Ecology and Society10 - PDF document

Ecology and Society10 http://www.ecologyandsociety.org/vol10/iss1/art31/ Research Predator-Resembling Aversive Conditioning for ManagingHabituated Wildlife Elsabé Louise Kloppers , Colleen Cassady St. Clair , and Thomas Eric Hurd 2 ABSTRACT. Wildlife habituation near urban centers can disrupt natural ecological processes, destroyhabitat, and threaten public safety. Consequently, management of habituated animals is typically invasiveand often includes translocation of these animals to remote areas and sometimes even their destruction.Techniques to prevent or reverse habituation and other forms of in situ management are necessary to balanceecological and social requirements, but they have received very little experimental attention to date. Thisstudy compared the efficacy of two aversive conditioning treatments that used either humans or dogs tocreate sequences resembling chases by predators, which, along with a control category, were repeatedlyand individually applied to 24 moderately habituated, radio-collared elk in Banff National Park during thewinter of 2001–2002. Three response variables were measured before and after treatment. Relative tountreated animals, the distance at which elk fled from approaching humans, i.e., the flight response distance,increased following both human and dog treatments, but there was no difference between the two treatments.The proportion of time spent in vigilance postures decreased for all treatment groups, without differencesamong groups, suggesting that this behavior responded mainly to seasonal effects. The average distancebetween elk locations and the town boundary, measured once daily by telemetry, significantly increasedfor human-conditioned elk. One of the co-variates we measured, wolf activity, exerted counteracting effectson conditioning effects; flight response distances and proximity to the town site were both lower whenwolf activity was high. This research demonstrates that it is possible to temporarily modify aspects of thebehavior of moderately habituated elk using aversive conditioning, suggests a method for reducinghabituation in the first place, and provides a solution for Banff and other jurisdictions to manage Key Words: Banff National Park; Canada; Cervus elaphus; behavior; aversive conditioning; dogs; INTRODUCTION An emerging problem in many urban areas andparks in North America is the habituation of wildlifeto humans and human use areas (Bounds and Shaw1994, Thompson and Henderson 1998). Habituationis defined in this context as the waning of an animal'sinnate response to repeated exposure to stimuli thatcarry no discernable biological consequence (e.g.,Whittaker and Knight 1998, Taylor and Knight2003). Animals that are habituated to humans thusdo not avoid contact with humans or areas in whichcontact with them is likely, unlike the behavior ofwilder conspecifics. Habituation of this sort typically produces several ecological and publicsafety problems. Ecologically, areas with human-habituated wildlife may exhibit a "refuge effect," inwhich concentrations of wildlife such as deer(Oidocoileus spp.) or elk (Cervus elaphus) areattracted to artificially fertilized lawns and gardensas food supplements (Lubow et al. 2002, Rubin etal. 2002). In such locations, they also receiveartificial refuge from predation (Isbell and Young1993, Riley et al. 1998) and often cease migratorybehavior (Geist 1982). This can lead to breakdownsin natural predator-prey relationships andhyperabundant populations (e.g., Ripple andBeschta 2004), causing declines in the diversity of 1University of Alberta, 2Banff National Park 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ associated wildlife and vegetation species (e.g.,Caughley 1981, Warren 1991, Soulé et al. 2003).As public safety problems, habituated animals canbe associated with wildlife-vehicle collisions (Etteret al. 2002), the transmission of diseases (Steere1994), the attraction of dangerous predators tohuman use areas (McCullough et al. 1997), and, incertain circumstances, aggression or attacksdirected at humans (e.g., Herrero 1985, Bounds andThe prevalence of both types of problems appearsto be rising in North America, particularly for elk(Thompson and Henderson 1998), and wasconspicuous by the mid 1990s in the town site ofBanff in Banff National Park, Canada.Concentrations of habituated elk used the town asa refuge from predation (e.g., McKenzie 2001),ceased to migrate (Woods 2001), and overgrazedthe native vegetation (White et al. 1998), withcascading effects on other herbivores such as moose(Hurd 1999) and beaver (Nietvelt 2001). Concernsfor public safety were also high because elk injuredan average of seven people per year between 1993and 2001 (G. Peers, unpublished manuscript) andattracted predatory carnivores to the town site(McNay 2002a). Interestingly, similar problems ofhabituation do not seem to afflict European red deer(also Cervus elaphus), even where average ungulatedensities are comparable to that of North America(Staines and Welch 1989). Possible reasons for thisinclude higher human hunting pressure (e.g.,Messmer et al. 1997, Morellet and Guibert 1999)and lower predation pressure immediatelysurrounding human use areas (e.g., Okarma 1995)in European jurisdictions. In combination, thesefactors may not create the same ungulate-humanoverlap and habituation within urban areas seen inDespite the growing need to prevent ungulatehabituation to humans and human use areas in Banffand other parts of North America, relatively littlepublished research has addressed methods by whichthis can be accomplished (Thompson andHenderson 1998). Reactionary methods typicallyused to manage "problem wildlife" includetranslocations (e.g., Baker and Fritsch 1997) ordestruction (e.g., Rondeau and Conrad 2003), butthese address the consequences, not the causes, ofhabituation. Moreover, such methods can havenegative ecological and social consequences (e.g.,Lee and Millar 2003, Rondeau and Conrad 2003)that may be ethically unacceptable to the public (Guynn and Yarrow 1997, Witmer and Whittaker2001). Finally, many protected areas in NorthAmerica must satisfy dual mandates to maintainecological integrity and visitor enjoyment (e.g.,National Park Service 1991, Government of Canada2000). This balance challenges managers to findalternate methods of redistributing and managing"wild" ungulate populations so that they do notexhibit conflict in human use areas, yet remainaccessible to carnivores within the ecosystem andto park visitors seeking to view wildlife. Oneapproach to achieving this balance is to create andmaintain higher levels of wariness in urban wildlifewithout removing them from the system altogether.Here we describe a research project predicated onthe assumption that it is possible to reverse andprevent the habituation process with predator-resembling aversive conditioning. Aversiveconditioning is a form of operant conditioning (forbackground, see Brush 1971, Davey 1981) that can(Domjan 2003). Because elk, like all animals, areadapted to conserve energy, access essentialresources, and avoid injury (Geist 1982), wehypothesized that habituated elk that