Peer Reviewed

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

MARK HEBBLEWHITE,'’’ D epartm ent o f B io logica l Sciences, University o f Alberta, Edmonton, A S T6E 2E9, Canada

EVELYN H. MERRILL, D epartm ent o f B ioiogioai Sciences, University o f Alberta, Edmonton, A B T6E 2E9, Canada

LUIGI E. MORGANTINI, W eyerhaeuser Ltd., Eorestiands Division, Edmonton, A B T5M 3N7, Canada, and D epartm ent o f Renewable Resources, University o f Alberta, Edmonton, A B T6G 2R2, Canada

CLIFFORD A. WHITE, B an ff W arden Service, Parks Canada, B an ff N ational Park, Banff, A B T IL 1K6, Canada

JAMES R. ALLEN, Alberta Sustainable Resource Development, Rooky M ounta in House, A B TOM ORO, Canada

ELDON BRUNS, Alberta Sustainable Resource Development, Rooky M ounta in House, A B TOM ORO, Canada

LINDA THURSTON, D epartm ent o f B ioiogioai Sciences, University o f Alberta, Edmonton, A B T6E 2E9, Canada

TOMAS E. HURD, B an ff W arden Service, Parks Canada, B an ff National Park, Banff, A B T IL 1K6, Canada

AbstractThere is grow ing concern that populations o f m igratory ungulates are declin ing globally. Causes o f declines in m igratory behavior can be d irect (i.e., differential harvest o f migrants) o r ind irect (i.e., habitat fragmentation orland-use changes). Elk fCervus elaphusj are an im portant big game species in North America whose m igratory behavior is changing in some montane ecosystems. We evaluated evidence and hypotheses fo r changes in m igratory behavior and population decline in one o f Canada’s largest elk populations, the Ya Ha Tinda. We com pared the ratio o f m igrant to resident elk (M:R) in the population and seasonal spatial distributions obtained from 22 w in ter and 13 sum m er he licopter surveys between 1972 and 2005. Timing o f m igration and the sum m er distribution fo r a sample o f radiocollared elk also was com pared fo r 1977-1980 (early period) and 2001-2004 (recent). The population M :R ratio decreased from 12.4 (SD = 3.22) in the early pe riod to 3.0 (SD = 1.63). The decrease was greater than expected based on population change. Declines in M :R also m irrored behavior o f radiocollared elk. M ore than 49% o f radiocollared elk we m onitored resided near the w inter range year-round b y 2001-2004, and m igrants were spending less time on sum mer ranges. We found w in ter range enhancements, access to hay fed to wintering horses, recolonization b y gray wolves fCanis lupusj, and management relocations o f e lk were m ost consistent w ith observed elk population grow th (adjusted fo r harvest and removals) and the change in m igratory behavior. However, we cou ld no t isolate the effects o f specific factors in time-series population modeling. We believe d irectly relating m igrant and resident demography to habitat and m ortality factors w ill be required to understand the mechanisms affecting m igratory behavior in this and other montane elk herds. (WILDLIFE SOCIETY BULLETIN 34(5): 1280-1294; 2006)

Key wordsAlberta, Canada, Canis lupus, Cervus elaphus, elk, habitat enhancement, habituation, migrantresldent ratio, migration, prescribed fire, winter range, wolf.

M igratory ungulates, such as w ildebeest {Connochaetes taurinus) in the Serengeti-M ara ecosystem, elk {Cervus elaphus) in the greater Yellowstone ecosystem, and barren- ground caribou {Rangifer tarandus) in the A rctic occupy a “keystone” ecosystem role, often defining and shaping ecosystems by their movem ents and effects on terrestrial processes such as herhivory and predator-prey dynamics (H ouston 1982, Fancy et al. 1989, Sinclair 2003, Johnson et al. 2005). Because o f their critical ecosystem role, concern for w orldwide changes in m igratory behavior o f ungulates is m ounting (Schaller 1988, Berger 2004, Johnson et al. 2005). T h e reasons for these changes usually are complex and variable hu t can he a result o f direct (e.g., overharvest) or indirect causes (e.g., habitat fragm entation).

In Africa encroaching cultivation and poaching threaten the Serengeti wildebeest m igration (Thirgood et al. 2004),

E-mall: mark.hebblewhlte@cfc.umt.edu ^ Present address: Wildlife Biology Program, College o f Forestry and Conservation, University o f Montana, Missoula, M T 59812, USA

while in Asia, m arket hun ting has been directly responsible for population declines o f m igratory M ongolian gazelles {Procapra gutturosa), Saiga antelope {Saiga tatarica), and T ibe tan antelope {Pantholops hodgsoniv, Schaller 1988, Lhagvasuren and M ilner-G u lland 1997, Arylov et al.2004). In N orth Am erica m any m igratory populations underw ent similar declines historically because o f direct overharvest; for example, elk (Toweill and Thom as 2002). Recently, however, m any populations have also faced indirect causes o f m igratory decline th a t has changed the relative benefits o f m igrating, favoring grow th o f nonm i- gratory segments o f populations. F or example, recent N orth A m erican m igratory declines have been attribu ted to differential hunting pressure on m igratory segments o f elk herds (Boyce 1989, Sm ith and Robbins 1994), habitat fragm entation associated w ith oil and gas developm ent for pronghorn {Antilocapra americana\ Berger 2004), diam ond m ine exploration for barren-ground caribou (Johnson et al.2005), hydroelectric developments for w oodland caribou {R. tarandus\ M ahoney and Schaefer 2002), and also by creation o f agricultural crops that, w hen com bined w ith hunting

1280 W ildlife Society Bulletin • 34(5)

sanctuaries, attract elk year-round (Burcham et al. 1999). R ecovering w o lf popu la tions {Canis lupus) also m ay influence ungulate m igrations. In western C anada a 10-year decline in the ratio o f m igrant to resident elk in the Bow Valley (BY) elk herd o f B anff N ational P ark (BN P) was correlated w ith hum an activity th a t created a predation refuge from wolves recolonizing these areas (W oods 1991, M cK enzie 2001, H ehhlew hite et al. 2005). Across m uch o f w estern N orth America, ungulate populations are faced w ith similar indirect and complex land-use changes th a t may threaten the long-term viahility o f m igratory populations (Sm ith and Rohhins 1994, Berger 2004, Johnson et al.2005).

W e examined population and m igratory dynamics o f the Ya H a T inda (Y H T) elk population in B N P from 1972 to 2005 to determ ine if m igration behavior had changed, evaluate the consequences o f m igratory change to popula­tion dynamics, and evaluate possible m anagem ent and ecological factors affecting m igration. In the 1970s, almost the entire Y H T population m igrated 25 -50 km w est to sum m er inside B N P (M organtini and H udson 1988). Since the late 1990s, concern has been m ounting th a t the decline in m igratory behavior o f the Y H T elk herd is m irroring that observed in the Bow Valley o f B N P a decade earlier. W e focused on the Y H T elk herd because it is the largest elk herd in B N P and one o f the largest m igratory herds in C anada (G unson 1997). T h e herd w inters outside o f B N P on the low-elevation grasslands o f the Y H T w inter range. A lthough this area was removed from B N P in 1931, Parks C anada retained ownership o f a 44-km ^ ranch for training and w intering 100-200 horses on the w inter range. D espite federal ownership by Parks Canada, however, m anagem ent o f natural resources is under A lberta provincial jurisdiction on the Y H T . T his transhoundary setting leads to differences in federal and provincial m anagem ent objectives (M organ­tin i 1995, sensu C lark et al. 2000), for example, controversy over horse versus elk overgrazing (M cG illis 1977, A G R A E arth and E nv ironm ental L td . 1998), differences in predator m anagem ent (protected inside B N P), and the restoration o f fire as a natural disturbance (W hite et al. 1998). W e took advantage o f this transhoundary study design to m ake a priori predictions about how 8 m anage­m ent and ecological factors would influence elk migration, and we tested w hether predictions were consistent w ith observed changes in the m igran t-to -residen t ratio in the population (M :R), seasonal spatial distribution, and elk population dynamics using a broad hypothetico-deductive approach.

Study AreaT h e study area included the front and m ain ranges o f the Canadian Rocky M ountains in B N P (51°30'N , 115°30'W ) and adjacent provincial lands, and was defined by movements o f the Y H T elk herd over a 6,000-km^ area (Fig. 1). Elevations ranged from 1,600 m in valley bottom s to 3,500 m. T h e study area lay along the eastern slopes o f the Rocky M ountains, w ith long, cold w inters and short growing

seasons during June-A ugust. V egetation was classified into 3 ecoregions: m ontane, suhalpine, and alpine. T h e m ontane ecoregion offered prim e elk w in te r h ab ita t and was dom inated by lodgepole pine {Pinus contortd) interspersed w ith Engelm ann spruce {Picea engelmannii)—sNi\[osN {Salix spp.) areas, aspen {Populus /rcOTw/offfcrj-parkland, and grasslands. Suhalpine and alpine ecoregions were composed o f Engelm ann spruce-suhalpine fir {Abies lasiocarpd)-\oA^e.- pole pine forest interspersed w ith w illow -shruh riparian com m unities, suhalpine grasslands, and avalanche terrain grading to open shruh-forh meadows in the alpine ecoregion. H olland and C oen (1983) provide a detailed description o f study area vegetation. Ya H a T inda means “m ountain prairie” in the Stoney-Sioux language, aptly describing the azonal, h igh-elevation , 20-km ^ m ontane rough fescue {Festuca campestris) grasslands along the north side o f the Red D eer River (Fig. 1). T h e Y H T represents one o f the m ost pristine and largest rough fescue m ontane grasslands left in A lberta (W illoughby 2001). T h e area was mixed w ith aspen forests, open conifer stands, w illow -hog hirch {Betula glandulosd) shruhlands and was surrounded by pine grading to spruce forests at h igher elevations. G rassland soils consisted o f azonal prairie types, including rich O rth ic Black and Eluviated Black C hernozem (M cG illis 1977, A G R A E arth and Environm ental L td. 1998).

