Transcending scale dependence in identifying habitat with resource selection functions

M ulti-scale resource selection m odeling is used to identify factors th a t limit species distributions across scales o f space and time. This m ulti-scale nature o f hab ita t suitability com plicates the translation o f inferences to single, spatial depictions o f hab ita t required for conservation o f species. W e estim ated resource selection functions (R SFs) across three scales for a threatened ungulate, w oodland caribou (Rangifer tarandus caribou), w ith two objectives: (1) to infer the relative effects o f two form s of anthropogenic disturbance (forestry and linear features) on w oodland caribou distributions a t m ultiple scales and (2) to estim ate scale-integrated resource selection functions (SRSFs) th a t synthesize results across scales for m anagem ent-oriented hab ita t suitability m apping. W e found a previously undocum ented scale-specific switch in w oodland caribou response to two form s o f anthropogenic disturbance. C aribou avoided forestry cut-blocks a t b road scales according to firstand second-order RSFs and avoided linear features a t fine scales according to th ird -order R SFs, corroborating predictions developed according to predator-m ediated effects o f each disturbance type. A dditionally, a single SR SF validated as well as each o f three single-scale R SFs when estim ating h ab ita t su itability across three d ifferent spatia l scales o f prediction . W e dem onstrate th a t a single SR SF can be applied to predict relative hab ita t suitability a t bo th local and landscape scales in support o f critical hab ita t identification and species recovery.