Habitat modeling using path analysis: delineating mountain goat habitat in the Washington Cascades

A 70-90% decline in mountain goat (Oreamnos americanus) populations in Washington State over the past few decades has spurred the need for an improved understanding of seasonal goat-habitat relationships. Habitat use data have been collected from 46 radio-collared mountain goats across their native range in Washington State. Using Geographical Information Systems (GIS), I explored relationships between use and availability of habitat. To overcome issues of autocorrelation, I compared actual mountain goat paths with available paths of matched identical spatial topology and used multi-scale path analysis to explore various ecologically informed relationships between landscape structure and the movements of mountain goats at the home range scale. I extracted used and available (randomized) paths at 4 scales of analysis using square extraction windows of 0.06, 4.4, 15.2, and 56.2 ha that were centered on each point along the path. Matched case logistic regression allowed me to determine the spatially and temporally explicit scales that were the strongest predictors of seasonal and year-round mountain goat habitat from a suite of predictor variables. I found that for year-round habitat, mountain goats chose both abiotic and biotic components of their landscape including; parkland, areas of high solar loading, terrain that is rugged, and terrain that allows escape from predators. This analysis represents one of the most extensive landscape-level habitat relationship studies conducted on mountain goats. Additionally, my methodological approach is applicable to other species-habitat association analyses