During the last 10 - PDF document

During the last 10…20 years,many areas haveexperienced an increase in the number of conflictsbetween black bears (Ursus americanushumans,and such conflicts have been dispropor-tional to human population growth (Beckmann2002,Beckmann and Berger 2003).This is espe-cially true in western North America,where rapidurban sprawl has encroached into areas adjacent to DETERRENT TECHNIQUES FOR BEARS Wildlife Society Bulletin 2004, 32(4):1141…1146Peer refereed Jon P. Beckmann, Carl W. Lackey, and Joel BergerAbstractThe general public often prefers nonlethality when dealing with problem black bears). We evaluated the efficacy of nonlethal deterrent techniques on 62 40 days. We conclude that in the Lake TahoeBasin the most common nonlethal deterrents, used by agencies responsible for black bear agement agencies currently use to alter behavior ofnuisanceŽbears,although exceptions clearly exist(Gilllin et al.1994,Ternent and Garshelis 1999,Clark et al.2002).A survey conducted by the Virginia Departmentof Game and Inland Fisheries in 2001 revealed that33 states currently manage black bears and respondto citizen complaints about nuisanceŽbears (D.Kocka,Virginia Department of Game and InlandFisheries,personal communication).Of those,26(79%) administer deterrent techniques with theaim of behavioral alteration of nuisanceŽindividu-als.The use of deterrent techniques,although not anew management tool,has been increasing rapidlyin both Canada and the United States,primarily inresponse to the publics request for nonlethal bearmanagement near urban…wildland interface areas.This is particularly true given the large increase inhuman…bear conflicts over the past 10…15 years(Beckmann and Berger 2003).Fifteen of 26 statesthat currently use deterrents began doing so in the1990s (D.Kocka,Virginia Department of Game andInland Fisheries,personal communication).In con-trast,only 4 states administered deterrents in the1960s and 1970s.The 6 most common techniquesused on trapped bears according to the 33 statessurveyed were 1) rubber buckshot,2) rubber slugs,3) pepper spray,4) cracker shells,5) dogs,and 6)loud noises (D.Kocka,Virginia Department of Gameand Inland Fisheries,personal communication).Although many states and other entities such asnational parks (e.g.,Yosemite National Park,Calif.)spend many dollars annually for such deterrents,todate no research has rigorously analyzed the effica-cy of these deterrents.We capitalized on the extent to which humanpopulation growth and coincident food stores inthe Lake Tahoe Basin in the Sierra Nevada Range inwestern Nevada offer an experimental setting inwhich to examine the effectiveness of deterrenttechniques on behavior of black bears.FromTahoe Basin increased by 26% and the number ofcomplaints by citizens concerning black bearsincreased by� 10-fold (Beckmann 2002,Beckmannand Berger 2003).Our goal was to examine theeffectiveness of the 6 most common deterrentsused on black bears.MethodsBlack bear distribution in Nevada is restricted tothe Sierra Nevada and nearby mountains thatinclude the Sweetwater,Pine Nut,and Wassukranges (Goodrich 1990),all of which were thefocus of our work.These Great Basin Desert moun-tain ranges contain areas with high granite peaksand deep canyons (Goodrich 1990).Desert floorsthat can be greater than 64 km wide separatemountain ranges in which bears occur.Further,desert basins often are large areas of unsuitabledesert habitat (e.g.,large expanses of sagebrushushArtemesiaspp.]) that bears do not use (Goodrich1990,Beckmann 2002,Beckmann and Berger).However,bears occasionally will make rel-atively short movements through areas of sage-brush to reach patchily distributed suitable habi-tats,consisting mainly of lodgepole (Pinus contor-) and Jeffrey pine (Pinus jeffreyi) forests in thisarid landscape (Grayson 1993).The human popu-lation inside the Lake Tahoe Basin w�as 50,000 peo-ple,with an additional 350,000 people living in thestudy area along the eastern front of the SierraNevada in Reno and Carson City,Nevada.Bears inthis region are at the eastern edge of their knownrange in the Great Basin,with the closest popula-tion to the east being about 750 km away in Utah.Although black bears are listed as a game species inNevada,there has never been a legal harvest.We captured bears in culvert traps (TetonWelding,Choteau,Mont.) and tranquilized andimmobilized them with a mixture of Telazol…Xylazine from 1 July 1997 to 1 April 2002.Weweighed each bear and fitted each adult with amortality-sensing radiocollar (Advanced TelemetrySystems,Isanti,Minn.).We determined age fromannuli of the first upper premolar (PM),the stan-dard tooth for age analysis in black bears (MatsonsLaboratory,Milltown,Mont.;Stoneberg and Jonkel1966),and we classified animals as cubs (years),juveniles (1.5…3 years),