experiencenegative predator-resembling stimuli such asanxiety, stress, pain, energy loss, and reducedforaging time unpredictably when they areapproached by humans or are close to human useareas should learn to avoid these contexts. Webelieved that such an association would be possiblebecause others have shown that human disturbancecan evoke predator-avoidance responses even in theabsence of predation events (Frid and Dill 2002,Beale and Monaghan 2004). To capitalize on likelycombinations of genetic disposition, learnedbehavior, and cultural transmission (sensuWhittaker and Knight 1998, Griffin 2004), weselected stimuli known to trigger natural predator-avoidance responses in nonhabituated "wild"ungulates: a fear of humans (Schultz and Bailey1978, MacArthur et al. 1982, Gander and Ingold1997) and of human-dog combinations (Martinettoand Cugnasse 2001, Miller et al. 2001). We thendrew from principles of stock herding (McConnelland Baylis 1985, Smith 1998) and learningpsychology (Domjan 2003) to use these stimuli ina chase sequence that we expected would resemblepredatory behavior by humans, human-dogcombinations, or wolves (Canis lupus). Ourobjectives were to (1) test the hypothesis thathabituated elk behavior can be reduced withaversive conditioning and (2) compare the efficacy 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ of two treatment types: human- and dog-basedaversive conditioning. To assess the relativeeffectiveness of these two techniques, we measuredthree response variables for the elk: flight responsedistance from an approaching human (e.g., Altmann1958), the proportion of time spent in vigilancepostures as a measure of predator-wariness (e.g.,Frid 1997), and proximity to the town boundary METHODS Study area and schedule Fieldwork was conducted in the town site andsurrounding area of Banff, Alberta (51°15’N, 116°30’W), within Banff National Park, Canada, duringthe winter of 2001–2002. Banff town site is situatedin the Bow Valley within the central CanadianRockies at an elevation of 1383 m. In the lower BowValley bottomlands, modest snowfall combinedwith occasional warm periods creates importantwinter habitat for ungulates (Holland and Coen1983, Holroyd and Van Tighem 1983, Woods1991). The town of Banff has a permanent humanpopulation of 7135, but park visitation approaches5 x 106 people per year (Banff-Lake Louise TourismBureau, personal communication). Our 466.5 hastudy area was composed of the urban land-use areaof Banff in addition to an adjacent golf course,montane wetlands, forests, and shrublands within 2km of the town boundary. The total elk populationin the town site area numbered 277 in the springsurvey of 2001, showing a continuing decline inresponse to management actions and increasedpredation by wolves since a high of 533 elk in 1994(Banff National Park, unpublished data). Duringthis 2001–2002 field season, up to 18 wolves (Canislupus) used the area surrounding the Banff town siteunpublished dataWe collected preconditioning (before) data on threeresponse variables (below) during the months ofSeptember and October and postconditioning (after)data between November and March. The winterseason was specifically chosen for this conditioningresearch to coincide with the time that partiallymigratory elk in Banff are most likely to concentratein the lower-elevation urban areas (Woods 1991,McKenzie 2001, see also Kilpatrick and Spohr2000), thus risking habituation to humans and their Radio-collaring and treatment assignment We radio-collared 19 moderately habituated elkbetween September and December of 2001 usingground-darting immobilization; including fiveanimals that had previously been collared, thisprovided a total sample size of 24 adult �( 2 yr)female elk that were without calves. We defined"moderately" habituated elk as those that had noknown histories of year-round town site residency,yet allowed human approach close enough forground-darting (40 m). Although the collaringprocess itself could be expected to affect elkbehavior (e.g., Mech et al. 2002), it was doneconsistently for elk in all treatment groups includingthe control, and is thus expected to have the samerelative impact on our response variables. It was oursubjective impression that the 24 study elkassociated in large mixed herds that moved freelythroughout the Banff town site and periphery bothat the time of collaring and throughout the winterresearch. Thus, we expected that elk from thedifferent treatment groups were subjected torelatively similar habitat, forage availability,predation pressure, and other environmental factorsthat could influence our dependent variables. Weevaluated this assumption using a home-rangeoverlap analysis to assess the relative similarity ofhabitat use between elk from the different treatmentgroups. We divided the 24 study elk into threetreatment groups of eight elk each: human, dog, andcontrol. The division was based haphazardly on thelocations of three loose groupings of radio-collaredelk on the first day of conditioning trials. On thatday, these arbitrary groups were all located 200m from the town boundary and were separated by1000 m. Prior and subsequent to our start day,these elk dissolved into one or more large mixedherds, and each elk was treated independently or ingroups of a few individuals. We evaluated theindependence of the treatment groups with anassociation analysis to determine how theproportion of time that individual elk spent with elkfrom their own treatment groups compared with theproportion spent with elk from the other treatment Aversive conditioning protocols We applied 10 15-min aversive conditioningtreatments per elk between November 2001 andMarch 2002 when elk were found within an arbitraryboundary around the town. To create long-lasting 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ and robust associations, we attempted to subject theelk to diligent yet unpredictable conditioning events(Brush 1971, Reynolds 1975), and researcherpersonnel and clothing were alternated betweensuccessive trials to help generalize the treatments toother humans (Rybarczyk et al. 2003). The humanconditioning treatment was applied by two peoplechasing the target elk for 15 min while using RG300 Clip Launcher starter pistols from MargoSupplies, High River, Alberta, Canada, to fire fivepyrotechnic screamers and five cracker shells overtheir heads. We fired the screamer shells first to startthe elk running and control their direction of travel,and then used the cracker shells toward the end ofthe treatment to maintain their sense of fear, anxiety,and confusion. If dense traffic, facilities, or peoplein close proximity posed a safety hazard during thetrial, we moved the elk more slowly to the edge oftown before firing any shells. The dog conditioningtreatment was applied for 15 min by one researcherand a professional dog handler with two bordercollies. The dogs silently herded the elk as directedby the dog handler using voice and whistlecommands, using stalking behavior and eye contactto emulate wolf hunting (McConnell and Baylis1985; see also Mech 1970). The border collies werespecifically chosen for this research becausepreliminary trials indicated