Habitat-Capability Model for Brown Bear in Southeastern Northeast Portion of Chichagof Island Were Illegal

HABITAT-CAPABILITYMODEL FOR BROWN BEAR IN SOUTHEASTALASKA

JOHNW. SCHOEN,Alaska Departmentof Fish and Game, 333 RaspberryRoad, Anchorage, AK 99518 RODNEYW. FLYNN,Alaska Departmentof Fish and Game, P.O. Box 20, Douglas, AK 99824 LOWELLH. SURING,1Alaska Region, U.S. Departmentof AgricultureForest Service, P.O. Box 21628, Juneau, AK 99802 KIMBERLYTITUS, Alaska Departmentof Fish and Game, P.O. Box 20, Douglas, AK 99824 LAVERNR. BEIER,Alaska Departmentof Fish and Game, P.O. Box 20, Douglas, AK 99824

Abstract: Habitat-capabilitymodels are necessary for evaluating the effects of forest management on the management of indicator species (including brown bears [Ursus arctos]) of the Tongass National Forest. Habitat-usedata from 95 radio-collaredbrown bears on Admiralty and Chichagof Islands were used to develop this habitat-capabilitymodel. Each of 20 habitatswas assigned a habitat-capabilityvalue based on bear habitatpreference or best professionaljudgment. The effects of human activity and resourcedevelopment on brown bears were estimated, based on best professional judgment, as reductions in habitatcapability within zones of human influence.

Int. Conf. Bear Res. and Manage.9(1):327-337

Once widely distributedacross western North the last 2 decades(see review in Zager and Jonkel America,brown/grizzly bears (Ursus arctos horribilis) 1983, Contrerasand Evans 1986), and several currentlyrange over a significantlyreduced portion of investigationshave recently been completed in the the continent. In 1975, they were declaredthreatened coastalforests of BritishColumbia and Alaska (e.g., in the UnitedStates south of Canada. Loss of habitat Hamilton1987, Schoenand Beier 1990). The brown to humanencroachment is a seriousproblem for bear bear has been selected as a managementindicator managementin the contiguous48 statesand elsewhere species(MIS) in therevised Tongass Land Management (Mattson1990, McLellan1990, Schoen1990, Servheen Plan(U.S. Dep. of Agric. For. Serv. 1991). Habitat- 1990). Throughoutthe world,the futureof manybear capabilitymodels are needed for each MIS on the populations,including brown bears, is inextricably TongassForest. Thesemodels will be used for project- linkedwith forestmanagement (Schoen 1991). level planning and are necessary for providing In North Americatoday, the largestpopulation of informationto evaluatethe cumulativeeffects of forest brownbears occurs in Alaska(Peek et al. 1987) with managementon wildlife habitats and populations. an estimated30,000-40,000 bears (Alaska Department Cumulativeeffects analysis is a relativelynew but of Fish andGame 1978). Brownbears are indigenous importantcomponent of forest planning(Christensen to SoutheastAlaska where they occur throughoutthe 1986,Weaver et al. 1986)and provides an approachfor mainlandcoast and on the islandsnorth of Frederick predictingthe long-termeffects of land-management Sound. Admiralty,Baranof, and ChichagofIslands activitieson brownbear habitat and populations. The have some of the highestbrown bear densities(e.g., modelto be describedevaluates quality of habitatfor 2.6 bears/km2on northernAdmiralty Island) in the brownbears, which is assumedto be relatedto long- world(Schoen and Beier 1990). term carryingcapacity. Habitatsare rated, using Brownbears are one of the specialfeatures of the habitat-preferencedata from Schoenand Beier(1990), Tongass NationalForest. Admiralty,Baranof, and on the basisof theirvalue to bearsduring late summer ChichagofIslands are one of the mostimportant brown when hyperphagicbears are most concentratedand bearhunting regions in Alaska. Tourismand outdoor vulnerableto humanactivities. recreationare also growing industriesin this area. Thisproject was fundedby the AlaskaDepartment of Visitorsto SoutheastAlaska as well as manyresidents Fish and Game and Alaska FederalAid in Wildlife are interestedin an opportunityto observethe brown RestorationProject W-22-7. Additionalsupport for the bear, a symbolof the Americanwilderness. projectwas providedby the Forest Service and the The declinein the rangeand numbers of brownbears GreensCreek Mine. D. Anderson,S. Herrero,B. duringthe past centuryin the contiguous48 stateshas McLellan, and 2 anonymous referees provided heightenedmanagement concern for the species and constructivecriticism and editorial review of the prompted an increase in brown bear research, manuscript, and L. McManus provided technical particularlyhabitat-related studies. Most researchon editing. bear-forestryrelationships has been conductedwithin

1 Present Address: Chugach National Forest, 3301 C St., #300, Anchorage, AK 99503. 328 Int. Conf.Bear Res. and Manage.9(1) 1994

