<<

Peer Reviewed Is the Migratory Behavior of Montane Herds in Peril? The Case of ’s Ya Ha Tinda Elk Herd

MARK HEBBLEWHITE,'’’^ Department of Biological Sciences, University of Alberta, Edmonton, AS T6E 2E9, Canada EVELYN H. MERRILL, Department of Bioiogioai Sciences, University of Alberta, Edmonton, AB T6E 2E9, Canada LUIGI E. MORGANTINI, Weyerhaeuser Ltd., Eorestiands Division, Edmonton, AB T5M 3N7, Canada, and Department of Renewable Resources, University of Alberta, Edmonton, AB T6G 2R2, Canada CLIFFORD A. WHITE, Banff Warden Service, Parks Canada, , Banff, AB TIL 1K6, Canada JAMES R. ALLEN, Alberta Sustainable Resource Development, Rooky Mountain House, AB TOM ORO, Canada ELDON BRUNS, Alberta Sustainable Resource Development, Rooky Mountain House, AB TOM ORO, Canada LINDA THURSTON, Department of Bioiogioai Sciences, University of Alberta, Edmonton, AB T6E 2E9, Canada TOM AS E. HURD, Banff Warden Service, Parks Canada, Banff National Park, Banff, AB TIL 1K6, Canada

Abstract

There is growing concern that populations of migratory are declining globally. Causes of declines in migratory behavior can be direct (i.e., differential harvest o f migrants) o r indirect (i.e., habitat fragmentation orland-use changes). fCervus Elk elaphusj are an important big game species in whose migratory behavior is changing in some montane ecosystems. We evaluated evidence and hypotheses for changes in migratory behavior and population decline in one of Canada’s largest elk populations, the Ya Ha Tinda. We compared the ratio of migrant to resident elk (M:R) in the population and seasonal spatial distributions obtained from 22 winter and 13 summer helicopter surveys between 1972 and 2005. Timing of migration and the summer distribution for a sample of radiocollared elk also was compared for 1977-1980 (early period) and 2001-2004 (recent). The population M:R ratio decreased from 12.4 (SD = 3.22) in the early period to 3.0 (SD = 1.63). The decrease was greater than expected based on population change. Declines in M:R also mirrored behavior of radiocollared elk. More than 49% of radiocollared elk we monitored resided near the winter range year-round by 2001-2004, and migrants were spending less time on summer ranges. We found winter range enhancements, access to hay fed to wintering horses, recolonization by gray fCanis lupusj, and management relocations of elk were most consistent with observed elk population growth (adjusted for harvest and removals) and the change in migratory behavior. However, we could not isolate the effects of specific factors in time-series population modeling. We believe directly relating migrant and resident demography to habitat and mortality factors will be required to understand the mechanisms affecting migratory behavior in this and other montane elk herds. (WILDLIFE SOCIETY BULLETIN 34(5): 1280-1294; 2006) Key words Alberta, Canada, Canis lupus, elaphus,elk, habitat enhancement, habituation, migrantresldent ratio, migration, prescribed fire, winter range, .

Migratory ungulates, such as {Connochaetes while in Asia, market hunting has been directly responsible taurinus) in the Serengeti-Mara ecosystem, elk {Cervus for population declines of migratory Mongolian elaphus) in the greater Yellowstone ecosystem, and barren- { gutturosa), Saiga {Saiga tatarica), and ground caribou {Rangifer tarandus) in the Arctic occupy a {Pantholops hodgsoniv, Schaller 1988, “keystone” ecosystem role, often defining and shaping Lhagvasuren and Milner-Gulland 1997, Arylov et al. ecosystems by their movements and effects on terrestrial 2004). In North America many migratory populations processes such as herhivory and predator-prey dynamics underwent similar declines historically because of direct (Houston 1982, Fancy et al. 1989, Sinclair 2003, Johnson et overharvest; for example, elk (Toweill and Thomas 2002). Recently, however, many populations have also faced al. 2005). Because of their critical ecosystem role, concern indirect causes of migratory decline that has changed the for worldwide changes in migratory behavior of ungulates is relative benefits of migrating, favoring growth of nonmi- mounting (Schaller 1988, Berger 2004, Johnson et al. 2005). gratory segments of populations. For example, recent North The reasons for these changes usually are complex and American migratory declines have been attributed to variable hut can he a result of direct (e.g., overharvest) or differential hunting pressure on migratory segments of elk indirect causes (e.g., habitat fragmentation). herds (Boyce 1989, Smith and Robbins 1994), habitat In Africa encroaching cultivation and poaching threaten fragmentation associated with oil and gas development for the Serengeti wildebeest migration (Thirgood et al. 2004), {Antilocapra americana\ Berger 2004), diamond mine exploration for barren-ground caribou (Johnson et al. ^ E-mall: [email protected] 2005), hydroelectric developments for woodland caribou {R. ^ Present address: Wildlife Biology Program, College of Forestry and Conservation, University of Montana, Missoula,tarandus\ Mahoney and Schaefer 2002), and also by creation M T 59812, USA of agricultural crops that, when combined with hunting

1280 Wildlife Society Bulletin • 34(5) sanctuaries, attract elk year-round (Burcham et al. 1999). seasons during June-August. Vegetation was classified into 3 Recovering wolf populations {Canis lupus) also m ay ecoregions: montane, suhalpine, and alpine. The montane influence migrations. In western Canada a 10-year ecoregion offered prime elk winter habitat and was decline in the ratio of migrant to resident elk in the Bow dominated by lodgepole pine {Pinus contortd) interspersed Valley (BY) elk herd of Banff National Park (BNP) was with Engelmann spruce {Picea engelmannii)—sNi\[osN {Salix correlated with human activity that created a predation spp.) areas, aspen {Populus /rcOTw/offfcrj-parkland, and refuge from wolves recolonizing these areas (Woods 1991, . Suhalpine and alpine ecoregions were composed McKenzie 2001, Hehhlewhite et al. 2005). Across much of of Engelmann spruce-suhalpine fir {Abies lasiocarpd)-\oA^e.- western North America, ungulate populations are faced with pole pine forest interspersed with willow-shruh riparian similar indirect and complex land-use changes that may communities, suhalpine grasslands, and avalanche terrain threaten the long-term viahility of migratory populations grading to open shruh-forh meadows in the alpine ecoregion. (Smith and Rohhins 1994, Berger 2004, Johnson et al. Holland and Coen (1983) provide a detailed description of 2005). study area vegetation. Ya Ha Tinda means “mountain W e examined population and migratory dynamics of the prairie” in the Stoney-Sioux language, aptly describing the Ya Ha Tinda (YHT) elk population in BNP from 1972 to azonal, high-elevation, 20-km^ montane rough fescue 2005 to determine if migration behavior had changed, {Festuca campestris) grasslands along the north side of the evaluate the consequences of migratory change to popula­ Red River (Fig. 1). The YHT represents one of the tion dynamics, and evaluate possible management and most pristine and largest rough fescue montane grasslands ecological factors affecting migration. In the 1970s, almost left in Alberta (Willoughby 2001). The area was mixed with the entire YHT population migrated 25-50 km west to aspen forests, open conifer stands, willow-hog hirch {Betula summer inside BNP (Morgantini and Hudson 1988). Since glandulosd) shruhlands and was surrounded by pine grading the late 1990s, concern has been mounting that the decline to spruce forests at higher elevations. soils in migratory behavior of the YHT elk herd is mirroring that consisted of azonal prairie types, including rich Orthic Black observed in the Bow Valley of BNP a decade earlier. We and Eluviated Black Chernozem (McGillis 1977, AGRA focused on the YHT elk herd because it is the largest elk Earth and Environmental Ltd. 1998). herd in BNP and one of the largest migratory herds in Elk were the most abundant ungulate in the study area Canada (Gunson 1997). The herd winters outside of BNP during the past 3 decades, ranging from 1,500 to 2,500 on the low-elevation grasslands of the YHT winter range. (Holroyd and Van Tighem 1983), and constituted Although this area was removed from BNP in 1931, Parks 70% of wolf diet (Hehhlewhite et al. 2004). The YHT elk Canada retained ownership of a 44-km^ ranch for training herd was partially migratory, with polymorphism for and wintering 100-200 horses on the winter range. Despite migrant and resident behavior. Migrant elk departed the federal ownership by Parks Canada, however, management winter range in May or June for summer ranges, returning to of natural resources is under Alberta provincial jurisdiction winter ranges from late Septemher-Decemher (Morgantini on the YHT. This transhoundary setting leads to differences 1988). Despite this movement into BNP in summer, elk in federal and provincial management objectives (Morgan­ from the YHT herd have shown little interchange with tini 1995, sensu Clark et al. 2000), for example, controversy other park elk herds (Morgantini and Hudson 1988, Woods over horse versus elk overgrazing (McGillis 1977, AGRA 1991). The YHT elk herd wintered outside BNP in the Earth and Environmental Ltd. 1998), differences in province of Alberta in 1 main and 2 secondary winter ranges predator management (protected inside BNP), and the (Fig. 1). The primary winter range for approximately 90% of restoration of fire as a natural disturbance (White et al. the elk herd (Hehhlewhite 2006) was the YHT (Wildlife 1998). We took advantage of this transhoundary study Management Unit [WMU] 418; Fig. 1). The 2 secondary design to make a priori predictions about how 8 manage­ ranges included the Panther-Dormer river corners (WMU ment and ecological factors would influence elk migration, 416) and Harrison-Lost Guide creek flats (WMU 420). and we tested whether predictions were consistent with Although elk dominated, white-tailed deer { observed changes in the migrant-to-resident ratio in the virginianus), {Alces alces), (O. hemionus), population (M:R), seasonal spatial distribution, and elk { canadensis), mountain {Oreamnos population dynamics using a broad hypothetico-deductive americanus), and a remnant herd of 5-8 mountain caribou approach. {R. tarandus) also occurred. Alternate prey species popula­ tion trends are less well-known, hut bighorn sheep have Study Area been relatively stable while deer species, moose, , and The study area included the front and main ranges of the caribou numbers appear to have declined in the study area Canadian Rocky Mountains in BNP (51°30'N, 115°30'W) since the mid-1980s (T. Hurd, Parks Canada, Banff and adjacent provincial lands, and was defined by movements National Park, unpublished data). See Hehhlewhite et al. of the YHT elk herd over a 6,000-km^ area (Fig. 1). (2004) for more information on wolf predation in this Elevations ranged from 1,600 m in valley bottoms to 3,500 multi-prey system. Other carnivores included grizzly hears m. The study area lay along the eastern slopes of the Rocky {Ursus arctos), black hears {U. americanus), cougars {Puma Mountains, with long, cold winters and short growing concolor), wolverines {Gulo gulo), and coyotes {Canis latrans).