that the elk respondedto their silent movements with a "flight response,"whereas elk responded to a different breed ofherding dog (the New Zealand huntaway) thatbarked continuously by stopping, turning, andconfronting the dogs aggressively, indicative of amobbing or "fight response" (see also Geist 1982).The control group received a "sham" treatment withtwo researchers standing silently within 50 m of theelk for the same 15-min period. No otherconditioning treatment was applied to the controlgroup, but radio-collaring and response variablemeasurement were conducted in the same way asDuring each conditioning trial using humans ordogs, we moved the animals as far and as quicklyas possible during the 15 min, typically at a runningpace if it was deemed appropriate for animal,human, and property safety. We considered this"chase sequence" component of the treatment to beparticularly important for emulating predationevents and maximizing elk energy loss and stress.If elk moved into dense hiding cover and we lostsight of them, we snow-tracked and continued thepursuit and application of noise and visual stimuli,again to better emulate predator hunting and stalking (e.g., Bateson and Bradshaw 1997). Ademonstration of our predator-resembling chase here by clicking on the linktitled An Elk of a Problem. All aversive conditioningtrials were tracked with a handheld globalfrom Trimble in Sunnyvale, California, USA, toaccurately measure the distance the elk weredisplaced. The average frequency of treatmentapplication, i.e., days between treatments, was alsoA final aspect of the conditioning protocol was theseparation of elk into treatable units. Because elkfrom different treatment groups were often foundinterspersed with one another, a treatment wasapplied only if the appropriate treatment animalscould be gently split away from the other animals.We split animals by walking slowly toward the elkand pushing them apart using subtle bodymovements and eye contact, which are standardlow-stress stock herding principles (Smith 1998; E.L. Kloppers, C. C. St. Clair, and T. E. Hurd,unpublished manuscript). On some occasions, thewhole herd reacted to the splitting, and weabandoned the conditioning trial. When wesucceeded in splitting the selected treatment animal(s) from the rest of the herd, they were conditionedin a direction away from the remaining herd and thetown site, and toward an area of suitable elk grazinghabitat. Although this splitting could be consideredanother form of conditioning, it was doneconsistently for all treatment groups including thecontrol, and we expected it to have the same relativeimpact on elk behavior. In Fig. 1, we providerepresentative photographs of habituated elk in theBanff townsite and examples of human and dog Response variables To assess conditioning-induced changes, wecollected data on each elk for three responsevariables: flight response distance, vigilance, anddistance to the town site boundary. We collectedthese data in two temporal phases: for six weeksprior to the application of the conditioningtreatments (before), and again during the latter halfof the conditioning period (after). Specific details 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ Fig. 1. Habituated elk in Banff National Park (A)prior to aversive conditioning research and during(B) human-based and (C) dog-based conditioning Flight response distancethroughout the winter, but at a minimum of 24 hfollowing the application of a conditioningtreatment to a given elk. Trials were conducted whenthe elk was neither bedded down nor traveling, andwas � 25 m away from vegetation cover. When theseconditions were met, one person approached thefocal elk directly from a minimum distance of 75 mfrom the elk. Flight response distance, i.e., theclosest distance the elk could be approached beforeit moved � 5 m in any direction, was measured witha digital range finder, the Bushnell Yardage Pro 500from Bushnell Corp., Overland Park, Kansas, USA,which is accurate to within 0.5 m. For analyticalpurposes, the preconditioning flight response datafor each elk were averaged into a "before" value. Toassess conditioning-induced change, we attemptedto measure the flight response of all elk followingtheir tenth aversive conditioning treatment.However, by that time, many elk had high levels ofwariness and concealment in forest cover thatprecluded the measurement of flight response,particularly for the human- and dog-conditionedelk. Thus, we derived our "after" value from anaverage of whatever flight response data wereavailable between the fifth and tenth conditioningtreatments for each elk. We acknowledge that thismetric undoubtedly provides a more conservativemeasure of treatment effects than might haveWe anticipated that several uncontrolled variablesmight also influence flight responses (LaGory1987). Thus, for each trial we recorded herd size,nearest neighbor distance, distance to cover, andrelative location within the herd, e.g., periphery,edge, center. Snow depth at the time of each flightresponse trial was calculated post hoc from anaverage of snow transect measurements around thetown site (H. Breniser, unpublished data). Becausethe proximity of wolves was likely to influence thebehavior of elk (e.g., Lima and Dill 1990), a relativeindex of wolf presence was also calculated post hocusing trackpad data from wildlife underpasses in thevicinity of the town site (A. Clevenger, unpublisheddata). Wolves and other animals concentrate theirmovements toward the town in these underpassesbecause a 2.4 m wildlife exclusion fence borders thehighway that parallels the Bow Valley throughBanff National Park (McGuire and Morrall 2000,Clevenger et al. 2001). To estimate wolf activityusing this information, we averaged the number of 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ wolf southbound passages, i.e., toward the town site,VigilanceVigilance trials were conducted opportunisticallythroughout the winter, but always more than 24 hafter a conditioning treatment was applied to thetarget elk. Trials were conducted when the elk was� 25 m away from vegetation cover and was notbedded down or traveling. The elk was observedfrom a vehicle or at long range with binoculars or aspotting scope to avoid detection and mitigate theinfluence of the observer. Each trial was 5 min inlength, and was abandoned if the elk bedded down,became aware of the observer's presence, or wasdisturbed by other human or wildlife activity in thearea. The amount of time was recorded for eachpredominant activity, e.g., feeding, scanning,grooming, defecating, moving, social interaction,and the proportion of time spent scanning or vigilantwas calculated. As for flight response, thepreconditioning vigilance data for each elk wereaveraged into a "before" value, and the vigilancedata between each elk's fifth and tenth conditioningtreatments were averaged into an "after" value. Theexact duration of time between average before andafter values differs among individual, and we referto this binary factor as simply the "time period."Again, we anticipated that other variables mightinfluence vigilance behavior, so we recorded herdsize, nearest neighbor