HABITATRELATIONSHIPS avalancheslopes. During early summer (mid-Jun Odum(1971:234) described habitat as theorganism's throughmid-Jul), most bearsmove to forestedslopes "address"or the place it inhabitsin fulfillingits life and alpine/subalpinemeadows where they forage on needs (e.g., food, cover, water). Harrisand Kangas newlyemergent vegetation. (1988) proposedthat the definitionof primaryhabitat Bearsconcentrate along low-elevation coastal salmon explicitlyextends beyond the individualto includean streams from mid-July through early September. area of sufficientsize or configurationto supporta During this late summerseason, 54% of all bear populationover time. We considerthat an effective relocationsoccurred in riparianforest habitat vegetated definitionof bear habitatmust also incorporatethe by a spruce-devils club (Picea sitchensis-Oplopanax influenceof humanactivities (Schoen 1990). horridus)community (Schoen and Beier 1990). During Thehabitat relationships of brown/grizzlybears vary this sameperiod, 66% of all bearrelocations occurred considerablyacross the diverse array of ecosystems withina 160-mband on eitherside of anadromousfish they inhabitfrom the easternRockies, through coastal streams(Schoen and Beier 1990). Althoughthis zone rainforests, and to the Arctic. The AlaskaDepartment includeda varietyof habitats,it was dominatedby the of Fish and Gamebegan brown bear investigations in riparianspruce-devils club community. Bears used this SoutheastAlaska in 1981 with particularemphasis on habitatfor fishingalong river banks, for foragingon habitatrelationships and the influenceof logging and succulentvegetation and berries,and for securityand mining activitieson bear populations. From 1981 thermalcover. through1988, 68 brownbears were radiocollaredon Althoughmost bears (>85%) are associatedwith northern Admiralty Island and 3,020 relocations anadromousfish streamsin late summer,some bears collected(Schoen and Beier 1990). Habitatuse by (primarilyfemales) do not use coastal fish streams radio-collaredbrown bears varied seasonally(P < (Schoenet al. 1986). Those bears (termed"interior" 0.01) (Table1) andis considereda responseto seasonal bears) remain in interior regions of the island differencesin food availabilityand quality. throughoutthe year, foragingprimarily on vegetation Most brown bears emerge from high-elevation and berriesin subalpineand avalancheslope habitat. (>300 m) dens betweenApril and May. After den By mid-September,most bears move to upperelevation emergence,many bears move to low-elevationold- (>300 m) forests, avalancheslopes, and subalpine growthforests, coastal sedge meadows, or south-facing meadowswhere they feed on currant(Ribes spp.) and devilsclub berries before denning. Table 1. Seasonal habitat use by 68 radio-collared brown Winterdenning begins in Octoberand November. bearsa on Admiralty Island, Southeast Alaska, 1982-88.b Mean elevationand slope of 121 den sites of radio- collaredbears from Admiraltyand ChichagofIslands Percentageof habitatuse were 640 m and 35? (Schoen et al. 1987). Fifty-two Summer percentof those dens occurredin old-growthforest on Habitattype Spring Early Late Fall Annual habitat. Althoughcave denning was common AdmiraltyIsland, many dens were excavatedunder Old-growth forest large-diameterold-growth trees or into the bases of Upland forest 55.9 28.2 24.5 30.6 28.4 largesnags (Schoen et al. 1987). Riparianforest 8.7 11.0 53.6 18.8 33.3 The seasonalfood habitsof Admiraltybrown bears were described Beach fringe 6.8 4.9 2.0 1.5 3.1 by McCarthy(1989). Duringspring, the diet of brownbears is dominatedby sedges(Carex forest 3.7 14.0 5.2 10.3 8.4 Subalpine spp.), other green vegetation, roots, and deer Nonforest (Odocoileus sitkensis). Sedges and salmon Avalanche slopes 12.4 15.7 5.5 23.2 11.3 (Oncorhynchusspp.) are the major foods consumed skunk 3.7 18.9 2.8 7.6 8.4 duringsummer, although cabbage(Lysichiton Alpine americanum),devils club berries, and other plants, Estuary 3.8 4.5 5.3 0.6 4.3 berries,and rootsare also used. Duringfall, salmon, Other 5.0 2.8 1.1 7.4 2.8 devilsclub berries, skunk cabbage, sedge, beach lovage dominatethe n relocations 161 772 1,285 340 2,558 roots(Ligusticum spp.), andcurrants diet. The distributionof bears correspondedclosely to the a Interiorbears were not included. seasonalabundance and quality of the food itemslisted b Schoen and Beier (1990). above. Because bears have relatively inefficient ALASKABROwNBROWN BEAR HABITAT MODEL *? Schoen et al. 329

carnivore digestive systems (Bunnell and Hamilton influence. Nine majorhabitat categories were identified 1983) and are active for only part of the year, they for use in this model: old-growth forest, beach-fringe must exploit the most productivefeeding sites available. old growth, subalpine forest, second-growth forest, This often brings bears into conflict with humansusing clearcuts, avalanche slopes, alpine, estuary, and other those same high-qualitylands (Schoen 1990). (Table 2). Forest habitats were further subdivided In Southeast Alaska, old-growth forest is used relative to upland or riparian status and presence or extensively throughout the year by brown bears for absence of anadromoussalmon. foraging, cover, and denning. Elsewhere, clearcut This model assumes that habitat quality is related to logging often results in the productionof an abundance brown bear preference for different habitats (e.g., of forage plants potentially valuable to bears during alpine, riparianold growth, clearcuts, second growth). early stages of forest succession (Lindzey and Meslow While recognizing potential problems associated with 1977, Mealey et al. 1977, Zager et al. 1983). population dynamics and interpretation of habitat Theoretically, these sites should provide good or availability(Johnson 1980, Van Home 1983, McLellan adequatehabitat for a generalist species like the brown 1986), we have used habitat preference of radio- bear. However, on Chichagof Island, clearcuts were collared bears on Admiralty Island as our measure of used minimally by bears; only 2.8% of 854 relocations of 27 radio-collaredbears occurred in clearcuts during Table 2. Descriptionof habitat used in the habitat- 1983 categories through 1986 (Schoen and Beier 1990). Although capabilitymodel for coastal brown bears, Southeast Alaska. clearcuts only encompassed about 6% of the entire Chichagof study area, they made up a much larger Habitat Description proportion of low-elevation to streams valleys adjacent Physiographiccategories -the areas used most extensively (> 60%) by bears in Beach Within150 late summer (Schoen and Beier 1990). Within the fringe m of meanhigh water individual home ranges of 14 radio-collared brown Estuaryfringe Within300 m of meanhigh wateralong bears on Chichagof Island, 8 bears were never located an estuary in clearcuts and 5 bears used clearcuts less and 1 bear Riparianzone Zone within160 m of a stream, used clearcuts more than their abundancewithin bear influencedby riparianhabitat home late summer from ranges during 1983 through Upland Areabetween the beachand estuary 1986 (Schoen and Beier 1990). fringesand the subalpine,excluding the We believe brown bears make limited use of riparianhabitat clearcuts in SoutheastAlaska because other sites (e.g., Forestcategories alpine/subalpinehabitat, wetlands, riparianold growth, Oldgrowth Foreststands greater than 300 yearsold avalancheslopes) provided more nutritiousforaging and better cover habitat than clearcuts (Schoen and Beier Subalpine Ecologicalsubalpine zone 1990). For example, devils club berries, currants,and Clearcut Stands0-25 yearsold salmonberries(Rubus which are spectabilis), foragedon Youngsecond Stands26-150 yearsold most extensively by bears (McCarthy 1989), are more growth abundantin riparianand avalancheslope habitatthan in Oldersecond growth Stands151-300 years old clearcuts. Because younger second-growth conifer Nonforestcategories stands (25-150 years old) in Alaska produce minimal understory vegetation, second growth provides poor Avalancheslopes Recurrentslide zone foraging habitat for herbivores and omnivores such as Alpine Ecologicalalpine community bears (Wallmo and Schoen 1980, Alaback 1982, Schoen and Beier Estuary Portionof an estuarybelow meanhigh 1990). water

