Linking Elk M ovement and Resource Selection to H unting Pressure in a Heterogeneous Landscape

E lk {Cervus elaphus) are increasing in fragm ented landscapes tha t result from exurban hum an developm ent th roughout western N orth America, w hich increases hum an-w iidiife conflicts and poses a challenge to w ildlife managers. E lk hunting m ust often he intensively m anaged to reduce population growth rate, crop depredation, and habituation to hum ans. However, little was know n about the indirect effect hunting has on anti-predator behavior, m ovem ent, resource selection, and hum an-e ik conflicts. W e outfitted elk w ith giohai positioning system (G PS) collars in the w iid iand-urhan interface (W U I) o f M issoula, M ontana, U SA , 2007-2009, to test the indirect effects o f hunting on elk. W e used data from 9 G PS-coiiared adult female elk during 3 hunting seasons w ith increasing h unting pressure (2007-2009) to test relationships between m ovem ent rates m easured by first passage tim e (FP T ) and resource selection. E lk m ovem ent rates were lower approximately 750 m from houses and trails, resulting in resource selection for areas approxi­ mately 1,200 m from houses and trails; this suggested th a t habituation to hum ans contributed to h u m an wildlife conflict. M ovem ent rates increased w ith increasing hunting pressure, and were lower in general versus focal hunting seasons and w ith archery versus rifle hunting. T hus, intensive hunting seasons in the W U I increased elk m ovem ent rates and exposure to hun ter predation risk, as predicted. These results support the hypothesis tha t elk modify their behavior in relation to tem poral and spatial variation in hum an predation risk. In the intensively m anaged W U I, our results dem onstrate th a t even small increases in the area hunted and increases in intensity can indirectly change resource selection and m ovem ent rates o f elk, potentially reversing recent trends o f increasing habituation in W U I elk populations. © 2012 T h e W ildlife Society. KEY W O R DS anti-predator behavior, elk, first passage time, habituation, hunting, movement, resource selection, wildland-urban interface. H um an activities can affect animals th rough direct and indirect pathways (Sinclair and Byrom 2006, Creel and Christianson 2008). For example, hum ans can cause popu­ lation declines either directly via over-harvest or indirectly th rough habitat fragm entation (Sawyer et al. 2007, W alker et al. 2007). H u n tin g by hum ans has direct effects on dem ography and indirect effects on animal behavior and life-history (Creel and Christianson 2008, D arim ont et al. 2009). For example, under harvest pressure by hum ans, wild hoar {Sus scrofa) invested m ore heavily in early reproduction at the cost o f adult survival (Toigo et al. 2008). Even the risk o f being killed by hum an hunters had a greater indirect effect on elk {Cervus elaphus) resource selection than did w olf {Canis lupus) predation risk (G ude et al. 2006). U nder w olf and hum an predation risk, elk selected refuge areas Received: 14 July 2011; Accepted: 31 March 2012 Published: 24 September 2012 ^E-mail: shawn. m. develand@gmail. com to avoid hum an predation risk in M ontana, U SA (Burcham et al. 1999, G ude et al. 2006, P roffitt et al. 2009). These examples dem onstrate tha t the risk o f m ortality from a predator, including hum ans, can cause animals to change resource selection strategies to avoid risk. H ow animals adjust their m ovem ent behavior in response to spatio-tem poral variation in predation risk, however, remains a neglected, hu t im portant, area o f study in hum an-w ildlife interactions (L im a and Bednekoff 1999). Previous studies dem onstrated tha t m ovem ent rates o f elk are influenced by predation risk and forage availability (Fortin et al. 2005a, Frair et al. 2005). T h e risk-allocation hypothesis suggests tha t animals should show m ore frequent an ti-predator behaviors under high risk, and forage during low risk (L im a and Bednekoff 1999, Ferrari et al. 2009). M any previous studies found th a t ungulates m odify vigilance in proportion to predation risk, for example (Laundre et al. 2001, F iley and Creel 2008). Fewer studies