E lk were the m ost abundant ungulate in the study area during the past 3 decades, ranging from 1,500 to 2,500 animals (H olroyd and Van T ighem 1983), and constituted 70% o f w olf diet (H ehhlew hite et al. 2004). T h e Y H T elk herd was partially m igratory, w ith polym orphism for m igrant and resident behavior. M igran t elk departed the w inter range in M ay or June for sum m er ranges, returning to w inter ranges from late Septem her-D ecem her (M organtini 1988). D espite this m ovem ent into B N P in summer, elk from the Y H T herd have shown little interchange w ith o ther park elk herds (M organtini and H udson 1988, W oods1991). T h e Y H T elk herd w intered outside B N P in the province o f A lberta in 1 m ain and 2 secondary w inter ranges (Fig. 1). T h e prim ary w inter range for approximately 90% o f the elk herd (H ehhlew hite 2006) was the Y H T (W ildlife M anagem ent U n it [W M U ] 418; Fig. 1). T h e 2 secondary ranges included the P an ther-D orm er river corners (W M U 416) and H arrison -L ost G uide creek flats (W M U 420).

A lthough elk dom inated, w hite-tailed deer {Odocoileus virginianus), moose {Alces alces), mule deer (O. hemionus), b ighorn sheep {Ovis canadensis), m ountain goats {Oreamnos americanus), and a rem nant herd o f 5 -8 m ountain caribou {R. tarandus) also occurred. A lternate prey species popula­tion trends are less well-known, hu t bighorn sheep have been relatively stable while deer species, moose, goat, and caribou numbers appear to have declined in the study area since the m id-1980s (T . H urd , Parks Canada, B anff N ational Park, unpublished data). See H ehhlew hite et al.(2004) for m ore inform ation on w o lf predation in this m ulti-prey system. O th er carnivores included grizzly hears {Ursus arctos), black hears {U. americanus), cougars {Puma concolor), wolverines {Gulo gulo), and coyotes {Canis latrans).

Hebblewhite et al. • M igratory Elk Declines 1281

, ' \ - ' . V '

B a i i f f J ^ ^ t i o n ^ l P a r l c v f .

V. N

Louise

30K J Io tn e le t s

LegendYa Ha Tinda

• Sum m er EIK Telemetry

B anff N ationa l Park

Roads

□ □ A B W M U 'S

Figure 1. Location of thie Ya Ha Tinda elk population study a rea on thie eas tern s lopes of Banff National Park, Alberta, C anada, sfiowing Ya H a Tinda Ranofi, major rivers, Alberta Wildlife M anagem ent Units (WMUs), and distribution of radiooollared elk from 2001 to 2004 during sum m er. A reas above 2 ,300 m are sfiaded gray.

MethodsW e tested for evidence o f m igratory changes and evaluated hypotheses underlying m igratory patterns in the Y H T population. W e synthesized data collected over the 1972- 2005 period to test for m igratory changes from 3 m ajor data sources: one early telem etry study from 1977 to 1980 (M organtini and H udson 1988), federal and provincial aerial survey and visual neck-handing data from 1970 to 2004, and one telem etry study during 2002-2004. Further, we identified 8 hypotheses as potential causes for m igratory declines tha t fell into 3 hroad categories: elk population m anagem ent, hahitat m anagem ent, and w olf p reda tion - rela ted (Tahle 1). Because m igratory elk h istorically rem ained in B N P until after the regular autum n elk harvest outside B N P ended (M organtini 1988), we predicted tha t autum n harvests would reduce residents m ore than m i­grants, thus increasing the M :R ratio (Tahle 1: H I [Hypothesis 1]) and, given sufficiently h igh harvests, reduce

elk population size (TVj.) and population grow th rate ( r j . Second, m ore than 1,000 elk were relocated from Y H T in the 1990s. I f no hias occurred during capture, we would expect no change in M :R following relocation. Thus, changes in M :R following relocation suggest relocation- influenced m igration (Tahle 1: H 2).

Prescrihed fires occurred over the past 2 decades on the sum m er range o f m igratory elk in B N P (W hite et al. 2003). T h e positive effects o f fire on eik (Boyce and M errill 1991, T aper and G ogan 2002) w ould favor m igrants, increasing M :R (Tahle 1: H 3). In contrast, we expected w inter range hahitat enhancem ents w ould benefit resident elk m ore than m igratory elk, thus decreasing the M :R ratio, because residents inhabited the enhanced w inter range year-round (Table 1: H 4). In either case, hahitat enhancem ents also w ould he expected to increase N f and I f com petition between elk and horses was lim iting access to forage, a reduction in num ber o f horses would he expected to increase

1282 W ildlife Society Bulletin • 34(5)

Table 1. H ypotheses (H1-H8) for effects of different c la s se s of m anagem en t actions and their p redicted effects on migratory behavior and population size of th e Ya H a Tinda (YHT) eik herd, Alberta, C anada, 1 9 7 0-2005 . M anagem ent actions are p redicted to increase (+) or d e c re a se (-) th e proportion of m igrants and overaii popuiation size (N). O bserved tren d s in th e ratio of migrant to resident eik (M:R) and popuiation size over th e 30-year period are p re sen ted for oom parison. Prediotions in boldface m atched observations.

Predicted effect onObserved

changeConsistent

with

Managem ent action Hypothesis M:R N ' Mechanism M:R M:R

H I : Eik harvest Differentiai harvest of resident eik shouid c a u s e M:R to increase.

+ - Eik harvest disproportionateiy red u ces residents b e c a u se m ost m igrants do not return to th e YHT for th e w hoie hunting se aso n .

+ No No

H2: Eik reiooations Removai of 1 ,044 eik from YHT cau sed m igrants to deoiine.

_ ? b C apture bias for m igrant eik wouid reduce m igrants and /o r disrupt iearned migratory behavior.

b Yes Yes'"

H3: Prescribed fire Burning on sum m er ranges shouid increase migrant eik.

+ + Burning increased forage in predom inantiy conifer burns avaiiabie to migrant eik (Saohro e t ai. 2005).

+ No Yes

H4: W inter range en h an ce m e n ts

W inter range en h an ce m e n ts increase resident eik num bers.

+ R esident eik remain on winter range aii year, benefiting from en h an ce d forage during sum m er a s a resuit of habitat e n h an ce m e n ts (Morgantini 1995).

+ Yes Yes

H5: W inter horse num bers Deoiining ho rse num bers reieased eik from range com petition (Moineneiy 2003).

+ Few er h o rses shouid increase eik an d d ec re a se M:R ratio b e c a u se resident eik increase due to carry-over effects of winter horse grazing in sum m er.

- n/a° Yes No'"

H6: Hay feeding Proionged a c c e s s to artifioiai food source contributed to migratory deoiine.

+ Hay feeding increased habituation of residen ts and reduced migration (Burcham et ai. 1999).

+ Yes Yes

H7: Woif predation Spatiai separa tion through migration reduced reiative predation risk for m igrants.

Migrant eik shouid have iower predation risk (Bergerud e t ai. 1934), but reooionizing w oives wouid stiii be p redicted to reduce eik popuiation size (Hebbiewhite 2005).

+ No No

H8: Provinoiai woif harvest Differentiai harvest of w oives in province surrounding YHT reduced predation on resident eik.

Woif protection in Banff Nationai Park wouid reverse th e reiative benefits of migration under th e spatiai separa tion hypothesis; overaii, eik N stiii deoiine (Hebbiewhite 2005).

+ Yes No

® Note p red ic ted e ffec ts o f hypotheses on W, and are the same.Overall popu la tion Increased bu t declined fo llow in g re locations. Th is p red ic tion w as on ly com pared to the postre loca tlon tim e period.

° H orses on ly s ta rted dec lin ing (Table 2) fo llow in g e lk re locations, and during th is period e lk num bers w ere s tab le (Fig. 2) as a result o f re lease from com pe tition a fte r re location .

elk N f and r . Residents may be expected to benefit m ore because residents rem ain on the w inter range year-round and would benefit m ost from carry-over effects o f reduced w inter horse grazing on sum m er forage availability (Table 1: H 5; M clnenely 2003). O pen access to hay fed to horses during late w inter may be associated w ith elk habituation to hum ans, w hich may reduce M :R over tim e (e.g., Burcham et al. 1999) hu t w ith uncertain effects on elk population dynamics (Tahle 1: H 6).

In addition, wolves were just recolonizing the study area during the late 1970s and were considered established by the early 1980s (M organ tin i 1988). M ig ration is broadly hypothesized to reduce predation risk for m igrant ungulates (Bergerud et al. 1984, Fryxell et al. 1988). I f true, m igration would he expected to increase the M :R ratio (Tahle 1: H 7). However, as an extension to this hypothesis, w o lf protection in B N P led to higher w olf survival between 1987 and 2000 than adjacent provincial areas w here wolves were harvested (Callaghan 2002). I f harvest was sufficiently high, survival differences could translate to higher relative w o lf densities inside B N P, w hich could reduce the M :R ratio (Table 1: H 8). In an additive fashion to any direct gradient, high hum an activity on the Y H T during sum m er may cause w olf avoidance (T heuerkauf et al. 2003), potentially benefiting resident dem ography and decreasing the M :R ratio similar to the BV elk herd (H ebbiew hite et al. 2005). Regardless, as an im portan t lim iting factor, predation by recolonizing wolves should reduce overall elk Nf and Vf (Hebbiew hite2005).

W e used a broad hypothetico-deductive fram ework to examine predictions o f these 8 hypotheses in com parison to observed population response and change in M :R ratio (Table 1). I f th e pred ic ted effect o f a m anagem ent hypothesis was consistent w ith observed changes in M :R population trend and helped explain elk population growth rate, we considered this strong evidence th a t the hypothesis influenced m igra to ry and pop u la tio n dynam ics. I f a m anagem ent hypothesis was related to M :R bu t not Nf or Vf, we considered th is w eaker evidence o f an overall m igratory effect. Finally, i f a m anagem ent hypothesis was consistent w ith elk Vf or Nf bu t not M :R , we concluded the m anagem ent hypothesis affected m igrants and residents equally.

