Risky business: Modeling mortality risk near the urban-wildland interface for a large carnivore

Abstract

We examined the spatial distribution of 382 black bear (Ursus americanus) mortalities in the Lake Tahoe Basin and Western Great Basin Desert (WGB) in Nevada, USA from 1997 to 2013. Of the 364 mortalities for which we could determine cause of death, vehicle collisions (n = 160) and direct removal of bears by management personnel (n = 132) were the two largest sources of mortality for bears in our study area at the confluence of the Sierra-Nevada Mountains and the Great Basin Desert. Here we use logistic regression and resource selection probability functions (RSPF) to model probability of mortality in the WGB based on anthropogenic and landscape variables. Further, we assessed the impact of spatial resolution on our analyses and resulting probability of mortality models. Human-induced mortalities of black bears were overwhelmingly concentrated near major roads (defined in our analyses as paved roads with two lanes or more), in the town of Incline Village, Nevada, and along the foothills of the Carson Range east of Lake Tahoe. Our results suggest mortality risk is associated with density of and distance to multiple forest types, human population density, landcover, recreation site density and distance, road density and distance, stream distance, hiking trail density, and waterbody distance. Our model results found environmental variables measured at coarse spatial resolutions such as distance to and density of forest best predicted black bear mortality risk, while anthropogenic variables measured at fine spatial resolutions like distance to and density of recreation site best predicted black bear mortality risk in our study area. Our results demonstrate that carnivore mortality as a phenomenon likely operates at multiple spatial resolutions and thus considering scale is important for modeling mortality risk on the landscape.