Managing for Grizzly Bear Security Areas in Banff

Home , Banff National Park, Kananaskis Country

MANAGINGFOR GRIZZLYBEAR SECURITYAREAS IN BANFF NATIONALPARK AND THE CENTRALCANADIAN ROCKY MOUNTAINS

MICHAELL. GIBEAU,Resources and the EnvironmentProgram, University of Calgary,2500 UniversityDrive N.W., Calgary, AB T2N 1N4, Canada, email:[email protected] STEPHENHERRERO, Faculty of EnvironmentalDesign, Universityof Calgary,2500 UniversityDrive N.W., Calgary, AB T2N 1N4, Canada BRUCEN. McLELLAN,British Columbia Ministry of Forests, RPO Number3, P.O.Box 9158, Revelstoke,BC VOE3KO, Canada JOHNG. WOODS,Parks Canada, Box 350, Revelstoke,BC VOE2SO,Canada

Abstract: The need for security areas in which grizzly bears (Ursus arctos) may rarely encounterhumans and maintain wary behavior is not explicitly addressedby cumulativeeffect modeling (CEM). In addition,CEM does not assess the value to bearsof small areasleft between zones of humandisturbance. We developed a predictiveGIS based model of adult female grizzly bear securityareas in the CentralCanadian Rocky Moun- tains to provide agency plannerswith a tool that addressesthese shortfalls. Our study areaincluded 4 majorjurisdictions: Alberta provincial lands, BritishColumbia provincial lands, KananaskisCountry improvement district in Alberta,and National Park lands in both provinces. Startingwith the total land base in eachjurisdiction, we progressivelyremoved areas of unsuitablehabitat (e.g., rock and ice), habitatwithin 500 m of high humanuse (>100 humanvisits/month), and areas of insufficient size based on an averagedaily feeding radius(polygons <9 km2). We identifiedthe remaining lands as secure areas.We then tested the hypothesisthat female grizzly bearuse of securityareas differs from the landscapeas a whole based on radio telemetry data. Of the 4 jurisdictions in the CentralCanadian Rocky Mountains,the largest percent of secure habitatwas on British Columbia provinciallands. Of the land surface areaof the Banff, Yoho, and KootenayNational Parks,48% is unsuitablefor grizzly bears, primarilybecause it is composed of rock and ice. This is unfortunate,because it is assumedthat these nationalparks form productivecore refugiafor grizzly bears. By reconstructingpast humanuse and forecastinginto the futurefor Banff National Parkand KananaskisCountry, we demonstrateprogressive loss of security areas. We found that an averageof 69% of the land within grizzly bear home ranges was secure using our sample of 28 radiocollaredadult females. Resource selection indices from these bears demonstratedselection of security areas within their home ranges. Existing mortalityand translocationdata, combinedwith our findings of low securityand high habitatfragmentation within some adultfemale home ranges,give quantita- tive substance to the assertion that grizzly bears in and aroundBanff National Park and KananaskisCountry exist in one of the most human- dominatedlandscapes where they still survive. Access and developmentmanagement are key to grizzly bear persistencein the region.

Ursus 12:121-130

Key words: Alberta,Banff NationalPark, British Columbia, Canadian Rocky Mountains,foraging radius, fragmentation, grizzly bear,habitat, home range, security areas, Ursus arctos