have tested the resource selection and m ovem ent predictions o f the 658 W ild life Society B ulle tin • 36(4) risk-allocation hypothesis. U nder the risk-allocation hypoth­ esis, animals are predicted to slow down in areas and at times w ith h igh forage and speed up in areas and tim es o f high predation risk (Ferrari et al. 2009). For example, M orales et al. (2005) identified “encam ped” states in elk in w hich step lengths were small and turning angles were high; these were associated w ith foraging areas, whereas “exploratory” states were those w ith long step length and small turning angles. M any resource selection studies have dem onstrated spatial avoidance o f predation risk (H ehhlew hite and M errill 2007), hu t from resource selection alone it is difficult to understand the mechanisms o f risk avoidance. G ude et al. (2006) dem ­ onstrated tha t elk avoided hum an predation risk m ore than w olf predation risk, hu t they were unahle to explain the behavioral mechanism explicitly. D id elk move m ore under hum an predation risk, or did they slow down and spend more tim e in safe patches as expected under the risk-allocation hypothesis? N ew m ovem ent analyses, such as first passage tim e (FPT; Fauchald and Tveraa 2003, F rair et al. 2005), provide novel approaches to test how predation risk influ­ ences the m ovem ent mechanisms o f selection (Ferrari et al. 2009). For example, Frair et al. (2005) used F P T to dem on­ strate tha t hum an activity resulted in increased w olf preda­ tion risk, affected elk m ovem ent rates (Frair et al. 2005), and ultimately, reduced elk survival and fitness (Frair et al. 2007). Knowing w hether changes in resource selection affect move­ m ents is im portant because m ovem ent could expose animals to increased risk o f being encountered or detected by pred­ ators (Yoder et al. 2004). In m any taxa, increased m ovem ent rates result in lower survival or fitness (Biro et al. 2003). T hus, understanding the effects o f hum an predation risk on m ovem ent will he especially im portant in fragm ented and heterogeneous hum an-dom inated landscapes. T here are few landscapes m ore fragm ented and rapidly changing in w estern N orth Am erica than low-elevation valley bottom s (Radeloff et al. 2005, H aggerty and Travis 2006) th a t are also critically im portant w inter ranges for ungulates (ToweiU and T hom as 2002). T h e expansion o f housing into ungulate w inter range results in a complex matrix o f land ownership th a t varies spatially in term s o f hum an predation risk. In such landscapes, hum an predation risk varies from partial to com plete exclusion, and is tem po­ rally regulated to m id-to-la te autum n hunting seasons. As habitat fragm entation encroaches on ungulate w inter range, ungulate exposure to hum ans is increasing, w hich increases the potential for habituation (Thom pson and H enderson 1998). H um ans also use differing hun ting modes, similar to coursing (rifle hunting), and sit-and-w ait predators (archery hunting), w hich may have different effects on risk allocation (Schm itz 2008). T hus, the w inter range m a­ trix o f tem poral and spatial variation in hum an predation risk wiU make elk m anagem ent m ore complex because elk can change bo th m ovem ent and/or resource selection in response to hum an hunting (Thom pson and H enderson 1998). Therefore, our objectives were to test the following predic­ tions o f the risk-allocation hypothesis: 1) F P T for elk will he lower (i.e., faster m ovem ent rates) in areas tha t allow hunting than in areas o f refiigia and hiding cover will reduce move­ m ent rates in hunted areas; 2) elk wiU move faster in areas o f rifle predation risk than in areas o f archery predation risk; 3) elk will increase m ovem ent rates w ith increm ental increases in hum an predation risk as predation risk varies annually; and 4) elk will select for areas away from houses, roads, and trails. W e tested these predictions over a 3-year period (2007-2009), during w hich hunting increased by nearly 30%, by m easuring the effect o f hun ting access, h u n t­ ing mode, distance to hum an influence (roads, trails, and houses), and the influence o f cover on F P T and resource selection.