In the long term, regenerating coniferous forests after logging, which is generally consistent with regional forest-management objectives, is necessary to maintain forests in a condition that resembles habitat currently used by this species in the region. However, this may not be suffi cient (Metsaranta, in press; Metsaranta et al. 2003) because of differences in post-logging and post-fi re successional pathways that may, in the case of logging, differentially 2007 J.M. Metsaranta and F.F. Mallory 585 favor other ungulates over caribou. In summer, mature conifer forests are not preferred over immature conifer forests in a pair-wise comparison. This may indicate that in the short-term, specifi c mitigation plans for forest harvesting (Tolko Industries 1999) that maintain buffer areas and access corridors around lakes where caribou are known to use islands and peninsulas to calve appear to be allowing adult female caribou to continue using those areas. However, this may be because not enough time has yet passed to see a signifi cant change in the populations of other ungulates, particularly moose, and the concurrent increase in predation pressure on caribou. The most-recent population surveys estimate that the mean density of moose in the study area ranges from 0.09 to 0.15 individuals km-2, reaching a maximum of 0.40 moose km-2 in some areas (Cross 1996, 2000). Although wolves are known to be present, there is no available estimate of their density. This density of moose present may not be able to support a high density of wolves on its own (Gasaway et al. 1992, Messier 1994, Messier and Crete 1985). Because of this, the wolf population is probably jointly supported by both moose and caribou, limiting caribou at low densities, but permitting a stable population (Rettie and Messier 2000). Unlike in summer, mature coniferous forests were preferred over immature coniferous forests during winter. Where caribou choose to be during winter might represent an implicit refuge from predation (Cumming 1996). The presence of other such refuges on this range is not known, suggesting that areas currently used during winter should be protected from disturbance. As a result of the uncertainty of moose and wolf populations, and the uncertainty as to the presence of alternate winter areas, the long-term effectiveness of mitigation plans needs to be monitored. Acknowledgments We thank Manitoba Conservation, Tolko Industries, and Manitoba Hydro for fi nancial support. We also thank Dale Cross for conducting much of the radiotelemetry work, the pilots at Jackson Air Services for skillful fl ying, and Becky Farguson for able assistance during two fi eld seasons. We also thank the numerous other Manitoba Conservation and Tolko Industries staff in The Pas, MB who contributed to this work over the years. J.M. Metsarana was supported by a Natural Science and Engineering Resource Council of Canada scholarship, as well as a Laurentian University Graduate Fellowship while conducting this work. The comments of two anonymous referees helped to improve the manuscript. Literature Cited Aebschier, N.J., P.A. Robertsen, and R.E. Kenward. 1993. Compositional analysis of habitat use from animal radiotracking data. Ecology 74:1313–1325. Benoit, A.D. 1996. A landscape analysis of woodland caribou habitat use in the Reed-Naosap Lakes Region of Manitoba (1973–1985). M.Sc. Thesis. Univeristy of Manitoba, Winnipeg, MB. 109 pp. Bergerud, A.T. 1996. Evolving perspectives on caribou population dynamics: Have we got it right yet? Rangifer (Special Issue 9):95–115. 586 Northeastern Naturalist Vol. 14, No. 4 Bergerud, A.T., and J.P. Elliot. 1986. Dynamics of caribou and wolves in northern British Columbia. Canadian Journal of Zoology 64:1515–1529. Bergerud, A.T., R. Ferguson, and H.E. Butler. 1990. Spring migration and dispersion of woodland caribou at calving. Animal Behaviour 39:360–368. Bradshaw, C.J.A., D.M. Hebert, A.B. Rippin, and S. Boutin. 1995. Winter peatland habitat selection by woodland caribou in northeastern Alberta. Canadian Journal of Zoology 73:1567–1574. Brown, K.G., C. Elliott, and F. Messier. 2000. Seasonal distribution and population parameters of woodland caribou in central Manitoba: Implications for forestry practices. Rangifer Special Issue 12:85–94. Carleton, T.J., and P. MaClennan. 1994. Woody vegetation responses to fi re versus clear-cutting logging: A comparative survey in the central Canadian boreal forest. Ecoscience 1:141–152. Cross, D.W. 1996. Moose population survey and age-sex data: Game Hunting Area 4. Manitoba Natural Resources, Regional Operations, The Pas, MB, Canada. Manuscript Report No. 96.8W. Cross, D.W. 2000. Moose Population Survey: GHA 7A - Northwest Region, January 2000. Manitoba Conservation, Northwest Region, The Pas, MB, Canada. Cumming, H.G. 1996. Managing for caribou survival in partitioned habitat. Rangifer Special Issue 9:171–179. Cumming, H.G., and D.B. Beange. 1987. Dispersion and movements of woodland caribou near Lake Nipigon, Ontario. Journal of Wildlife Management 51:69–79. Darby, W.R., and W.O. Pruitt. 1984. Habitat use, movements, and grouping behaviour of woodland caribou, Rangifer tarandus caribou, in southeastern Manitoba. Canadian Field Naturalist 98:184–190. Dice, L.R. 1945. Measures of the amount of ecological association between species. Ecology 26:297–302. Edmonds, E.J. 1988. Population status, distribution, and movements of woodland caribou in west-central Alberta. Canadian Journal of Zoology 66:817–826. Ferguson S.H., and P.C. Elkie. 