As omnivores and apex predators,grizzly bears pos- ourNational Parks to supportbears has been significantly sess little resiliency (Weaveret al. 1996) and are one of reducedby widespreadhuman presence (Gibeau 1998). the first species to be lost from an areaas a result of land Humanaccess is one of the most influentialfactors af- development activities. The status of the grizzly bear fecting grizzly bearhabitat security (Interagency Grizzly populationand habitatare indicatorsof ecological integ- Bear Committee 1998). Althoughgrizzly bear mortality rity in Banff, Yoho, and KootenayNational Parks and the can be regulatedand influencedby changes in humanat- significantly largerregional ecosystem, the CentralCa- titudes, it seems unlikely that humanswill toleratemuch nadian Rocky Mountains,upon which grizzly bears de- contactwith animalslike grizzly bears:first, because they pend. The combination of a bear's biological traits, directly compete with humans for space and foods interactingwith people's proclivity to develop and use (Mattson 1990) and second, because interactions with grizzly bearhabitat, usually resultsin compromisedgriz- them can be potentiallyhazardous (Herrero 1985). Thus, bear zly populationsand habitat. By maintaininga healthy thereis a strongcase for preserving"security" areas where grizzly bear population,we suggest that most other ele- grizzly bears will be relativelyfree from encounterswith ments and of the processes terrestrialecosystem will also humans;that is, where bears can meet their energeticre- be maintained. quirements while at the same time choosing to avoid in the Typically past, aesthetically pleasing lands not people (Mattson1993). Such securityareas would foster suitablefor othereconomic uses were set aside as reserves the wary behavior in grizzly bears that most managers with little considerationfor whether they provided high consider desirable(Mattson 1993), given that habituated habitatfor sensitive quality species. By defaulthowever, bearshave a significantlyelevated mortality risk (Mattson our National Parks have become the core refugia even et al. 1992). though,as Caughley(1994) pointedout, species in trouble Since the development of a cumulative effects model end not in up living the habitatmost favorable to them, (CEM) for grizzly bears in the mid-1980s (Weaveret al. but in the habitat least favorableto the agent of decline. 1987, U.S. Departmentof Agriculture[USDA] ForestSer- Suchis the case with grizzlybears. As we open the twenty- vice 1990), there have been significant advances in un- first are century,managers recognizing that the ability of derstanding the management of grizzly bears and 122 Ursus 12:2001 substantialaccumulation of empiricaldata about their ecol- status for several years" (USDA Forest Service 1995). ogy (see Gibeauet al. 1996, Mattsonet al. 1996, Weaver Security areas help reduce the incidence of habituated et al. 1996, Mace andWaller 1997, andreferences therein). bears, bearskilled in self-defense, and bears removedby However, the CEM does not explicitly addressthe need managementagencies because of unacceptablebehavior. for habitatsecurity wherebybears maintainwary behav- In an erawhere humans significantly influence the land- ior, nor does it assess the value of small areas left be- scape, even within areasconsidered to be refugia,agency tween zones of human disturbance. To address this managersneed effective tools to assist in delineationand problem, Mattson (1993) developed the idea of micro- planningof importantand productivesites for bears. To scale securityareas where bears can foragefor 24-48 hours thatend, we developed a predictivemodel of securityar- safe from human disturbance. Micro-securityareas in- eas in the CentralCanadian Rocky Mountains,where adult clude a core zone (based on foraging radius) surrounded female grizzly bears will have a low human encounter by a disturbance-freebuffer zone (based on displacement ratebased on an averagedaily feeding radius. We further distances). tested the hypothesisthat female grizzly bearuse of secu- The InteragencyGrizzly Bear Committee (IGBC) in rity areas differed from use of the landscape as a whole the UnitedStates endorsed an approachfor providinghabi- based on radio telemetrydata. tat security for grizzly bears at the larger scale of bear managementunits (BMU; IGBC 1998). A BMU is a dis- crete areaof contiguoushabitat for managementpurposes STUDYAREA that would also incorporatethe year-roundneeds of an The CentralCanadian Rocky Mountains,with an area adult female grizzly bear. Based on researchby Mace of roughly41,000 km2,straddles the continentaldivide of and Waller(1997), interagencymanagers in the Northern Albertaand British Columbia. Topographicfeatures in- ContinentalDivide grizzly bear ecosystem in northwest clude rugged mountainslopes, steep-sided ravines, and Montana specified that >68% of a BMU be in "secure flat valley bottoms. The climate is continentalwith long,