Other Miscellaneous(e.g., muskeg,rock, HABITAT-CAPABILITYMODEL roads) This model was designed to evaluate bear habitaton Streamcategories a single- or multiple-watershedscale in 2 stages. First Fish Anadromousfish present at the habitat level, the model calculates a habitat- No fish No anadromousfish present capability index (HCI). Second, the model calculates reductions in habitat capability within zones of human 330 Int. Conf.Bear Res. and Manage.9(1) 1994 habitatcapability for brown bears in SoutheastAlaska. Table 3. Habitat capability for brown bear habitats during the late summer season in Southeast Alaska. The ecological basis for inferring habitat quality from preference data is found in habitat-selection theory Usea Availabilityb Preferencec (Rosenzweig 1981, Fagen 1988). As stated by Habitat (%) (%) index HCId Densitye Ruggiero et al. (1988), "Habitatpreferences are based Upland forest on evolved behavior and thus relate directly to the probability of persistence. Therefore, habitat Old growth 24.5 55 0.31 0.34 0.32 preferences must be viewed as reliable information Subalpine 5.2 10 0.34 0.37 0.35 about the environments needed for population Old 2nd growth 0.10f 0.10 persistence, and should be considered a valid basis for ------0.00 0.00 decisions." Young 2nd management growth Clearcut -- -- - 0.10g 0.10 Habitat Capability The habitat-level model has M distinct habitattypes Riparianforest = Each of the that occupy an area Ai(i 1, 2,...M). Old growth 53.6 5 0.91 1.00 0.95 habitat was a categories assigned habitat-capability Fish ------1.00 0.95 index (HCI) based on habitat preference or best ------professionaljudgment (Table 3). Ivlev's (1961) index No fish 0.40 0.38 of electivity was used as the measure of habitat Old 2nd growth preference by brown bears for the habitat-capability Fish 0.30f 0.29 model. To transformIvlev's indices (which range from No fish ------0.10of 0.10 -1 to +1) to positive numbers, we calculated (Ei) as follows: Ei = r,/(ri + pi); where Ei = the transformed Young 2nd growth index of electivity or habitat-preferenceindex, ri = the Fish 0.20f 0.19 of observed use of habitat i proportion category No fish 0.00f 0.00 (relocations of radio-collared bears), and pi = the Clearcut proportion of habitat category i in the study area (availability). Habitat-capabilityindices (HCIs) were Fish ------0.50f 0.48 computed by dividing the preference index for each No fish ------0.20f 0.19 habitat category by the maximum value for the index Beach-fringe forest 2.0 3 0.40 0.44 0.42 (HCIi = Ei/Emax), so the highest value habitat has an HCI value of 1.0. Estuary-fringeforest ------0.60f 0.57 Availabilityof habitatswithin the 365-km2Admiralty Avalanche slope 5.5 5 0.52 0.57 0.54 area was estimated extrapolation from a study by Alpine 2.8 10 0.22 0.24 0.23 habitat-database derived for a 300-km2subsection of 5.3 2 0.73 0.79 0.75 this study area. The original availabilitydata (collected Estuary for a deer study) were determined from a random Other 1.1 10 0.10 0.11 0.10 sample of 2,495 points systematically overlaid on a Habitatuse radio-collaredbrown bears on Island (n 1:12,000-scale aerial photographs. These were: old by Admiralty = relocations). 8.1%; 9.6%; and 1,285 growth, 75.6%; subalpine, alpine, b Availability of habitats on Admiralty Island study site. other, 6.6% (Schoen and Kirchhoff 1990). In the bear c Transformationof Ivlev's (1961) electivity coefficient (E). d model, we recognized a greater variety of habitat Habitat-capabilityindex (scaled from 0-1). e site. categories than in the original study. Old-growthforest Bear density (per km2) by habitat from Admiralty study f based on best was further subdivided into upland, beach fringe, and HCI determination professional judgment. g from Schoen and Beier (1990) and best professional and the relative abundanceof each habitatwas Extrapolated riparian, judgment. estimated. We also estimatedthe relative abundanceof avalancheslopes and estuaries. to bears and that critical To simplify our habitat-capabilitymodel, we assumed have unique importance the late summer season was the most criticalor limiting seasons may vary regionally. However, the late is when the period for brown bears in Southeast Alaska. We summer season (mid-Jul through mid-Sep) food acknowledgethat other seasons (e.g., springwhen bears most abundant, high-quality (e.g., spawning Brown bears are most are feeding on new growth of sedges at tidewater)also salmon) is available. ALASKABROWN BEAR HABITAT MODEL * Schoen et al. 331 concentrated along low-elevation valley bottoms and to be substantially affected by forest management. coastal salmon streams at this time. These are also the However, to minimize loss of denning habitat as a areas of highest human use and where the most intense consequence of logging, Schoen et al. (1987) resourcedevelopment activities occur (e.g., logging and recommendedavoiding logging on mid-volume (20-30 road building). We believe that brown bears are most mbf/acre), hemlock-sprucestands on slopes greaterthan vulnerableto human-inducedmortality (aside from legal 20? at elevations above 300 m in or adjacentto areas of hunting) at this time and place. Late summer habitat brown bear concentrations. use by radio-collared bears, habitat availability, an The number of brown bears and composition of index of habitat preference, and a habitat-capability habitats in the Admiralty study area were used to index are presented in Table 3. Habitat-use estimate bear density in each habitat. For this part of determinationsexcluded "interior"bears because those the model, we assumedthat the density of bears in each bears represented a relatively small proportion of the habitatwas proportionalto the HCI value. The density northern Admiralty Island study population of brown bears on the Admiralty study site was (approximately 10%) and may be somewhat unique to estimated, in a mark-recapturestudy, at 1 bear/2.6 km2 Admiralty Island. Furthermore, those bears are (Schoen and Beier 1990). After excluding the relatively isolated from most forest management "interior"segment (10%) of the population, 127 bears activities. were estimatedto inhabitthe 365-km2 study area. The Several additionalhabitats are listed for which we did density of bears in the best habitat (HCI = 1.0) not have preference data from Admiralty Island. was computed as follows: Although these habitatsdid not occur on the Admiralty m study site or were not delineated, they are important Dmax=NI HCIi Ai; i=1 because they are the result of forest-management activities (e.g., clearcuts and second-growthforest) or where D max= the density of bears in the best habitat, are used extensively by bears and subject to a N = the numberof bears in the study area, HCIi = the disproportionateamount of logging (e.g., riparianold habitat-capabilityindex for habitat i, Ai = the area of growth). Although we had empirical data on bear habitati. The density of bears in the other habitatswas preference for riparian habitat in general, we further calculatedby multiplyingDmax by each HCI value. The subdivided riparianinto 2 categories (streamswith and number of bears in an area can then be without anadromous fish) based on best professional determinedby the following relationship: judgment (Table 3). m Because clearcuts (0-24 years) and second-growth N= E D, Ai; forests (25-150 years) were not available within the i=i Admiraltystudy area, their suitabilitywas rankedbased where N = the number of bears in an area, Di = the on professionaljudgment (Table 3). The avoidance of density of bears in habitat i, and Ai = the area of clearcuts by radio-collaredbears on Chichagof Island habitat i. As the mix of habitats is changed by forest (Schoen and Beier 1990) justifies their low rankings managementactivities, we can estimate the effect on relative to old growth. Because of the virtual absence habitat capability by calculating N for the new set of of understory vegetation in Southeast Alaska second habitatconditions and comparing it with N. Although growth (Wallmo and Schoen 1980, Alaback 1982), we we have chosen to express habitatcapability in terms of ranked the habitat capability of second growth as 0. the numberof bears in an area, we could also evaluate We distinguished an older category of second growth the effects of managementactivities by examining only (151-300 years). The habitatcapability of older second the percentagechange in habitatcapability. growth was estimated intermediate between young We simulatedthe effects of timber harvest on bears second growth and old growth because of the increasing by running the habitat portion of the model on a productionof forage plants as the standsage. Clearcuts hypothetical65,587-ha watershed of which 31,580 ha and second growth in ripariansites with salmon streams were available for timber harvest (Table 4). Timber were given higher value than upland sites because of harvest was restrictedto low to mid-elevation, upland, the availabilityof spawning salmon during late summer old-growth forest. Riparianareas, beach-fringe forest, (Table 3). and estuary-fringe forests were not harvested. As a Although availability of suitable den sites is an result of a 50% harvest, the brown bear population importantcomponent of brown bear habitat,we assume declined by 16% after 10 years and 23% following 50 it is not limiting in most circumstancesand is unlikely years (Table 4). This difference 50 years after logging 332 Int. Conf.Bear Res. and Manage.9(1) 1994