Hebblewhite et al. • Migratory Elk Declines 1281 , ' \ - '. V ' B aiiff J^^tion^l Parlcvf .

V. N

Legend Louise Ya Ha Tinda • Summer EIK Telemetry

Banff National Park

R oads 30 KJIotnelets □ □ AB WMU'S

Figure 1. Location of thie Ya Ha Tinda elk population study area on thie eastern slopes of Banff National Park, Alberta, Canada, sfiowing Ya Ha Tinda Ranofi, major rivers, Alberta Wildlife Management Units (WMUs), and distribution of radiooollared elk from 2001 to 2004 during summer. Areas above 2,300 m are sfiaded gray.

Methods elk population size (TVj.) and population growth rate (rj. W e tested for evidence of migratory changes and evaluated Second, more than 1,000 elk were relocated from YHT in hypotheses underlying migratory patterns in the YHT the 1990s. If no hias occurred during capture, we would population. W e synthesized data collected over the 1972- expect no change in M:R following relocation. Thus, 2005 period to test for migratory changes from 3 major data changes in M:R following relocation suggest relocation- sources: one early telemetry study from 1977 to 1980 influenced migration (Tahle 1: H2). (Morgantini and Hudson 1988), federal and provincial Prescrihed fires occurred over the past 2 decades on the aerial survey and visual neck-handing data from 1970 to summer range of migratory elk in BNP (White et al. 2003). 2004, and one telemetry study during 2002-2004. Further, The positive effects of fire on eik (Boyce and Merrill 1991, we identified 8 hypotheses as potential causes for migratory Taper and Gogan 2002) would favor migrants, increasing declines that fell into 3 hroad categories: elk population M:R (Tahle 1: H3). In contrast, we expected winter range management, hahitat management, and wolf predation- hahitat enhancements would benefit resident elk more than related (Tahle 1). Because migratory elk historically migratory elk, thus decreasing the M:R ratio, because remained in BNP until after the regular autumn elk harvest residents inhabited the enhanced winter range year-round outside BNP ended (Morgantini 1988), we predicted that (Table 1: H4). In either case, hahitat enhancements also autumn harvests would reduce residents more than mi­ would he expected to increase N f and If competition grants, thus increasing the M:R ratio (Tahle 1: HI between elk and horses was limiting access to forage, a [Hypothesis 1]) and, given sufficiently high harvests, reduce reduction in number of horses would he expected to increase

1282 Wildlife Society Bulletin • 34(5) Table 1. Hypotheses (H1-H8) for effects of different classes of management actions and their predicted effects on migratory behavior and population size of the Ya Ha Tinda (YHT) eik herd, Alberta, Canada, 1970-2005. Management actions are predicted to increase (+) or decrease (-) the proportion of migrants and overaii popuiation size (N). Observed trends in the ratio of migrant to resident eik (M:R) and popuiation size over the 30-year period are presented for oomparison. Prediotions in boldface matched observations.

Observed Consistent Predicted effect on change with

Management action Hypothesis M:R N' Mechanism M:R M:R

HI: Eik harvest Differentiai harvest of resident eik shouid + - Eik harvest disproportionateiy reduces residents + No No cause M:R to increase. because most migrants do not return to the YHT for the whoie hunting season. H2: Eik reiooations Removai of 1,044 eik from YHT caused _? b Capture bias for migrant eik wouid reduce migrants b Yes Yes'" migrants to deoiine. and/or disrupt iearned migratory behavior. H3: Prescribed fire Burning on summer ranges shouid + + Burning increased forage in predominantiy conifer burns + No Yes increase migrant eik. avaiiabie to migrant eik (Saohro et ai. 2005). H4: Winter range Winter range enhancements increase + Resident eik remain on winter range aii year, benefiting + Yes Yes enhancements resident eik numbers. from enhanced forage during summer as a resuit of habitat enhancements (Morgantini 1995). H5: Winter horse numbers Deoiining horse numbers reieased eik from + Fewer horses shouid increase eik and decrease - n/a° Yes No'" range competition (Moineneiy 2003). M:R ratio because resident eik increase due to carry-over effects of winter horse grazing in summer. H6: Hay feeding Proionged a c c e s s to artifioiai food source + Hay feeding increased habituation of residents and + Yes Yes contributed to migratory deoiine. reduced migration (Burcham et ai. 1999). H7: Woif predation Spatiai separation through migration reduced Migrant eik shouid have iower predation risk + No No reiative predation risk for migrants. (Bergerud et ai. 1934), but reooionizing woives wouid stiii be predicted to reduce eik popuiation size (Hebbiewhite 2005). H8: Provinoiai woif harvest Differentiai harvest of woives in province Woif protection in Banff Nationai Park wouid reverse + Yes No surrounding YHT reduced predation the reiative benefits of migration under the spatiai on resident eik. separation hypothesis; overaii, eik N stiii deoiine (Hebbiewhite 2005).

® Note predicted effects of hypotheses on W, and are the same. Overall population Increased but declined following relocations. This prediction was only compared to the postrelocatlon time period. ° Horses only started declining (Table 2) following elk relocations, and during this period elk numbers were stable (Fig. 2) as a result of release from competition after relocation. elk N f and r^. Residents may be expected to benefit more when migrant and residents mix in large groups. Association because residents remain on the winter range year-round matrices confirmed thorough mixing of migrants and and would benefit most from carry-over effects of reduced residents during winter (Hebbiewhite 2006). Capture winter horse grazing on summer forage availability (Table 1: locations between studies were similar; one corral trap in H5; Mclnenely 2003). Open access to hay fed to horses both studies was located <1 km apart, and during 2002- during late winter may be associated with elk habituation to 2004 we used a second trap 3 km east of the first trap to humans, which may reduce M :R over time (e.g., Burcham et minimize potential capture bias. During both periods we al. 1999) hut with uncertain effects on elk population relocated radiocoUared elk biweekly, either from the ground dynamics (Tahle 1: H6). or aerially from fixed- or rotary-wing aircraft. During the In addition, wolves were just recolonizing the study area early 1977-1980 period, we resighted 9 neck-banded elk an during the late 1970s and were considered established by the average of 3.3 times/summer but 2 were never sighted again, early 1980s (Morgantini 1988). Migration is broadly suggesting potential sightability bias (see Morgantini 1988). hypothesized to reduce predation risk for migrant ungulates Because of different sampling intensities (GPS collars vs. (Bergerud et al. 1984, Fryxell et al. 1988). If true, migration neck-bands), we used collared animals to assess M:R ratio would he expected to increase the M:R ratio (Tahle 1: H7). and simple watershed distribution patterns between studies. However, as an extension to this hypothesis, wolf protection Research was conducted according to University of Alberta in BNP led to higher wolf survival between 1987 and 2000 Care Protocol 353112, Parks Canada EIA permit than adjacent provincial areas where wolves were harvested BNP-00047531, and Alberta research and collection (Callaghan 2002). If harvest was sufficiently high, survival permits GP4618 and CN087. differences could translate to higher relative wolf densities inside BNP, which could reduce the M:R ratio (Table 1: Changes in Migration Behavior H8). In an additive fashion to any direct gradient, high W e evaluated the M:R ratio, seasonal (spring and autumn) human activity on the YHT during summer may cause wolf migration dates, and the distribution of radiocollared elk. avoidance (Theuerkauf et al. 2003), potentially benefiting We calculated population-level M:R ratio using the resident demography and decreasing the M:R ratio similar maximum number of elk observed from air or ground to the BV elk herd (Hebbiewhite et al. 2005). Regardless, as during summer on the YHT winter range as a proportion of an important limiting factor, predation by recolonizing the following winter’s aerial count. W e compared popula­ wolves should reduce overall elk Nf and Vf (Hebbiewhite tion-level M:R between early (1977-1987) and late (1988- 2005). 2004) periods using an unbalanced /-test. As a second We used a broad hypothetico-deductive framework to measure, we compared M:R ratio of both radiocollared and examine predictions of these 8 hypotheses in comparison to neck-banded elk between the early and late intensive study observed population response and change in M:R ratio periods of 1977-1980 and 2002-2004. Despite winter herd (Table 1). If the predicted effect of a management mixing and our capture precautions, we tested for bias by hypothesis was consistent with observed changes in M:R comparing the M:R ratio of captured elk to the population population trend and helped explain elk population growth M:R ratio during each year using chi-square tests. rate, we considered this strong evidence that the hypothesis We defined migration as seasonal movement between influenced migratory and population dynamics. If a allopatric home ranges and estimated migration date as the management hypothesis was related to M:R but not Nf or midpoint between subsequent telemetry locations on Vf, we considered this weaker evidence of an overall alternate migratory ranges (Craighead et al. 1972). We migratory effect. Finally, if a management hypothesis was compared spring and autumn migration date by calculating consistent with elk Vf or Nf but not M:R, we concluded the the probability of early and late migration dates differing management hypothesis affected migrants and residents under the Z-approximation to the normal distribution equally. (Sokal and Rohlf 1995). We had no information on duration of migration for the early period. Thus, we Elk Capture and Monitoring assumed duration was similar to late period GPS-collar W e captured elk during 2 separate studies approximately 20 estimates of 5 days (M. Hebbiewhite, University of years apart, using 1 corral trap during the 1971-1980 study Montana, unpublished report). Whether the proportion of and 2 corral traps during the 2002-2004 study. W e deployed collared elk (both radio and visual neck-bands) on summer visual neck collars on 11 adult females in 1971-1973 and 11 ranges identified by Morgantini (1988) changed between radiocollars (Telonics Inc., Mesa, Arizona) in 1977-1980 (7 early and late periods was tested using a chi-square test adult F, 1 M and 2 F yearlings, 1 M calf), such that during (Sokal and Rohlf 1995). 1977-1980 there were 22 marked elk in the population. In 2002-2004 we marked 59 elk (50 adult F, 9 F yearlings) Spatiai Distribution with very high frequency radiocollars (LOTEK Inc., Parks Canada or the Alberta Fish and Wildlife Division Aurora, Ontario, Canada) and 20 elk (18 adult F, 2 F conducted aerial surveys in rotary-wing aircraft (Bell 206B yearlings) with Global Positioning System (GPS) 2200 or JetRanger, Fort Worth, Texas) every winter since 1972 3300 collars (LOTEK Inc). W e captured all elk on the main except 1974, 1975, 1984, and 1989 (Table 2), and YHT winter range between January and March of each year approximately every third summer since 1977 (1977, 1978,