Fig. 1. The Central Canadian Rocky Mountains divided into 4 government jurisdictions: (1) National Parks including, Banff, Yoho, and Kootenay; (2) British Columbia provincial lands, (3) Alberta provincial lands, and (4) Kananaskis Country improvement district. MANAGINGFOR GRIZZLY BEAR SECURITY * Gibeau et al. 123 cold winters and short, cool summers. The aspect and for analysis in MapInfo Professional?(MapInfo Corpo- elevationof the mountainoustopography modifies climate ration,Troy, New York,USA) and Idrisi?(Clark Univer- somewhat. Topography,soil, and local climate strongly sity, Worcester, Massachusetts,USA) geographic influence plantcommunities. The landscapecan be clas- informationsystems including: - sified into majorecoregions: montane (1,300-1,600 m), 1. Elevation A 1:50,000scale digitalelevation model subalpine(1,600-2,300 m), and alpine (>2,300 m). The compiledfrom NationalTopographic System series maps montane region is dominated by dry grasslands, wet was used to derive an elevation cut at 2,500 m. That el- shrubland,and forestsof lodgepole pine (Pinus contorta), evation was the upperlimit of useful grizzly bear forag- Douglas-fir(Pseudotsuga menziesii), white spruce(Picea ing through analysis of a grizzly bear habitat model glauca), and aspen (Populus tremuloides). Subalpinear- (Kansasand Riddell 1995) for Banff, Yoho,and Kootenay eas are forested with mature stands of lodgepole pine, NationalParks. More than99% of grizzly beartelemetry Engelmanspruce (Picea engelmanii),subalpine fir (Abies locations (n = 7,380) between 1994-1998 from the East- lasiocarpa), and subalpine larch (Larix lyallii), inter- ern Slopes Grizzly Bear Projectwere located below the spersedby areasof wetlandshrub. A mosaic of low shrubs 2500 m level. and herbs characterizealpine areas. 2. SatelliteImage - A LandsatThematic Mapper im- Managementof the landscapeis divided into 4 major age was used to derivea layerrepresenting vegetated area. governmentaljurisdictions: (1) NationalParks, including The unsuitableareas of snow and ice, rock and bare soil, Banff, Yoho, and Kootenay; (2) British Columbia pro- and water were reclassified out of the image leaving the vincial lands, (3) Alberta provincial lands, and (4) area actuallyavailable to bears. Alberta'sKananaskis Country (Fig. 1). Differingagency 3. HumanActivity - Humanactivity data from Cen- mandatesoversee preservation,tourism, forestry, oil and tral CanadianRocky Mountainsused in a habitateffec- gas extraction, mining, agriculture,and stock grazing. tiveness model (Gibeau 1998) formed the basis of this Native councils, towns and municipalities, commercial layer. The most recent data availablefrom all 4 jurisdic- developers,and residentialowners diversify land admin- tions were compiledto form a single map of humanactiv- istrationeven further,making this one of the most inten- ity across the region. These maps categorized vector, sively developed landscapes in the world where grizzly point, andpolygon dataof all motorizedroads, trails, and bears still survive. facilities into high and low use (USDA Forest Service 1990, Gibeau 1998) based on visitation records and ex- area >100 humanvisits/month METHODS pertopinion. Any receiving during the active bear foraging season (May-Oct) was Much of the basis for area security analysis relies on consideredto be high use. These data became the basis the radiusand defining averagedaily foraging subsequent for delineatingthe types and intensityof humanuses and area for an adult female bear. daily requirements grizzly their associateddisturbance buffers. Ourmapping could We used a subset of radio data from intensive telemetry not keep up with continued development, especially on of adult female bears tracking wary (Herrero 1985:51) Albertaprovincial lands, therefore,these maps only por- between 1994-1998 the Eastern Griz- gathered by Slopes tray a minimum amountof human activity on the land- Bear in the CentralCanadian Moun- zly Project Rocky scape. Historicalhuman use mapsfor KananaskisCountry tains to establisha mean movementdistance. These daily and Banff National Parkwere compiled from circa 1950 datawere calculatedfrom the lineardistance of >2 largest recordsand personal interviews (Banff-Bow Valley Study and do not reflect total distance traveled relocations/day 1996). Futurehuman activity projections for Kananaskis over the course of a which be much day, may greater.The Countrywere based on a build-outscenario from the ex- calculated radius all bears average daily foraging among isting recreationalpolicy and a 6% annualvisitor growth sampledis one half of the mean movementdistance. daily rate for Banff National Park (Banff-Bow Valley Study Therefore,the minimum area is daily requirement simply 1996). The majorassumption of the humanactivity maps the area of a circle (n r2)based on the ra- daily foraging is thatthey accuratelyreflect humanuse at an ecosystem dius. scale. Multi-annualhome range data for adultfemale grizzly bears based on radio-telemetrywere obtained from the Analysis Eastern Slopes Bear Project east of the ContinentalDi- The initial stage of GIS analysis was to remove areas vide = (n 19) and the concurrentWest Slopes Bear Re- from the landscapethat were unsuitablefor foraging for search Project(Woods et al. 1997) in a contiguous study bears. This was accomplishedby combiningthe vegetated area = west of the ContinentalDivide (n 9). cover map with the 2500 m elevation cut. All vegetated Severaltypes of geographicinformation were compiled lands below 2500 m were consideredavailable lands for 124 Ursus 12:2001 securityareas. test was selected because of the ability to partitionbe- Next, all high humanuse featuresfor the past, present, tween and among variables(Sokal and Rohlf 1981). The and future were buffered by a 500 m zone of influence assumptionsthat all expected values be >1 and <20% be following standardsdeveloped by IGBC (1998). We con- less than 5 was met (Zar 1984, Krebs 1989). Given the sider this a minimum zone of human influence in com- goodness-of-fit test rejects the null hypothesisthat use is parisonto researchresults from Yellowstone National Park equivalentto availability,the statisticaltest reviewed by (Mattsonet al. 1987, 1992; Reinhartand Mattson 1990; Neu et al. (1974) andByers et al. (1984) provideda method Green et al. 1997). Once the buffered high human use for furtherevaluation of resource selectivity (Alldredge data were prepared,these files were imposed on the map and Ratti 1986, 1992; White and Garrott 1990). of available/unsuitablelands. This removed all areaswithin Bonferroniconfidence intervals (Neu et al. 1974, McLean the zone of influence from furtherconsideration. Interim et al. 1998) led to conclusions aboutwhether security ar- maps depicting habitatpatches of all sizes were gener- eas were used more, in proportionto, or less than their atedonce the areaof availablelandscape outside the zone availability. of influence had been determined. We used Spearmancorrelation coefficients to investi- After removing unsuitablelands and lands within 500 gate the degree of associationbetween home range size, m of high humanuse, we eliminatedall remainingpoly- the proportionof the land base in securityareas, and the gons smaller than the minimum daily area requirement amountof land not secure due to high human influence for an adult female grizzly. Based on this analysis, re- (SAS InstituteIncorporated, Cary, North Carolina, USA). sults were tabulatedinto 4 categories:(1) unsuitableland- Results of all statisticaltests were consideredsignificant scape, (2) land base in security area, (3) land not secure at P< 0.01. due to high human use, (4) and land not secure due to small size. Historicaland future scenarios for Kananaskis Country and Banff National Park were preparedin the RESULTS same manner. We analyzed 227 daily movement episodes from 16 We used the programCALHOME (Kie et al. 1996) to wary adult female grizzly bears to producea mean daily = calculateminimum convex polygon (MCP) home ranges movement distanceof 3.4 km (range0.2-16.3. km, SE for adult female grizzly bears. We chose the 99% MCP 0.18 km). Therefore,the calculatedaverage daily forag- area to measurepotential female grizzly bear occupancy,ex- ing radiuswas 1.7 km with a 9.0 km2minimum daily cluding only extremeoutliers. The home rangepolygons requirement. were then convertedto GIS map layers. For each adult Available areas for grizzly bears were ones left after bare soil and female home range we prepareda tabulationof the un- those covered with rock, ice, water, and Thus 48% of the Na- suitable landscape, land base in security area, land not >2500 m elevation were removed. for for secure due to high humanuse, and land not secure due to tionalPark landscape was unsuitable foraging griz- small size. zly bears. This contrasts with only 12% unsuitable on 21% unsuitable in Alberta's Comparisonof the observedtelemetry data from weekly Alberta provincial lands, aircraftrelocations and the securityanalysis map provided KananaskisCountry, and 27% unsuitableon British Co- use and expected values to evaluateresource selection of lumbiaprovincial lands. areas and land not secure due to security areas. In the context of this study, availability We defined security in the CentralCa- was defined as the home range of an individual animal. humanuse or size for eachjurisdiction a of the avail- Both Johnson(1980) and White and Garrott(1990) point nadianRocky Mountainsby comparing map area of out that the home range of an animal alreadyrepresents able landscapewith a minimumdaily requirement on an adultfemale's radius some prior selection because of territorialityor food re- 9.0 km2based daily foraging with an intensive sources. Availability of security differed for each indi- (Fig. 2). Alberta'sKananaskis Country, had the of land base vidual bear depending on its home range configuration. recreationmandate, greatestpercent This form of third-orderselection (Johnson 1980) per- (40%) not secure due to humanuse. Albertaprovincial then British tains to the use made of security areas within the home lands followed closely behind with 33%, lands and National Parks range. Radio telemetry data from each collared indi- Columbia provincial (28%) female vidual were overlaid onto the security analysis map to (25%). The percentof landbase where adult griz- low of encounterswith determinethe proportionsof securityused and available zly bearshave a probability people amountof land avail- within the home range. (secure) dependson the productive extent of humaninfluence. The G-testfor goodness-of-fit (Zar1984) was employed able to a bear and the loss of areas is evident re- to test the hypothesisthat adult female grizzly bearsused A progressive security by human with the 1950s, securityareas in proportionto their availability. The G- constructingpast use, beginning MANAGING FOR GRIZZLY BEAR SECURITY * Gibeau et al. 125

