Vegetation, predation and natural disturbance may interact in complex ways to influence habitat use by elk (Cervus elaphus), yet these interactions are poorly understood. I used field studies and statistical modeling of Global Positioning System (GPS) biotelemetry data to address how the movement patterns and habitat selection of female elk vary with landscape characteristics at multiple scales in Yellowstone National Park, USA. First, I used a state-space statistical approach to model the internal behavioral state and the proximate movement response of elk to available forage biomass, landscape composition, topography, and wolf (Canis lupus)density. Second I used logistic regression to examine habitat use among individual elk at broad (home range) and fine (700-m radius) scales. Finally, I used fecal pellet counts to examine if dense aggregations of coarse wood and pine saplings provide browsing refugia for aspen (Populus tremuloides) seedlings in areas burned by wildfires in 1988.
The state-space analysis showed a consistent, strong pattern of increased movement by elk during the crepuscular hours. While landscape factors influenced both the behavioral state and the proximate movement response, the specific variables included in the best-fitting models varied substantially among individual elk and were not explained by animal or home range characteristics. In the habitat-use analysis, elk avoided areas of high wolf density and selected for areas that provided a mixture of high-biomass grasslands and high-cover stands of young coniferous forest. The importance of some covariates in individual-based models could be explained by characteristics of the animal (age) and its home range (amount and spatial configuration of open area). The results of both the state-space and habitat-use analyses indicate that a careful consideration of individual variability may help produce more general models of habitat use. Variability in pellet counts was not explained by the density of logs or pine saplings. Aspen seedlings were heavily browsed throughout the study area and were less likely to be found in areas with greater elk use. These results suggest that fire-induced coarse wood piles and pine sapling thickets may not create broad-scale browsing refugia for aspen in the landscape of central Yellowstone National Park.