1284 Wildlife Society Bulletin • 34(5) Table 2. Time series of eik data and potentiai factors influencing temporai popuiation dynamics of thie Ya Ha Tinda (YHT) eik hierd (Wiidiite Management Unit |WMU] 418), 1970-2005, Aiberta, Canada. Eik data inoiude winter eik count (A/f) and maximum summer count of residents (Nn-t) on YHT. Popuiation factors inoiude thie totai Jun-Aug precipitation, Banff Nationai Park (BNP) woif abundance index, number of eik transiooated, totai number of eik hiarvested, winter hiorse numbers, oumuiative area (km^) burned in thie provinoiai and BNP portions of thie study area, and oumuiative area (hia) of winter range hiabitat enhianoement projects.

Cumulative W M U 418 Maximum Jun-Aug Cumulative winter range winter elk summer resident precip. No. of No. of Total elk No. of burn area enhancements Year^ count Nf^ count N r (mm)'= BNP wolves'* elk relocated harvest® horses' (km^) (km^)

1973807 277.2 9 133 0.0 0.0 1974356 78.4 5 124 0.0 0.0 1975351 78.8 11 33 0.0 0.0 1976334 194.2 4 53 0.0 0.0 1977 358 34 125.3 4 56 136 0.0 0.0 1978 278 25 97.3 5 92 0.0 0.0 1979 422 25 203.4 5 135 0.0 0.0 1980378 88 5 74 0.0 0.0 1981 174.3 10 170 0.0 0.0 1982217 354.4 7 130 0.0 0.3 1983 200 182.1 16 136 0.0 0.5 19841058 75 221.1 23 160 0.0 0.3 1985 620 50 127.2 21 126 0.0 1.0 1986 77.9 20 76 4.2 1.3 1987 758 75 298.5 18 124 4.2 3.0 1988 918 209.3 25 150 4.2 6.9 1989 1075 180.4 29 170 130 6.5 7.7 1990 1052 140.6 30 131 133 21.6 3.0 1991 1285 245.6 35 63 196 21.6 3.3 1992 179.5 24 65 171 21.6 3.5 19932099 262.6 30 65 139 21.6 3.3 1994 1074 257.1 35 2299 67 190 43.0 9.0 1995 1534 370.6 24 132 67 152 43.0 9.3 19961642 99.6 25 324 73 43.0 9.5 1997 952 163.6 35 139 67 146 43.0 9.8 1998 901 313.4 31 135 121 43.0 10.0 1999 976 129.1 25 85 37 153 45.7 10.3 2000 843 200 178.6 25 91 155 63.3 10.5 2001 931 150 187.3 26 65 144 63.3 10.8 2002 991 324 73.9 36 93 147 130.1 11.0 2003 916 259 83.2 32 107 127 130.1 11.3 2004 848 267 29 113 95 130.1 11.5 Mean 829.1 121.7 188.6 19.1 29.8 105.5 160.9 25.3*' 4.3*' SD 412.01 106.77 84.89 11.49 74.45 40.13 27.3 33.3 4.7

® Year 2004 refers to biological year 2003-2004 from 1 Jun 2004 to 31 May 2005. Maximum number of resident eik counted on winter range WMU 418 during summer (1 Jun-31 Aug). ° Totai precipitation (mm) for Jun, Jui, and Aug reported at Blue Hiii Tower Environment Canada weather station. Woif popuiation index derived in Hebbiewhite (2006). ® Eik harvest includes aii age classes, and eik killed by guides, resident, nonresident, and native hunters. * Number of horses wintered at YHT includes brood mares and horses being trained. ® Only eik that did not return to YHT following relocation (Aiberta Fish and Wiidiite Division, unpublished data). Mean annual area burned or enhanced.

1979, 1980, 1982, 1983, 1984, 1985, 1987, 1991, 1998, winter aerial surveys 1-2 days after heavy snowfalls during 2003, 2004). We conducted surveys 100-200 m above the morning (0800-1200 hours) on sunny or flat-light days ground level at 50-70 km/hour. We conducted summer during January or February to maximize sightability of eik surveys in July during the morning (0600-1200 hours) on (Alien 2005). We photographed large herds (>50) for clear sunny days when elk were on high-elevation summer counting. Continuous participation since 1972 by one range and sightability was highest (Anderson et ai. 1998). author during winter surveys and by another author during During summer surveys, we surveyed aii alpine and summer surveys during both periods (early and recent) suhalpine summer eik ranges and key winter ranges ensured data consistency. Only winter popuiation counts, identified by Morgantini and Hudson (1988). Telemetry not spatiai data, were avaiiabie for aerial surveys from 1972 data from both early and late periods confirmed no major to 1977. After 1977 we recorded herd size, general herd summer ranges were missed during surveys (Morgantini and composition (male, female, mixed), activity, and location Hudson 1988, M. Hebbiewhite and L. Morgantini, and later transcribed these data to Universal Transverse University of Montana, unpublished report). We flew Mercator coordinates. We considered locations accurate

Hebblewtiite et al. • Migratory Elk Declines 1285 1 8 km away. During the first year approximately 50% (223) EU^N rvloM 1 6 returned to YHT. By the second year elk were moved far 2000 -^ISERRatio ^ enough away that return rate was <10% (Alberta Fish and 1 4 O Wildlife, unpublished data). Thus, we adjusted number of O 1 2 O elk translocated by 50% during the first year (Table 2). [7] 1500 ■43 16 rt W ithin BNP Parks Canada burned an average of 5.04 km^ 031 (0-25.4 km^; Tahle 2) per year of predominantly coniferous 8 P i S pine and spruce forest (81% conifer; White et al. 2003, 6 H Sachro et al. 2005) for a total of >88.0 km^ burned since 4 1986. In the provincial portion of the study area, one