oCalgary 51?

50'

~.s>fHtv ;is/( -l-< )\X( )RKS

Fig. 2. The Central Canadian Rocky Mountains categorized by security area classes, 1999. and forecasting future use for KananaskisCountry and Banff National Park(Fig. 3). The average size of secure habitatpatches in Banff National Park from 218 ranged 360 - - 130 km2in the 1950s, to 56 km2 currently,and finally 43 km2 330 - ---120 in the future. t^X4~\ tA :securttyarea Banff N.P.n Correspondingto the decreasein patchsize ' '^ (km2) - [ L", r7"/ / uAvagesiaof 300 _scurity - we found an increase in the numberof patches from 13 ^ - * arnaKanas Countryy (krnr) -110 E Numberof awcrty to 39 both and into in the future. areas Banff N.P. historically currently 270 - 100 v) _ Number o( secunty areas The same patternexisted throughoutKananaskis Coun- KananasisCountry < 240 - --80 < try. Fragmentationand insularizationof habitat within 3 210 -70 3 bothjurisdictions was evident, accompaniedby a loss in (D the abilityto foster the warybehavior in grizzly bearsthat 06 180 60 0 3 0 251 1 most managersconsider desirable. (? 150- -0 A o46465 _ _1 - -* - 50 E a) :3 Our model identified an average of 69% (range 32- z SE = of the availableland within a 120 -40 100%, 4.2%) individual CD home rangesof adultfemale grizzly bearsas secure(Table 90- 30