.50;or adults (3 years).At time of capture and administration of deterrents,the history of each bear was unknown due to a lackof recent study of the population immediately pre-ceding onset of this study.However,subsequentmonitoring and data collection indicated that all 62truly were nuisanceŽbears,as all of them wereinside urban areas multiple times (Beckmann andBerger 2003).We also documented rapid changesin the ecology and behavior of urban bears as aconsequence of their foraging almost entirely ongarbage (Beckmann and Berger 2003We tested the effectiveness of the 6 most com-mon deterrents used by state agencies in the Wildlife Society Bulletin 2004,32(4):1141…1146 United States (D.Kocka,Virginia Department ofGame and Inland Fisheries,personal communica-tion).We randomly assigned 62 collared bears cap-tured in urban areas in the Lake Tahoe Basin of theSierra Nevada to 1 of 3 groups:experimental group,which received deterrents (=21);experimental+dog group,which were chased by dogs in additionto receiving other deterrents (=20);or controlgroup,which did not receive deterrents (=21).Experimental treatment consisted of bears beinghit with pepper spray,12-gauge rubber buckshot,and a rubber slug,and exposed to cracker shellsand yelling.Our methods did not allow examina-tion of each deterrent individually.We were inter-ested only in the combination of these methods.Bears in the treatment+dog group were chased byhounds in addition to other deterrents.Each bearwas chased by the same hounds to standardize thetreatment.We released control bears in a silentŽmanner,with no physical or audible deterrents.Clark et al.(2002) suggested that the capture andhandling experience also should be considered anaversive agent.Because all bears in our analyseswere captured and handled in the same manner,weassumed that this would not bias the data.Wemoved individual bears varying distances from thecapture site to administer deterrents.Distancemoved ranged from 1…75 km,and distance wasincluded as a continuous variable in the model.We measured effectiveness as time (in days)required for the bear to return to the urban patch(RUP) after treatment application,or silentŽreleasein the case of controls.We located animals weekly,weather permitting,from a Cessna 206 fixed-wingairplane (Cessna Aircraft Company,Wichita,Kans.).Most flights occurred from 0500…1600 PacificStandard Time.We assigned Universal TransverseMercator coordinates to each location from aGlobal Positioning System unit onboard the aircraft.We entered all locations into coverage maps anddefined urban areas by town and city delineation(incorporated city limits) in ArcView 3.2.We con-sidered an individual to have returned (RUP) thefirst time we located it inside the same urban patchfrom which it had been removed.If an individualreturned to the urban area in the weeklong timeperiod between flights,we averaged date of return.We modeled deterrent effectiveness using a mul-tivariate survival analysis with a Cox proportionalhazards model (PROC PHREG in SAS statistical soft-ware;SAS Institute 2001).This analysis is a log like-lihood model and produces a hazard ratio:the con-ditional probability that the event of interest (RUP)occurs in the given time interval (Schoenfeld 1980,Klein and Moeschberger 1997,Song and Lee 2000,Persson 2002).Analysis of survival data,in this caseRUP (i.e.,failureŽor deathŽ),requires special tech-niques because data are almost always incompleteand parametric assumptions may be unjustifiable.For example,5 individuals failed to RUP by the timeof our analyses;thus their status was unknown.ŽThe problem is onerous because the 5 bears maynever come back,or they could return at anyunknown future time.These 5 survival times (8% ofthe observations) were censored.The remaining 57bears had known survival times,referred to asevent times.Methods for survival analysis mustaccount for both censored and noncensored data.The Cox proportional hazards model is an excel-lent tool for making inferences on the populationaverage effect of covariates with incomplete failuretime data (Schoenfeld 1980,Klein andMoeschberger 1997,Song and Lee 2000,PerssonPredictor variables in the saturated 2-way inter-active model included age,weight,season,sex,dis-tance moved,and treatment (3 levels).We assignedcategorical predictor variables dummy variables forthe proportional hazard regression model.We com-pared potential models,beginning with the saturat-ed 2-way interactive model,using information-theo-retic methods to direct model selection.We calcu-lated Akaikes Information Criterion (AIC) andadjusted these for small sample sizes (AICgested by Hurvich and Tsai (1989) and Anderson etal.