Table 4. Effect of varying levels of timber harvest on controlillegal kills, woundingloss, anddefense of life populations of brown bears as estimated by the habitat- or kills. Once a roadhas been built for one capability model.a property developmentproject, it often results in additional developments,which increase human-bear interactions, Percentof brownbear population and reduces the area's for maintainedfollowing timber harvest ultimately capability Percentof viablebear populations (McLellan 1990). areaharvested 10 50 supporting years years The dense rain forestof SoutheastAlaska provides 0 100 100 moresecurity cover for bearsthan more open habitats 25 92 88 in the Rocky Mountainsor northernAlaska. Road activitiesin the Greens Creek of 50 84 77 building drainage AdmiraltyIsland displaced fewer bearsthan expected, 75 76 65 presumablybecause of the securitycover providedby the denseforest andBeier In Southeast a Timber harvest was restrictedto low to old- (Schoen 1990). mid-elevation,upland, bears remain closer to growth forests. Riparian areas, beach fringe, and estuary fringe Alaska, may development forests were not harvested. Total area was 65,587 ha; 31,580 ha activitiesthan they do in the RockyMountains because were available for timber harvest under this scenario. of the dense forest cover. As those bears become habituatedto humansand/or associate humans with food reflectsthe lowerhabitat value of second-growthforest. (e.g., garbage),human-bear interactions will increase All the simulatedharvest scenarios resulted in declining andresult in higherbear mortality. Human garbage has bearpopulations. These results appear reasonable but beenimplicated as one of the majorcontributors to bear underestimatethe total impacton bears because no attackson humansand ultimately the reasonthat many riparianforest (the most valuable habitat) was harvested garbage-habituated"problem" bears must be destroyed and the influenceof human-inducedmortality was not (Herrero1985:52, Herrero and Fleck 1990). yet calculated. The combinationof increasedroad access and bears becominghabituated to garbagedumps (and people) is Human-InducedMortality a majorconcern of bearmanagers in the coastalforests After estimating habitat capability, the model of BritishColumbia and SoutheastAlaska (Archibald incorporatesthe effects of human-inducedmortality as 1983, Archibaldet al. 1987, Schoen 1990). For a second step in the analysis. These factors are example, the brown bear season on northeastern assumedto have a landscape-leveleffect and may ChichagofIsland was closedunder an emergencyorder reducehabitat capability regardless of the habitat.This of the AlaskaDepartment of Fish and Game on 30 stage of the model should be considereda working September 1988 because of high bear mortality hypothesis. resultingfrom increased road access and the inadequate Large carnivores,like brown bears, which range garbagedisposal policies of severalsmall communities over extensiveareas (from 1,400 to 40,000 ha) should and logging camps. Titus and Beier (1991) found a be consideredcreatures of landscapesrather than of direct correlation(r = 0.93, P < 0.001) between specifichabitat types per se (Harrisand Kangas1988, autumnbrown bear kill and cumulativekilometers of Schoen 1990). Aside from habitatimpacts, resource road constructionon northeasternChichagof Island development(e.g., logging, mining, hydroelectric duringthe period 1978 to 1989. The numberof illegal development,tourism) must also be evaluatedin terms bears taken there duringthat period is a significant of human-bearinteractions (Peek et al. 1987, Mattson unknown. Clearly,the impactsof humanactivity and 1990, McLellan 1990, Schoen 1990). Resource developmenton bearsneed to be incorporatedinto any developmentin brown bear habitat(generally wild, analysisof the effectsof land-managementactivities on undevelopedareas) significantly improves human access brownbears (Schoen 1990). andconsequently increases disturbance as well as direct We subdividedthe effects of human activity and human-inducedmortality of bears (Pearson 1977, developmentinto differentlevels of impact. These Craigheadet al. 1982, Schoen1990). In general,roads relationshipswere estimated, based on bestprofessional are detrimentalto bears because they increase judgment, as reductions in habitat capability (or opportunitiesfor human-bearinteractions (Elgmork potentialcarrying capacity) within zones of human 1978, Zager1980, Archibaldet al. 1987, McLellanand influence/disturbance(Table 5). Thesereduction factors Shackleton1988, Schoen1990). Althoughit is possible shouldbe consideredas relativevalues (e.g., high, 0- to manage legal huntingof bears, it is difficultto 0.3; medium, 0.4-0.7; light, 0.8-1.0) rather than ALASKABROWN BEAR HABITAT MODEL * Schoen et al. 333

