Improving habitat and connectivity model predictions with multi-scale resource selection functions from two geographic areas

摘要: Habitat loss and fragmentation are the most pressing threats to biodiversity, yet assessing their impacts across broad landscapes is challenging. Information on habitat suitability sometimes available in form of a resource selection function model developed from different geographical area, but its applicability unknown until tested. We used Mexican spotted owl as case study demonstrate how models geographic areas affect our predictions for suitability, landscape resistance, connectivity. identified suitable habitats core dispersal movement species. applied two multi-scale models—a local non-local model—to area northern Arizona. converted into resistance surfaces simulations connectivity corridors species, created composite by averaging models. While both performed well, best part where it was built, worse that beyond extent data train it. The improved performances over cases. With rigorous testing, built empirical other can be useful. Averaging multiple improve performance, effectiveness subject performance reference