Elk Capture and MonitoringW e captured elk during 2 separate studies approximately 20 years apart, using 1 corral trap during the 1971-1980 study and 2 corral traps during the 2002-2004 study. W e deployed visual neck collars on 11 adult females in 1971-1973 and 11 radiocollars (Telonics Inc., M esa, Arizona) in 1977-1980 (7 adult F, 1 M and 2 F yearlings, 1 M calf), such th a t during 1977-1980 there were 22 m arked elk in the population. In 2002-2004 we m arked 59 elk (50 adult F, 9 F yearlings) w ith very h igh frequency radiocollars (L O T E K Inc., Aurora, O ntario , Canada) and 20 elk (18 adult F, 2 F yearlings) w ith G lobal Positioning System (G PS) 2200 or 3300 collars (L O T E K Inc). W e captured all elk on the main Y H T w inter range between January and M arch o f each year

w hen m igrant and residents mix in large groups. Association matrices confirm ed thorough m ixing o f m igrants and residents during w in ter (H ebbiew hite 2006). C apture locations between studies were similar; one corral trap in bo th studies was located < 1 km apart, and during 20 0 2 - 2004 we used a second trap 3 km east o f the first trap to m inim ize potential capture bias. D uring bo th periods we relocated radiocoUared elk biweekly, either from the ground or aerially from fixed- or rotary-w ing aircraft. D uring the early 1977-1980 period, we resighted 9 neck-banded elk an average o f 3.3 tim es/sum m er bu t 2 were never sighted again, suggesting potential sightability bias (see M organtin i 1988). Because o f different sampling intensities (G PS collars vs. neck-bands), we used collared animals to assess M :R ratio and simple w atershed distribution patterns between studies. Research was conducted according to University o f A lberta Anim al Care Protocol 353112, Parks C anada E IA perm it B N P -0 0 0 4 7 5 3 1 , and A lberta research and collection perm its G P4618 and CN 087.

Changes in Migration BehaviorW e evaluated the M :R ratio, seasonal (spring and autum n) m igration dates, and the distribution o f radiocollared elk. W e calculated population-level M :R ratio using the maxim um num ber o f elk observed from air or ground during sum m er on the Y H T w inter range as a proportion o f the following w inter’s aerial count. W e com pared popula­tion-level M :R between early (1977-1987) and late (1988-2004) periods using an unbalanced /-test. As a second measure, we com pared M :R ratio o f bo th radiocollared and neck-banded elk between the early and late intensive study periods o f 1977-1980 and 2002-2004. D espite w inter herd m ixing and our capture precautions, we tested for bias by com paring the M :R ratio o f captured elk to the population M :R ratio during each year using chi-square tests.

W e defined m igration as seasonal m ovem ent between allopatric hom e ranges and estim ated m igration date as the m id p o in t betw een subsequent te lem etry locations on alternate m igratory ranges (Craighead et al. 1972). W e com pared spring and autum n m igration date by calculating the probability o f early and late m igration dates differing under the Z -approxim ation to the norm al distribution (Sokal and R ohlf 1995). W e had no inform ation on duration o f m igration for the early period. T hus, we assumed duration was similar to late period G PS-collar estim ates o f 5 days (M . H ebbiew hite , U niversity o f M ontana, unpublished report). W h ether the proportion o f collared elk (both radio and visual neck-bands) on sum m er ranges identified by M organtin i (1988) changed between early and late periods was tested using a chi-square test (Sokal and R ohlf 1995).

Spatiai DistributionParks C anada or the A lberta Fish and W ildlife Division conducted aerial surveys in rotary-w ing aircraft (Bell 206B JetRanger, F ort W orth , Texas) every w inter since 1972 except 1974, 1975, 1984, and 1989 (Table 2), and approximately every th ird sum m er since 1977 (1977, 1978,

1284 W ildlife Society Bulletin • 34(5)

Table 2. Time se ries of eik d a ta an d potentiai facto rs influencing tem porai popuiation dynam ics of thie Ya H a Tinda (YHT) eik hierd (Wiidiite M anagem ent Unit |WMU] 418), 1970-2005 , Aiberta, C an ad a . Eik d a ta inoiude winter eik cou n t (A/f) and m axim um sum m er count of residen ts (Nn-t) on YHT. Popuiation fac to rs inoiude thie totai Jun-A ug precipitation, Banff Nationai Park (BNP) woif ab u n d a n c e index, num ber of eik transiooated , totai num ber of eik hiarvested, winter hiorse num bers, oumuiative a rea (km^) burned in thie provinoiai and BNP portions of thie study area, and oumuiative area (hia) of winter range hiabitat enhianoem ent projects.

Year^

W MU 418 winter elk count Nf^

Maximum summer resident

count N r

Jun-Augprecip.(mm)'=

No. of BNP wolves'*

No. of elk relocated

Total elk harvest®

No. of horses'

Cumulative burn area

(km^)

Cumulative winter range

enhancements (km^)

1973 807 277.2 9 133 0.0 0.01974 356 78.4 5 124 0.0 0.01975 351 78.8 11 33 0.0 0.01976 334 194.2 4 53 0.0 0.01977 358 34 125.3 4 56 136 0.0 0.01978 278 25 97.3 5 92 0.0 0.01979 422 25 203.4 5 135 0.0 0.01980 378 88 5 74 0.0 0.01981 174.3 10 170 0.0 0.01982 217 354.4 7 130 0.0 0.31983 200 182.1 16 136 0.0 0.51984 1058 75 221.1 23 160 0.0 0.31985 620 50 127.2 21 126 0.0 1.01986 77.9 20 76 4.2 1.31987 758 75 298.5 18 124 4.2 3.01988 918 209.3 25 150 4.2 6.91989 1075 180.4 29 170 130 6.5 7.71990 1052 140.6 30 131 133 21.6 3.01991 1285 245.6 35 63 196 21.6 3.31992 179.5 24 65 171 21.6 3.51993 2099 262.6 30 65 139 21.6 3.31994 1074 257.1 35 2299 67 190 43.0 9.01995 1534 370.6 24 132 67 152 43.0 9.31996 1642 99.6 25 324 73 43.0 9.51997 952 163.6 35 139 67 146 43.0 9.81998 901 313.4 31 135 121 43.0 10.01999 976 129.1 25 85 37 153 45.7 10.32000 843 200 178.6 25 91 155 63.3 10.52001 931 150 187.3 26 65 144 63.3 10.82002 991 324 73.9 36 93 147 130.1 11.02003 916 259 83.2 32 107 127 130.1 11.32004 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

® Y ear 2004 re fers to b io log ica l yea r 2 0 03 -2 004 from 1 Jun 2004 to 31 M ay 2005.M axim um num ber o f res ident e ik coun ted on w in te r range W M U 418 during sum m er (1 Jun-31 Aug).

° Tota i p rec ip ita tion (mm) fo r Jun, Jui, and Aug reported a t Blue Hiii T ow er E nvironm ent C anada w ea the r sta tion.W o if popu ia tion index derived in H ebb iew h ite (2006).

® Eik harvest inc ludes aii age classes, and e ik k illed by gu ides, resident, nonresident, and native hunters.* N um ber o f horses w in te red at YH T inc ludes b rood m ares and horses be ing tra ined.® O nly e ik th a t d id not return to YH T fo llow in g re loca tion (A iberta Fish and W iid iite D ivision, unpub lished data).

Mean annual area burned o r enhanced.

1979, 1980, 1982, 1983, 1984, 1985, 1987, 1991, 1998, 2003, 2004). W e conducted surveys 100-200 m above ground level at 50-70 km /hour. W e conducted sum m er surveys in July during the m orning (0600-1200 hours) on clear sunny days w hen elk were on high-elevation sum m er range and sightability was highest (A nderson et ai. 1998). D u rin g sum m er surveys, w e surveyed aii alpine and suhalpine sum m er eik ranges and key w in te r ranges identified by M organtin i and H udson (1988). Telem etry data from bo th early and late periods confirmed no major sum m er ranges were missed during surveys (M organtini and H u d so n 1988, M . H ebb iew h ite and L. M o rg an tin i, U niversity o f M ontana, unpublished report). W e flew

w inter aerial surveys 1-2 days after heavy snowfalls during the m orning (0800-1200 hours) on sunny or flat-light days during January or February to maximize sightability o f eik (Alien 2005). W e photographed large herds (> 5 0 ) for counting. C ontinuous participation since 1972 by one author during w inter surveys and by another author during sum m er surveys during bo th periods (early and recent) ensured data consistency. O nly w inter popuiation counts, not spatiai data, were avaiiabie for aerial surveys from 1972 to 1977. A fter 1977 we recorded herd size, general herd com position (male, female, mixed), activity, and location and later transcribed these data to Universal Transverse M ercator coordinates. W e considered locations accurate

Hebblew tiite et al. • M igratory Elk Declines 1285

EU^N rvloM-^ IS E R R a tio ^2000

O

[7] 1500

O

H

r-- wt-' I - I -<>. o\

1 8

1 6

1 4

1 2 O■43

16 rt031

8 P i

S6

4

7

0

Figure 2. W inter aerial survey c o u n ts of eik from 1973 to 2004 in Wiidiite M anagem ent Unit 418, Aiberta, C a n ad a (■), and m igrant-to- resident ratio (M:R) of eik (• ) . End of th e ia te-seaso n eik hunt and eik reiooations are show n.

only to 500 m because o f m apping differences over tim e. Agency biologists occasionally conducted surveys in another agency’s jurisdiction. W h en surveys overlapped in the same year, we used only agency area-specific data. Because aerial survey area sometimes varied, we used the 90% kernel o f aerial eik sightings to define core seasonal survey areas. For each season we com pared the spatiai distribution o f eik between tim e periods using m uitipie-range perm utation procedures (M R PP; Berry and M ieike 1983) in program B L O S S O M (C ade and R ichards 2001). T h e M R P P compares in tra-group Euclidean distances w ith distances calculated at random (Berry and M ieike 1983) and tests the hypothesis th a t the spatiai distribution o f locations does not differ between > 2 sampling occasions.

Factors Influencing MigrationW e obtained num ber o f eik harvested during regular hunting seasons by resident and nonresident (outfitter) hunters from 1972 to 2004 from W M U 418 (Fig. 1), from registered hunter phone surveys (A iberta Fish and W ildlife, unpub lished reports) from 1986 to 2004, and from registered harvests prior to 1986 (Table 2). H un ting by F irst N ations is unreported and unregulated in Canada, hut we obtained field estimates o f First N ations harvest during years w ith field research and by Parks C anada ranch staff (Parks Canada, Ya H a T inda Ranch, unpublished data). Late-season hunts occurred after aerial surveys in January- Fehruary o f 1969-1975, and elk harvest during these hunts were recorded at gam e-check stops and by registration (Alberta F ish and W ildlife, unpublished annual harvest reports).

C oncern for overgrazing grew in the early 1990s as the Y H T elk herd exceeded 2,200 elk (Fig. 2), and A lberta Fish and W ildlife Division (AB FW ) relocated elk instead o f allowing controversial late-season hunts to m itigate over- grazing concerns (G unson 1997). F rom 1994 to 1999, 1,273 elk (Table 2) were relocated from Y H T to locations 20-100

km away. D uring the first year approximately 50% (223) returned to Y H T . By the second year elk were moved far enough away th a t return rate was < 10% (Alberta F ish and W ildlife, unpublished data). T hus, we adjusted num ber o f elk translocated by 50% during the first year (Table 2).