1). Bears with the highest degree of human influence 60-- 20 within theirhome rangestended to have the least security 30- 10 (bearsES24, ES26, ES47). Several bearshad lower than 0 0 average security despite average or lower than average Historical Current Future humaninfluence due to a high percentof land not secure because the remainingpotentially secure areas were too Fig. 3. Analysis of grizzly bear security areas for Banff small (bearsES18, ES41, ES61). National Park and Kananaskis Country for past (1950s), and IndividualG-tests for goodness of fit were significant present (1999), future (2020s). (P < 0.01) for 26 of 28 bears,rejecting the hypothesisthat 126 Ursus 12:2001

Table 1. Percent of the available land base in security policies, and differentgrizzly bearmanagement. A large classes for 28 adult female bears in the Central grizzly percent (48%) of the land surface of the combined na- Canadian Rocky Mountains, 1999. tionalparks (Banff, Kootenay, and Yoho) in our studyarea Not secure Not secure is unsuitablefor grizzlybears, primarily because it is com- Home range due to due to posed of rock and ice. This is unfortunate,because it has humanuse small size Bear ID size (km2) Secure been assumed that national parks form productivecore ES17 111 55 39 7 refugia for difficult to maintain species such as grizzly ES18 236 38 42 20 bears. In the Alberta lands of the CentralCa- ES24 279 32 57 11 provincial ES26 427 46 46 8 nadianRocky Mountains,the percentof unsuitableland ES28 307 72 21 7 was much less (12%), indicating more productiveland, ES30 268 59 38 3 but conflicts with human uses are to ES31 79 50 41 10 multiple predicted ES32 328 53 37 10 increase. BritishColumbia provincial lands also have less ES33 218 67 29 4 unsuitableland (27%) thanNational Parks, but theretoo, ES36 617 60 28 12 humanuses are to rise. ES37 884 53 40 7 multiple expected ES40 206 84 12 4 The concept of secure habitat,an area where an adult ES41 189 67 17 16 female grizzlybear can meet herdaily foragingneeds with 4 ES46 321 67 29 a low of disturbanceby people, was centralto ES47 245 34 60 7 probability ES55 180 52 36 12 ourresearch. Mattson et al. (1992) foundthat grizzly bears ES57 282 73 27 0 with access to secure habitatmaintained desirable wary 34 16 ES61 232 50 behavior, had low probabilitiesof becoming habituated ES62 150 63 34 3 WS35 49 81 9 10 or food-conditioned,and had significantlyless mortality WS42 131 81 6 13 than did non-waryadult females. Resource selection in- WS44 220 100 0 0 dices from our of radiocollaredindividuals sub- 100 0 0 sample WS71 78 bearsselected WS74 35 100 0 0 stantiatedthat adult female grizzly security WS282 78 99 0 1 areas,or at least avoided areasthat were not secure. WS284 59 99 0 1 Ourresults characterized habitat security at the level of 0 1 WS289 109 99 bears' home and on WS295 37 100 0 0 an adultfemale grizzly range average at the landscapelevel. For 28 adultfemale bearsthrough- of 69% of the home was adult female grizzly bears use security areas in propor- out the region, an average range Female bears within Banff National Park,how- tion to their availability. Furtherevaluation of resource secure. 60% withintheir home selection using Bonferroniconfidence intervals indicated ever,averaged only security ranges. Park that 18 of 26 bearsused secure areasmore thanexpected. Although female bears in Banff National currently habitat than Park land use is Six bears used secure areas in proportionto availablity, have less secure average, more thanland use of the surround- while only 2 bears used secure areas less than expected. probably predictable One of the 2 bears that used secure areas less than ex- ing areas (ParksCanada 1997). The of secure habitatwithin a jurisdic- pected was habituatedto human presence (M. Gibeau, largestpercent tion was on British Columbia provincial lands (68%). personalobservation). Alberta's Kananaskis (52% secure habitat)and The proportionof land in securityareas was negatively Country Alberta lands did not meet the current correlatedwith both home range size (rs = -0.559, P < provincial (63%) level of 68% consideredto be securityset 0.01) and the amountof land not secure due to high hu- target adequate the USDA Forest Service in the NorthernCon- man influence (rs = -0.556, P < 0.01). This association by (1995) tinentalDivide bear in northwestMon- between security and home range size demonstratesthat grizzly ecosystem tana. Kananaskis dedicated primarily to smallerhome ranges tend to be more secure. Country, recreationand resource development, also had the largest percentof land in the zone of humaninfluence. This area DISCUSSION is also dedicatedto maintainingall wildlife populations, thus these conflict, and this conflict could in- Our results quantitativelycharacterized grizzly bear objectives crease. the combined National Parks (68%) and habitat at the landscape level for the CentralCanadian Only BritishColumbia lands (68%)met the USDA's Rocky Mountains.This areasupports a grizzlybear popu- provincial level. A more refined of Banff National lation with, we assume, a significant degree of genetic target analysis however, revealedthat only 12 of 27 bear manage- exchange within it, and isolation from populationsnorth Park, ment units exceed levels (Eastern Slopes and south. Four major land use jurisdictions are in the (48%) target area. Each has a uniqueland base, differentmanagement Project,unpublished data). MANAGINGFOR GRIZZLY BEAR SECURITY * Gibeau et al. 127