distance, distance to cover,and relative location within the herd, i.e., periphery,edge, center. Average snow depth and wolf presencevalues for the time of each trial were calculated asProximity to town siteTo assess displacement from the urban areafollowing conditioning, we recorded one morningvisual sighting or radio-telemetry location per elkdaily between September 2001 and March 2002.These locations were recorded prior to dailymeasurements of flight response or vigilancebehavior or the application of conditioningtreatments. From those locations, elk distance to theclosest point on the town site boundary wascalculated using an ArcGIS Spatial Analyst(Environmental Systems Research Institute,Redlands, California, USA; McCoy and Johnston2000). Average snow depth and wolf presencevalues for the time of each location were calculatedas per flight response above. Again, the preconditioning distances for each elk wereaveraged into a "before" value, and the distancesafter each elk had received 10 conditioningtreatments were averaged into an "after" value per Statistical analyses All analyses were conducted using SPSS 11.5(Norusis 2002). Because of the relatively smallsample size of eight animals per group used by thisstudy, we set a = 0.10 to balance Type I and TypeII errors. One-way ANOVA was used in preliminaryanalyses to compare the average distance moved ineach trial among treatments; the proportion of timeeach elk associated with elk of the same and othertreatment groups, i.e., an association analysis; andthe proportions of Minimum Convex Polygonhome-range overlap within and among the differenttreatment groups, i.e., home-range overlap analysis.Analyses of all three response variables wereconducted using repeated-measures linear mixedmodels (Norusis 2002). In the flight response andvigilance models, the variables group size, nearestneighbor, distance to cover, location in the group,snow depth, and wolf presence were tested forsignificant univariate correlations and inclusion aspossible co-variates. The variables snow depth andwolf presence were tested for use in the modeldescribing proximity to town site. All biologicallyplausible two-way interactions were also tested byadding them, one at a time, to the final main effectsmodels and assessing their significance. Flightresponse data were log-transformed, and vigilancedata were square-root-transformed to achieve RESULTS Elk received an average of 8.0 aversive conditioningtrials (± 1.2 trials SE) at the time of postconditioningresponse variable measurement, with no significantdifference between treatment groups (t = -1.27, P =0.22, n = 16). During the 15-min conditioning trials,the dog and human treatments moved the elkaverage distances of 1219 m (± 80.3 m SE) and 1148m (± 50.8 m SE) respectively, whereas the controlgroup elk drifted an average of 49 m (± 13.7 m SE)during the sham treatment. We found no significantdifference in trial distance between the human- anddog-conditioned groups (F1,88 = 0.62, P = 0.43), butboth the human and dog treatments moved elk 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ Fig. 2. Elk flight response distance (m ± SE) before (white column) and after (black column) aversive significantly farther than the control group (F1,80 ³ 229.4, P £ 0.001). The association analysisexamined the proportion of time that elk spent withelk of their own treatment group (40.2 % average,± 2.1 % SE) in comparison with the time spent withelk of the other treatment groups (35.5 % average,± 1.7 % SE). Five of the six two-way comparisonsdid not significantly differ (P ³ 0.367), although thehuman-conditioned treatment group spent a lowerproportion of time with elk of the control group thanwith each other (P = 0.001). Average elk MinimumConvex Polygon (MCP) home-range size was2857.5 ha, and there was no significant difference in home range size among treatment groups (F2,23 =0.042, P = 0.96). The MCP home ranges of the 24individual research elk overlapped an average of81.3% (± 2.9 % SE), and there was no significantdifference between the proportions of home rangeoverlap within and among treatment groups (F5,551 =0.38, P = 0.86). Elk were conditioned at an averagefrequency of one treatment every 9.8 days (± 1.2 Overall, elk flight response distance increasedsignificantly following treatment application (F 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ Fig. 3. Relationship between flight response distance and nearby wolf activity for elk before (circle) and 1,28 = 23.120, P 0.0001), and there was a significantdifference in that increase among the treatments, asshown by the significant interaction betweentreatment and time period (Fig. 2; F2,22 = 10.86, P = 0.001). As we had predicted, the human- and dog-conditioned groups both differed significantly fromthe control group after conditioning (F1,14 ³ 5.63,P £ 0.037). However, there was no significantdifference between the human- and dog-conditioned groups (F1,14 = 1.10, P = 0.31). Onaverage, the elk in the human- and dog-conditionedgroups increased their flight response distance from an approaching human by 22.1 m (± 5.5 m SE), anincrease of 47.4%. The flight response of elk in thecontrol group negligibly increased by 6.4 m (± 5.5m SE). The variables group size, distance to cover,nearest neighbor, location in the group, snow depth,and wolf presence were tested for significantunivariate correlations. Snow depth was correlatedwith time (Pearson's r = 0.60) and excluded fromthe model. To test for possible inclusion as co-variates in the model, each remaining variable wasentered as a fixed-effect co-variate, singly and incombination with other co-variates, but only wolf 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ Fig. 4. Proportion of time elk spent vigilant (% ± SE), before (white column) and after (gray column) activity significantly and negatively affected flightresponse distance (F1,17 = 4.87, P = 0.041). Thismeant that elk exhibited shorter flight responsedistances when recent wolf activity was high. Therewas also a significant interaction between wolfactivity and time period (Fig 3; F1,22 = 5.79, P =0.025), showing that flight responses declined withincreasing wolf activity more strongly after Contrary to our expectation, vigilance declined for all treatment groups after conditioning (Fig. 4;F1,21 = 10.11, P = 0.005), and there was nosignificant difference in the relative decline amongtreatments (time period by treatment interaction;F2,21 = 0.18, P = 0.83). There was, however, a trendindicating that the vigilance of the conditionedanimals declined less (42%) than that of the controlanimals (62%). The same variables were tested forcorrelations and inclusion as in the flight responsemodel. Although snow depth was a significant co-variate, we excluded it from the model because ofa strong correlation with time (Pearson's r = 0.77). 