File S1 - Perception of Human-Derived Risk Influences Choice at Top of the Food Chain

Table S1. Adult grizzly bear resting sites confirmed during field visitation of GPS radiocollar location clusters in 2010 and 2011 in west-central Alberta, Canada. Table S2. Model structure and deviance for top 3 resting-site selection models (RSFs) for grizzly bear resting on reclaimed mines in west-central Alberta, Canada. Model assessment was done by ranking AICc values (Δi) and weights (wi) describing model likelihood. Model complexity (number of parameters) is given by Ki. The top resting-site selection models were selected from candidate food, security and combined food and security models also selected via Δi and wi. Only the top models from the latter categories are given below with the full set of models available in the Table S7. The best overall model is given in bold. Table S3. Model structure and deviance for top 3 resting-site selection models (RSFs) for grizzly bear resting in protected areas in west-central Alberta, Canada. Model assessment was done by ranking AICc values (Δi) and weights (wi) describing model likelihood. Model complexity (number of parameters) is given by Ki. The top resting-site selection models were selected from candidate food, security and combined food and security models also selected via Δi and wi. Only the top models from the latter categories are given below with the full set of models available in Table S8. The best overall model is given in bold. Table S4. Model structure and deviance for top 3 resting-site selection models (RSFs) for grizzly bear resting on non-mined Crown (public) land in west-central Alberta, Canada. Model assessment was done by ranking AICc values (Δi) and weights (wi) describing model likelihood. Model complexity (number of parameters) is given by Ki. The top resting-site selection models were selected from candidate food, security and combined food and security models also selected via Δi and wi. Only the top models from the latter categories are given below with the full set of models available in Table S9. The best overall model is given in bold. Table S5. Model structure and deviance for candidate models for grizzly bear resting on reclaimed mines in west-central Alberta, Canada. Model assessment was done by ranking AICc values (Δi) and weights (wi) describing model likelihood. Model complexity (number of parameters) is given by Ki. The top resting-site selection models were selected from candidate food, security and combined food and security models also selected via Δi and wi. Table S6. Model structure and deviance for candidate models for grizzly bear resting in protected areas in west-central Alberta, Canada. Model assessment was done by ranking AICc values (Δi) and weights (wi) describing model likelihood. Model complexity (number of parameters) is given by Ki. The top bedding site selection models were selected from candidate food, security and combined food and security models also selected via Δi and wi. Table S7. Model structure and deviance for candidate models for grizzly bear resting on non-mined Crown (public) land in west-central Alberta, Canada. Model assessment was done by ranking AICc values (Δi) and weights (wi) describing model likelihood. Model complexity (number of parameters) is given by Ki. The top bedding site selection models were selected from candidate food, security and combined food and security models also selected via Δi and wi. Table S8. Model structure and deviance for top GLM models testing the influence of season, land designation and time of day on selection ratios for grizzly bear resting in west-central Alberta, Canada. Model assessment was done by ranking AICc values (Δi) and weights (wi) describing model likelihood. Model complexity (number of parameters) is given by Ki. Table S9. Model structure, deviance, significance and goodness-of-fit (Wald χ2) for top GLM models testing the influence of season, land designation and time of day on vertical and horizontal cover selection ratios for grizzly bear resting in west-central Alberta, Canada. Model assessment was done by ranking AICc values (Δi) and weights (wi) describing model likelihood. Model complexity (number of parameters) is given by Ki. The full set of candidate models including the null models is provided below.
(DOC)