Table 5. Reductions in brown bear habitat capability within fromlong distances(Schoen and Beier 1990, Titusand zones of human Southeast activity/disturbance in Alaska. Beier1991). Thosebears become habituated to humans and humanfoods and are more proneto interactwith Habitat-reductionfactor humans,thus decreasingtheir probability of survival. withinzone of influencea We estimatedsignificant habitat-reduction factors for Humanactivity/landscape modification <1.6 km 1.6-8.0 km landfillswithout incineration (Table 5). Road access was considereddetrimental to bears. Human communities Arterialand collector roads accessible to vehicleswere > 1,000 0.0 0.3 estimatedto have greaterimpacts on bears than local roads and roadsclosed 501-1,000 0.0 0.5 to vehiculartraffic (Table 5). We believe that roads closed administratively(e.g., 11-500 0.3 0.6 with gates or excavatedpits) would still have some <10 0.5 0.8 level of off-road vehicle traffic. Although less detrimentalto bearsthan roads accessibleto vehicles, Landfill without effective incineration 0.0 0.5 roadsclosed temporarily (e.g., withgates) pose greater Forest Service cabin/developed 0.8 1.0 impactsthan permanently closed roads (e.g., through campground bridgeremoval). We believethat all roads, regardless of still have the Permanentcamp site 0.2 0.5 closure, potential for supporting additionalhuman foot traffic, which also influences Temporary camp site 0.5 0.8 bearpopulations. Access Point (airstrip, dock, float 0.8 1.0 In this model, some habitat-capabilityindices plane lake) requiredprofessional judgment for determiningtheir value and all reductionsin Arterial and collector roads 0.4 0.7 (Table3) habitatcapability accessible to vehicles and connected within zones of human activity/disturbancerequired to ferry access or town professionaljudgment (Table 5). Brownbear studies from a area of SoutheastAlaska Local roads accessible to vehicles 0.6 0.9 high road-density (Titusand Beier 1991) were used to evaluatesome of Roads closed temporarily 0.8 1.0 these attributesthat influencethe habitatcapability as relatedto model evaluation. we tested Roads closed permanently 0.9 1.0 Specifically, whetherradio-collared brown bear telemetrylocations a Habitat-capabilityindex (HCI) multipliedby this factor equals bear exhibitedany patternrelated to distancefrom primary potential within the specified zone. Derivation of reduction factors roads, secondaryroads, blocked roads, and salmon (ranked on a relative scale) are based on best professionaljudgment. streams.To makethis evaluation, we chose a subsetof 58 radio-collaredbrown bears captured from 1989 to specificquantifiable values derived from empirical data. 1991on the northeastportion of ChichagofIsland. We We estimatedthat largercommunities would have selectedaerial telemetry locations from 15 July to 15 greaterimpacts than smallercommunities (Table 5). Septemberto coincidewith the late summerseason of For example,brown bears are rarelyobserved in or the habitat-capabilitymodel. Two adjacent,uncut, and adjacentto majorcities or towns in SoutheastAlaska, largelyunroaded watersheds that effectively form one whereasbears are much more frequentlyencountered watershed(total = 185 km2)were comparedwith a nearsmall villages. This indicatesthat suitable habitat watershed(90 km2) that had undergone the most is not used adjacentto theseareas because the bearsare extensive clearcutlogging on the 1,000-km2 study area. killed or Even displaced. thoughthe habitatmay be Each watershed had a single, major salmon spawning suitable,value to bearsis decreasedby humanactivity. stream flowing much of its length. Chum We similarlyestimated that permanent camp sites would (Oncorhynchusketa) and pink (0. gorbuscha) salmon have more impactsthan temporarycamps (Table 5). were the most importantspecies for brownbears. Of We also assumedthat industrialcamp sites frequented 26 anadromousfish streams on the study area, those 2 transientworkers by (manywith limitedexperience in streamshad the highest and fourth highest numbers of Alaska)would be less inclinedto toleratebears than spawning pink and chum salmon. long-termresidents of permanentcommunities. After data screening, 58 aerial location In Southeast telemetry Alaska,landfills without effective fuel- estimateswere availablefor analysisfrom 29 brown firedincineration and/or bear-proof fencing attract bears bearsduring the late summers of 1990 and 1991. Mean 334 Int. Conf.Bear Res. and Manage.9(1) 1994 distance from a brown bear telemetrypoint to primary habitat variables in the model was modified while the roads did not differ between watersheds owing to a other variables were held constant. Variable values primary road oriented near the uncut watershed associated with lowest and highest habitat capability (Table 6). Brown bears were much closer to secondary were used. The resulting estimate of habitatcapability and blocked roads in the roaded watershed, indicating from each run of the model was recorded and the that they did not avoid those locations. That attribute percentage of change determined (Table 7). A high resulted in more frequentbear-human encounters. percentage of change indicates the variable has a high The most important result was that brown bear potential to affect the estimate of habitat capability. locations were much farther away from the salmon Conversely, a low percentage of change indicates that streamin the highly roaded and clearcutwatershed than changes in the variabledo not result in large differences in the uncut and pristine watershed. We believe that a in the estimatesof habitatcapability. lack of cover and forested streambuffers contributedto The successionalstage (i.e., clearcut, second growth, this result. This pattern fits the professionaljudgment old growth) variablehas the greatesteffect on estimates of the capability model whereby the capability is of habitat capability for brown bear (Table 7). reduced in clearcut habitat and salmon spawning Variablesrepresenting riparian habitats with fish present streams. have an intermediate influence on the estimates. Brown bears continued to make use of salmon Presence of estuary fringe has a moderate effect. streams in heavily logged watersheds. They seldom Riparianhabitats without fish and beach fringe habitats used the clearcut habitat, but made frequent use of have the least relative influence on estimates of habitat roads and the patches of remainingforest. That results capability. Because of our Geographic Information in more frequent bear-humanencounters and increases System (GIS) program, we were unable to identify as mortality rates, thereby reducing the habitat capability large a riparian corridor as identified in the model. as suggested by the model. Brown bear mortalityon Therefore, the sensitivity of riparian habitat is the northeastportion of Chichagof Island supports the underestimated. reduction in capabilitywithin zones of human activity. For example, 2 of 4 brown bears killed outside the Table 7. of the sensitivity of variables included in season 1990 and 1991 on the Analysis legal hunting during the habitat-capability model for brown bear in Southeastern northeast portion of Chichagof Island were illegal. Alaska.a Two of those 4 bear deaths were associated with communitiesand a landfill, and 2 were shot and left to Results lie along a primaryroad. Highvalue Low value