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ Fig. 5. Average elk daily distance (m ± SE) to the closest point on the town boundary, before (white column)and after (black column) aversive conditioning treatments were applied to the three treatment groups in The distance from the daily position of each elk tothe town boundary increased significantly afterconditioning for all treatment groups (Fig. 5; F1,31 =11.51, P = 0.002). As for flight response, there wasa significant time period by treatment interactionthat revealed a much greater increase for the human-conditioned group (F4,27 = 2.97, P = 0.037). Thisgroup differed significantly from both the dog andcontrol treatments (F2,19 � 4.12, P 0.034), with nosignificant difference between the dog and controltreatments (F2,18 = 0.402, P = 0.675). Snow depth and wolf activity were tested for univariatecorrelations and possible inclusion in the model, butsnow depth was highly correlated with time(Pearson's r = 0.934) and thus excluded. Wolfactivity was not significant once entered with theother variables and so was excluded from the finalmodel. However, there was a significant interactionbetween wolf activity and treatment (Fig. 6; F3,37 =3.35, P = 0.029), showing that elk occurred at greaterdistances from town in the human treatment when 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ Fig. 6. Relationship between the proximity of elk to the town boundary and nearby wolf activity for elkthat were subjected to human (red-rimmed circle), dog (green square), and control (blue circle) aversive DISCUSSION Our results suggest that aversive conditioning iscapable of modifying some aspects of elk behaviortoward humans and human use areas. They alsosuggest that dog and human conditioning treatmentscan achieve similar levels of success for our primarymeasure of habituation, flight response distance.This variable, which showed a significant increaseafter both conditioning treatments, upheld ourpredictions that elk would be more wary of humansfollowing aversive conditioning. The similarity inresponse to both treatments for this variable maystem from the magnitude and consistency of the conditioning effort. In particular, both treatmentsinvolved chases with sufficient distance (1215 m),frequency (1/9.8 d), and duration (15 min) toemulate or exceed natural predation events bywolves. Prior research in Riding Mountain NationalPark (Carbyn 1983) showed that the length of lethalwolf chase sequences ranged from 20 to 260 m indistance. In Banff, the length of these sequencesaveraged 180 m, with a range of 10 to 1700 m (M.Hebblewhite, unpublished data; Parks Canada,unpublished data), although both studiesacknowledged a potential sample bias towardshorter sequences. Neither of those studies includednonlethal, i.e., unsuccessful, chase sequences, 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ which would presumably be much longer (M.Hebblewhite, personal communication). Ouraverage chase sequence length of 1215 m thereforeexceeded that of recorded natural predation events,but was perhaps comparable in length to nonlethalThe average frequency of our conditioningtreatments, which occurred once every 9.8 d, mayhave been similarly important to their efficacybecause it also exceeded the average frequency ofvisitation by wolves to an elk herd (1/13.4 d; Weaver1994). Although our conditioning treatments did notemulate all aspects of hunting by wolves, the energyloss and stress from our longer and more frequentconditioning events appeared to be severe enoughto trigger elk escape and avoidance responses.Longer hunting sequences by humans havegenerated extraordinarily large volumes of the stresshormone cortisol in red deer (Cervus elaphus),particularly when the animals dashed away only tobe repeatedly found and chased again by hunters(Bateson and Bradshaw 1997). Therefore, althoughour aversive conditioning chase sequences usinghumans and dogs did not present any directmortality risk for elk, the procedure may haveemulated enough aspects of wolf or human huntingfor elk to perceive us as potential predators anddisplay typical antipredator responses (e.g., BealeOur interpretation that our chase sequencesemulated predation events is consistent with thetheory that human-caused disturbance stimuli canbe considered a form of predation risk (Frid and Dill2002, Beale and Monaghan 2004), causing animalsto modify their behavior to maximize their security,decrease stress, and maintain their reproductivefitness (Geist 1982, Lima and Dill 1990, Frid andDill 2002). The predator-resembling nature of thechase sequence in our human and dog conditioningtreatments may also explain why the elk in ourexperiment showed no signs of habituating to theconditioning stimuli, whereas, in other studies,wildlife quickly habituated to the use of auditory(Bomford and O'Brien 1990, Belant et al. 1996,Bender 2003) or visual stimuli (Espmark andLangvatn 1985, Beringer et al. 2003, VerCauterenet al. 2003). A lack of perceived predation risk inthose contexts offers a potential explanation for theminimal increase in flight response for the controlelk in our study. This missing stimulus may alsoexplain increased signs of habituation and evenaggression by some control group elk to researchers during the four-month treatment period. Therapidity of the changes in the treated elk and theapparent change in the other direction by somecontrol elk suggests that elk can habituate to humanpresence in a nonthreatening environment quickly,which has management implications for Banff andother areas in which large numbers of tourists,golfers, and residents have benign encounters withThe decrease in the vigilance behavior of all the elkduring our treatment period is inconsistent with ourpredictions, but perhaps consistent with predictableseasonal changes in elk energy budgets related tosnow depth (e.g., Goodson et al. 1991). Our baselinedata were collected in the fall, when elk have higherenergy reserves that may support greater vigilanceinvestments and flight response distances (Parker etal. 1984). In contrast, our conditioning period andrelated data collection partially coincided with thelate winter period, when elk typically begincatabolism of body reserves (DelGiudice et. al 1991,DelGiudice et al. 2000). At that time, they maysacrifice their security, particularly with decreasedvigilance, to increase feeding time (e.g., Moen 1976,Gates and Hudson 1979, Lima and Dill 1990, Parkeret al. 1996). The strength of these seasonal effectson vigilance relative to treatment effects couldconceivably have diminished our ability to detectreal differences among the treatment groups (TypeII error). To avoid these sources of potential errorfor this type of research in the future, one wouldideally collect all vigilance data within the sameseason. Although we could not detect a significantdifference among treatments, control animalsexhibited a 20% greater decline in vigilance afterconditioning than did treated animals (Fig. 3). Thisimplies that elk maintained higher vigilance levelsunder human and dog conditioning than they didwithout it, offering qualified support for the utilityof aversive conditioning for modifying habituatedelk behavior. Unfortunately, the logistical necessityof using vigilance data between the fifth and tenthconditioning treatment of each elk, rather than datafollowing the tenth treatment, may have contributedto the nonsignificant difference among treatmentOur final result, that the human-conditioned elkexhibited