These results give quantitativesubstance to the asser- humanencounter rate for adult female grizzly bears at a tion that grizzly bears in and aroundBanff National Park much more refined scale than the habitat effectiveness and KananaskisCountry exist in one of the most human- model. Securityareas help reducethe numberof habitu- dominatedlandscapes where they still survive. This was atedbears, bears killed out of self-defense,and bears killed also demonstratedin the habitateffectiveness model ap- by managementagencies because of unacceptablebehav- plied to Banff, Yoho, and Kootenay National Parks ior. (Gibeau 1998). The effect of this development has been Managementagencies in southernCanada now realize high grizzly bear mortality,even in Banff National Park, that grizzly bears requirespecial considerationto main- especially in the female cohortduring the 1970s andearly tainhealthy populations. Many management options have 1980s (Benn 1998). Because Banff National Park had been evaluatedin the United States and need only to be the lowest percentof its lands not secure due to high hu- applied and refined in southernCanada. Much of this man use between 1971-96, 95 out of 107 mortalitiesoc- experience can be summarized,arguably, into manage- curred in the zone of human influence. Outside of the ment of humanaccess (IGBC 1998). NationalParks, Benn (1998) found that 89% of 172 mor- The need to control human access has been acknowl- talities in Alberta and 71% of 303 in British Columbia edged in Banff National Park, althoughthe mechanisms were within the zone of human influence. This further for doing so are not fully apparent. Managementstrate- supportsthe function of secure habitatin bufferinggriz- gies such as those employed in Denali National Park zly bears from mortality. (Schirokauerand Boyd 1998) and Yellowstone National Analysis of securityareas (for only Banff NationalPark Park(Gunther 1994, 1998) have provensuccessful. Strict and KananaskisCountry) clearly demonstratedthe de- control of humanaccess has been achieved throughsea- creasing size over time of relatively undisturbedhabitat sonal area and/or trail closures, day use and party size units. This habitatfragmentation has occurredthrough- restrictions,and limiting travel to mid-day only. For ex- out Banff National Park and KananaskisCountry, but is ample,in Yellowstone'sPelican Valley bear management dramaticin the Bow River Valley. The decreasingsize of area, the "areais closed April 1 throughJuly 3. From security areas was paralleledby a significantdecrease in July 4 throughNovember 10, the area is open to day-use total amountof security area available throughoutBanff only between the hours of 9 a.m. and 7 p.m." (Gunther National Parkand KananaskisCountry. 1998:3). The explicit acknowledgmentto manage areas We can surmise,but cannot conclusively demonstrate, for grizzly bearsecurity has been key to managementsuc- thatcompromised habitat security and habitatfragmenta- cess (NationalParks Service 1982). tion are decreasinggrizzly bear populationsize and pos- To facilitateuse by bears, these securityareas of no or sibly long-termpopulation viability. Doak (1995) reached low humanuse shouldremain in place for at least 10 years a similarconclusion regardingthe linkage between habi- (IGBC 1998). In our opinion, however, given the long tat degradationand potential population depression in the life span of grizzly bears,this durationmust be measured Yellowstone ecosystem. We cannot demonstratedemo- in generationsand not merely in calendaryears. and graphic viability effects because baseline data from The NationalParks alone can not sustaina regionalgriz- the do not exist. past However, existing mortalityand zly bearpopulation. Some of the best chances for grizzly translocationdata (Benn 1998) combined with our find- bear persistence come from outside National Parks of low ings securityand high habitatfragmentation within (McLellanet al. 1999), and hence a cooperativeand co- adult female home ranges, present converging evidence ordinatedmanagement approach is critical. The nucleus for bear grizzly populationdecline across the region. of an interagencygrizzly bear planning committee has been formedin the CanadianRockies, althoughits exist- ence seems tenuouswithout continued commitment from MANAGEMENTIMPLICATIONS all jurisdictions. Thereis a clear need for complementary In the habitateffectiveness was past, modeling the pri- managementguidelines that provide security for grizzly marytool used to measurethe impactof humanactivities bears as they cross jurisdictionalboundaries. on bears ForestService Gibeau (USDA 1990, 1998). The Although the concept of providing security areas ap- model with a broadscale providedmanagers measureof pears logical and relatively straightforward,perhaps the the and of a bear's quality quantity foragingopportunities most challenging steps ahead will be implementingnew when human activities were considered. The model fell managementprescriptions to achieve securityfor grizzly short, however, in the human encounter estimating rate bears. As humans, we have never been particularlyac- and risk that is as mortality equally important foraging cepting of land-use policy that restrictsindividual rights for opportunities populationpersistence. Security area and privileges. In the end, grizzly bears may prove to be analysis now provides managerswith a measure of the the ultimatechallenge