7 prescribed burn of 7.01 km^ was conducted in 1994 in W M U 420, and one human-caused fire burned >60.0 km^ 0 I- during autumn 2001 in W M U 416. Fires only occurred in t-' I- r-- w <>. o\ areas inhabited by migrant elk during summer. Because of delayed effects of fire on elk forage (Sachro et al. 2005), we used an index of cumulative area burned (White et al. 2005) Figure 2. Winter aerial survey counts of eik from 1973 to 2004 in Wiidiite Management Unit 418, Aiberta, Canada (■), and migrant-to- to investigate fire’s effects on elk (Tahle 2). resident ratio (M:R) of eik (•). End of the iate-season eik hunt and eik Hahitat enhancement projects have been implemented in reiooations are shown. the YHT beginning in 1986 (reviewed by Gunson 1997). From 1987 to 1990, 3.25 km^ of shrub-encroached only to 500 m because of mapping differences over time. grasslands were mowed during July to reduce shrub Agency biologists occasionally conducted surveys in another (primarily hog hirch) encroachment and 1.78 km^ of the agency’s jurisdiction. W hen surveys overlapped in the same mowed area was also fertilized one time. Shrub-mowing has year, we used only agency area-specific data. Because aerial been standard ranch policy since 1982, with an average of survey area sometimes varied, we used the 90% kernel of 0.25 km^ mowed/year on a rotational basis during June- aerial eik sightings to define core seasonal survey areas. For August (R. Smith, Parks Canada, Ya Ha Tinda Ranch, each season we compared the spatiai distribution of eik personal communication). In 1990, 0.33 km^ of shrub- between time periods using muitipie-range permutation encroached grasslands in W M U 416 were also mowed. In procedures (MRPP; Berry and Mieike 1983) in program 1988, 3.16 km^ of mature conifer adjacent to the winter BLOSSOM (Cade and Richards 2001). The MRPP range was logged and seeded with nonnative grasses to compares intra-group Euclidean distances with distances enhance elk winter forage. W e expected an elk response calculated at random (Berry and Mieike 1983) and tests the from fertilizing, mowing, and logging because of demon­ hypothesis that the spatiai distribution of locations does not strated short-term increases in grass production (reviewed differ between >2 sampling occasions. by Morgantini 1995). Similar to fire, we used cumulative area of treated habitats to examine effects of winter hahitat Factors Influencing Migration enhancements on elk (Tahle 2). We obtained number of eik harvested during regular The number of horses pastured on the YHT ranch during hunting seasons by resident and nonresident (outfitter) winter (Nov-May) has averaged 150 until recently when hunters from 1972 to 2004 from W M U 418 (Fig. 1), from numbers have declined <100 (Table 2). Horses have been registered hunter phone surveys (Aiberta Fish and Wildlife, fed hay {Agrostis-Dactylis-Phleum spp. mix) during late unpublished reports) from 1986 to 2004, and from winter (Feh-Apr) since at least the late 1970s (Parks registered harvests prior to 1986 (Table 2). Hunting by Canada, unpublished data), and elk have access to hay First Nations is unreported and unregulated in Canada, hut provided for horses. Despite scant quantitative data, hay we obtained field estimates of First Nations harvest during feeding increased since the early 1990s when overgrazing years with field research and by Parks Canada ranch staff concerns resurfaced, accompanied with an increased fre­ (Parks Canada, Ya Ha Tinda Ranch, unpublished data). quency of hay depredations by elk (L. Morgantini, Late-season hunts occurred after aerial surveys in January- University of Alberta-Edmonton, and E. Bruns, Alberta Fehruary of 1969-1975, and elk harvest during these hunts Fish and Wildlife, Rocky Mountain House, personal were recorded at game-check stops and by registration communication). (Alberta Fish and Wildlife, unpublished annual harvest Wolves were extirpated throughout all of the Canadian reports). Rockies in the 1950s by poisoning and trapping (Gunson Concern for overgrazing grew in the early 1990s as the 1992), were considered rare during 1977-1980 (Morgantini YHT elk herd exceeded 2,200 elk (Fig. 2), and Alberta Fish 1988), but had naturally recovered by the mid-1980s and Wildlife Division (ABFW) relocated elk instead of (Paquet 1993). Currently 30-50 wolves in 4-5 packs overlap allowing controversial late-season hunts to mitigate over- the YHT elk population (Hebbiewhite 2006). Winter wolf grazing concerns (Gunson 1997). From 1994 to 1999, 1,273 numbers have been surveyed in the BNP portion of the elk (Table 2) were relocated from YHT to locations 20-100 study area through radiotelemetry or winter snowtrack

1286 Wiidiife Society Buiietin • 34(5) surveys since the mid-1940s (Table 2; reviewed in Hebbie­ collinearity >0.5 and developed a set of exploratory white 2006). Unfortunately, similar wolf trend data do not candidate generalized linear models (GLMs) of factors exist for adjacent provincial areas. Despite the potential for affecting elk Vf (Burnham and Anderson 1998). Generalized effects of harvest in Alberta, Alberta population trends linear models were of the general form should be coupled with wolf numbers inside BNP because 'N/+! all BNP wolf packs use adjacent Alberta lands (Callaghan rt In 2002). For example, average 100% annual wolf minimum Nf complex polygon territory size in our study was 1,229 km^ Po + ^\elkN {i) + p2^2 (^) + . . . + P„,Xot(/) + e, (« = 5; Hebbiewhite 2006), indicating the large spatial scales where / = 1 to 32 years, N is population size in year /, = involved with wolf populations. Thus, we assumed wolf population growth in year /, po is the intercept, p 2 • • • Pm are population trends in adjacent Alberta areas were the same as coefficients of independent variables X2 .. . and £ is in BNP, especially in regard to recovery from extirpation random error where ^(e;) = 0. We estimated GLMs using during the duration of the study. the identity link by maximum likelihood estimation in Stata W e included effects of summer precipitation (following 8.0 (StataCorp 2004). W e ranked models using Akaike’s Portier et al. 1998) and winter severity (Hebbiewhite 2005) Information Criterion adjusted for small sample size (AICJ, in population models. We obtained total June-August and where model selection uncertainty arose, a cutoff of precipitation (mm) from Environment Canada for Blue Hill AAIC,, = 2 was used to estimate the top model set (Burnham tower 20 km southeast of YHT for 1972-2004 (Table 2). and Anderson 1998). We ranked relative importance of W e used the North Pacific Oscillation (NPO) climate index covariates using Akaike weights (o;) following Burnham (Trenbertb and HurreU 1994) as an index of winter severity and Anderson (1998:141). In the lexicon of Burnham and for elk (Hebbiewhite 2005; available from ). to reveal insights for further research. Elk Population Dynamics We also examined population models of elk counts Because up to 90% of the regional elk population winters in unadjusted for harvest = ln(AYi-i/A(.)] to evaluate WMU 418 (YHT; Hebbiewhite 2006), we considered effects of harvest on population dynamics. W e estimated K WMU 418 winter counts of elk as representative of the by solving for W wben r/ = 0 in the harvest-adjusted and raw YHT elk population. We determined population growth elk models. W e used Akaike weights to select the top model rate (rj of elk wintering in W M U 418 from aerial counts set for r, and raw and constructed unconditional and adjusted for relocation and harvest by parameter estimates for coefficients to estimate K (Burnham and Anderson 1998). Comparing estimates of K from and r, = ln[(W+i + H,+r)/{N, - LH,)], raw Tf_ffffff models compares the effect of harvest on elk N where is adjusted elk population growth rate, is the (Sinclair and Caugbley 1994). number of elk “removed” (harvested) before winter surveys during year / + 1, and LHf is the num ber o f elk “removed” Results (late-season bunts and translocated elk) during year / after Ratio of Migrant to Resident Elk (M:R) population surveys (Merrill and Boyce 1991). The numer­ The average M:R observed during population surveys in the ator (W +i + 7/;_|_i) is the prebunt elk population during W+i> early period (12.4; SD = 3.22, n = 6) was higher than during whereas the denominator (W — LH^ is the p o st-late-bu nt/ the late period (3.0; SD = 1.67, n = 5, 4 ,0.05 = 4.35, P = relocation population size during year Nf. Adjusting for elk 0.002). If the M:R bad not declined, we would expect 47 harvest and relocation approximates population dynamics in residents during the late period, much lower than the the absence of bunting (Taper and Gogan 2002). We average of 246 residents observed (Fig. 2; xf = 705.1, P < assumed no poaching loss, no crippling loss, and additive 0.0001). There was slight evidence for capture bias in the harvest rates. M:R ratio between the captured and population estimates We modeled bow elk density (W)> horse numbers, wolf {x^ = 7.16, P < 0.03) such that the captured sample bad a numbers, habitat variables, summer precipitation, and the 6% bias towards residents. Despite this slight bias, the M:R winter NPO index (Table 2) affected over the 32-year ratio of the collared sample of elk generally was consistent time series. Given the a priori importance of density with the population M:R ratio from surveys. In 1977-1980, dependence, we retained elk Nf in all models and assumed 0 of 22 collared elk were residents. By 2003-2004, 49% of linear density dependence. Despite debate regarding density 79 collared elk were year-round residents (Fig. 3). dependence, linear density dependence provides a useful first step in analysis of population dynamics (Sinclair and Migration Dates Caugbley 1994). W e examined factors at one time lag and Migration dates were normally distributed, and the average included nonlinear climatic effects by including quadratic spring migration date of 9 radiocollared female elk terms. W e tested for the following interactions: 1) NPO and monitored in 1978 fell within the range of spring migration wolf numbers (Hebbiewhite 2005), 2) NPO and elk density dates of 79 female elk during the late period (Z-test, P = (Portier et al. 1998), and 3) summer rainfall and elk density 0.53). Autumn migration was one month earlier during the (Clutton-Brock et al. 1982). We screened variables for late period (Z-test, P = 0.04; Table 3).

Hebbiewhite et al. • Migratory Elk Declines 1287 Distrihution shifts observed in aerial surveys were mirrored hy distributions of radiocollared elk (Fig. 3). Spatial distrihution of collared elk differed between periods {x^ = 20.2, P = 0.003), with the largest increase in elk Early % occurring in year-round resident elk on the YHT Ranch and Late % the greatest decline in the elk in the Lake Louise and areas (Fig. 3).