significantly greater increases in thedistance between their daily positions and the Banfftown-site boundary, further supports our initialprediction that aversive conditioning can modify elkbehavior by teaching avoidance of human use areas, 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ corroborating the results of Nolte et al. (2003). Ourresult recommends human over dog conditioningand corroborates Frid and Dill's (2002) hypothesisthat human disturbance is a form of predation risk.Our human-conditioned elk subsequently avoidedthose areas in which conditioning had occurred,suggesting some memory of those aversive stimuliand the behavior that successfully avoided them. Analternative, but not mutually exclusive, explanationfor the subsequent avoidance of the town by theconditioned elk might reflect the energetic cost ofconditioning to the animal, particularly whenapplied in winter. During the late winter period withlower temperatures and deeper snow, elk couldnormally be expected to conserve energy byreducing their movement and staying in the areaswith the highest relative nutritional quality (Gatesand Hudson 1979, Parker et al. 1984, Sweeney andSweeney 1984, McCorquodale 1993), such as thoseoffered in the town site (McKenzie 2001). When wepersistently conditioned the elk more than akilometer away during that critical period, theenergetic cost of repeatedly moving to and from thetown site may have outweighed the benefits ofreturning to it (see also Bunnell and Gillingham1985, Bradshaw et al. 1998). These results suggestthat aversive conditioning can potentially reduce theurban "refuge effect" by both increasing theperceived predation risk and reducing the energeticDespite the significant treatment effect we detectedfor proximity to the town boundary, our associationanalysis showed some evidence that all treatmentgroups responded to our conditioning, indicatingsome herd-level influences (see also Galef 1995,Ralphs and Provenza 1999). Because elk receiveboth energetic and security benefits from herdinginto larger groups, particularly in late winter (Moen1976, Geist 1982, Hebblewhite and Pletscher 2002),it is likely that this phenomenon reduced some ofthe variation we might otherwise have detectedamong treatments. An additional explanation forherd-level effects could relate to emphatic learning(e.g., Klopfer 1957) or social facilitation (e.g.,Harlow 1932), whereby individuals not exposed tothe conditioning treatment learn from theantipredatory behavior of other conditioned herdmembers (see Muller-Schwarze 1991, Griffin2004). Nevertheless, regardless of the potential forherd-level effects to influence this research, the highvariation we documented among individualsencourages further individual-based research, andstrengthens the apparent success of our human- based aversive conditioning protocols for managinghyperabundant wildlife in areas with high humanAn important co-variate that was apparent in ouranalyses of both flight response distance andproximity to town was the activity of wolves. Inboth cases, wolf activity appeared to reduce theefficacy of our conditioning treatments, as might beexpected when elk had the option of challenging usvs. real predators. Flight responses decreased whenwolf activity was high, suggesting that elk werereluctant to flee from us in the vicinity of predators.This effect was much stronger after conditioning,effectively acting in opposition to our treatmenteffects. Proximity to town also responded to wolfactivity, this time exhibited as an interaction withtreatment. This meant that the strong effect ofhuman conditioning was sharply reduced by highwolf activity. Dog and control treatments, which didnot show significant postconditioning effects onproximity, were not similarly affected by wolfactivity. Again, this result indicates that oureffective (human) conditioning would have beeneven more effective were it not for the counteractingeffects of wolves. Banff and other jurisdictions maythus bear in mind that more conditioning effort willlikely be needed when it pushes animals in thedirection of predators relative to comparable CONCLUSION These results have two main implications formanagers. First, because the dog and humansimilar levels of success in modifying elk flightresponses, the choice of which treatment to applymight be based on economic efficiency and locallogistics for jurisdictions contending only with thebehavior, and not the location, of urban wildlife.Presumably, professional dog handlers who arewilling to use their highly trained border collies forwild elk management are both scarce and expensive.In our system, we calculated the monthly cost ofconditioning with dogs to be $4300.00 CAD, 15 %higher than the cost of conditioning with humans.Human conditioning in many jurisdictions couldalso be done without hiring additional staff, whichwould further lower the associated costs. For thesereasons, we chose to use humans for a second seasonin which we sought to identify the optimal frequencyof conditioning for elk with differing levels of initial 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ habituation and assessed the extinction ofconditioned responses (Kloppers et al., unpublishedmanuscriptA second aspect of the choice of conditioningmedium for managers of national parks and urbanareas is the nature of public perception. For us, thepublic response to the quiet, friendly appearance ofthe dog treatment was generally positive, but wereceived some noise complaints about the crackerand screamer shells associated with the humantreatment. Because the predator-resembling chasesequence appears to be the critical component ofconditioning and the method used, i.e., crackershells or dogs, is secondary, managers could alsochoose a variety of other methods to complementthe chase sequence and suit their situation. Forexample, wardens in Banff National Park typicallyuse raised hockey sticks with bags tied to one endto condition elk by emulating the antler displayscommon among sparring elk (Geist 1982, Jenningset al. 2002), and Canadians who observe this aregenerally quite accepting of this form of threatdisplay. Ultimately, budgets, the local situation, andassociated management priorities will dictatewhether the human or dog aversive conditioningtreatment is the most appropriate choice for a givenOur research suggests that predator-resemblingaversive conditioning can provide both an effectiveand socially appealing alternative to traditionallyinvasive management techniques for habituatedwildlife. By proactively addressing the causalfactors rather than just treating the symptoms,aversive conditioning can prevent habituation fromoccurring, thereby achieving ecological and publicsafety objectives while maintaining wildlifeviewing opportunities and public support. Predator-resembling aversive conditioning ultimately offersmanagers of urban and protected areas a solutionthat balances ecological and human dimensions in Responses to this article can be read online at: http://www.ecologyandsociety.org/vol10/iss1/art31/responses/ Acknowledgments: We acknowledge the generous financial andlogistical support of Banff National Park and the University of Alberta; the capable and enthusiasticfield assistance of J. Mills, P. Ashby, R. Gareau, C.Aikens, and J. Zehnder with herding dogs Blue andDuke; the insightful advice of E. Merrill, S. Bayley,C. White, R. Hudson, and four anonymousreviewers; and the operational support of G. Peers,R. LeBlanc, I. Syme, D. Herman, D. Eastcott, T.Shury, T. Davidson, T. Gui, and J. Truscott. Someessential equipment was provided by the CanadaFoundation for Innovation via a New OpportunitiesGrant to C. C. St. Clair. This research was fundedthrough Parks Canada MOA #01-0068, inaccordance with the University of Alberta AnimalCare Committee Permit #354111. LITERATURE CITED Altmann, M. 1958. The flight distance in free-ranging big game. Journal of Wildlife Management 22Baker, S. V., and F. Fritsch. 