in whetherhumans can coexist with 128 Ursus 12:2001 nature. reportto the Banff Bow Valley Task Force by the Eastern ACKNOWLEDGMENTS SlopesGrizzly Bear Project. University of Calgary,Alberta, Canada. In the EasternSlopes Projectwe thankM. Dupuis, M. D.J. ANDJ.M. PEEK.1997. Jalkotzy,R. Leblanc,C. Mamo,J. Paczkowski,I. Ross, J. GREEN,G.I., MArtSON, Springfeeding on ungulate carcasses by grizzly bears in Yellowstone Saher,T. Shury, S. Stevens, andM. Urquhart,who made National Park. Journalof Wildlife Management61:1040- majorcontributions to the captureand monitoring of griz- 1055. bears. We contributionsmade in the zly acknowledge GUNTHER,K.A. 1994. Bearmanagement in YellowstoneNational West Slopes Projectby J. Flaa, J. Hooge, F. Hovey, D. Park, 1960-93. InternationalConference on Bear Research Lewis, D. Mair, D. McTighe, R. Smith, and M. Super. and Management9:549-560. We also thankthe BritishColumbia Conservation Officer . 1998. Bear managementarea program,Yellowstone Service, Alberta EnvironmentalProtection, and Parks National Park. Information paper No. BMO-5, Bear Canadafor both technical and logistic support. We also ManagementOffice, YellowstoneNational Park. contributions members of the Eastern HERRERO,S. 1985. Bear attacks:Their causes and avoidance. acknowledge by and New New USA. Bear Colum- Lyons Burford, York, York, Slopes Grizzly ProjectSteering Committee, INTERAGENCYGRIZZLY BEAR COMMITTEE. 1998. bear/ bia Basin Fish and Wildlife Grizzly Compensation Program; motorized access management taskforce report. U.S. Friendsof MountRevelstoke and Glacier National Parks; Department of the Interior, Fish and Wildlife Service, the Grizzly Bear ConservationStrategy (BC); Ministry Missoula, Montana,USA. of Environment,Lands and Parks(BC); Ministryof For- JOHNSON,D.H. 1980. The comparisonof usage and availability ests (BC), Forest Renewal (BC), and the SouthernB.C. measurementsfor evaluationof resourcepreference. Ecology Guides and Guide OutfittersAssociation. Special thanks 61:65-71. ANDR.N. RIDDELL.1995. bearhabitat model to D. Mattsonfor plantingthe original seeds of this con- KANSAS,J.L., Grizzly for the four contiguous mountain national parks: Second cept and J. Weaverfor nurturingthose ideas. S. Jevons iteration. for Canadian Parks Service, and J. Wierzchowski GIS and techni- Report Calgary, provided computer Alberta,Canada. constructivecomments were cal support.Thoughtful and KIE,J.G., J.A. BALDWIN,AND C.J. EVANS.1996. Calhome: a providedby C. McLaughlinand 2 anonymousreviewers. programfor estimatinganimal home ranges.Wildlife Society Bulletin 24:332-344. KREBS,C.J. 1989. Ecological methodology. Harperand Row, LITERATURECITED New York,New York,USA. bear ALLDREDGE,J.R., ANDJ.T. RAT. 1986. Comparisonof some MACE,R.D., ANDJ.S. WALLER.1997. Final report:grizzly statistical techniques for analysis of resource selection. ecology in the Swan Mountains.Montana Fish, Wildlife,and Journalof WildlifeManagement 50:157-165. ParksDepartment, Helena, Montana,USA. , AND . 1992. Further comparison of some MATTSON,D.J. 1990. Human impacts on bear habitat use. statisticaltechniques for analysisof resourceselection. InternationalConference on BearResearch and Management Journalof WildlifeManagement 56:1-9. 8:33-56. BANFF-BOWVALLEY STUDY. 1996. Banff-Bow Valley: At the . 1993. Background and proposed standards for in the Yellowstone crossroads.Technical report of the Banff-Bow Valley Task managing grizzly bear habitat security ParkStudies Unit Force. Preparedfor Ministerof CanadianHeritage, Ottawa, ecosystem.Cooperative Report.University Ontario,Canada. of Idaho, Moscow, Idaho, USA. ANDR.R. KNIGHT.1992. Yellowstone BENN,B. 1998. Grizzly bear mortalityin the CentralRockies , B.M. BLANCHARD, human andwhitebark Ecosystem, Canada. Masters Degree Project. Faculty of grizzlybear mortality, habituation, pine Wildlife 56:432-442. Environmental Design, University of Calgary, Alberta, seed crops. Journalof Management Canada. , S. HERRERO,R.G. WRIGHT,AND C.M. PEASE. 1996. of Mountain bears. BYERS, C.R., R.K. STEINHORST, AND P.R. KRAUSMAN. 1984. Science and management Rocky grizzly 10:1013-1025. Clarification of a technique for analysis of utilization- ConservationBiology R.R. ANDB.M. BLANCHARD.1987. The effects availabilitydata. Journal of Wildlife Management48:1050- , KNIGHT, 1053. of developmentsand primaryroads on grizzly bears habitat National International CAUGHLEY,G. 1994. Directionsin conservationbiology. Journal use in Yellowstone Park,Wyoming. on Bear Researchand 7:259-273. of Animal Ecology 63:215-244. Conference Management R.M. W.L. 1998. DOAK,D.F. 1995. Source-sinkmodels andthe problemof habitat MCLEAN,S.A., M.A. RUMBLE, KING,AND BAKER. Evaluation of resource selection methods with different degradation: General models and applications to the of Journalof Wildlife Yellowstone grizzly. ConservationBiology 9:1370-1379. definitions availability. Management 62:793-801. GIBEAU, M.L. 1998. Grizzly bear habitat effectiveness model J.G. for Banff, Yoho, and Kootenay National Parks, Canada. MCLELLAN,B.N., F.W. HOVEY,R.D. MACE, WOODS,D.W. Ursus 10:235-241. CARNEY,M.L. GIBEAU,W.L. WAKKINEN, AND W.F. KASWORM. andcauses of bear in the interior , S. HERRERO,J.L. KANSAS, ANDB. BENN. 1996. Grizzly 1999. Rates grizzly mortality of British Montana, bearpopulation and habitatstatus in Banff NationalPark: A mountains Columbia, Alberta, MANAGINGFOR GRIZZLY BEAR SECURITY* Gibeau et al. 129

