Modelling the spatial distribution of human-caused grizzly bear mortalities in the Central Rockies ecosystem of Canada

We examined the spatial patterns of 297 human-caused grizzly bear mortalities from 1971 to 2002 within the Central Rockies ecosystem (CRE) of Canada to explore relationships between mortalities and variables reflecting human development, terrain, and vegetation. Using logistic regression, we modelled the distribution of grizzly bear mortalities based on local landscape attributes as well as examining variation among demographic status, seasons, and mortality type. Grizzly bear mortalities were concentrated in three main regions of the CRE: (1) Lake Louise; (2) Banff town site; and (3) Alberta Provincial lands near the Red Deer River. We found no evidence for environmental differences in mortality locations between sexes or seasons, while sub-adult male and legal harvest mortalities were more dispersed than other mortalities. Models describing the relative risk of mortality were positively associated with human access, water, and edge features, while negatively associated with terrain ruggedness and greenness indices. Model predictions fit well with independent data. Overall, relatively little of the landscape was secure from human-caused mortality for grizzly bears. This would be most directly remedied by controlling human access.