(1994).We used these values (pare candidate models to select the most parsimo-nious model that accurately represented the data(Anderson et al.2000).We ran these complex models to try and developa predictive model for managers to assess whenapplication of deterrents is most likely to be suc-cessful.We included 6 ecological and biologicalparameters of individuals that we believed to be apriori relevant.We also analyzed RUP data using amore simplistic analysis of variance (ANOVA)model to compare treatment levels.The ANOVAmodel also had the added benefit of not censoringdata,as the Cox proportional hazards regressionmodel had done.For the 5 cases in which bearshad not returned (RUP) at the time of the analysis,we used number of days they had been gone. Deterrent techniques for bears € Beckmann et al. SD are presented unless otherwise noted.ResultsIn 57 of the 62 cases (92%),bears returned to theurban patch where they were captured.Of the 62bears,33 (53%) returned in less than 30 days,17(27%) returned between 31 and 180 days,7 (11%)returned between 181 and 365 days,and 5 (8%) hadnot retur�ned in 365 days (Table 1).Mean numberof days for RUP was not significantly differentbetween the 3 treatments (ANOVA;F=0.61,0.5468).However,mean RUP for the treatment+dog group was 154202 days compared to 88.476.5 days for the treatment-only group and 64.6103.9 days for the control group.Only the treat-ment level was selected in the most parsimoniousmodel (hazard ratiotreatment=0.454;=0.0061;Table2) when analyzing data using Cox proportional haz-ards regression.Thus,an individual was 0.454 lesslikely to RUP in the presence of deterrents than intheir absence,and treatment level alone predictedRUP.Dogs chased individuals in 4 of the 5 cases inwhich they had yet to RUP.Mean number of dayselapsed between application of the deterrent andtime of analysis for the 5 bears that had not RUPwas 481 days (range 424…641 days;Table 1).In allmodels no other predictor variables or any interac-tions accounted for significant variation in RUP.DiscussionWe evaluated 6 ecological and biological param-eters of bears we believed might prove relevant todeterring nuisanceŽbears from returning.Wefound that 92% of the time,black bears returned tothe urban patch from which they were removed.Although it would be ideal if treated bears did notreturn to urban areas,thus eliminating human…bearconflicts in those regions,the more realistic issue isnot whether bears return but when.Our data indi-=44) of the bears returned in days.Based on assessment of 6 ecological and bio-logical parameters,we were unable to produce amodel that predicted which individuals and underwhat circumstances this combination of deterrentswould be a useful management strategy.However,deterrents did have a positive effect on modelselection:only treatment was left in the most parsi-monious model,and change in AIC was� 2.0 versusthe next best model (Burnham and Anderson 1998;Table 2).However,a majority of bears generallyreturned within 1 month,suggesting that deter-rents were not very effective at altering behavior ofbears based on RUP data.Bears chased by hounds did return slightly lateron average than bears not chased by dogs,either inthe control group or in the treatment-only group.However,differences were not statistically signifi-cant.Further indication that dogs may not be aneffective deterrent was the tremendous variation inRUP for bears chased by dogs (range:5…641 days).In fact,bears within all groups showed tremendousvariation in RUP (overall range:1…641 days).Thisextreme within-group variance suggested thateffectiveness of deterrents was likely based on indi-vidual variation among bears and how they Wildlife Society Bulletin 2004,32(4):1141…1146 Table 1. Time period (days) and number of black bears () that returned to the urban patch (RUP) from whichthey were captured in western Nevada from July 1997 to April2002 following 3 different treatment levels of deterrents.Control bears were released without any deterrents; treatmentbears were released after exposure to pepper spray, 12-gaugerubber buckshot, a rubber slug, and cracker shells and yelling;the other deterrents and chased by hounds.ControlTreatmentTreatment+dog Days to RUP(= 21)(= 21)(169831-180296181-3653220-4;؀.;　365014 Table 2. Multi-model inference of behavioral responses of) captured in western Nevada from July1997 to April 2002 to deterrent techniques. The best model foreach of the respective number of parameters is presented.Overall, the model containing only treatment was the most par-Number of days to returnto urban area# parameters{T}100.542{T, Dis}22.560.330{T, S, Dis}34.130.071{T, S, W, Dis}45.940.027{T, S, G, W, Dis}510.210.018{T, A, S, G, W, Dis}613.400.0115T = treatment (control, deterrent, deterrent + dog); A =age; S = season (spring, summer, winter, fall); G = sex (male,female); W = weight; Dis = distance moved from capture site towhere deterrents were administered.AIC is the rank of each model by rescaling the modelwith a minimum AIC value to zero (AIC = AIC… minAIC). AICweights are the likelihood of the model given the data (Akaike responded to such disturbances.Our models didnot allow us to incorporate individual variation inthe level of boldnessŽor shynessŽof each bear.There are obvious limitations in approaches tostudying large wild carnivores.First,although 62collared individuals may be a reasonable sample,larger sample sizes are needed for adequate powerto detect the true effectiveness of some deterrents.Second,with 62 bears we were unable to establisha group that received only dogs without the otherdeterrents.This would have created too many cat-egories for the limited sample.Additional researchshould examine the effectiveness of each deterrentindividually.Management implicationsWe recommend that any group dealing with nui-sanceŽblack bears conduct a cost…benefit analysisto decide whether monetary investments in deter-rents are worth it.Ifagencies define success ofdeterrents a priori as never having to deal with anuisanceŽbear again,our data suggest that thisobjective will most likely always fail.If the goal isto establish positive public relations or to avoiddealing with an individual bear for several weeks ormonths,then deterrents may be an effective man-agement tool.The Nevada Division of Wildlife hashad fewer negative responses from the local mediaand public in the Lake Tahoe Basin about using non-lethal deterrents compared to the 5-year periodbefore these techniques were in use (C.Healy,Nevada Division of Wildlife,personal communica-tion).Use of nonlethal deterrents may have theadded benefit of increasing public awareness ofhuman…bear conflicts created by availability ofurban food sources.For example,2 homeownerassociations and a private campground on thesouth shore of Lake Tahoe spent a combined$100,000 on 350 bear-proof garbage containers inresponse to the use of nonlethal deterrents onbears in Nevada (M.Paulson,Tahoe VillageHomeowners Association,personal communica-Our results indicate that the most commonlyused deterrents to alter behavior of nuisanceŽblack bears are not effective when used in combi-nation.Use of dogs and other deterrents providedlimited evidence for a longer time period beforebears returned to the urban area,although due toextreme within-group variance this difference wasnot significant.Further,even use of dogs in combi-nation with other deterrents was not very effectiveat altering bear behavior beyond 1 month.Ourstudy suggests that bears that were human-food(i.e.,garbage) conditioned and habituated to livingnear or in urban…wildland interface areas wereunlikely to alter their behavior in response to thedeterrent techniques currently adopted by moststate and federal agencies.A more effective strategy to reduce human…bearconflicts may be aggressive public education,as isbeing done in numerous areas,states,and parks(Beckmann 2002)Areas that contain black bearsshould pass laws,ordinances,and regulationsagainst intentional or unintentional feeding ofbears or other wildlife that may inadvertentlyattract bears.These areas should pass ordinancesrequiring private landowners and businesses toobtain and use bear-proof garbage containers.Forexample,the combination of an aggressive publicoutreach campaign through a bear working groupconsisting of state and federal agencies,empiricaldata,and county ordinances requiring bear-proofdumpsters has led to the first decline since 1994 inthe number of complaints concerning bears inNevada from 2002…2003.In 2002 Juneau,Alaskacreated several ordinances requiring bear-proofdumpsters that have resulted in fewer conflicts (D.Garcia,City of Juneau,personal communication).Additionally,an aggressive public education cam-paign and citation program in Yosemite NationalPark,California reduced the monetary damagecaused by bears by� 70% since 1998 (S.Matthews,Wildlife Conservation Society Yosemite Program,personal communication).In these instances,acombination of ordinances and public educationhas been successful in reducing the number of con-flicts between bears and humans.Future researchshould focus on the effectiveness of such publiceducation campaigns and bear-proof garbage con-tainers in reducing conflicts.Acknowledgments.We thank the University ofNevada,Nevada Agricultural Experiment Station,Nevada Division of Wildlife,and the United StatesNational Science Foundation for funding.We alsoare extremely grateful for the many hours of safeflights provided by M.Wiklanski,J.Kelly,and theentire staff at El Aero Services in Carson City,Nevada.We thank J.Nelson,who graciously volun-teered many hours of his time.We also thank