W ith in B N P Parks C anada burned an average o f 5.04 km^ (0-25 .4 km^; Tahle 2) per year o f predom inantly coniferous pine and spruce forest (81% conifer; W h ite et al. 2003, Sachro et al. 2005) for a total o f > 8 8 .0 km^ burned since 1986. In the provincial portion o f the study area, one prescribed burn o f 7.01 km^ was conducted in 1994 in W M U 420, and one hum an-caused fire burned > 6 0 .0 km^ during autum n 2001 in W M U 416. Fires only occurred in areas inhabited by m igrant elk during summer. Because o f delayed effects o f fire on elk forage (Sachro et al. 2005), we used an index o f cumulative area burned (W hite et al. 2005) to investigate fire’s effects on elk (Tahle 2).

H ah ita t enhancem ent projects have been im plem ented in the Y H T beginning in 1986 (reviewed by G unson 1997). F rom 1987 to 1990, 3.25 km^ o f shrub-encroached grasslands w ere m ow ed during July to reduce shrub (prim arily hog hirch) encroachm ent and 1.78 km^ o f the m owed area was also fertilized one time. Shrub-m ow ing has been standard ranch policy since 1982, w ith an average o f 0.25 km^ m owed/year on a rotational basis during Ju n e - A ugust (R. Sm ith, Parks Canada, Ya H a T inda Ranch, personal com m unication). In 1990, 0.33 km^ o f shrub- encroached grasslands in W M U 416 were also mowed. In 1988, 3.16 km^ o f m ature conifer adjacent to the w inter range was logged and seeded w ith nonnative grasses to enhance elk w inter forage. W e expected an elk response from fertilizing, mowing, and logging because o f dem on­strated short-term increases in grass production (reviewed by M organtin i 1995). Similar to fire, we used cumulative area o f treated habitats to examine effects o f w inter hahitat enhancem ents on elk (Tahle 2).

T h e num ber o f horses pastured on the Y H T ranch during w inter (N ov-M ay) has averaged 150 until recently w hen num bers have declined < 1 0 0 (Table 2). H orses have been fed hay {Agrostis-Dactylis-Phleum spp. mix) during late w inter (Feh-A pr) since at least the late 1970s (Parks Canada, unpublished data), and elk have access to hay provided for horses. D espite scant quantitative data, hay feeding increased since the early 1990s w hen overgrazing concerns resurfaced, accom panied w ith an increased fre­quency o f hay depredations by elk (L . M o rg an tin i, University o f A lberta-E dm onton, and E. Bruns, A lberta F ish and W ildlife, Rocky M o u n ta in H ouse, personal com m unication).

W olves were extirpated th roughout all o f the C anadian Rockies in the 1950s by poisoning and trapping (G unson1992), were considered rare during 1977-1980 (M organtini 1988), bu t had naturally recovered by the m id-1980s (Paquet 1993). C urrently 30-50 wolves in 4 -5 packs overlap the Y H T elk population (H ebbiew hite 2006). W in ter w olf num bers have been surveyed in the B N P portion o f the study area th rough radiotelem etry or w inter snowtrack

1286 W iidiife Society Buiietin • 34(5)

surveys since the m id-1940s (Table 2; reviewed in H ebbie­w hite 2006). U nfortunately, similar w olf trend data do not exist for adjacent provincial areas. D espite the potential for effects o f harvest in Alberta, A lberta population trends should be coupled w ith w o lf num bers inside B N P because all B N P w o lf packs use adjacent A lberta lands (Callaghan2002). For example, average 100% annual w o lf m inim um complex polygon territory size in our study was 1,229 km^ (« = 5; H ebbiew hite 2006), indicating the large spatial scales involved w ith w olf populations. T hus, we assumed w olf population trends in adjacent A lberta areas were the same as in B N P, especially in regard to recovery from extirpation during the duration o f the study.

W e included effects o f sum m er precipitation (following P ortier et al. 1998) and w inter severity (H ebbiew hite 2005) in population models. W e obtained total June-A ugust precipitation (mm) from E nvironm ent C anada for Blue H ill tower 20 km southeast o f Y H T for 1972-2004 (Table 2). W e used the N orth Pacific Oscillation (N P O ) climate index (T renbertb and HurreU 1994) as an index o f w inter severity for elk (H ebbiew hite 2005; available from < b ttp ://w w w . cgd.ucar.edu/cas/jburrell/indices.btm l>).

Elk Population DynamicsBecause up to 90% o f th e regional elk population w inters in W M U 418 (Y H T; H ebbiew hite 2006), we considered W M U 418 w inter counts o f elk as representative o f the Y H T elk population. W e determ ined population growth rate ( r j o f elk w intering in W M U 418 from aerial counts and adjusted for relocation and harvest by

r, = ln[(W +i + H ,+ r)/{N , - LH ,)],

w here is adjusted elk population grow th rate, is the num ber o f elk “ rem oved” (harvested) before w inter surveys during year / + 1, and LHf is the num ber o f elk “ rem oved” (late-season bunts and translocated elk) during year / after population surveys (M errill and Boyce 1991). T h e num er­ator (W +i + 7/;_|_i) is the prebunt elk population during W+i> whereas the denom inator (W — LH^ is the p o st- la te -b u n t/ relocation population size during year Nf. A djusting for elk harvest and relocation approximates population dynamics in the absence o f bun ting (Taper and G ogan 2002). W e assumed no poaching loss, no crippling loss, and additive harvest rates.

W e m odeled bow elk density (W)> horse numbers, w olf numbers, habitat variables, sum m er precipitation, and the w inter N P O index (Table 2) affected over the 32-year tim e series. G iven the a priori im portance o f density dependence, we retained elk Nf in all models and assumed linear density dependence. D espite debate regarding density dependence, linear density dependence provides a useful first step in analysis o f population dynamics (Sinclair and Caugbley 1994). W e examined factors at one tim e lag and included nonlinear climatic effects by including quadratic terms. W e tested for the following interactions: 1) N P O and w olf num bers (H ebbiew hite 2005), 2) N P O and elk density (Portier et al. 1998), and 3) sum m er rainfall and elk density (C lu tton-B rock et al. 1982). W e screened variables for

collinearity > 0 .5 and developed a set o f exploratory candidate generalized linear models (G L M s) o f factors affecting elk Vf (Burnham and A nderson 1998). G eneralized linear models w ere o f the general form

rt In'N/+!

NfPo + ^\elkN {i) + p2 2 ( ) + . . . + P„,Xot(/) + e,

w here / = 1 to 32 years, N is population size in year /, =population grow th in year /, po is the intercept, p2 • • • Pm are coefficients o f independent variables X2 .. . and £ is random error w here ^ ( e ; ) = 0. W e estim ated G L M s using the identity link by maximum likelihood estim ation in Stata 8.0 (S tataC orp 2004). W e ranked models using Akaike’s Inform ation C riterion adjusted for small sample size (A IC J , and w here model selection uncertainty arose, a cutoff o f AAIC,, = 2 was used to estim ate the top model set (Burnham and A nderson 1998). W e ranked relative im portance o f covariates using A kaike w eights (o;) following Burnham and A nderson (1998:141). In the lexicon o f B urnham and A nderson (1998), our analyses were exploratory and m eant to reveal insights for further research.

W e also examined population models o f elk counts unadjusted for harvest = ln(AYi-i/A(.)] to evaluateeffects o f harvest on population dynamics. W e estim ated K by solving for W w b e n r/ = 0 in the harvest-adjusted and raw elk models. W e used A kaike w eights to select the top modelset for r, and raw and constructed unconditionalparam eter estimates for coefficients to estim ate K (Burnham and A nderson 1998). C om paring estimates o f K from and raw Tf_ffffff models compares the effect o f harvest on elk N (Sinclair and Caugbley 1994).

ResultsRatio of Migrant to Resident Elk (M:R)T h e average M :R observed during population surveys in the early period (12.4; SD = 3.22, n = 6) was higher than during the late period (3.0; SD = 1.67, n = 5, 4 ,0.05 = 4.35, P = 0.002). I f the M :R bad not declined, we w ould expect 47 residents during the late period, m uch lower than the average o f 246 residents observed (Fig. 2; xf = 705.1, P < 0.0001). T here was slight evidence for capture bias in the M :R ratio between the captured and population estimates {x = 7.16, P < 0.03) such th a t the captured sample bad a 6% bias towards residents. D espite this slight bias, the M :R ratio o f the collared sample o f elk generally was consistent w ith the population M :R ratio from surveys. In 1977-1980, 0 o f 22 collared elk were residents. By 2003-2004, 49% o f 79 collared elk were year-round residents (Fig. 3).

Migration DatesM igration dates were normally distributed, and the average spring m ig ra tion date o f 9 rad iocollared fem ale elk m onitored in 1978 fell w ith in the range o f spring m igration dates o f 79 female elk during the late period (Z-test, P = 0.53). A utum n m igration was one m onth earlier during the late period (Z -test, P = 0.04; Table 3).

Hebbiewhite et al. • M igratory Elk Declines 1287

Early %

Late %

^ 0.3

F igures. Distribution of radioooiiared eik during the eariy (1978; n = 20) and iate (2002-2003; n = 79) periods within m ajor sum m er range a reas in Banff Nationai Park, C an ad a , identified by Morgantini and H udson (1988). W inter range: LL, Lake Louise; RD, R ed Deer; P, Panther; CW, Ciearwater; O, Other; YHT, Ya Ha Tinda. S tandard errors oaiouiated assum ing binomiai errors.

Spatial DistributionFrom 1977 to 2004, Parks Canada flew 9 surveys, A B F W flew 16 surveys, o f w hich 3 years overlapped, and 4 years had no surveys flown hy either agency, resulting in a total o f 22 w inter surveys. W e grouped w inter aerial surveys into 3 periods w ith a balanced 7 surveys each: 1977-1986 (early), 1986-1997 (m id-), and 1998-2004 (late periods; Fig. 4). T h e 90% adaptive kernel core area for elk locations during w inter surveys was 1,418 km^ (Fig. 4). A shift in w inter elk distrihution occurred across the 3 tim e periods (M R P P A = 1.49, P < 0.0001; Fig. 4), w ith m ore elk observed outside B N P near Y H T during the latter periods.