2005. Use of lake areas in winter by woodland caribou. Northeastern Naturalist 12:45–66. Gasaway, W.C., R.O. Stephenson, J.L. Davis, P.E.K. Shepherd, and O.E. Burris. 1992. The role of predation in limiting moose at low densities in Alaska and Yukon: The implications for conservation. Wildlife Monographs No. 120. Fuller, T.K., and L.B. Keith. 1981. Woodland caribou population dynamics in northeastern Alberta. Journal of Wildlife Management 45:197–213. Hirsch, K. 1991. A chronological overview of the 1989 fi re season in Manitoba. Forestry Chronicle 67:358–365. Hooge, P.N., and B. Eichenlaub. 1999. Animal movement extension to Arcview, ver. 2.04. Alaska Biological Science Center, US Geological Survey, Anchorage, AK. Johnson, D.H. 1980. The comparison of usage and availability measurements for evaluating resource selection. Ecology 61:65–71 Landis, J., and G.G. Koch. 1977. The measurement of observer agreement for categorical data. Biometrics 33:159–174. Mahoney, S.P., and J.A. Virgil. 2003. Habitat selection and demography of a nonmigratory woodland caribou population in Newfoundland. Canadian Journal of Zoology 81:321–324. 2007 J.M. Metsaranta and F.F. Mallory 587 Manitoba Conservation. 2005. Manitoba’s conservation and recovery strategy for woodland caribou (Rangifer tarandus caribou). Manitoba Conservation - Wildlife Branch, Winnipeg, MB, Canada. Messier, F. 1994. Ungulate population models with predation: A case study with the North American moose. Ecology 75:478–488 Messier, F., and M. Crete. 1985. Moose-wolf dynamics and the natural regulation of moose populations. Oecologia 65:503–512. Metsaranta, J.M. 2002. Habitat utilization by woodland caribou (Rangifer tarandus caribou): An assessment of use in disturbed and undisturbed habitats in westcentral Manitoba. M.Sc. Thesis. Laurentian University, Sudbury, ON, Canada. 127 pp. Metsaranta, J.M. In press. Assessing the length of the post-disturbance recovery period for woodland caribou habitat after fire and logging in west-central Manitoba. Rangifer Special Issue. Metsaranta, J.M., F.F. Mallory, and D.W. Cross. 2003. Vegetation characteristics of forest stands used by caribou and those disturbed by fire or logging in Manitoba. Rangifer Special Issue 14:255–266. Mosnier, A., J-P. Ouellet, L. Sirois, and N. Fournier. 2003. Habitat selection and home-range dynamics of the Gaspe caribou: A hierarchical analysis. Canadian Journal of Zoology 81:1174–1184. Ouellet, J.P., J. Ferron, and L. Sirois. 1996. Space and habitat use by the threatened Gaspe caribou in southeastern Quebec. Canadian Journal of Zoology 74:1922– 1933 Poole, K.G., D.C. Heard, and G. Mowat. 2000. Habitat use by woodland caribou near Takla Lake in central British Columbia. Canadian Journal of Zoology 78:1552–1561 Rettie, W.J., and P.D. McLoughlin. 1999. Overcoming radio-telemetry bias in habitat- selection studies. Canadian Journal of Zoology 77:1175–1184. Rettie, W.J., and F. Messier. 2000. Heirarchical habitat selection by woodland caribou: Its relationship to limiting factors. Ecography 23:466–478. Rettie, W.J., and F. Messier. 2001. Range use and movement rates of woodland caribou in Saskatchewan. Canadian Journal of Zoology 79:1933–1940. Rettie, W.J., J.W. Sheard, and F. Messier. 1997. Identification and description of forested vegetation communities available to woodland caribou: Relating wildlife habitat to forest cover data. Forest Ecology and Management 93:245–260. Schaeffer, J.A., and W.O. Pruitt. 1991. Fire and woodland caribou in southeastern Manitoba. Wildlife Monographs 116:1–39. Schaeffer, J.A., C.M. Bergman, and S.N. Luttich. 2000. Site fidelity of female caribou at multiple spatial scales. Landscape Ecology 15:731–739. Seaman, D.E., and R.A. Powell. 1996. An evaluation of the accuracy of kerneldensity estimators for home-range analysis. Ecology 77:2075–2085 Seip, D.R. 1992. Factors limiting woodland caribou populations and their interrelationships with wolves and moose in southeastern British Columbia. Canadian Journal of Zoology 70:1494–1503. Shoesmith, M.W., and D.R. Storey. 1977. Movements and Associated Behaviour of Woodland Caribou in Central Manitoba. Proceedings of the International Congress of Game Biologists 13:51–64 588 Northeastern Naturalist Vol. 14, No. 4 Stardom, R.R.P. 1975. Woodland caribou and snow conditions in southeast Manitoba. Pp 436–461, In J.R. Luick, P.C. Lent, D.R. Klein, and R.G. White (Eds.). Proceedings of the First International Reindeer/Caribou Symposium. Biological Papers of the University of Alaska Special Report 1. University of Alaska, Fairbanks, AK. Stuart-Smith, A.K., C.J.A. Bradshaw, S. Boutin, D.M Hebert, and A.B. Rippen. 1997. Woodland caribou relative to landscape patterns in northeastern Alberta. Journal of Wildlife Management 61:622–633. Strong, W.L., and C.C. Gates. 2006. Herbicide-induced changes to ungulate-forage habitat in western Alberta, Canada. Forest Ecology and Management 222:469– 475. Tolko Industries. 1999. Forest management/woodland caribou mitigation plan: Naosap/Peterson operating areas. Tolko Industries, Manitoba Woodlands, The Pas, MB. Wittmer, H.U., B.N. McLellan, and F.W. Hovey. 2006. Factors infl uencing varation in site fi delity of woodland caribou (Rangifer tarandus caribou) in southeastern British Columbia. Canadian Journal of Zoology 84:537–545. Zijdenbos, A.P., B.E. Dawant, R.A.Margolin, and A.C. Palmer. 1994. Morpohometric analysis of white-matter lesions in MR Images: Method and validation. IEEE Transactions on Medical Imaging 13:716–724.