1997. New territoryfor deer management: human conflicts on thesuburban frontier. Wildlife Society Bulletin 25Bateson, P., and E. L. Bradshaw. 1997.Physiological effects of hunting red deer (Cervuselaphus). Proceedings of the Royal Society ofLondon Series B-Biological Sciences 264Beale, C. M., and P. Monaghan. 2004. Humandisturbance: people as predation-free predators.Journal of Applied Ecology41Belant, J. L., T. W. Seamans, and C. P. Dwyer. 1996. Evaluation of propane exploders as white-tailed deer deterrents. Crop Protection 15Bender, H. 2003. Deterrence of kangaroos fromagricultural areas using ultrasonic frequencies:efficacy of a commercial device. Wildlife SocietyBulletin 31Beringer, J., K. C. VerCauteren, and J. J.Millspaugh. 2003. Evaluation of an animal- 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ activated scarecrow and a monofilament fence forreducing deer use of soybean fields. Wildlife SocietyBulletin 31Bomford, M., and P. H. O'Brien. 1990. Sonicdeterrents in animal damage control: a review ofdevice tests and effectiveness. Wildlife SocietyBulletin18Bounds, D. L., and W. W. Shaw. 1994. Managingcoyotes in U.S. National Parks: human-coyoteinteractions. Natural Areas Journal 14(4):280-284.Bradshaw, C. J. A., S. Boutin, and D. M. Hebert.1998. Energetic implications of disturbance causedby petroleum exploration to woodland caribou.Canadian Journal of Zoology 76Brush, R. 1971. Aversive conditioning andlearning. Academic Press, New York, New York,Bunnell, F. L., and M. P. Gillingham. 1985.Foraging behavior: dynamics of dining out. Pages53-79 in R. J. Hudson and R. G. White, editors.Bioenergetics of wild herbivores. CRC Press, BocaCarbyn, L. N. 1983. Wolf predation on elk inRiding Mountain National Park, Manitoba. Journalof Wildlife Management47Caughley, G. 1981. What is overabundance? Pages159-187 in P. A. Jewell and S. Holt, editors.Problems in management of locally abundant wildmammals. Academic Press, New York, New York,Clevenger, A. P., B. Chruszcz, and K. E. Gunson. 2001. Highway mitigation fencing reduces wildlife-vehicle collisions. Wildlife Society Bulletin 29Davey, G. 1981. Animal learning and conditioning.DelGiudice, G. D., F. J. Singer, and U. S. Seal. 1991. Physiological assessment of winternutritional deprivation in elk of YellowstoneNational Park. Journal of Wildlife Management 55 DelGiudice, G. D., K. D. Kerr, L. D. Mech, andU. S. Seal. 2000. Prolonged winter undernutritionand the interpretation of urinary allantoin:creatinineratios in white-tailed deer. Canadian Journal ofZoology78Domjan, M. 2003. The principles of learning andbehavior. Fifth edition. Thomson Wadsworth,Espmark, Y., and R. Langvatn. 1985.Development and habituation of cardiac andbehavioural responses in young red deer calves(Cervus elaphus) exposed to alarm stimuli. Journalof Mammalogy 66Etter, D. R., K. M. Hollis, T. R. Van Deelen, D. R.Ludwig, J. E. Chelsvig, C. L. Anchor, and R. E.Warner. 2002. Survival and movements of white-tailed deer in suburban Chicago, Illinois. Journal ofWildlife Management66Frid, A. 1997. Vigilance by female Dall's sheep:interactions between predation risk factors. AnimalBehaviour 53Frid, A., and L. M. Dill. 2002. Human-causeddisturbance stimuli as a form of predation risk.Conservation Ecology Galef, B. G. 1995. Why behaviour patterns thatanimals learn socially are locally adaptive. AnimalBehaviour49Gander, H., and P. Ingold. 1997. Reactions of malealpine chamois Rupricapra r. rupicapra to hikers,Biological Conservation79Gates, C. C., and R. J. Hudson. 1979. Effects ofposture and activity on metabolic responses ofwapiti to cold. Journal of Wildlife Management 43Geist, V. 1982. Adaptive behavioural strategies.Pages 219-277 in J. W. Thomas and D. E. Toweill,editors. Elk of North America: ecology andmanagement. Stackpole Books, Harrisburg,Goodson, N. J., D. R. Stevens, and J. A. Bailey. 1991. Effects of snow on foraging ecology and 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ nutrition of bighorn sheep. Journal of WildlifeManagement55Government of Canada. 2000. Canada NationalParks Act: an act respecting the National Parks ofCanada.Griffin, A. S. 2004. Social learning about predators:a review and prospectus. Learning and Behavior 32Guynn, J. R., and G. K. Yarrow. 1997. Ecosystemmanagement and wildlife management: compatibleor conflicting? Proceedings of the Eastern WildlifeDamage Management Conference, Jackson,Mississippi7: http://wildlifedamage.unl.edu/handbook/Chapters/sech2. doc Harlow, H. F. 1932. Social facilitation of feedingin the albino rat. Pedagogical Seminary and Journalof Genetic Physiology41Hebblewhite, M. and D. H. Pletscher. 2002.Effects of elk group size on predation by wolves.Canadian Journal of Zoology 80Herrero, S. 1985. Bear attacks: their causes andavoidance.Holland, W. D., and G. M. Coen. 1983. Ecological(biophysical) land classification of Banff and JasperNational Parks. Volume 1. Summary. AlbertaHolroyd, G. L., and K. J. Van Tighem. 1983.Ecological (biophysical) land classification of Banffand Jasper National Parks. Volume 3. The wildlifeinventory. Canadian Wildlife Service, Edmonton,Hurd, T. E. 1999. Factors limiting moose numbersand their interaction with elk and wolves in theCentral Rocky Mountains, Canada. Thesis.University of British Columbia, Vancouver,Isbell, L. A., and T. P. Young. 1993. Humanpresence reduces predation in a free-ranging vervetmonkey population in Kenya. Animal Behaviour 45Jennings, D. J., M. P. Gammell, C. M. Carlin, andT. J. Hayden. 2002. Does lateral presentation of the palmate antlers during fights by fallow deer (Damadama L.) signify dominance or submission?Ethology 108:Kilpatrick, H. J., and S. M. Spohr. 2000. Spatialand temporal use of a suburban landscape by femalewhite-tailed deer. Wildlife Society Bulletin 28Klopfer, P. H. 1957. An experiment in empathiclearning in ducks. The American Naturalist 91LaGory, K. E. 1987. The influence of habitat andgroup characteristics on the alarm and flightAnimal Behaviour 35:Lee, M. E., and R. Millar. 2003. Managing elk inthe wildland-urban interface: attitudes of Flagstaff,Arizona residents. Wildlife Society Bulletin 31Lima, S. L., and L. M. Dill. 1990. Behaviouraldecisions made under the risk of predation: a reviewand prospectus. Canadian Journal of Zoology 68Lubow, B. C., F. J. Singer, T. L. Johnson, and D.C. Bowden. 