M.Herzog,M.Beck,and A.Beckmann for helpful com-ments.M.Peacock,B.Longland,P.Brussard,and S. Deterrent techniques for bears € Beckmann et al. Livingston provided keen insights.We thank S.Sheaand G.Bracket for assistance in the field and fortheir expertise in handling hounds.Finally,wethank the 2 reviewers,J.Clark and K.Peirce,whohelped strengthen this manuscript.Research wasconducted under UNR Animal Care and UseProtocol #A99/00-02.LiteratureCitedANDERSON,D.R.,K.P.BW.L.T.2000.Nullhypothesis testing:problems,prevalence,and an alternative.Journal of Wildlife Management 64:912-923.,D.R.,K.P.BG.C.W.1994.AIC modelselection in overdispersed capturerecapture data.Ecology75:1780,J.P.2002.Changing dynamics of a population of blackbears (Ursus americanus):causes and consequences.Disser-tation,University of Nevada,Reno,USA.,J.P.,J.B.2003.Using black bears to test ideal-free distribution models experimentally.Journal of Mammalo-gy 84:594,J.P.,J.B.2003.Rapid ecological and behav-ioural changes in carnivores:the responses of black bearsUrsus americanus) to altered food.Journal of Zoology 261:,K.R.,D.R.A.1998.Model selection andinference:a practical information-theoretic approach.Springer-Verlag,New York,New York,USA.,R.J.1983.Micro encapsulated lithium chloride bait aver-sion did not stop coyote (Canis latrans) predation on sheep.Journal of Wildlife Management 47:1010,R.J.,D.E.ZP.J.SAVARIE.1996.Effectiveness oflarge livestock protection collars against depredating coyotes.Wildlife Society Bulletin 24:123,J.E.,F.T.VANM.R.PELTON.2002.Correlates of suc-cess for on-site releases of nuisance black bears in Great SmokyMountains National Park.Wildlife Society Bulletin 30:104,R.E.,J.,G.E.CONNOLLY,R.J.BR.T.S.1978.Coyotes,sheep and lithium chloride.Proceedings of the Verte-brate Pest Conference 8:190,C.M.,P.M.HC.M.P.1994.Evaluationof an aversive conditioning technique used on female grizzlybears in the Yellowstone Ecosystem.International Conferenceon Bear Research and Management 9:503,J.M.1990.Ecology,conservation,and management oftwo western Great Basin black bear populations.Thesis,Uni-versity of Nevada,Reno,USA.RAYSON,D.K.1993.The deserts past:a natural prehistory of theGreat Basin.Smithsonian Institution Press,Washington,D.C.,ERRERO,S.2002.Bear attacks;their causes and avoidance.Revisededition.Lyons Press,New York,New York,USA.URVICH,C.M.,.1989.Regression and time series modelselection in small samples.Biometrika 76:297,D.E.,R.C.RJ.R.JOWSEY.1983.Coyote predationon sheep,and control by aversive conditioning inSaskatchewan.Journal of Wildlife Management 36:16,J.P.,M.L.M.1997.Survival analysis:tech-niques for censored and truncated data.Springer,New York,New York,USA.,I.2002.Essays on the assumption of proportional hazardsin Cox regression.Dissertation,Uppsala University,Uppsala,Sweden..2001.SAS software:usage and reference.Version8.02.SAS Institute,Cary,North Carolina,USA.,D.1980.Chi-squared goodness-of-fit test for the pro-portional hazards regression model.Biometrika 67:145,H.H.,S.L.2000.Comparison of goodness of fit testsfor the Cox proportional hazards model.Communications instatistics:simulation and computation 29:187TONEBERG,R.P.,C.J.J.1966.Age determination of blackbears by cementum layers.Journal of Wildlife Management 30:,M.A.,D.L.G.1999.Taste-aversion condition-ing to reduce nuisance activity by black bears in a Minnesotamilitary reservation.Wildlife Society Bulletin 27:720Jon P.Beckmann(right) is a research ecologist for the WildlifeConservation Society in the greater Yellowstone ecosystem. Hereceived his B.S. in wildlife and fisheries biology and a second-ary major in natural resources and environmental sciences fromKansas State University in 1996. He received his Ph.D. in ecol-ogy, evolution, and conservation biology from the University ofNevada, Reno in 2002. He has been an active member of TWSimpacts of anthropogenic factors on behavior of mammals, par-ticularly carnivores. Carl W.Lackey(left) received his B.S. in nat-ural resources and wildlife management from the University ofNevada, Reno in 1990. He has been working for the NevadaDepartment of Wildlife since 1993 and has held his current posi-tion as wildlife biologist for the last 8 years. His responsibilitiesinclude black bears, mountain lions, furbearers, mule deer, andupland game. His primary interests are human…bear interactionsand population ecology. Joel Bergerscientist with the Wildlife Conservation Society. He completedhis Ph.D. at the University of Colorado, was a research associatefor the Smithsonian Institution for 7 years, and then was on thefaculty at University of Nevada, Reno for 16 years. He nowfocuses on applied issues in conservation, working pri-marily in the greater Yellowstone ecosystem.Whittaker Wildlife Society Bulletin 2004,32(4):1141…1146