Sensitivity Analysis Variable Mean Mean % (namein database) Populationindex Populationindex Change An analysis of the sensitivity of the model was conductedto determinethe responsivenessof the model Successional stageb 182.35 0.40 33.99 0.07 81 the to changes in the value of the variables. Each of Riparian(with fish)c 338.35 0.74 147.39 0.32 56 Estuary fringed 235.65 0.52 166.86 0.36 29 brown bear Table 6. Mean distances (km) from telemetry Riparian(without fish)e 167.71 0.36 147.39 0.32 12 locations to primary roads, secondary roads, blocked roads, Beach 186.55 0.41 167.28 0.37 10 and salmon spawning streams in an uncut and largely fringef unroaded watershed versus a highly clearcut watershed a was used for this because the during late summer, Chichagof Island, Alaska. The Kadashanquadrangle analysis GIS data set is complete for the whole quadrangle. Total land surface of the is 48,087 ha. Unmodified habitat capability is Watershedtype quadrangle 167.49 brown bears; mean habitat-capabilityindex is 0.37. b Uncut Clearcut pa Successional stage: old growth; poletimber. c Riparian(with fish): riparianarea adjacentto a stream with fish; Primary road 3.0 ? 1.8 2.7 ? 2.6 0.082 not a riparianarea. d Secondary road 7.8 ? 2.0 2.4 ? 1.7 <0.001 Estuary fringe: within 300 m of an estuary; beyond 300 m of an estuary. Blocked road 2.7 ? 2.1 0.9 ? 1.5 0.001 e Riparian(without fish): riparianarea adjacentto a stream without Salmon spawning stream 0.5 ? 1.0 1.2 ? 1.2 0.025 fish; not a riparianarea. f Beach fringe: within 150 m of the mean high tide line; beyond 150 a Based on t-tests after testing for equal variances. m of the mean high tide line. ALASKABROWN BEAR HABITAT MODEL * Schoen et al. 335