Between 1977 and 2004, Parks C anada and A B F W flew 12 and 1 sum m er surveys, respectively, resulting in 12 years o f sum m er survey data. W e grouped surveys into 2 relatively balanced periods w ith 7 and 5 surveys, respectively, 1977- 1986 and 1986-2004 (Fig. 5) to align w ith w inter periods. T h e 90% adaptive kernel core area for sum m er observations was 2,708 km^ (Fig. 5). Sum m er elk distributions shifted (M R P P A = 1.41, P = 0.0006), w ith noticeable declines o f elk in the front ranges o f B N P and increases in elk near theY H T (Fig. 5).

D is tr ih u tio n shifts observed in aerial surveys w ere m irrored hy distributions o f radiocollared elk (Fig. 3). Spatial distrihution o f collared elk differed between periods {x^ = 20.2, P = 0.003), w ith the largest increase in elk occurring in year-round resident elk on the Y H T Ranch and the greatest decline in the elk in the Lake Louise and Red D eer areas (Fig. 3).

Population DynamicsT h e best models o f elk W included density dependence and either a negative effect o f sum m er rainfall or cumulative burn area, or a positive effect o f w inter range enhancem ent (Tables 4, 5). O f the 3 retained covariates, the effect o f fire was m ost variable and 95% confidence intervals overlapped zero (Tahle 4). W e found high (r > 0.7) collinearity between elk W and cumulative hectares o f w inter range enhancem ent ( r = 0.80), w inter w o lf numbers ( r = 0.78), and num ber o f horses ( r = 0.71). T h e h igh collinearity between elk Nf and w inter range enhancem ent suggests caution is w arranted w hen interpreting the top models (Tahle 4), although param eter estim ates should rem ain relatively unbiased (Freckleton 2002). Accordingly, we considered models as exploratory. U sing the sum o f A kaike weights for each variable (B urnham and A nderson 1998:141), we ranked param eters in order o f influence on elk population grow th rate: sum m er rainfall ( ^ O ; = 0.545), cumulative burn area ( ^ O ; = 0.526), w inter range hahitat enhance­m ents ( ^ O ; = 0.332), previous-w inter w o lf numbers ( ^ O ; = 0.141), the rain X elk num ber interaction (^ ( f l; = 0.103), a nonlinear effect o f rainfall ( ^ O ; = 0.055), N P O (X)®; = 0.023), N P O X elk num bers ( ^ O ; = 0.004), hay ( ^ O ; = 0.004), N P O X w olf num bers ( ^ O ; = 0.003), and a nonlinear effect o f N P O (^ ( f l; = 0.003). O ther variables had E ® ; < 0 .0001. Based on A kaike weights, only rainfall, cumulative area burned, and hahitat enhancem ents appeared related to elk r .

Solving rt for Nf = 0 using unconditional param eter estimates (po and p,x;) for all top models resulted in K = 1,285 (95% C l = 1,098-1,471) w hen adjusted for hunting and m anagem ent removals, and K = 954 (95% C l 779 - 1,124; T ah le 5) w ith o u t ad justing for h u n tin g and m anagem ent removals (results for Tt-ra-w are not shown; H eh h lew h ite 2006). W ith h u n tin g and m anagem ent removals, the Y H T elk herd was about 25% lower, or 331 fewer elk on average, than w ithout harvest or removals.

Table 3. Midpoints of spring and autum n migration d a te s of radioooiiared eik in th e Ya Ha T inda popuiation for eariy (1978) an d iate periods (2002, 2003), Banff Nationai Park, Aiberta, C anada.

Period Year

Spring migration Autumn migration

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 S ep 25.3 13.5 16®

2003 1 Jun 13.2 15.6 41 30 Got 27.2 17.1 33®A verage iate 4 Jun 11.5 14.5 61 2 Got 27.1 16.0 54

' A u tum n N is con s is ten tly iow er than spring N due to m orta lity , rad ioco iia r fa ilure, etc.

1288 W iidiife Society Buiietin • 34(5)

H a ; T in d a ''

Ya HaT inda

a t i o nlO

LegendY a H a T inda

mElk G ro u p SEzo

1 ■ 10 11-5D 51--HW

101 -5 0 0

501 * 2500

Rrvers

| _ _ | 9 0 ? t G o re $ ir f w y A rea

] BaniT NBUanai ParK

RiDacfI I E levaE ldn > 2 2 5 Q m

Figure 4. W inter elk distribution (Feb-Mar) during (a) early (1977-1985), (b) mid- (1986-1994), and (c) late (1995-2004) study periods in th e Ya Ha T inda elk population, Banff National Park, Alberta, C anada. The a rea within whioh 90% of all aerial observations ooourred is show n in th e outline. N um ber of surveys flown per period w as equal.

Evaluating PredictionsEvidence from bo th our hypothetico-deductive fram ework and population dynamics models suggest observed trends in M :R ratio and population dynamics were consistent w ith predictions o f hypotheses 4, 6, and 8, namely, w inter range enhancem ents, habituation due to hay feeding, and a w olf protection gradient in B N P (Tables 1, 4). However, we could not rule out potential effects o f elk relocation (Tahle 1: H 2). Observed elk population trends were opposite the predicted effects o f elk harvest, prescrihed burning, or horse num bers (Tahle 1: H I , H 3, and H 5). M igratory changes also were opposite o f predictions if m igration reduced w olf p red a tio n rela tive to residen ts (T ah le 1: H 7 ). O ur population models revealed tha t only elk N , prescrihed burns, sum m er rainfall, and perhaps hahitat enhancem ents (Tahle 5) affected elk

DiscussionO ur com parison o f m igratory and population dynamics o f the Y H T elk herd strongly suggests m igration behavior has changed dramatically since the 1970s. T h e proportion o f the population m igrating into B N P declined hy approximately 75%, and m igrant elk now return to the w inter range almost one m onth earlier. These changes cannot he explained hy changes in average between the early (TV/ = 608) and late (TV/ = 917) periods because M :R declined as TV/ increased. T h e shift in elk distrihution was m ost pronounced from the fron t ranges o f B N P to the Y H T in w inter, and a corresponding increase from < 3 0 elk in 1977 to > 3 0 0 elk sum m ering on Y H T during 2002-2004 (Fig. 4). W hile the increase in resident elk occurred during a period o f general

population growth, the increase at Y H T in sum m er was greater than expected due to population grow th rate alone. Therefore, despite small sample sizes o f collared elk during the early period, changes in collared and population samples’ M :R ratio and distrihution revealed the same trends o f declining m igration and distrihution shifts to year-round residence on the w inter range.

O ur m anagem ent hypotheses predictions th a t were the m ost consistent w ith changing m igratory behavior were those benefiting resident over m igrant elk. These included w inter range enhancem ent, access to hay, and possibly w olf avoidance o f the Y H T during summer. Resident elk would have benefited from w inter range enhancem ents year-round hy sum m ering on im proved ranges w ithout m igrating. T he im portance o f sum m er nutrition to elk condition and reproduction is now well docum ented (C ook et al. 2004). W in te r range enhancem ents may have m ade w inter ranges m ore nutritious during sum m er than high-elevation sum m er ranges, given trade-offs w ith w olf predation risk (H ehhle­w hite 2006). W hile elk feeding on hay during w inter may provide energetic benefits, we believe an im portan t effect o f hay feeding is as an attractant th a t leads to elk habituation to hum ans and loss o f traditional behavior (Burcham et al. 1999, Sm ith 2001, Kloppers et al. 2005). H abituation to hum ans from hay feeding also w ould benefit elk in w olf avoidance o f hum an activity on the w inter range. Num erous studies have docum ented carnivore avoidance o f high hum an activity (e.g., T heuerkauf et al. 2003). In B N P the tow n site created a predation refuge th a t enhanced elk survival and recruitm ent (H ehhlew hite et al. 2005), leading to m igratory declines. W hile hum an use may he lower at the

H ebbiewhite et al. • M igratory Elk Declines 1289

5?

L « g e r d

Va Tlndv

i E lk G ro u p S(Z9

^ S2&' 29-50« re-iso

0 2 4

Rhw rs

! ! □ 00 CQK Survey Ij Banff Naligrul Pjrk

R w ds

I I E le u a tic ti > 2 2 5 n m

Figure 5. S um m er elk distribution of elk (Jui-Aug) during (a) early (1977-1986) and (b) iate (1986-2004) study periods in th e Ya H a Tinda elk population, Banff Nationai Park, Alberta, C anada. The a rea within whioh 90% of all aerial observations ooourred is show n in th e outline. N um ber of surveys flown per period w as equal.

Y H T than in B N P, direct hum an-caused w olf m ortality from legal hunting and trapping 10 m onths o f the year and some illegal killing during the rest o f the year (Hehhlew hite2006) may reinforce w olf avoidance o f hum an activity (T h eu erk au f et al. 2003) and foster developm ent o f predation refugia, even if hum an use is lower than in theBV.

In contrast to research elsewhere on elk and fire, we found little evidence th a t large prescrihed fires were effective at increasing m igratory elk numbers. In fact. F ron t Range elk herds th a t had access to the largest prescrihed hum s w ithin B N P declined the most, and the effect o f fire on population grow th was weakly negative, not positive as predicted. O ur results may relate to how we m easured effects o f burning using a cumulative area burned following W h ite et al.(2005). Further, it was difficult to completely isolate effects o f burning in our study because the am ount o f area burned was correlated w ith declining elk and increasing w olf populations. D espite these caveats, however, we propose the hypothesis th a t in the presence o f w olf predation, effects

o f fires are weaker or even negative on elk (W hite et al.2005). A lm ost all previous studies dem onstrating positive effects o f fire on elk populations occurred in the absence o f w olf predation (e.g.. T aper and G ogan 2002). O ur results at least suggest an interaction between predation hy wolves and hahitat restoration th rough fire tha t has im portant m an­agem ent implications for ecosystem m anagem ent in national parks (W hite et al. 1998).

W e suggest the hypothesized dem ographic benefits o f m igration (Bergerud et al. 1984, Fryxell et al. 1988; Tahle I: H 7) may not exist for m igrants in the Y H T elk herd: hy all counts, residents seem to he doing relatively better. In further support, during the 1980s elk resided along the front-range areas o f B N P during w inter. However, hy 2000 w intering elk populations w ith in these areas had declined or shifted to the Y H T . W hile these trends support the existence o f a predation refugia, a com parison o f w olf predation on resident elk relative to forage trade-offs is required to empirically test for this effect (Hehhlewhite2006). In the absence o f experimental approaches, other

Table 4. Popuiation grow th rate (r?) m odel seieotion for th e Ya H a Tinda eik popuiation, Alberta, C an ad a , w inters 1 9 70-2004 . Following Burnham and A nderson (1998), w e report from generalized linear m odels, N, K, iogdikeiihood (LL), AAiCc (AiCc = A kaike’s information Criterion ad justed for small sam ple size), and AiC w eights. W e only report m odels within 0 -2 AAiC^.

