Estimating wildlife utilization distributions using randomized shortest paths

Abstract

Incorporating hard and soft barriers into individual space-use measures in wildlife tracking studies remains an ongoing challenge in movement ecology. Randomized shortest paths are proposed as a new tool for estimating wildlife utilization distributions—termed the RSPUD. The RSPUD model requires a single parameter (θ) which controls the trade-off between random exploration of the landscape and deterministic movement along the least-cost path. Synthetic data are used to demonstrate the flexibility of RSPUDs across a range of scenarious that could be encountered with wildlife tracking data. A case-study using GPS tracking data of an individual caribou (Rangifer tarandus) in British Columbia, Canada is used to demonstrate the method in an applied setting. The synthetic data examples highlight the properties of RSPUD across a range or scenarios. In the empirical data, we found that the RPSUD method more appropriately delineated individual space use and the avoidance of obvious barriers on the landscape in comparison to the Brownian bridge model. The calculation of RSPUDs provides new opportunities for more sophisticated spatial analysis of individual measures of space use (home range and utilization distributions), by explicitly taking into consideration the role of hard and soft barriers constraining individual movement opportunities. The approach should be attractive to movement ecologists as it takes a similar approach to existing methods (e.g., Brownian bridges and biased random bridges). The method is made accessible to other researchers as part of the wildlifeTG R package.