Movement patterns of gray wolves (Canis lupus) have been studied in much of their current range in North America (Ballard et al. 1997; Fritts and Mech 1981; Messier 1985a). Most studies were of territorial wolves that prey on ungulates including deer (Odocoileus), elk (Cervus elaphus), moose (Alces alces), and sheep (Ovis). Although some of these ungulates may undergo seasonal migrations, they are of lesser magnitude than the migrations of barren-ground caribou (Rangifer tarandus groenlandicus). Consequently, most studies have concluded that wolves maintain relatively stable annual territories.
In the Northwest Territories, Nunavut, Yukon Territory, Alaska, and northern Quebec, caribou herds are not sedentary but migrate between the boreal forest where they winter to calving grounds on the tundra (Hemming 1971; Kelsall 1968; Messier et al. 1988). In many of these northern habitats, caribou are the only ungulates that occur at densities sufficient to support wolves, so wolves occupying these areas prey primarily on caribou (Kuyt 1972; Stephenson and James 1982). Wolves associated with these herds are not thought to be territorial but move seasonally with the caribou. However, from parturition (mid- to late May) until pups can travel with the adults (September-October), movements of wolves are restricted to the area near their den sites. Furthermore, most wolves den near tree line and do not follow caribou to their calving grounds (Heard and Williams 1992; Kuyt 1972; Parker 1973). Thus, during the denning period and when least mobile, wolves may be forced to search large areas for prey. Little information is available on the movement patterns of wolves inhabiting ranges of migratory caribou herds (Ballard et al. 1997; Kuyt 1962). Spurred by the recent discovery and development of a diamond-resource industry, we used satellite-- tracking methods to collect information on annual and seasonal movements of wolves associated with the largest contiguous wilderness area on the continent.
MATERIALS AND METHODS
Study area.-Our study area was located in the central Canadian Arctic, 300 km NE of Yellowknife, Northwest Territories (Fig. 1). The region recently has experienced intense exploration and mining activity associated with the diamond industry. Wolves were captured in a 30,000-km^sup 2^ area centered around Lac de Gras (64deg27'N, 110deg35'W). However, the study included the northern boreal forest during winter as wolves moved south with caribou. Climate was semiarid, characterized by short cool summers and long cold winters. Annual precipitation averaged 300 mm; about 50% fell as snow. Summer temperatures averaged 10degC, with winter temperatures often <-30degC. The northern part of the study area consisted of low Arctic tundra. Dwarf shrubs (Salix and Betula glandulosa) occurred in drainages. Other common shrubs included Vaccinium uliginosum, V. vitisidaea, and Empetrum nigrum. Heath tundra was common throughout the area. The southern part of the study area encompassed forest tundra and the northern boreal forest. The dominant tree species included Picea mariana, P. glauca, and Pinus banksiana. Many lakes occurred throughout the area, as is characteristic of the rocky upland regions of the Canadian Precambrian Shield. Topography was gently rolling with numerous rock outcrops and glacial-fluvial features such as eskers, kames, drumlins, and raised beaches. The permafrost layer was discontinuous.
The Bathurst caribou herd, estimated at 349,000 caribou +/- 95,000 SE in 1996 (A. Gunn et al., in litt.), migrated annually, leaving the northern boreal forest in April and reaching the calving grounds near Bathurst Inlet by early June. The herd dispersed south by late June and reached tree line by late autumn or early winter. Muskoxen (Ovibos moschatus) occurred sporadically in the northern part of the study area. Other potential prey included Arctic hare (Lepus arcticus), Arctic ground squirrel (Spermophilus parryii), and several small mammals (Clethrionomys, Dicrostonyx, Lemmus, and Microtus).
Capture and marking of wolves.-In June 1997 and 1998, wolves at dens were located using a small fixed-wing aircraft. After they were located, wolves were captured from a helicopter using net guns (Helicopter Wildlife Management Inc., Salt Lake City, Utah), and were immobilized with a hand injection of Telazol(R) (10 mg/ kg, A. H. Robins, Richmond, Virginia-Ballard et al. 1991). The capture process was approved by a University of Saskatchewan Animal Care Committee (protocol number 98003 1). Standard measurements and body weight were determined for each immobilized wolf. Gender was determined, and a unique identifying number was applied using ear tags and a tattoo applied on the left and right buccal margins of the upper lip. We planned to place a collar-mounted satellite transmitter (Telonics Inc., Mesa, Arizona) on 1 breeding adult in each pack handled. Breeding females equipped with transmitters were selected based on development of their mammary glands. Males that were equipped with transmitters were selected based on large body size and better body condition relative to other males within the pack. Thus, compared with females, we are less certain if all collared males were dominant.