Model rank and structure N K LL AAlCc AiC weight

1: ElkA/® + Rain‘d 0.33 25 3 7.979 0 0 .1632: ElkA/ + Burn‘s + HE"' 0 .35 25 4 9 .244 0 .325 0 .1393: ElkA/ + HE 0.33 25 3 7.644 0 .669 0 .1174: ElkA/ + Burn 0.28 25 3 7.344 1.268 0 .087

® ElkW is the postha rvest e lk N,.Average sum m er rainfall (mm) m easured at B lue Hill tow e r, 20 km sou theast o f Y a Ha Tinda.C um u la tive area bu rned (km ).C um u la tive area a ffec ted by w in te r range hab ita t enhancem ents (km^).

1290 W iidiife Society Buiietin • 34(5)

Table 5. M odel-averaged p aram eter es tim ates and unconditional SEs for the to p harvest rem oval-ad justed Vf eik popuiation grow th rate m odels for the Ya H a Tinda eik herd, Alberta, C an ad a , 1970-2004 .

Parameter

n

P

adj Model

SE

Intercept 0.440* 0 .0904Elk A/7 -0 .0 0 0 3 4 * 0 .000045Rain'^ -0 .0 0 0 3 4 * 0 .000127Burn° -0 .0 0 0 9 0 .0027Habitat enhancem ent'^ 0 .0154 0 .0144

® Elk N t is raw e lk coun t In /',_raw rnodel, and postha rvest e lk coun t In the r, m odel.

A verage sum m er rainfall (mm) m easured at Blue Hill Tower, 20 km sou theast o f Ya Ha Tinda.

° C um u la tive area bu rned (km^) In B anff N ational Park.C um u la tive area o f Ya Ha T inda w in te r range enhancem ent (km^).

* Param eter es tim a tes are s ta tis tica lly s ign ifican t at P = 0.05.

tools, such as resource selection functions (Boyce and M cD ona ld 1999), landscape-linked sim ulation models (Turner et al. 1994), or habitat-linked dem ographic studies 0ohnson et al. 2004), will he required to understand the mechanisms o f how predation risk and hahitat enhancem ent interact to influence m igratory behavior.

A n im portan t m anagem ent factor not directly tested was one o f the m ost pervasive and difflcuit to quantify impacts: the effects o f hum an recreation on eik behavior. In the 1970s, M organtin i and H udson (1979) docum ented dis­placem ent o f resident elk on the Y H T hy m otorized use, and m otorized hum an use was restricted in 1986. Recrea­tional activity is now predom inantly equestrian-hased, w hich appears to disturb elk less at Y H T despite overall increases in hum an use (M . H ehhlew hite, personal obser­vation). Increased hum an activity, equestrian-hased or o therw ise, com bined w ith d irect hum an-caused w o lf m ortality, may repel wolves (T heuerkauf et al. 2003), creating predation refugia (W hite et al. 1998). F urther study o f interactions between hum ans, wolves, and elk on the Y H T w inter range is needed to confirm w hether refugia are leading to reduced m igration and w hether a refuge is spatiai (i.e., B anff townsite; H ehhlew hite et ai. 2005) or only tem porai (e.g., T heuerkauf et al. 2003). As an im m ediate m anagem ent im plication, aversive conditioning similar to w hat has been used on eik in the B anff tow n site (Kloppers et ai. 2005) may he necessary to counteract potentiai predation refugia at the Y H T .

T h e only climatic effect we found was th a t increased sum m er rainfall decreased eik r , similar to findings o f C iu tton-B rock et ai. (1982). Increased precipitation during June-A ugust often produces snow in the Rocky M ountains and may delay spring p lant phenology critical for calf survival and popuiation dynamics (Post and Klein 1999). W e speculate the m ain effect o f rainfall on eik may he th rough reduced calf survival during w etter, colder summers because o f the frequency o f spring and sum m er snowfall during w etter summers (HoUand and C oen 1983). W in ter severity, m easured hy the N P O , also was unrelated to at / = 0 -2 lags. Nearby in the BV, severe w inters interacted w ith

high densities to reduce because the rate at w hich woives idUed eik increased w ith w inter severity (H ehhlew hite2005). A lthough the N P O correlates strongly w ith climate on the eastern slope o f the Rockies (T renherth and H urreii 1994), azonal clim atic conditions characterizing Y H T (M o rg an tin i 1995) m ay have w eakened th e clim atic signature o f N P O . Alternately, because the popuiation did not spend m uch tim e near K, density-ciim ate interactions may have not occurred.

O ur popuiation models also have im portant implications for long-term controversies surrounding range m anagem ent at the Y H T . W e found our assum ption o f simple linear density dependence was w arranted, similar to eik studies elsewhere (e.g., C iu tton-B rock et ai. 1982, M errill and Boyce 1991, Luhow and S m ith 2004) and estim ated carrying capacity (K) based on this density dependence. In com parison to studies elsewhere (M errill and Boyce 1991, Luhow and Sm ith 2004), our estimates o f K represent ecological carrying capacity w ith predation rather than food- based K (Sinclair and C augbley 1994). H u n tin g and relocations reduced long-term N hy an average o f 22% from approximately 1,285 to approximately 985, closer to sustainable range capacity assessments o f K (e.g., A G R A E arth and Environm ental 1998). W ith or w ithout hunting or relocations, long-term equilibrium for the population is tow ard an N well below the m aximum observed num ber o f elk o f approximately 2,200. This peak in elk numbers occurred after a series o f interm ediate precipitation sum ­mers, and im m ediately after fires in B N P and w inter range enhancem ents, and m ay represent a short-term overshoot o f K. In this context, elk m anagem ent (hunting and relocation) was effective at reducing elk W closer to the 1,000 elk recom m ended based on range assessments for threatened rough fescue conservation (M cGillis 1977, A G R A E arth and Environm ental 1998). However, at the tim e range assessments were done, 170-200 horses were w intered at Y H T . W ith recent declines o f w intered horses at Y H T , it m ay he w o rth w h ile rev isiting range assessm ents for conservation o f threatened fescue grasslands (M clnenely2003).

Differences in resource m anagem ent policies between federal and provincial agencies across jurisdictional bound­aries have facilita ted creation o f spatial gradients in predation risk and hahitat tha t appear to favor resident elk over m igrants. N ational park policies protect wolves, while provincial policies include liberal w o lf harvests to prom ote elk population goals (G unson 1997). Inside B N P, m anage­m en t seeks to reduce the negative effects o f hum an recreation (Parks Canada 1997), while the province o f A lberta has a m ore liberal recreation policy for the Y H T area (Anonym ous 1986). D irect w olf m ortality and indirect w o lf avoidance o f higher hum an activity at Y H T are, therefore, em ergent properties o f the present transhoundary m anagem ent policy framework. Similarly, Parks C anada seeks to restore long-term ecological conditions th rough application o f prescrihed fire to elk sum m er ranges (W hite et al. 1998), w hile the A lberta governm ent had a more

H ebbiewhite et al. • M igratory Elk Declines 1291

conservative forest fire suppression program , albeit w ith a growing prescribed fire program . A lberta provincial habitat enhancem ent policy has instead been focused on elk w inter range enhancem ent, whereas Parks C anada’s m ain objectives for the w inter range have been horse grazing and hay feeding (Parks C anada 1987). These contrasting m anage­m ent objectives pose a significant difficulty to development o f a com m on in terest approach to the transhoundary m anagem ent o f the Y H T elk herd (C lark et al. 2000). Historically, there was been little effective coordination o f m anagem ent activities across the park boundary, though recen t coord ina tion efforts should be con tinued and strengthened (e.g.. Parks C anada 2002). For example, the B ighorn cutblocks and the prescribed fire programs were im plem ented hy provincial and federal agencies w ithout regional assessment o f their effects on the Y H T elk herd. W e contend th a t transhoundary m anagem ent m ust be coordinated th rough developm ent o f a com mon interest approach, such as m aintaining m igration across the spatial extent o f this elk herd (C lark et al. 2000).

Transhoundary m anagem ent o f m igratory elk herds will be increasingly im portan t because the factors th a t changed m igration o f the Y H T elk herd occur elsewhere across western N orth Am erica (e.g.. W h ite and G arro tt 2005). O ur analyses indicate th a t isolating factors responsible for m igratory changes w ith certainty will be difficult in complex m anagem ent settings. W e suggest the re is sufficient evidence to indicate th a t recolonization by wolves, w inter range habitat enhancem ents, and habituation to hay have contributed to m igratory change. Therefore, these factors m erit prim ary consideration in future m anagem ent o f the Y H T elk herd. W ith recovering w olf populations present or im m inent in m any areas o f w estern U nited States, m any elk herds will face this new factor as an influence on m igratory behavior. P ark and wildlife managers should be alert for m igratory changes in elk populations, given the im portant ecosystem ramifications o f m igration and the implications o f changes in m igration for park m anagem ent. For example, in B N P w olf and grizzly bear population viability ultim ately

Literature CitedAGRA Earth and Environmental Ltd. 1998. P an ther C orners U pper Red

Deer range inventory. R eport on file with Aiberta Fish an d Wiidiife Division, Edm onton, C anada.

Alien, J. R. 2005. U se of sightabiiity m odels an d resouroe seieotion funotions to en hanoe aerial popuiation surveys of eik (Cervus elaphus) in Aiberta. Thesis, University of Aiberta, Edm onton, C anada.

A nderson, C. R., D. S. Moody, B. L. Smith, F. G. Lindzey, and R. P. Lanka. 1998. D evelopm ent and evaluation of sightabiiity m odels for sum m er eik surveys. Journal of Wiidiife M anagem ent 6 2 :1055-1066 .

Anonym ous. 1986. N ordegg-R ed D eer River sub-regionai integrated resouroe plan. Aiberta Forestry, R esouroe Evaluation and Planning Division, Edm onton, C anada.

Aryiov, Y., V. B adm aev, A. Bekenov, J. Chim eg, A. Entwistie, Y. A. Craohev, B. Lhagvasuren, A. Lusohekina, D. Maiion, E. J. Miiner- Cuiiand, an d V. Ukrainsky. 2004. The S aiga an te lope— teetering on the brink but stiii o au se for hope. Oryx 38 :2 5 0 -2 5 1 .