Population Dynamics ^ 0.3 The best models of elk W included density dependence and either a negative effect of summer rainfall or cumulative burn area, or a positive effect of winter range enhancement (Tables 4, 5). O f the 3 retained covariates, the effect of fire was most variable and 95% confidence intervals overlapped zero (Tahle 4). We found high (r > 0.7) collinearity between elk W and cumulative hectares of winter range enhancement (r= 0.80), winter wolf numbers (r= 0.78), and Figures. Distribution of radioooiiared eik during the eariy (1978; n = 20) number of horses (r= 0.71). The high collinearity between and iate (2002-2003; n = 79) periods within major summer range areas elk Nf and winter range enhancement suggests caution is in Banff Nationai Park, Canada, identified by Morgantini and Hudson (1988). Winter range: LL, Lake Louise; RD, Red Deer; P, Panther; CW, warranted when interpreting the top models (Tahle 4), Ciearwater; O, Other; YHT, Ya Ha Tinda. Standard errors oaiouiated although parameter estimates should remain relatively assuming binomiai errors. unbiased (Freckleton 2002). Accordingly, we considered models as exploratory. Using the sum of Akaike weights for Spatial Distribution each variable (Burnham and Anderson 1998:141), we From 1977 to 2004, Parks Canada flew 9 surveys, ABFW ranked parameters in order of influence on elk population flew 16 surveys, of which 3 years overlapped, and 4 years had growth rate: summer rainfall (^O ; = 0.545), cumulative no surveys flown hy either agency, resulting in a total of 22 burn area (^O ; = 0.526), winter range hahitat enhance­ winter surveys. We grouped winter aerial surveys into 3 ments (^O ; = 0.332), previous-winter wolf numbers (^O ; periods with a balanced 7 surveys each: 1977-1986 (early), = 0.141), the rain X elk number interaction (^(fl; = 0.103), a 1986-1997 (mid-), and 1998-2004 (late periods; Fig. 4). nonlinear effect of rainfall (^O ; = 0.055), NPO (X)®; = The 90% adaptive kernel core area for elk locations during 0.023), NPO X elk numbers (^O ; = 0.004), hay (^O ; = winter surveys was 1,418 km^ (Fig. 4). A shift in winter elk 0.004), NPO X wolf numbers (^O ; = 0.003), and a distrihution occurred across the 3 time periods (MRPP A = nonlinear effect of NPO (^(fl; = 0.003). Other variables had 1.49, P < 0.0001; Fig. 4), with more elk observed outside E® ; < 0 .0001. Based on Akaike weights, only rainfall, BNP near YHT during the latter periods. cumulative area burned, and hahitat enhancements appeared Between 1977 and 2004, Parks Canada and ABFW flew related to elk r^. 12 and 1 summer surveys, respectively, resulting in 12 years Solving rt for Nf = 0 using unconditional parameter of summer survey data. W e grouped surveys into 2 relatively estimates (po and p,x;) for all top models resulted in K = balanced periods with 7 and 5 surveys, respectively, 1977- 1,285 (95% Cl = 1,098-1,471) when adjusted for hunting 1986 and 1986-2004 (Fig. 5) to align with winter periods. and management removals, and K = 954 (95% Cl 779- The 90% adaptive kernel core area for summer observations 1,124; Tahle 5) without adjusting for hunting and was 2,708 km^ (Fig. 5). Summer elk distributions shifted management removals (results for Tt-ra-w are not shown; (MRPP A = 1.41, P = 0.0006), with noticeable declines of Hehhlewhite 2006). With hunting and management elk in the of BNP and increases in elk near the removals, the YHT elk herd was about 25% lower, or 331 YHT (Fig. 5). fewer elk on average, than without harvest or removals.

Table 3. Midpoints of spring and autumn migration dates of radioooiiared eik in the Ya Ha Tinda popuiation for eariy (1978) and iate periods (2002, 2003), Banff Nationai Park, Aiberta, Canada.

Spring migration Autumn migration

Period Year Date SD(Date) Range N Date SD(Date) Range N

Eariy 1978 3 Jun 14.20 17.1 9 5 Nov 3.54 33.1 7® Late 2002 9 Jun 14.4 12.2 20 30 Sep 25.3 13.5 16® 2003 1 Jun 13.2 15.6 41 30 Got 27.2 17.1 33® Average iate 4 Jun 11.5 14.5 61 2 Got 27.1 16.0 54

' A utum nN is consistently iower than springN due to mortality, radiocoiiar failure, etc.

1288 Wiidiife Society Buiietin • 34(5) H a; Ya Ha T in d a'' Tinda

lO a t i o n

Legend

Ya Ha Tinda

Elk G ro u p SEzo 1 ■ 10 m 11-5D 51--HW

101 -5 0 0

501 * 2500

Rrvers

|_ _ | 90?t Gore $irfwy Area

] BaniT NBUanai ParK RiDacf^ I I ElevaEldn >225Qm

Figure 4. Winter elk distribution (Feb-Mar) during (a) early (1977-1985), (b) mid- (1986-1994), and (c) late (1995-2004) study periods in the Ya Ha Tinda elk population, Banff National Park, Alberta, Canada. The area within whioh 90% of all aerial observations ooourred is shown in the outline. Number of surveys flown per period was equal.

Evaluating Predictions population growth, the increase at YHT in summer was Evidence from both our hypothetico-deductive framework greater than expected due to population growth rate alone. and population dynamics models suggest observed trends in Therefore, despite small sample sizes of collared elk during M:R ratio and population dynamics were consistent with the early period, changes in collared and population samples’ predictions of hypotheses 4, 6, and 8, namely, winter range M:R ratio and distrihution revealed the same trends of enhancements, habituation due to hay feeding, and a wolf declining migration and distrihution shifts to year-round protection gradient in BNP (Tables 1, 4). However, we residence on the winter range. could not rule out potential effects of elk relocation (Tahle Our management hypotheses predictions that were the 1: H2). Observed elk population trends were opposite the most consistent with changing migratory behavior were predicted effects of elk harvest, prescrihed burning, or horse those benefiting resident over migrant elk. These included numbers (Tahle 1: H I, H3, and H5). Migratory changes winter range enhancement, access to hay, and possibly wolf also were opposite of predictions if migration reduced wolf avoidance of the YHT during summer. Resident elk would predation relative to residents (Tahle 1: H7). Our have benefited from winter range enhancements year-round population models revealed that only elk N^, prescrihed hy summering on improved ranges without migrating. The burns, summer rainfall, and perhaps hahitat enhancements importance of summer nutrition to elk condition and (Tahle 5) affected elk reproduction is now well documented (Cook et al. 2004). Winter range enhancements may have made winter ranges Discussion more nutritious during summer than high-elevation summer Our comparison of migratory and population dynamics of ranges, given trade-offs with wolf predation risk (Hehhle­ the YHT elk herd strongly suggests migration behavior has white 2006). While elk feeding on hay during winter may changed dramatically since the 1970s. The proportion of the provide energetic benefits, we believe an important effect of population migrating into BNP declined hy approximately hay feeding is as an attractant that leads to elk habituation to 75%, and migrant elk now return to the winter range almost humans and loss of traditional behavior (Burcham et al. one month earlier. These changes cannot he explained hy 1999, Smith 2001, Kloppers et al. 2005). Habituation to changes in average between the early (TV/ = 608) and late humans from hay feeding also would benefit elk in wolf (TV/ = 917) periods because M:R declined as TV/ increased. avoidance of human activity on the winter range. Numerous The shift in elk distrihution was most pronounced from the studies have documented carnivore avoidance of high front ranges of BNP to the YHT in winter, and a human activity (e.g., Theuerkauf et al. 2003). In BNP the corresponding increase from <30 elk in 1977 to >300 elk town site created a predation refuge that enhanced elk summering on YHT during 2002-2004 (Fig. 4). While the survival and recruitment (Hehhlewhite et al. 2005), leading increase in resident elk occurred during a period of general to migratory declines. While human use may he lower at the

Hebbiewhite et al. • Migratory Elk Declines 1289 5?

L « g e r d Va Tlndv

Elk Group S(Z9 ^ S^2& i ' 29-50 « re-iso Rhwrs ! ! □ 00^ CQK Survey I j ^ Banff Naligrul Pjrk R w ds I I Eleuaticti >225nm

0 2 4

Figure 5. Summer elk distribution of elk (Jui-Aug) during (a) early (1977-1986) and (b) iate (1986-2004) study periods in the Ya Ha Tinda elk population, Banff Nationai Park, Alberta, Canada. The area within whioh 90% of all aerial observations ooourred is shown in the outline. Number of surveys flown per period was equal.

YHT than in BNP, direct human-caused wolf mortality of fires are weaker or even negative on elk (White et al. from legal hunting and trapping 10 months of the year and 2005). Almost all previous studies demonstrating positive some illegal killing during the rest of the year (Hehhlewhite effects of fire on elk populations occurred in the absence of 2006) may reinforce wolf avoidance of human activity wolf predation (e.g.. Taper and Gogan 2002). Our results at (Theuerkauf et al. 2003) and foster development of least suggest an interaction between predation hy wolves and predation refugia, even if human use is lower than in the hahitat restoration through fire that has important man­ BV. agement implications for ecosystem management in national In contrast to research elsewhere on elk and fire, we found parks (W hite et al. 1998). little evidence that large prescrihed fires were effective at We suggest the hypothesized demographic benefits of increasing migratory elk numbers. In fact. Front Range elk migration (Bergerud et al. 1984, Fryxell et al. 1988; Tahle I: herds that had access to the largest prescrihed hums within H7) may not exist for migrants in the YHT elk herd: hy all BNP declined the most, and the effect of fire on population counts, residents seem to he doing relatively better. In growth was weakly negative, not positive as predicted. Our further support, during the 1980s elk resided along the results may relate to how we measured effects of burning front-range areas of BNP during winter. However, hy 2000 using a cumulative area burned following White et al. wintering elk populations within these areas had declined or (2005). Further, it was difficult to completely isolate effects shifted to the YHT. While these trends support the of burning in our study because the amount of area burned existence of a predation refugia, a comparison of wolf was correlated with declining elk and increasing wolf predation on resident elk relative to forage trade-offs is populations. Despite these caveats, however, we propose required to empirically test for this effect (Hehhlewhite the hypothesis that in the presence of wolf predation, effects 2006). In the absence of experimental approaches, other

Table 4. Popuiation growth rate (r?) model seieotion for the Ya Ha Tinda eik popuiation, Alberta, Canada, winters 1970-2004. Following Burnham and Anderson (1998), we report from generalized linear models, N, K, iogdikeiihood (LL), AAiCc (AiCc = Akaike’s information Criterion adjusted for small sample size), and AiC weights. We only report models within 0-2 AAiC^.

