A statistical framework for modelling migration corridors

Abstract Management of animal populations requires spatially explicit knowledge movement corridors, such as those used during seasonal migrations. Global Positioning System (GPS) tracking data allow for mapping corridors from directly observed movements, but are absent many populations. We developed a novel statistical corridor modelling approach that predicts cost‐distance models fit to migration data. Unlike existing predictive approaches, this does not require the ad hoc transformation habitat suitability surfaces into resistance surfaces. tested ability recover parameters generate simulated movements. then GPS three migrating mule deer Odocoileus hemionus herds in Idaho and Wyoming model function elevation, slope, aspect, percent shrub, date peak green‐up, snow‐off human footprint. assessed fitted using validation tracks same herd well other herds. The reproduced predicted by migratory populations, described direction, magnitude confidence levels effects environmental variables on corridors. Within‐herd indicated more accurate at predicting than null models, cross‐herd some accurately migrations In addition practical benefit management, our framework sets stage evaluating fundamental questions about fitness trade‐offs, navigation, learning, fidelity constraints influence corridor‐generating behaviour. Models can inform management planning conservation across taxa, including potential restoration Our is also readily applied non‐migratory