As species ranges shift in response to climate change and affected species are more threatened by local extirpations, studying impacts at lagging range margins will likely provide insights into future conditions range-wide. Some of the most drastic impacts in North America are forecasted in the Great Basin, where species unable to seek new habitat in the naturally fragmented landscape will be vulnerable to climatic extremes, vegetation community shifts, and anthropogenic land use change. One of the species under threat from climate change is the greater sage-grouse (Centrocercus urophasianus), a sagebrush obligate with considerable habitat constraints that make it susceptible to habitat loss and fragmentation impacts as sagebrush systems contract at their southern range margin. In this study we evaluated the local and regional factors directly impacting sage-grouse habitat selection in several study areas along their lagging range edge. We used >130,000 GPS locations from >100 birds across four study areas in southern Utah and Nevada from 2014-2020 in habitat selection analyses using mixed effects logistic regressions and random forest models to evaluate the impacts of land cover and topography and infer habitat suitability and functional impacts of habitat management actions.
Results/Conclusions
Our results show significant, complex effects of terrain, land cover, and interactions between them on sage-grouse habitat selection, including impacts common across study areas and particular thresholds and idiosyncrasies within study areas. Most notably, rugged terrain and contiguous sagebrush may break predator sightlines and complicate the effects of tree cover and conifer expansion into sagebrush systems by allowing sage-grouse safe habitat closer to trees than previously reported. The variation across study areas indicates a benefit of including place-based impacts in sage-grouse conservation plans. We found that random forests outperformed traditional resource selection analyses in predictive capability on-out-of-sample data, suggesting wildlife movement and habitat selection research could benefit from exploring machine learning methods for detecting patterns and selecting the most important environmental variables and interactions. Incorporating new understandings of local, place-based ecological drivers will refine regional and range-wide models and support efforts to most effectively create habitat and plan for climate-driven shifts in suitability across their range.