Model rank and structure N KLL AAlCc AiC weight

1: ElkA/® + Rain‘d 0.33 25 3 7.979 0 0.163 2: ElkA/ + Burn‘s + HE"' 0.35 25 4 9.244 0.325 0.139 3: ElkA/ + HE 0.33 25 3 7.644 0.669 0.117 4: ElkA/ + Burn 0.28 25 3 7.344 1.268 0.087

® ElkW is the postharvest elkN,. Average summer rainfall (mm) measured at Blue Hill tower, 20 km southeast of Ya Ha Tinda. Cumulative area burned (km ). Cumulative area affected by winter range habitat enhancements (km^).

1290 Wiidiife Society Buiietin • 34(5) Table 5. Model-averaged parameter estimates and unconditional SEs high densities to reduce because the rate at which woives for the top harvest removal-adjusted Vf eik popuiation growth rate models for the Ya Ha Tinda eik herd, Alberta, Canada, 1970-2004. idUed eik increased with winter severity (Hehhlewhite 2005). Although the NPO correlates strongly with climate n adj Model on the eastern slope of the Rockies (Trenherth and Hurreii

Param eter P SE 1994), azonal climatic conditions characterizing YHT (Morgantini 1995) may have weakened the climatic Intercept 0.440* 0.0904 Elk A/7 -0 .0 0 0 3 4 * 0.000045 signature of NPO. Alternately, because the popuiation did Rain'^ -0 .0 0 0 3 4 * 0.000127 not spend much time near K, density-ciimate interactions Burn° -0.0009 0.0027 may have not occurred. Habitat enhancement'^ 0.0154 0.0144 Our popuiation models also have important implications ® Elk Nt is raw elk count In /',_raw rnodel, and postharvest elk count for long-term controversies surrounding range management In the r, m odel. at the YHT. We found our assumption of simple linear Average summer rainfall (mm) measured at Blue Hill Tower, 20 density dependence was warranted, similar to eik studies km southeast of Ya Ha Tinda. ° Cumulative area burned (km^) In Banff National Park. elsewhere (e.g., Ciutton-Brock et ai. 1982, Merrill and Cumulative area of Ya Ha Tinda winter range enhancement (km^). Boyce 1991, Luhow and Smith 2004) and estimated * Parameter estimates are statistically significant at P = 0.05. carrying capacity (K) based on this density dependence. In comparison to studies elsewhere (Merrill and Boyce 1991, Luhow and Smith 2004), our estimates of K represent tools, such as resource selection functions (Boyce and ecological carrying capacity with predation rather than food- McDonald 1999), landscape-linked simulation models based K (Sinclair and Caugbley 1994). Hunting and (Turner et al. 1994), or habitat-linked demographic studies 0ohnson et al. 2004), will he required to understand the relocations reduced long-term N hy an average of 22% mechanisms of how predation risk and hahitat enhancement from approximately 1,285 to approximately 985, closer to interact to influence migratory behavior. sustainable range capacity assessments of K (e.g., AGRA An important management factor not directly tested was Earth and Environmental 1998). W ith or without hunting one of the most pervasive and difflcuit to quantify impacts: or relocations, long-term equilibrium for the population is the effects of human recreation on eik behavior. In the tow ard an N well below the maximum observed number of 1970s, Morgantini and Hudson (1979) documented dis­ elk of approximately 2,200. This peak in elk numbers placement of resident elk on the YHT hy motorized use, occurred after a series of intermediate precipitation sum­ and motorized human use was restricted in 1986. Recrea­ mers, and immediately after fires in BNP and winter range tional activity is now predominantly equestrian-hased, enhancements, and may represent a short-term overshoot of which appears to disturb elk less at YHT despite overall K. In this context, elk management (hunting and relocation) increases in human use (M. Hehhlewhite, personal obser­ was effective at reducing elk W closer to the 1,000 elk vation). Increased human activity, equestrian-hased or recommended based on range assessments for threatened otherwise, combined with direct human-caused wolf rough fescue conservation (McGillis 1977, AGRA Earth mortality, may repel wolves (Theuerkauf et al. 2003), and Environmental 1998). However, at the time range creating predation refugia (White et al. 1998). Further assessments were done, 170-200 horses were wintered at study of interactions between humans, wolves, and elk on YHT. W ith recent declines of wintered horses at YHT, it the YHT winter range is needed to confirm whether refugia may he worthwhile revisiting range assessments for are leading to reduced migration and whether a refuge is conservation of threatened fescue grasslands (Mclnenely spatiai (i.e., Banff townsite; Hehhlewhite et ai. 2005) or only 2003). temporai (e.g., Theuerkauf et al. 2003). As an immediate Differences in resource management policies between management implication, aversive conditioning similar to federal and provincial agencies across jurisdictional bound­ what has been used on eik in the Banff town site (Kloppers aries have facilitated creation of spatial gradients in et ai. 2005) may he necessary to counteract potentiai predation risk and hahitat that appear to favor resident elk predation refugia at the YHT. over migrants. National park policies protect wolves, while The only climatic effect we found was that increased provincial policies include liberal wolf harvests to promote summer rainfall decreased eik r^, similar to findings of elk population goals (Gunson 1997). Inside BNP, manage­ Ciutton-Brock et ai. (1982). Increased precipitation during ment seeks to reduce the negative effects of human June-August often produces snow in the Rocky Mountains recreation (Parks Canada 1997), while the province of and may delay spring plant phenology critical for calf Alberta has a more liberal recreation policy for the YHT survival and popuiation dynamics (Post and Klein 1999). area (Anonymous 1986). Direct wolf mortality and indirect We speculate the main effect of rainfall on eik may he wolf avoidance of higher human activity at YHT are, through reduced calf survival during wetter, colder summers therefore, emergent properties of the present transhoundary because of the frequency of spring and summer snowfall management policy framework. Similarly, Parks Canada during wetter summers (HoUand and Coen 1983). Winter seeks to restore long-term ecological conditions through severity, measured hy the NPO, also was unrelated to at / application of prescrihed fire to elk summer ranges (White = 0-2 lags. Nearby in the BV, severe winters interacted with et al. 1998), while the Alberta government had a more

Hebbiewhite et al. • Migratory Elk Declines 1291 conservative forest fire suppression program, albeit with a depend on the density of elk as their primary prey (Carbone growing prescribed fire program. Alberta provincial habitat and Gittleman 2002) and are thus dependent on continued enhancement policy has instead been focused on elk winter transhoundary migrations. Our research provides an illus­ range enhancement, whereas Parks Canada’s main objectives tration of the vital role that areas outside of protected areas for the winter range have been horse grazing and hay have in ecosystem management of national parks (Groom et feeding (Parks Canada 1987). These contrasting manage­ al. 1999). National parks with transhoundary populations of ment objectives pose a significant difficulty to development migratory ungulates must increase cooperative management of a common interest approach to the transhoundary with adjacent agencies to ensure key park processes are management of the YHT elk herd (Clark et al. 2000). maintained. Historically, there was been little effective coordination of management activities across the park boundary, though Acknowledgments recent coordination efforts should be continued and Funding from the following agencies made this study strengthened (e.g.. Parks Canada 2002). For example, the possible (in alphabetical order): Alberta Cooperative Con­ Bighorn cutblocks and the prescribed fire programs were servation Research Unit, Alberta Conservation Association, implemented hy provincial and federal agencies without Alberta Enhanced Career Development, Alberta Sustain­ regional assessment of their effects on the YHT elk herd. able Resource Development, Centre for Mathematical We contend that transhoundary management must be Biology, Challenge Grants in Biodiversity, Foothills Model coordinated through development of a common interest Forest, Foundation for North American Wild Sheep, approach, such as maintaining migration across the spatial Marmot, Mountain Equipment Co-op Enviro-fund, Parks extent of this elk herd (Clark et al. 2000). Canada, Patagonia, Rocky Mountain Elk Foundation, Transhoundary management of migratory elk herds will be Sunpine Forest Products Ltd., University of Alberta, and increasingly important because the factors that changed Weyerhauser, Inc. M. Hebbiewhite was supported by a migration of the YHT elk herd occur elsewhere across Canon National Parks Science Scholarship for the Amer­ western North America (e.g.. White and Garrott 2005). Our analyses indicate that isolating factors responsible for icas. W e especially thank the 2 generations of Parks Canada migratory changes with certainty will be difficult in complex wardens who assisted in all aspects of the research, management settings. We suggest there is sufficient provincial fish and wildlife officers and biologists, and most evidence to indicate that recolonization by wolves, winter importantly, the Ya Ha Tinda Ranch staff for their range habitat enhancements, and habituation to hay have assistance over the duration of the study. Notably, we thank contributed to migratory change. Therefore, these factors T. Skjonsberg, P. Jacobsen, J. Nylund, C. Hayes, R. Smith, merit primary consideration in future management of the and other wardens and ranch staff whose help ensured YHT elk herd. W ith recovering wolf populations present or collection of long-term data. We thank J. Wasylyk for imminent in many areas of western United States, many elk transcribing aerial survey data to GIS. W e thank pilot M. herds will face this new factor as an influence on migratory Dupuis for safe fixed-wing telemetry monitoring, and pilots behavior. Park and wildlife managers should be alert for from Alpine helicopters, Canmore, Alberta. W e also thank migratory changes in elk populations, given the important the mountain horses who carried us on our surveys, notably ecosystem ramifications of migration and the implications of Bandit (Ban), Cody, and others. Finally, we thank M. S. changes in migration for park management. For example, in Boyce, S. Boutin, C. Gates, F. Messier, and C. Cassady-St. BNP wolf and grizzly bear population viability ultimately Clair for constructive comments.