Berger, J. 2004. The last mile: how to sustain iong-distanoe migration in m am m als. C onservation Biology 18 :320-331 .

Bergerud, A. T., hi. E. Butler, and D. R. Miiier. 1984. Anti predator taotios

depend on the density o f elk as their prim ary prey (Carbone and G ittlem an 2002) and are thus dependent on continued transhoundary m igrations. O ur research provides an illus­tration o f the vital role tha t areas outside o f protected areas have in ecosystem m anagem ent o f national parks (G room et al. 1999). N ational parks w ith transhoundary populations o f m igratory ungulates m ust increase cooperative m anagem ent w ith adjacent agencies to ensure key park processes are m aintained.

AcknowledgmentsFunding from the following agencies m ade this study possible (in alphabetical order): A lberta Cooperative C o n ­servation Research U nit, A lberta Conservation Association, A lberta E nhanced Career D evelopm ent, A lberta Sustain­able Resource D evelopm ent, C en tre for M athem atical Biology, Challenge G rants in Biodiversity, Foothills M odel Forest, F oundation for N o rth A m erican W ild Sheep, M arm ot, M ountain E quipm ent C o-op Enviro-fund, Parks C anada, Patagonia, Rocky M oun ta in E lk Foundation, Sunpine Forest Products L td., University o f A lberta, and W eyerhauser, Inc. M . H ebbiew hite was supported by a C anon N ational Parks Science Scholarship for the A m er­icas. W e especially thank the 2 generations o f Parks Canada w ardens w ho assisted in all aspects o f th e research, provincial fish and wildlife officers and biologists, and most im portantly , the Ya H a T in d a R anch sta ff for the ir assistance over the duration o f the study. Notably, we thank T . Skjonsberg, P. Jacobsen, J. Nylund, C . Hayes, R. Smith, and other wardens and ranch staff whose help ensured collection o f long-term data. W e thank J. W asylyk for transcribing aerial survey data to G IS . W e thank pilot M . D upuis for safe fixed-wing telem etry m onitoring, and pilots from A lpine helicopters, Canm ore, A lberta. W e also thank the m ountain horses w ho carried us on our surveys, notably B andit (Ban), Cody, and others. Finally, we thank M . S. Boyce, S. Boutin, C . G ates, F. M essier, and C. Cassady-St. Clair for constructive comments.

of calving caribou— dispersion in m ountains. C anadian Journal of Zooiogy 62 :156 6 -1 5 7 5 .

Berry, K. J ., an d P. W. Mieike. 1983. Com putation of finite popuiation param ete rs and approxim ate probabiiity values for m ulti-response perm utation p ro ced u res (MRPP). Com m unioations in S tatistics— Simuiation and C om putation 12 :83-107 .

B oyce, M. 8 . 1989 . T tie J a c k so n eik tierd: in tensive wiidiife m an ag em en t in Nortin America. C am bridge University, C am bridge, United Kingdom.

Boyce, M. S., and L. L. M oDonaid. 1999. Relating popuiations to tiabitats using resouroe seieotion funotions. T rends in Eooiogy and Evoiution 14 :268-272 .

Boyce, M. S., an d E. hi. Merriii. 1991. Effects of 1988 fires on ungulates in Yeiiowstone Nationai Park. P rooeed ings of thie Tail T im bers Fire Eooiogy C onference 17 :121-132 .

Burohiam, M., W. D. Edge, an d C. L. M arcum. 1999. Eik u se of private land refuges. Wiidiife Sooiety Buiietin 27 :8 3 3 -8 3 9 .

Burnhiam, K. P., an d D. R. A nderson. 1998. Model seieotion and inference: a praotioai inform ation-tiieoretio approaoin. Springer- Veriag, New York, New York, USA.

1292 W iidiife Sooiety Buiietin • 34(5)

C ade, B. S., and J. D. R ichards. 2001. B lossom statistical softw are. U.S. Geoiogioai Survey, M idoontinent Eooiogioai S cience Center, Fort Coiiins, Colorado, USA.

Callaghan, C. 2002. The eooiogy of gray wolf (Canis lupus) habitat use, survival, and persis ten ce in th e central Rocky M ountains, C anada. D epartm ent of Zoology, University of Guelph, Guelph, Ontario, C anada.

C arbone, C., and J. L. Gittleman. 2002. A com m on rule for th e scaling of carnivore density. Science 295 :227 3 -2 2 7 6 .

Clark, T. W., D. C asey, and A. Halverson. 2000. Developing sustainab le m an ag em en t policy for th e National Elk Refuge, Wyoming. Yale School of Forestry and Environmental S tud ies Bulletin Series, Yale University, New Haven, C onnecticut, USA.

Clutton-Brock, T. H., F. E. G uineas, and S. D. Albon. 1982. Red deer: behavior and ecology of tw o sexes. University of Chicago, Chicago, Illinois, USA.

Cook, J. G., B. K. Jo h n so n , R. C. C ook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004. Effects of su m m er-au tum n nutrition and parturition d a te on rep ro d u c tio n a n d survival of elk. Wildlife M onographs 155.

C raighead, J. J ., G. Atwell, and B. W. O ’Gara. 1972. Elk migration In and near Yellowstone National Park. Wildlife M onographs 29.

Fancy, S. G., L. F. Rank, K. R. W hitten, and W. L. Regelln. 1989. S easonal m ovem ents of caribou In arctic Alaska a s determ ined by satellite. C anad ian Journal of Zoology 67 :6 4 4 -6 5 0 .

Freckleton, R. P. 2002. On th e m isuse of residuals in ecology: regression of residuals vs. multiple regression. Journal of Animal Ecology 7 1 :542-545 .

Fryxell, J. M., J . Greever, and A. R. E. Sinclair. 1988. W hy are migratory ungulates so ab u n d an t? American Naturalist 131 :781-798 .

Groom , M., D. B. Je n sen , R. L. Knight, S. G atew ood, L. S. Mills, D. Boyd-H eger, and M. E. Soule. 1999. Buffer zones: benefits and d an g e rs of com patible stew ardsh ip . P a g e s 1 7 1 -1 9 8 in M. E. Soule and J. Terborgh, editors. Continental conservation: scientific foun­dations of regional reserve netw orks. Island, W ashington, D.C., USA.

G unson, J. R. 1992. Historical and p resen t m an agem en t of w olves In Alberta. Wildlife Society Bulletin 2 0 :330-339 .

G unson , J . R. 1997. Elk m an ag em en t plan for Alberta. Wildlife M anagem ent Division, Alberta Environmental Protection, Natural R esource Service, Edm onton, C anada.

Hebbiewhite, M. 2005. Predation Interacts with the North Pacific Oscillation (NPO) to Influence w estern North Am erican elk population dynam ics. Journal of Animal Ecology 74 :226-233 .

Hebbiewhite, M. 2006. Linking predation risk and forage to partially migratory ungulate population dynam ics. Dissertation, University of Alberta, Edm onton, C anada.

Hebbiewhite, M., P. 0 . P aquet, D. H. P letscher, R. B. L essard, and 0 . J . C allaghan. 2004. D evelopm ent an d application of a ratlo-estlm ator to estim ate wolf killing ra tes and variance In a multlple-prey system . Wildlife Society Bulletin 3 1 :933-946 .

Hebbiewhite, M., C. A. White, C. NIetvelt, J. M. McKenzie, T. E. Hurd, J. M. Fryxell, S. Bayley, and P. C. P aquet. 2005. H um an activity m ed ia tes a trophic c a s c a d e ca u se d by wolves. Ecology 7 6 :2 1 3 5 - 2144.

Holland, W. D., and G. M. C oen. 1983. Ecological (biophysical) land classification of Banff an d J a s p e r National P arks. Volume 1: sum m ary. Alberta Institute of Pedology, Edm onton, C anada.

Holroyd, G. L., an d K. J. Van Tighem. 1983. Ecological (biophysical) land classification of Banff and Ja sp e r National Parks. Volume 3: the wildlife Inventory. C anad ian Wildlife Service, Edm onton, Alberta, C anada.

H ouston, D. B. 1982. The Northern Yellowstone elk: ecology and m anagem en t. Macmillan, New York, New York, USA.

Jo h n so n , C. J ., M. S. Boyce, R. L. C ase , D. H. Cluff, R. J. G au, A. G unn, an d R. M ulders. 2 005 . C um ulative e ffec ts of hum an developm ents on arctic wildlife. Wildlife M onographs 160.

Jo h n so n , C. J ., M. S. Boyce, C. C. Schw artz, an d M. A. Haroldson. 2004. Modeling survival: application of the multiplicative hazards m odel to Yellowstone grizzly bear. Journal of Wildlife M anagem ent 6 8 :966-974 .

K loppers, E. L., C. C. S t Clair, and T. E. Hurd. 2005. Predator- resem bling aversive conditioning for m anaging habituated wildlife.

Ecology an d Society 10:31. < h ttp ://w w w .ecologyandsociety .org/ v o l10 /lss1 /art31 /> . A cce sse d 2000 Aug 31.

Lhagvasuren, B., an d E. J . Mllner-Gulland. 1997. The s ta tu s and m an ag em en t of the Mongolian gazelle Procapra qutturosa p opula­tion. Oryx 3 1 :127-134 .

Lubow, B. C., and B. L. Smith. 2004. Population dynam ics of the Ja ck so n elk herd. Journal of Wildlife M anagem ent 6 8 :810-829 .

M ahoney, S. P., and J. A. S chaefer. 2002. Hydroelectric developm ent an d th e disruption of migration In caribou. Biological Conservation 107 :147-153 .

McGillis, J . R. W. 1977. R ange m an ag em en t of th e Ya H a Tinda. C anad ian Wildlife Service, Edm onton, Alberta, C anada.

Mclnenely, L. E. 2003. S easonal effects of defoliation on m ontane rough fe scu e (Festuca cam pestris Rydb.). University of Alberta, Edm onton, C anada.

McKenzie, J . A. 2001. The selective advan tag e of urban habitat use by elk in Banff National Park. University of Guelph, Guelph, Ontario, C anada.

Merrill, E. H., and M. S. Boyce. 1991. S um m er range an d elk population dynam ics in Yellowstone National Park. P a g e s 2 6 3 -2 7 3 in R. B. Kelter and M. S. Boyce, editors. The G reater Yellowstone Ecosystem : redefining A m erica’s w ilderness heritage. Yale University, New H aven, C onnecticut, USA.

Morgantini, L. E. 1988. Behavioral adaptive stra teg ies of wapiti (Cervus elaphus) In the C anadian Rocky M ountains. University of Alberta, Edm onton, C anada.