Model Verification riparian habitat, this number represents a small This model and the associatedcomputer program underestimate.However, the numberof brownbears have been verifiedthrough use of the GIS data base estimatedon AdmiraltyIsland based on empiricaldata availablefor the TongassNational Forest in Southeast was withinthe rangeof 1,200 to 1,700 bears(Schoen Alaska. The purposeof thisverification process was to andBeier 1990). Thesecomparisons suggest the model ensurethat the habitat-capabilitysection of the model is performingwithin reasonable bounds. providesreasonable results on severaltest areas. First, we ran the model on the northernAdmiralty Island studysite to confirmthat the computer code was correct CONCLUSIONS and thatthe availabilityof habitattypes was similarto In SoutheastAlaska, industrial-scalelogging is our originalestimates. Next we ran the modelon the affectingthousands of hectaresof brownbear habitat Kadashan quadrangle(Sitka C4) on southeastern annually. To ensurethe conservationof brownbears, ChichagofIsland. Using densitiesderived from Table we must begin comprehensiveforest planningon a 3, the modelgenerated a densityof 1 bear/2.9km2 for landscapescale with a time perspectiveof at least a the Chichagoftest site. This value lies within our hundredyears (Schoen 1991). The model described estimatedrange of 1.0 bear/2.6km2 to 1 bear/5.2km2 here evaluatesbear habitaton 2 levels. The habitat based on previous brown bear studies in that area level is derivedlargely from empiricaldata on bear (Schoen and Beier 1990). We also compared habitatpreference. The influenceof humanactivity on populationestimates for brownbears generated by the bear mortality and disturbanceis based on best modelto an independentmeasure of the populationfor professionaljudgment and should be considereda AdmiraltyIsland. The modelestimated a populationof working hypothesis. This habitat-capabilitymodel 1,440 brown bears on AdmiraltyIsland (Table 8). provideswildlife and forest managers with an effective Because of the GIS limitationdescribed above for tool for systematicallyassessing the cumulativeeffects

Table 8. Verificationof the habitat-capabilitymodel for brown bear in Southeast Alaska on AdmiraltyIsland.

Habitatcapability Area Totalnumber of Habitatcategories km2 % Meanindex bears No./km2 % Total area 4,359.00 0.34 1,439.26 0.33 Habitat (index >0.01) Forest

Old growth, upland 2,585.00 59 0.34 831.22 0.32 58 Old growth, riparian/withfish 4.20 t 1.00 4.00 0.95 t Old growth, riparian/withoutfish 5.50 t 0.40 2.10 0.38 t Forest beach fringe 181.60 4 0.44 76.13 0.42 5 Forested estuary fringe 12.10 t 0.63 7.21 0.60 1 Subalpine forest 643.40 15 0.37 229.88 0.36 16 Nonforest

Avalanche slope 446.50 10 0.57 242.70 0.54 17 Alpine 143.30 3 0.24 33.48 0.23 2 Estuary 0.41 t 0.79 0.31 0.76 t Clearcut 2.10 t 0.10 0.21 0.12 t Other 119.80 3 0.11 12.02 0.10 1 Total habitat 4,143.70 1,439.26 0.35 Nonhabitat = (index 0.00) 251.50 6 0.00 0.00 0.00 0 336 Int. Conf.Bear Res. and Manage.9(1) 1994

of timber management on brown bear populations in insights. Proc. Int. Conf. Bear Res. and Manage. 8:25-32. SoutheastAlaska. IVLEV,V.S. 1961. Experimental feeding of fishes. Yale Univ. Press, New Haven, Conn. 302pp. JOHNSON, D.H. 1980. The comparison of usage and LITERATURECITED availability measurements for evaluating resource ALABACK,P.B. 1982. Dynamicsof understorybiomass in preference. Ecology 61:65-71. Sitkaspruce-western hemlock forest of southeastAlaska. LINDZEY,F.G., ANDE.C. MESLOW. 1977. Population Ecology63:1932-1948. characteristicsof black bears on an island in Washington. ALASKA DEPARTMENTOF FISH AND GAME. 1978. Petition J. Wildl. Manage. 41:408-412. to remove the brown bear Ursus arctos L. in Alaska from MATTSON,D. 1990. Human impacts on bear use of habitat. Appendix II to the Convention on InternationalTrade and Int. Conf. Bear Res. and Manage. 8:33-56. Endangered Species of Wild Fauna and Flora. On file, MCCARTHY,T. 1989. Food habits of brown bears on Alaska Dep. Fish and Game, Juneau. 71pp. Admiralty Island, southeast Alaska. M.S. Thesis, Univ. ARCHIBALD,W.R. 1983. Problem analysis: grizzly bears Alaska, Fairbanks. 84pp. and coastal development with particular reference to MCLELLAN,B. 1986. Grizzly bear habitatuse analysis and intensive forestry. Fish and Wildl. Bull. B-26. Minist. use of timber sites. Pages 163-166 in G.P. Contrerasand Environ., Victoria, B.C. 24pp. K.E. Evans, eds. Proc. Grizzly Bear HabitatSymposium. , R. ELLIS,AND A.N. HAMILTON.1987. Responses of U.S. Dep. Agric. For. Serv., Gen. Tech. Rep. INT-207. grizzly bears to logging truck traffic in the KimsquitRiver 1990. Relationships between human industrial Valley, British Columbia. Int. Conf. Bear Res. and activity and bears. Int. Conf. Bear Res. and Manage. Manage. 7:251-257. 8:57-64. BUNNELL,F.L., AND T. HAMILTON. 1983. Forage , AND D.M. SHACKLETON.1988. Grizzly bears and digestibility and fitness in grizzly bears. Int. Conf. Bear resource-extraction industries: effects of roads on Res. and Manage. 5:179-185. behaviour, habitat use and demography. J. Appl. Ecol. CHRISTENSEN,A.G. 1986. Cumulative effects analysis: 25:451-460. origins, acceptance, and value to grizzly bear management. MEALEY, S.P., C.J. JONKEL, AND R. DEMARCHI. 1977. Pages 213-221 in G.P. Contreras and K.E. Evans, eds. Habitatcriteria for grizzly bear management. Pages 276- Proc. Grizzly Bear HabitatSymposium. U.S. Dep. Agric. 287 in T. Peterle, ed. Proc. Int. Cong. Game Biol. 13. For. Serv., Gen. Tech. Rep. INT-207. ODUM,E. 1971. Fundamentalsof ecology. W.B. Saunders CONTRERAS, G.P., AND K.E. EVANS, EDS. 1986. Co., Philadelphia, Pa. 574pp. Proceedings of the Grizzly Bear HabitatSymposium. U.S. PEARSON,A.M. 1977. Habitat, managementand the future and Dep. Agric. For. Serv., Gen. Tech. Rep. INT-207. of Canada'sgrizzly bears. Pages 33-40 in T. Mosquin 252pp. C. Suchal, eds. Proc. Symp. on Canada's Threatened CRAIGHEAD,J.J., J.S. SUMNER, AND G.B. SCAGGS. 1982. Species and Habitats. Ottawa, Ont. A definitive system for analysis of grizzly bear habitatand PEEK,J.M., M.R. PELTON,H.D. PICTON,J.W. SCHOEN,AND other wilderness resources. Wildlife-Wildlands Inst. P. ZAGER. 1987. Grizzly bear conservation and 15:160-169. Monog. No. 1, Univ. MontanaFound., Missoula. 279pp. management:a review. Wildl. Soc. Bull. ELGMORK,K. 1978. Human impact on a brown bear ROSENZWEIG,M.L. 1981. A theory of habitat selection. population. Biol. Conserv. 13:81-103. Ecology 62:327-335. K.B. FAGEN,R. 1988. Population effects of habitat change: a RUGGIERO,L.F., R.S. HOLTHAUSEN,B.G. MARCOT, AND E.C. MESLOW. 1988. quantitativeassessment. J. Wildl. Manage. 52:41-46. AUBRY, J.W. THOMAS, for HAMILTON,A.N. 1987. Classification of coastal grizzly Ecological dependency:the concept and its implications and bear habitatfor forestry interpretationsand the role of food research and management. Trans. North Am. Wildl. in habitatuse by coastal grizzly bears. M.S. Thesis, Univ. Nat. Resour. Conf. 53:115-126. a review British Columbia. 86pp. SCHOEN,J.W. 1990. Bear habitat management: Conf. Bear Res. and HARRIS, L.D., AND P. KANGAS. 1988. Reconsideration of and futureperspective. Int. Manage. the habitat concept. Trans. North Am. Wildl. and Nat. 8:143-154. Resour. Conf. 53:137-143. 1991. Forest management and bear conservation. Y. W. HERRERO,S. 1985. Bear attacks, their causes and Pages 97-101. in N. Maruyama, B. Bobek, Ono, the avoidance. Winchester Press, Piscataway, N.J. 287pp. Regelin, L. Bartos, and P. Ratcliffe, eds. Proc. of V , ANDS. FLECK. 1990. Injury to people inflicted by InternationalSymposium on Wildlife Conservation. Int. black, grizzly or polar bears: recent trends and new Congress of Ecology. Tsukuba and Yokohama, Japan. ALASKABROWN BEAR HABITAT MODEL * Schoen et al. 337