Literature Cited

AGRA Earth and Environmental Ltd. 1998. Panther Corners Upper Red of calving caribou—dispersion in mountains. Canadian Journal of Deer range inventory. Report on file with Aiberta Fish and Wiidiife Zooiogy 62:1566-1575. Division, Edmonton, Canada. Berry, K. J., and P. W. Mieike. 1983. Computation of finite popuiation Alien, J. R. 2005. Use of sightabiiity models and resouroe seieotion parameters and approximate probabiiity values for multi-response funotions to enhanoe aerial popuiation surveys of eik (Cervus elaphus) permutation procedures (MRPP). Communioations in Statistics— in Aiberta. Thesis, University of Aiberta, Edmonton, Canada. Simuiation and Computation 12:83-107. Anderson, C. R., D. S. Moody, B. L. Smith, F. G. Lindzey, and R. P. Boyce, M. 8. 1989. Ttie Jackson eik tierd: intensive wiidiife management in Nortin America. Cambridge University, Cambridge, Lanka. 1998. Development and evaluation of sightabiiity models for United Kingdom. summer eik surveys. Journal of Wiidiife Management 62:1055-1066. Boyce, M. S., and L. L. MoDonaid. 1999. Relating popuiations to Anonymous. 1986. Nordegg-Red Deer River sub-regionai integrated tiabitats using resouroe seieotion funotions. Trends in Eooiogy and resouroe plan. Aiberta Forestry, Resouroe Evaluation and Planning Evoiution 14:268-272. Division, Edmonton, Canada. Boyce, M. S., and E. hi. Merriii. 1991. Effects of 1988 fires on ungulates Aryiov, Y., V. Badm aev, A. Bekenov, J. Chim eg, A. Entwistie, Y. A. in Yeiiowstone Nationai Park. P rooeedings of thie Tail Tim bers Fire Craohev, B. Lhagvasuren, A. Lusohekina, D. Maiion, E. J. Miiner- Eooiogy Conference 17:121-132. Cuiiand, and V. Ukrainsky. 2004. The Saiga antelope—teetering on Burohiam, M., W. D. Edge, and C. L. Marcum. 1999. Eik use of private the brink but stiii oause for hope. 38:250-251. land refuges. Wiidiife Sooiety Buiietin 27:833-839. Berger, J. 2004. The last mile: how to sustain iong-distanoe migration in Burnhiam, K. P., and D. R. Anderson. 1998. Model seieotion and . Conservation Biology 18:320-331. inference: a praotioai information-tiieoretio approaoin. Springer- Bergerud, A. T., hi. E. Butler, and D. R. Miiier. 1984. Anti predator taotios Veriag, New York, New York, USA.

1292 Wiidiife Sooiety Buiietin • 34(5) Cade, B. S., and J. D. Richards. 2001. Blossom statistical software. Ecology and Society 10:31. < http://www.ecologyandsociety.org/ U.S. Geoiogioai Survey, Midoontinent Eooiogioai Science Center, Fort vol10/lss1/art31/>. Accessed 2000 Aug 31. Coiiins, Colorado, USA. Lhagvasuren, B., and E. J. Mllner-Gulland. 1997. The status and Callaghan, C. 2002. The eooiogy of gray wolf (Canis lupus) habitat use, management of the Mongolian Procapra qutturosa popula­ survival, and persistence in the central Rocky Mountains, Canada. tion. Oryx 31:127-134. Department of Zoology, University of Guelph, Guelph, Ontario, Lubow, B. C., and B. L. Smith. 2004. Population dynamics of the C anada. Ja ck so n elk herd. Journal of Wildlife M anagem ent 68:810-829. Carbone, C., and J. L. Gittleman. 2002. A common rule for the scaling Mahoney, S. P., and J. A. Schaefer. 2002. Hydroelectric development of carnivore density. Science 295:2273-2276. and the disruption of migration In caribou. Biological Conservation Clark, T. W., D. Casey, and A. Halverson. 2000. Developing sustainable 107:147-153. management policy for the National Elk Refuge, Wyoming. Yale McGillis, J. R. W. 1977. Range management of the Ya Ha Tinda. School of Forestry and Environmental Studies Bulletin Series, Yale C anadian Wildlife Service, Edm onton, Alberta, C anada. University, New Haven, Connecticut, USA. Mclnenely, L. E. 2003. Seasonal effects of defoliation on montane Clutton-Brock, T. H., F. E. G uineas, and S. D. Albon. 1982. Red deer: rough fescue (Festuca campestris Rydb.). University of Alberta, behavior and ecology of two sexes. University of Chicago, Chicago, Edmonton, Canada. Illinois, USA. McKenzie, J. A. 2001. The selective advantage of urban habitat use by Cook, J. G., B. K. Jo h nson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. elk in Banff National Park. University of Guelph, Guelph, Ontario, Bryant, and L. L. Irwin. 2004. Effects of summer-autumn nutrition and C anada. parturition date on reproduction and survival of elk. Wildlife Merrill, E. H., and M. S. Boyce. 1991. S um m er range and elk population Monographs 155. dynamics in Yellowstone National Park. Pages 263-273 in R. B. Craighead, J. J., G. Atwell, and B. W. O’Gara. 1972. Elk migration In Kelter and M. S. Boyce, editors. The Greater Yellowstone Ecosystem: and near Yellowstone National Park. Wildlife Monographs 29. redefining America’s wilderness heritage. Yale University, New Fancy, S. G., L. F. Rank, K. R. Whitten, and W. L. Regelln. 1989. Haven, Connecticut, USA. Seasonal movements of caribou In arctic as determined by Morgantini, L. E. 1988. Behavioral adaptive strategies of wapiti (Cervus satellite. Canadian Journal of Zoology 67:644-650. elaphus) In the Canadian Rocky Mountains. University of Alberta, Freckleton, R. P. 2002. On the misuse of residuals in ecology: Edmonton, Canada. regression of residuals vs. multiple regression. Journal of Animal Morgantini, L. E. 1995. The Ya Ha Tinda: an ecological overview. Ecology 71:542-545. Canadian Heritage, Parks Canada, Calgary, Alberta, Canada. Fryxell, J. M., J. Greever, and A. R. E. Sinclair. 1988. Why are migratory Morgantini, L. E., and R. J. Hudson. 1979. Human disturbance and ungulates so abundant? American Naturalist 131:781-798. habitat selection In elk. Pages 132-139 in M. S. Boyce and L. D. Groom, M., D. B. Je n sen , R. L. Knight, S. G atew ood, L. S. Mills, D. Hayden-WIng, editors. North American elk: ecology, behavior, and Boyd-Heger, and M. E. Soule. 1999. Buffer zones: benefits and management. University of Wyoming, Laramie, USA. dangers of compatible stewardship. Pages 171-198 in M. E. Soule Morgantini, L. E., and R. J. Hudson. 1988. Migratory patterns of the and J. Terborgh, editors. Continental conservation: scientific foun­ wapiti, Cervus elaphus, In Banff National Park, AB. Canadian Fleld- dations of regional reserve networks. Island, Washington, D.C., USA. Naturallst 102:12-19. Gunson, J. R. 1992. Historical and present management of wolves In Paquet, P. C. 1993. Summary reference document—ecological studies Alberta. Wildlife Society Bulletin 20:330-339. of recolonizing wolves In the central Canadian Rocky Mountains. G unson, J. R. 1997. Elk m anagem ent plan for Alberta. Wildlife Prepared by John/Paul and Associates for Parks Canada, Banff Management Division, Alberta Environmental Protection, Natural National Park Warden Service, Alberta, Canada. Resource Service, Edmonton, Canada. Parks Canada. 1987. Ya Ha Tinda management plan. Parks Canada, Hebbiewhite, M. 2005. Predation Interacts with the North Pacific Department of Canadian Heritage, Banff, Alberta, Canada. Oscillation (NPO) to Influence western North American elk population Parks Canada. 1997. Banff National Park management plan. Parks dynamics. Journal of Animal Ecology 74:226-233. Canada, Department of Canadian Heritage, Banff, Canada. Hebbiewhite, M. 2006. Linking predation risk and forage to partially Parks Canada. 2002. Regional forest management strategy: Banff migratory ungulate population dynamics. Dissertation, University of National Park. Environmental screening report. Parks Canada, Alberta, Edmonton, Canada. Department of Canadian Heritage, Banff, Alberta, Canada. Hebbiewhite, M., P. 0 . P aquet, D. H. Pletscher, R. B. L essard, and 0 . Portier, C., M. Festa-Blanchet, J.-M. Galllard, J. T. Jorgenson, and N. J. Callaghan. 2004. Development and application of a ratlo-estlmator G. Yoccoz. 1998. Effects of density and weather on survival of to estimate wolf killing rates and variance In a multlple-prey system. bighorn sheep lambs (Ovis canadensis). Journal of Zoology, London Wildlife Society Bulletin 3 1 :933-946. 245:271-278. Hebbiewhite, M., C. A. White, C. NIetvelt, J. M. McKenzie, T. E. Hurd, J. Post, E., and D. R. Klein. 1999. Caribou calf production and seasonal M. Fryxell, S. Bayley, and P. C. P aquet. 2005. Hum an activity range quality during a population decline. Journal of Wildlife mediates a trophic cascade caused by wolves. Ecology 76:2135- Management 63:335-345. 2144. Sachro, L. L., W. L. Strong, and C. C. Gates. 2005. Prescribed burning Holland, W. D., and G. M. Coen. 1983. Ecological (biophysical) land effects on summer elk forage availability In the sub-alplne zone, Banff classification of Banff and Jasper National Parks. Volume 1: National Park. Journal of Environmental Management 77:183-193. summary. Alberta Institute of Pedology, Edmonton, Canada. Schaller, G. B. 1988. Wildlife of the Tibetan Plateau. University of Holroyd, G. L., and K. J. Van Tighem. 1983. Ecological (biophysical) Chicago, Chicago, Illinois, USA. land classification of Banff and Jasper National Parks. Volume 3: the Sinclair, A. R. E. 2003. population regulation, keystone wildlife Inventory. C anadian Wildlife Service, Edm onton, Alberta, processes and ecosystem dynamics. Philosophical Transactions of C anada. the Royal Society of London Series B, Biological Sciences 358:1729- Houston, D. B. 1982. The Northern Yellowstone elk: ecology and 1740. management. Macmillan, New York, New York, USA. Sinclair, A. R. E., and G. Caughley. 1994. Wildlife ecology and Johnson, C. J., M. S. Boyce, R. L. C ase, D. H. Cluff, R. J. G au, A. management. Blackwell Science, London, United Kingdom. Gunn, and R. Mulders. 2005. Cumulative effects of human Smith, B. L. 2001. Winter feeding of elk In western North America. developm ents on arctic wildlife. Wildlife M onographs 160. Journal of Wildlife M anagem ent 65:173-190. Johnson, C. J., M. S. Boyce, C. C. Schwartz, and M. A. Haroldson. Smith, B. L., and R. L. Robbins. 1994. Migrations and m anagem ent of 2004. Modeling survival: application of the multiplicative hazards the Jackson elk herd. U.S. Department of the Interior, National model to Yellowstone grizzly bear. Journal of Wildlife Management Biological Survey, Washington, D.C., USA. 68:966-974. Sokal, R. R., and F. J. Rohlf. 1995. Biometry. W. H. Freem an, New Kloppers, E. L., C. C. S t Clair, and T. E. Hurd. 2005. Predator- York, New York, USA. resembling aversive conditioning for managing habituated wildlife. StataCorp. 2004. Stata 8.0. StataCorp, College Station, Texas, USA.

