Species distribution models for Denali National Park and Preserve, Alaska

Abstract

The objective of this study is to explore the use of existing data to model the distribution of four species in Denali National Park; caribou, moose, grizzly bear, and wolf. Radiolocation data consisting of 1331 locations collected over three years for female caribou, 1329 locations collected over three years for female moose, 6579 locations collected over ten years for grizzly bears, and 2686 locations collected over three years for wolves were obtained from park biologists. A geographic information system was used to derive landscape characteristics associated with the animal locations and random locations placed in the same area. Caribou models were developed at three spatial scales with three different algorithms. Classification tree models showed a high prediction success, correctly classifying 75 to 94 percent of randomly withheld animal locations. Fall models for female caribou had the poorest prediction ability while summer models for female grizzly bears performed best. Topographic landscape characteristics such as elevation and terrain ruggedness were important classifiers for most of my prediction models. Distribution maps were developed for individual and multiple species during different seasons. Areas of moderate elevation along the north side of the Alaska Range are important for all our study animals.