, AND L.R. BEIER. 1990. Brown bear habitat U.S. DEP. AGRIC.FOR. SERV. 1991. Tongass land preferences and brown bear logging and mining management plan revision. Supplement to the Draft relationships in Southeast Alaska. Alaska Dep. Fish and EnvironmentalImpact Statement. U.S. Dep. Agric. For. Game. Fed. Aid in Wildl. Restor. Final Report. Proj. Serv., Alaska Region. Manage. Bull. R10-MB-149. W-22-1 through 6 and W-23-1 through3. Juneau. 90pp. VANHORNE, B. 1983. Density as a misleading indicator of ___, , J.W. LENTFER, AND L.J. JOHNSON. 1987. habitatquality. J. Wildl. Manage. 47:893-901. Denning ecology of brown bears on Admiralty and WALLMO,O.C., ANDJ.W.SCHOEN. 1980. Response of deer Chichagof Islands, Southeast Alaska, and implicationsfor to secondary forest succession in Southeast Alaska. For. management. Int. Conf. Bear Res. and Manage. 7:293- Sci. 26:448-462. 304. WEAVER, J., R. ESCANO, D. MATTSON, AND T. PUCHLERZ. _ , ANDM.D. KIRCHHOFF.1990. Seasonal habitatuse 1986. A cumulative effects model for grizzly bear by Sitka black-taileddeer on AdmiraltyIsland, Alaska. J. managementin the Yellowstone ecosystem. Pages 234-246 Wildl. Manage. 54:371-378. in G.P. Contreras and K.E. Evans, eds. Proc. Grizzly , J.W. LENTFER,AND L.R. BEIER. 1986. Differential Bear Habitat Symposium. U.S. Dep. Agric. For. Serv., distributionof brown bears on AdmiraltyIsland, Southeast Gen. Tech. Rep. INT-207. Alaska: A preliminary assessment. Int. Conf. Bear Res. ZAGER,P.E. 1980. The influence of logging and wildfire on and Manage. 6:1-5. grizzly bear habitat in northwestern Montana. Ph.D. SERVHEEN,C. 1990. The status and conservation of bears Thesis, Univ. Montana, Missoula. 131pp. of the world. Int. Conf. Bear Res. and Manage. Monogr. , ANDC.J. JONKEL. 1983. Managing grizzly bear Ser. No. 2. 32pp. habitatin the northernRocky Mountains. J. For. 81:524- TITUS,K., ANDL.R. BEIER. 1991. Population and habitat 526. ecology of brown bears on Admiralty and Chichagof , ,_ ANDJ.HABECK. 1983. Logging and wildfire Islands. Alaska Dep. Fish and Game. Fed. Aid in Wildl. influence on grizzly bear habitatin northwesternMontana. Restor. Prog. Rep. Proj. W-23-4. Juneau. 32pp. Int. Conf. Bear Res. and Manage. 5:124-132.