Monitoring wolves.-Two different models of satellite collars were deployed (ST-10 and ST-- 14, Telonics Inc.-Walton et al., in press). All collars contained a conventional very-high frequency transmitter to permit aerial locations. Both collar types were programmed to operate for 1 year and transmit more frequently during summer than winter. The collars had 7-h transmitting periods set for the time of day when satellite overpasses occurred most frequently (Burger 1995; Fancy et al. 1988). The interval between transmitting periods differed between collar types because of the lower power demand of the ST-10 collars. In summer 1997, the ST-14 collars had a transmitting period every 48 h during the first 83 days, whereas ST-10 collars had a transmitting period every 24 h during the first 97 days after deployment. After that, transmission periods occurred every 14 days for the ST-- 14 collar and every 5 days for the ST-10 collars. In early May 1998, both collar types deployed in 1997 reverted to their original duty cycles until they were removed (early June 1998). In 1998, only ST-10 collars were deployed, and they had 1 transmitting period every 24 h for the first 122 days after deployment, changing to 1 period every 4 days in winter, before reverting to the original duty cycle in early May 1999.
Locations for all satellite collars were obtained monthly from Service Argos, Inc. (Landover, Maryland). Wolves also were located with small fixed-wing aircraft or helicopters occasionally throughout summer and autumn to document use of den sites. Because only I pack received >1 satellite collar, we used data from only 1 satellite-collared wolf per pack for the following analyses. Therefore, reported estimates of range size for each individual wolf are not synonymous with pack territory sizes. We considered each wolf-year to be independent and, therefore, included range sizes for 2 wolves during both years. Data on all other wolves spanned only 1 year. Locations received from Service Argos, Inc. were used in analyses if they were of location class 1 or better (ca. =<1 km accuracy-Argos 1996; cf. Ballard et al. 1995; Keating et al. 1991).
Home-range size and excursions.-For each wolf, size of annual ranges were estimated from collar deployment (early June) to 31 May the following year. Summer was defined as the period from arrival at the denning area until departure from the summer range in autumn. Winter included locations from the time of departure from the summer range until the wolf returned to a denning area the following spring. We estimated time of departure from the summer range as the middate between the last location known to be within the summer range and the 1st date in which the wolf had moved >=50 km from the den site (and did not return to the denning area until the following year). We assessed precision of this method by plotting distance from the den for each location throughout the year and chose the middate between the 2 locations in which there was a distinct departure from the den site. In all situations, middates were similar. Timing of return to the summer range was estimated by taking the middate from the lst location <50 km from that year's den site (arrival) and in which the individual continued to show fidelity to the den site, to the last location >50 km (departure).
Annual and seasonal range sizes were calculated using the minimum-convex polygon method, modified to include 95% of the points closest to the median location for each wolf (Tracker, Version 1.1, Radio Location Systems AB, Huddinge, Sweden). For comparative purposes, we also calculated range sizes with the 95% fixed kernel method (Worton 1989) using leastsquares cross-validation to estimate the smoothing parameter (The Home Ranger, Version 1.5, Ursus Software, Revelstoke, British Columbia, Canada). That kernel method was considered less biased and more accurate than the adaptive kernel method (Seaman et al. 1999; Seaman and Powell 1996). We selected the best quality locations from each transmission period that were >= 18 h apart to include in the analyses. Locations obtained >=18 h apart were considered independent because wolves can move large distances and could potentially move anywhere within the seasonal home range in 18 h. For example, during summer, we documented 1 satellite-collared female that moved 92 km in <22 h.
Many investigators suggest that 30-120 locations are necessary to adequately describe annual territory sizes of wolves (Ballard et al. 1998; Carbyn 1983; Messier 1985a). Furthermore, when using kernel methods, >=50 locations are preferred (Seaman et al. 1999). Using a subsample of wolves (n = 8) we plotted summer range size (95% minimum convex polygon) versus sample size and found that 25-27 locations were required to reach an asymptote. Because we were only describing range sizes and not defining a territory, we selected all wolves for which we had >=28 locations.
Excursion movements were observed during a short period during summer. We defined an excursion as any location that was > 10 km from the 95% minimum convex polygon boundary. The approximate duration of those excursions was calculated from the middate between the last location within the boundary and the Ist location >10 km outside the boundary to the middate of the last location of the excursion and the Ist location in which the wolf had moved back inside the boundary. Duration of an excursion was only calculated for trips when there was a pre- and postexcursion location =<5 days (Messier 1985b). The average straight-line distance for each excursion was calculated by determining the distance between each excursion location and the closest segment of the home-range boundary and then calculating the mean distance for those locations.
Figure 1: Locations of wolves captured in the central Canadian Arctic during June 1997 and 1998.
Table 1: Seasonal and annual home range sizes of individual satellite-collared wolves in the central Canadian Arctic, Northwest Territories, 1997-1999.
Figure 2: Mean direction (degrees) from den site and average distance (km) from the closest segment of the summer range boundary for 8 female and 3 male wolves during 15 excursions, 19 June-12 July 1997-1998 in the central Canadian Arctic.