A non-Euclidean movement model to quantify landscape connectivity with open spatial capture-recapture models

Open Population Spatial Capture-Recapture (OPSCR) models provide a unifying framework to simultaneously model demography and movement while accounting for imperfect detection of individuals. In OPSCR models, movements of individual home ranges between primary occasions usually follow a random walk process that neglects the role of the landscape. Here, we developed a non-Euclidean OPSCR model to explicitly estimate the extent to which home range shifts are shaped by spatial descriptors of the landscape, also referred to as landscape connectivity. We used simulations to validate the robustness of the model and then applied it to a 5-year, noninvasive genetic monitoring dataset of brown bears (Ursus arctos) in the Pyrenees mountain range (France, Spain, and Andorra). We found that male bears' home range movements were smaller close to main roads. The estimated resistance of the distance to roads was negative (-1.49 95% CrI [-2.86, -0.33]), meaning that the cost of moving their home ranges was higher close to roads. Our new OPSCR model provides a data-driven tool to assess the impact of landscape fragmentation on population connectivity at the scale of home range movement using noninvasive spatial capture-recapture data.