Morgantini, L. E. 1995. The Ya H a Tinda: an ecological overview. C anad ian Heritage, Parks C anada, Calgary, Alberta, C anada.

Morgantini, L. E., and R. J. H udson. 1979. H um an d istu rbance and habitat selection In elk. P ag es 1 3 2 -1 3 9 in M. S. Boyce and L. D. Hayden-W Ing, editors. North American elk: ecology, behavior, and m anagem ent. University of Wyoming, Laramie, USA.

Morgantini, L. E., and R. J. H udson. 1988. Migratory patterns of the wapiti, Cervus elaphus, In Banff National Park, AB. C anadian Fleld- Naturallst 102 :12-19 .

P aquet, P. C. 1993. Sum m ary reference docu m en t— ecological stud ies of recolonizing w olves In th e central C anad ian Rocky M ountains. P repared by John/P au l and A ssocia tes for Parks C anada, Banff National Park W arden Service, Alberta, C anada.

Parks C an ad a . 1987. Ya H a T inda m an agem en t plan. Parks C anada, D epartm ent of C anad ian Heritage, Banff, Alberta, C anada.

Parks C anada. 1997. Banff National Park m an ag em en t plan. Parks C an ad a , D epartm ent of C anadian Heritage, Banff, C anada.

Parks C anada. 2002. Regional forest m an ag em en t strategy: Banff National Park. Environm ental screen ing report. P arks C an ad a , D epartm ent of C anad ian Heritage, Banff, Alberta, C anada.

Portier, C., M. Festa-B lanchet, J.-M . Galllard, J. T. Jo rgenson , and N.G. Y occoz. 1998. Effects of density and w eather on survival of bighorn sh e e p lam bs (Ovis canadensis). Journal of Zoology, London 245 :2 7 1 -2 7 8 .

Post, E., and D. R. Klein. 1999. Caribou calf production and seasonal ran g e quality during a population decline. Journal of Wildlife M anagem ent 6 3 :335-345 .

Sachro , L. L., W. L. S trong, an d C. C. G ates. 2005. P rescribed burning effects on sum m er elk forage availability In the sub-alp lne zone, Banff National Park. Journal of Environmental M anagem ent 7 7 :183-193 .

Schaller, G. B. 1988. Wildlife of the T ibetan Plateau. University of Chicago, Chicago, Illinois, USA.

Sinclair, A. R. E. 2003. Mammal population regulation, keystone p ro c e sse s an d eco sy stem dynam ics. Philosophical T ransactions of th e Royal Society of London Series B, Biological S c ien ces 3 5 8 :1 7 2 9 - 1740.

Sinclair, A. R. E., an d G. Caughley. 1994. Wildlife ecology and m anagem ent. Blackwell Science, London, United Kingdom.

Smith, B. L. 2001. W inter feeding of elk In w estern North America. Journal of Wildlife M anagem ent 6 5 :173-190 .

Smith, B. L., and R. L. Robbins. 1994. Migrations and m an ag em en t of th e Ja ck so n elk herd. U.S. D epartm ent of th e Interior, National Biological Survey, W ashington, D.C., USA.

Sokal, R. R., and F. J. Rohlf. 1995. Biometry. W. H. Freem an, New York, New York, USA.

StataC orp . 2004. S ta ta 8.0. S tataC orp , College Station, Texas, USA.

Hebbiewhite et al. • M igratory 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 M anagem ent 6 6 :106-122 .

Theuerkauf, J ., W. Jedrzejew skI, K. Sohm ldt, and R. Gula. 2003. Spatlotem poral segregation of w olves from h um ans In th e Blalowleza Forest (Poland). Journal of Wildlife M anagem ent 6 7 :706-716 .

Thirgood, S., A. M osser, S. Tham , G. Hoporaft, E. M w angom o, T. MIengeya, M. Kllewo, J. Fryxell, A. R. E. SInolaIr, and M. Borner. 2004. C an parks p ro test migratory ungulates? The e a se of the Serengeti w ildebeest. Animal Conservation 7 :113-120 .

Toweill, D. E., and J. W. T hom as. 2002. North Amerloan elk: eooiogy and m anagem en t. Sm ithsonian Institution, W ashington, B.C., USA, London, United Kingdom.

Trenberth, K. E., and J. W. Hurrell. 1994. Deoadal a tm o sp h e re -o o ean variations In th e Paolflo. Climate Dynamlos 9 :303-319 .

Turner, M. G., Y. Wu, L. L. W allaoe, W. H. Rom m e, an d A. Brenkert. 1994. Simulating winter Interaotlons am ong ungulates, vegetation, and fire In northern Yellowstone Park. Eooiogioai Applloatlons 4 :4 7 2 - 496.

White, C. A., T. L. Hurd, M. Hebbiewhite, an d I. R. Pengelly. 2005. Mitigating fire sup p ressio n , highway, and habitat fragm entation effeots In th e Bow Valley eoosystem : preliminary evaluation of a before-after-oontrol-lm paot (BACI) design with path analysis. P ag es 2 5 -3 9 in J. Innes and G. Jo n e s , editors. Monitoring th e effeotlveness of bioiogioai oonservatlon. Ministry of Forests, Ministry of W ater, Air, Lands, Vanoouver, British Colum bia, C anada.

White, C. A., C. E. O lm sted, and C. E. Kay. 1998. A spen, elk, an d fire In the Rooky M ountain national parks of North Amerloa. Wildlife Sooiety Bulletin 26 :4 4 9 -4 6 2 .

White, C. A., I. R. Pengelly, M .-P. R ogeau, an d D. Zell. 2003. L andsoape fire regim es and vegetation restoration In Banff National Park. Parks C a n ad a Oooaslonal P aper B N P -2003-01 . Parks C anada, Banff, Alberta, C anada.

White, P. J ., and R. A. Garrott. 2005. Yellow stone’s ungulates after w olves— expeotatlons, realizations, and prediotions. Bioiogioai C o n ­servation 125 :141-152 .

Willoughby, M. 2001. The rough fesoue dom inated oom m unlty ty p es In the foothills of north-oentral Alberta. Volume T /017. Publlo Lands Division, Sustainab le R esouroe D evelopm ent, Edm onton, Alberta, C anada.

W oods, J . C. 1991. Eooiogy of a partially migratory elk population. University of British Columbia, Vanoouver, C anada.

M ark Hebbiewhite (left) Is an a ssis ta n t professor In the Wildlife Biology Program In the D epartm ent of Eoosystem and C onservation Solenoes, University of M ontana. He reoelved his B.So. from the University of G uelph (1995), his M.S. from the University of M ontana (2000), and his Ph.D. In eooiogy from th e University of Alberta (2006). He h as oonduoted researoh on th e relationships betw een wolves, elk, an d hab itat, an d Is In terested In w lld llfe-habltat relationsh ips, population dynam los, and oonservatlon biology of ungulates. EvelynH. Merrill (not plotured) Is a professor of eooiogy In the D epartm ent of Bioiogioai Solenoes, University of Alberta. S h e obtained her B.A. from St. Law renoe University (1974), her M.S. from University of Idaho (1978), an d her Ph.D. from the University of W ashington (1987). She w orked for Idaho D epartm ent of Fish and C a m e prior to entering aoadem io positions a t th e University of W yoming and University of W Isoonsln-S tevens Point. Her ourrent a rea of researoh a d d re sse s m eohanlsm s of animal behavior In p red a to r-p rey relationships and d i s e a s e s p r e a d In h e te r o g e n e o u s e n v iro n m e n ts . L u lg l E. M o rg a n tin i (se o o n d from right) Is th e C h ie f B io lo g is t for W eyerhaeuser C om pany In Alberta and an adjunot p rofessor a t the University of Alberta. He holds a D ootorate In Solenoe from the University of R om e (Italy; 1974), and an M.So. (1979) and a Ph.D.

(1988) In wildlife m an agem en t from th e University of Alberta. In 2001 he reoelved th e Wildlife H abitat C a n ad a Forest S tew ardsh ip Award for his leadership In oarlbou oonservatlon and. In 2002, the Alberta Emerald Award for Life-Long Environmental S tew ardship . His Interests Inoiude m any a s p e o ts of elk eooiogy an d behavior, an d th e a s se s sm e n t and mitigation of oil and g a s developm ents an d forestry on wildlife. Clliford A. White (seoond from left) ob tained his B.S. from University of M ontana (1977), his M.S. from C olorado S ta te University (1985), an d his Ph.D. In forest eooiogy from University of British Colum bia (2001). He h as w orked with Parks C a n ad a sinoe 1973 and Is presently the E oosystem M anagem ent C oordinator for Banff National Park. His researoh foous h as been on th e long-term roles of h u m ans In determ ining fire regim es and wildlife populations In th e C anad ian Rookies, Inoluding the oumuiative environm ental effeots of highway mitigation, wildlife oorrldor restoration, and presorlbed burning. James R. Allen (oenter) Is a wildlife m an agem en t biologist with th e Alberta Fish an d Wildlife Division In Rooky M ountain H ouse, Alberta. He reoelved his Fish and Wildlife Teohnloal Diploma from Lakeland College (1975), a B.So. from th e University of W aterloo (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 34-year oareer a s a wildlife m an agem en t biologist with th e Alberta Fish and Wildlife Division. He reoelved his B.So. (1967) and M.S. (1969) d e g re e s In zoology from th e University of Calgary. He oon o en tra ted his efforts on wildlife hab itat pro teotlon an d th e m a n a g e m e n t of la rge m a m m als In so u th e rn A lberta . Linda Thurston (not plotured) h as w orked In the Rooky M ountains In C a n ad a and th e United S ta te s on various a s p e o ts of wolf behavior and eooiogy. S he holds a B.S. In eooiogy, behavior, and evolution from University of Callfornla-San Diego (1995), and an M.S. In wildlife biology from T exas A&M University (2002). Her Interests lie In p redato r-p rey dynam los and oonfllot resolution In wildlife Issues. S he h as m ost reoently w orked a s a biologist with Idaho Fish and C am e. Tomas E. Hurd (not plotured) Is ourrently leading a Parks C an ad a feasibility s tudy to Investigate th e potential for a new national park In th e S outh O kanagan a rea of British Columbia. He previously held the position of Wildlife Biologist, Banff National Park. He holds a B.So. In zoology (1985) and an M.So. In forest so lenoes (1999) both from the University of British Columbia.

Associate Editor: Whittaker.

1294 W ildlife Sooiety Bulletin • 34(5)