Hebbiewhite et al. • Migratory Elk Declines 1293 Taper, M. L., and P. J. G ogan. 2002. The northern Yellowstone elk: density d ep en d en o e and ollmatio oondltlons. Journal of Wildlife Management 66:106-122. Theuerkauf, J., W. JedrzejewskI, K. Sohmldt, and R. Gula. 2003. Spatlotemporal segregation of wolves from humans In the Blalowleza Forest (Poland). Journal of Wildlife Management 67:706-716. Thirgood, S., A. Mosser, S. Tham, G. Hoporaft, E. Mwangomo, T. MIengeya, M. Kllewo, J. Fryxell, A. R. E. SInolaIr, and M. Borner. 2004. Can parks protest migratory ungulates? The ease of the Serengeti wildebeest. Animal Conservation 7:113-120. Toweill, D. E., and J. W. Thomas. 2002. North Amerloan elk: eooiogy and management. Smithsonian Institution, Washington, B.C., USA, London, United Kingdom. Trenberth, K. E., and J. W. Hurrell. 1994. Deoadal atmosphere-ooean variations In the Paolflo. Climate Dynamlos 9:303-319. Turner, M. G., Y. Wu, L. L. Wallaoe, W. H. Rom m e, and A. Brenkert. 1994. Simulating winter Interaotlons among ungulates, vegetation, and fire In northern Yellowstone Park. Eooiogioai Applloatlons 4:472- 496. White, C. A., T. L. Hurd, M. Hebbiewhite, and I. R. Pengelly. 2005. Mitigating fire suppression, highway, and habitat fragmentation effeots In the Bow Valley eoosystem: preliminary evaluation of a (1988) In wildlife management from the University of Alberta. In 2001 before-after-oontrol-lmpaot (BACI) design with path analysis. Pages he reoelved the Wildlife Habitat Canada Forest Stewardship Award for 2 5 -3 9 in J. Innes and G. Jones, editors. Monitoring the effeotlveness his leadership In oarlbou oonservatlon and. In 2002, the Alberta of bioiogioai oonservatlon. Ministry of Forests, Ministry of Water, Air, Emerald Award for Life-Long Environmental Stewardship. His Interests Lands, Vanoouver, , Canada. Inoiude many aspeots of elk eooiogy and behavior, and the White, C. A., C. E. O lm sted, and C. E. Kay. 1998. A spen, elk, and fire In assessment and mitigation of oil and gas developments and forestry the Rooky Mountain national parks of North Amerloa. Wildlife Sooiety on wildlife. Clliford A. White (seoond from left) obtained his B.S. Bulletin 26:449-462. from University of Montana (1977), his M.S. from Colorado State White, C. A., I. R. Pengelly, M.-P. R ogeau, and D. Zell. 2003. University (1985), and his Ph.D. In forest eooiogy from University of Landsoape fire regimes and vegetation restoration In Banff National British Columbia (2001). He has worked with Parks Canada sinoe Park. Parks Canada Oooaslonal Paper BNP-2003-01. Parks Canada, Banff, Alberta, Canada. 1973 and Is presently the Eoosystem Management Coordinator for White, P. J., and R. A. Garrott. 2005. Yellowstone’s ungulates after Banff National Park. His researoh foous has been on the long-term wolves—expeotatlons, realizations, and prediotions. Bioiogioai Con­ roles of humans In determining fire regimes and wildlife populations In servation 125:141-152. the Canadian Rookies, Inoluding the oumuiative environmental effeots Willoughby, M. 2001. The rough fesoue dominated oommunlty types In of highway mitigation, wildlife oorrldor restoration, and presorlbed the foothills of north-oentral Alberta. Volume T/017. Publlo Lands burning. James R. Allen (oenter) Is a wildlife m anagem ent biologist Division, Sustainable Resouroe Development, Edmonton, Alberta, with the Alberta Fish and Wildlife Division In Rooky Mountain House, C anada. Alberta. He reoelved his Fish and Wildlife Teohnloal Diploma from Woods, J. C. 1991. Eooiogy of a partially migratory elk population. Lakeland College (1975), a B.So. from the University of Waterloo University of British Columbia, Vanoouver, Canada. (1985), and an M.So. from the University of Alberta (2005). Eldon Bruns (right) Is presently enjoying a w ell-deserved retirem ent from a Mark Hebbiewhite (left) Is an assistant professor In the Wildlife 34-year oareer as a wildlife management biologist with the Alberta Biology Program In the Department of Eoosystem and Conservation Fish and Wildlife Division. He reoelved his B.So. (1967) and M.S. Solenoes, University of Montana. He reoelved his B.So. from the (1969) degrees In zoology from the University of Calgary. He University of Guelph (1995), his M.S. from the University of Montana oonoentrated his efforts on wildlife habitat proteotlon and the (2000), and his Ph.D. In eooiogy from the University of Alberta (2006). management of large mammals In southern Alberta. Linda He has oonduoted researoh on the relationships between wolves, elk, Thurston (not plotured) has worked In the Rooky Mountains In and habitat, and Is Interested In wlldllfe-habltat relationships, Canada and the United States on various aspeots of wolf behavior population dynamlos, and oonservatlon biology of ungulates. Evelyn and eooiogy. She holds a B.S. In eooiogy, behavior, and evolution H. M errill (not plotured) Is a professor of eooiogy In the Department from University of Callfornla-San Diego (1995), and an M.S. In wildlife of Bioiogioai Solenoes, University of Alberta. She obtained her B.A. biology from Texas A&M University (2002). Her Interests lie In from St. Lawrenoe University (1974), her M.S. from University of Idaho predator-prey dynamlos and oonfllot resolution In wildlife Issues. She (1978), and her Ph.D. from the University of Washington (1987). She has most reoently worked as a biologist with Idaho Fish and Came. worked for Idaho Department of Fish and Came prior to entering aoademio positions at the University of Wyoming and University of Tomas E. Hurd (not plotured) Is ourrently leading a Parks Canada WIsoonsln-Stevens Point. Her ourrent area of researoh addresses feasibility study to Investigate the potential for a new national park In meohanlsms of animal behavior In predator-prey relationships and the South Okanagan area of British Columbia. He previously held the disease spread In heterogeneous environments. L u lg l E. position of Wildlife Biologist, Banff National Park. He holds a B.So. In M o rg a n tin i (seoond from right) Is the Chief Biologist for zoology (1985) and an M.So. In forest solenoes (1999) both from the Weyerhaeuser Company In Alberta and an adjunot professor at the University of British Columbia. University of Alberta. He holds a Dootorate In Solenoe from the University of Rome (Italy; 1974), and an M.So. (1979) and a Ph.D. Associate Editor: Whittaker.

1294 Wildlife Sooiety Bulletin • 34(5)