Washington and Idaho. Journal of Wildlife Management. Department of Agriculture, Forest Service, Missoula, 63:911-920. Montana,USA. NATIONALPARKS SERVICE. 1982. Final environmental impact 1995. FlatheadNational Forest plan amendment#19: statement, grizzly bear management program. U.S. Allowable sale quantity and objectives and standardsfor Department of the Interior, National Park Service, grizzly bear habitat management. U.S. Department of YellowstoneNational Park,Wyoming, USA. Agriculture,Forest Service, Kalispell, Montana,USA. NEU, C.W., C.R. BYERS,J.M. PEEK,AND V. BOY. 1974. A WEAVER,J.L., R.E.F. ESCANO, AND D.S. WINN.1987. A framework techniquefor analysisof utilization-availabilitydata. Journal for assessing cumulative effects on grizzly bears. of Wildlife Management38:541-545. Transactionsof the North American Wildlife and Natural PARKSCANADA. 1997. Banff National Park managementplan. Resources Conference52:364-376. Ministryof CanadianHeritage, Ottawa, Ontario, Canada. , P.C. PAQUET,AND L.F. RUGGERIO.1996. Resilience and REINHART,D.P., ANDD.J. MArrSON.1990. Bear use of cutthroat conservationof large carnivoresin the Rocky Mountains. trout spawning streams in Yellowstone National Park. ConservationBiology 10:1013-1025. InternationalConference on BearResearch and Management WHITE,G.C., ANDR.A. GARROTT. 1990. Analysis of wildlife 8:343-350. radio-trackingdata. Academic Press, San Diego, California, SCHIROKAUER,D.W., ANDH.M. BOYD.1998. Bear-humanconflict USA. managementin Denali National Park and Preserve, 1982- WOODS,J.G., B.N. MCLELLAN,D. PAETKAU, M. PROCTOR,C. 94. Ursus 10:395-403. STROBECK.1997. West Slopes BearResearch Project, Second SOKAL,R.R., ANDF.J. ROHLF. 1981. Biometry:the principleand ProgressReport. Parks Canada, Box 350, Revelstoke,British practice of statistics in biological research.Second edition. Columbia,Canada. W.H. Freeman,San Francisco,California, USA. ZAR,J.H. 1984. Biostatisticalanalysis. Second edition.Prentice- U.S. DEPARTMENTOFAGRICULTURE FOREST SERVICE. 1990. CEM- Hall, Englewood Cliffs, New Jersey,USA. A model for assessing effects on grizzly bears. U.S.