Landscape evaluation of grizzly bear habitat in western Montana

We present a method for evaluating the cumulative effects of human activity on grizzly bear (Ursus arctos) habitat in the Northern Continental Divide Ecosystem of western Montana. Using logistic regression, we modeled the rel ative probabilities of female grizzly bear resource selection from telemetr y data, TM satellite imagery (greenness), elevation, human activity points, roads, and trails. During spring, adult female grizzly bears were positive ly associated with low- and mid-elevation habitats. Logistic regression coe fficients were negative for all road and human activity variables. Summer a nd fall coefficients were also negative for road, human activity, and trail variables. During summer and fall, females were positively associated with mid to high elevations. Coefficients were positive for greenness during al l seasons. Extrapolations of seasonal potential and realized habitat models were made to other areas on the western side of the region where no teleme try data existed. During spring, much of the Bob Marshall Wilderness exhibi ted a relatively low probability of use by female grizzly bears, but the co nverse was observed during summer and fall. The mapping and extrapolation p rocess highlighted areas where habitat restoration would have the greatest benefit. These areas were typically low-elevation spring habitats with high road densities and private lands where urbanization occurred. We recommend that habitat management agencies implement reductions in road densities in seasonal habitat and implement methods to maintain habitat function on pri vate lands.