2002. Dynamics of interacting elkpopulations within and adjacent to Rocky MountainNational Park. Journal of Wildlife Management 66MacArthur, R. A., V. Geist, and R. H. Johnston. 1982. Cardiac and behavioural responses ofmountain sheep to human disturbance. Journal ofWildlife Management 46Martinetto, K., and J.-M. Cugnasse. 2001.Reaction distance in Mediterranean mouflon (Ovusgmelini musimon x ovis sp.) in the presence of hikerswith a dog on the Caroux Plateau (Herault, France).Revue d'Ecologie—la Terre et la Vie 56McConnell, P. B., and J. R. Baylis. 1985.Interspecific communication in cooperativeherding: acoustic and visual signals from humanshepherds and herding dogs. Zeitschrift furTierpsychologie67:McCorquodale, S. M. 1993. Winter foragingbehavior of elk in the shrub-steppe of Washington.Journal of Wildlife Management57 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ McCoy, J., and K. Johnston. 2000. Using ArcGISSpatial Analyst. Environmental Systems ResearchMcCullough, D. R., K. W. Jennings, N. B. Gates,B. G. Elliott, and J. E. DiDonato. 1997.Overabundant deer populations in California.Wildlife Society Bulletin 25McGuire, T. M., and J. F. Morrall. 2000. Strategichighway improvements to minimize environmentalimpacts within the Canadian Rocky MountainNational Parks. Canadian Journal of CivilEngineering27:McKenzie, J. A. 2001. The selective advantage ofurban habitat use by elk in Banff National Park. Thesis. University of Guelph, Guelph, Canada. http://www.collectionscanada.ca/obj/s4/f2/dsk3/ftp04/ MQ61927.pdf McNay, M. E. 2002a. A case history of wolf-humanencounters in Alaska and Canada. AlaskaDepartment of Fish and Game Wildlife Technical http://wildlife.alaska.gov/pubs/techpubs/research_pdfs/ techb13_full.pdf McNay, M. E. 2002b. Wolf-human interactions inAlaska and Canada: a review of the case history.Wildlife Society Bulletin30Mech, L. D. 1970. The wolf: the ecology andbehavior of an endangered species. Doubleday/Natural History Press, Garden City, New York,Mech, L. D., and S. M. Barber. 2002. A critiqueof wildlife radio-tracking and its use in nationalparks: a report to the U.S. National Park Service. U.S. Geological Survey, Northern Prairie WildlifeResearch Center, Jamestown, North Dakota, USA. http://www.npwrc.usgs.gov/resource/wildlife/radiotrk/ radiotrk.htm Messmer, T. A., S. M. George, and L. Cornicelli.1997. Legal considerations regarding lethal andnonlethal approaches to managing urban deer.Wildlife Society Bulletin 25Miller S., R. Knight, and C. Miller. 2001. Wildliferesponses to pedestrians and dogs. Wildlife Society Bulletin29Moen, A. N. 1976. Energy conservation by white-Ecology 57Morellet, N., and B. Guibert. 1999. Spatialheterogeneity of winter forest resources used byForest Ecology and Management 123Muller-Schwarze, D. 1991. The chemical ecologyApplied Animal BehaviourScience 29National Park Service. 1991. Natural resourcesmanagement guidelines. U.S. Department of theNietvelt, C. G. 2001. Herbivory interactionsbetween beaver (Castor canadensis) and elk (Cervuselaphus) on willow (Salix spp.) in Banff NationalPark, Alberta. Thesis. University of Alberta, http://www.nlc-bnc.ca/obj/s4/f2/dsk3/ftp04/MQ60482. pdf Nolte, D. L., K. C. Vercauteren, K. R. Perry, andS. E. Adams. 2003. Training deer to avoid sitesthrough negative reinforcement. Pages 95-104 in K.A. Fagerstone and G. W. Witmer, editors.Proceedings of the Tenth Wildlife DamageManagement Conference. Wildlife DamageManagement Working Group of the WildlifeSociety, Fort Collins, Colorado, USA. Available http://www.aphis.usda.gov/ws/nwrc/is/03pubs/nolte0311. pdf Norusis, M. 2002. SPSS 11.0 guide to data analysis. Prentice Hall, Englewood Cliffs, New Jersey, USA.Okarma, H. 1995. The trophic ecology of wolvesand their predatory role in ungulate communities offorest ecosystems in Europe. Acta Theriologica 40Parker, K. L., C. T. Robbins, and T. A. Hanley.1984. Energy expenditures for locomotion by muledeer and elk. Journal of Wildlife Management 48Parker, K. L., M. P. Gillingham, T. A. Hanley,and C. T. Robbins. 1996. Foraging efficiency:energy expenditure versus energy gain in free-ranging black-tailed deer. Canadian Journal of 10 http://www.ecologyandsociety.org/vol10/iss1/art31/ Zoology74Ralphs, M. H., and F. D. Provenza. 1999.Conditioned food aversions: principles andpractices, with special reference to socialfacilitation. Proceedings of the Nutrition Society 58Reynolds, G. 1975. A primer of operantconditioning. Scott, Foresman and Co., Glenview,Riley, S. P. D., J. Hadidian, and D. A. Manski. 1998. Population density, survival, and rabies inracoons in an urban national park. CanadianJournal of Zoology76Ripple, W. J., and R. L. Beschta. 2004. Wolvesand the ecology of fear: can predation risk structureBioScience 54Rondeau, D., and J. M. Conrad. 2003. Managingurban deer. American Journal of AgriculturalEconomics 85Rubin, E. S., W. M. Boyce, C. J. Stermer, and S.G. Torres. 2002. Bighorn sheep habitat use andselection near an urban environment. BiologicalConservation104Rybarczyk, P., J. Rushen, and A. M. de Pasille. 2003. Recognition of people by dairy calves usingcolour of clothing. Applied Animal BehaviourScience81Schultz, R. D., and J. A. Bailey. 1978. Responsesof national park elk to human activity. Journal ofWildlife Management42Smith, B. 1998. Moving 'em: a guide to low stressanimal handling. Graziers Hui, Kamuela, Hawaii,Soulé, M. E., J. A. Estes, J. Berger, and C. M. DelRio. 2003. Ecological effectiveness: conservationgoals for interactive species. Conservation Biology 17Staines, B. W., and D. Welch. 1989. Impact of redand roe deer on Scottish woodlands. Pages 128-130in R. J. Putman, editor. Mammals as pests. ChapmanSteere, A. C. 1994. Lyme disease: a growing threat to urban populations. Proceedings of the NationalAcademy of Sciences of the United States ofAmerica 91Sweeney, J. M., and J. R. Sweeney. 1984. Snowdepths influencing winter movements of elk.Journal of Mammalogy65Taylor, A. R., and R. L. Knight. 2003. Behavioralresponses of wildlife to human activity: terminologyand methods. Wildlife Society Bulletin 31Thompson, M., and R. Henderson. 1998. Elkhabituation as a credibility challenge for wildlifeprofessionals. Wildlife Society Bulletin 26VerCauteren, K. C., S. E. Hygnstrom, M. J.Pipas, P. B. Fioranelli, S. J. Werner, and B. F.Blackwell. 2003. Red lasers are ineffective fordispersing deer at night. Wildlife Society Bulletin 31Warren, R. J. 1991. Ecological justification forcontrolling deer populations in eastern nationalparks. North American Wildlife and NaturalResources Conference 56Weaver, J. L. 1994. Ecology of wolf predationamidst high ungulate diversity in Jasper NationalPark, Alberta. Thesis. University of Montana,White, C. A., C. E. Olmsted, and C. E. Kay. 1998.Aspen, elk, and fire in the Rocky Mountain nationalparks of North America. Wildlife Society Bulletin 26Witmer, G. W., and D. G. Whittaker. 2001.Dealing with nuisance and depredating black bears.Western Black Bear Workshop 7:73-81. Available http://www.aphis.usda.gov/ws/nwrc/is/01pubs/01-88. pdf Whittaker, D., and R. Knight. wildlife responses to humans. Wildlife SocietyBulletin26Woods, J. G. 1991. Ecology of a partially migratoryelk population. Thesis. University of British