Comparison of methods for estimating omnidirectional landscape connectivity

Abstract

Maintaining and improving ecological connectivity is an important component of wildlife conservation. Omnidirectional circuit theory algorithms model the flow of electric current across a resistance grid from all directions, making them particularly useful for modeling connectivity of multiple or widespread species, or when source and destination sites are not specified, such as in the case of continuously distributed species. We compared three published omnidirectional connectivity methods—the point-based method, the wall-to-wall method, and the Omniscape method. We compared the three omnidirectional connectivity methods using ten simulated resistance grids, representing commonly encountered landscape features. We then used three Canada lynx (Lynx canadensis) snow-track datasets from Ontario, Canada to validate the ability of each method to detect landscape use. Current density maps generated by all three methods were highly correlated for most simulated landscapes, with the exception of uniform resistance grids. In uniform grids, without landscape features to guide current, each method revealed unique node-placement biases. All three methods produced similar connectivity maps for lynx, showing higher current at lynx presence locations compared to absence locations. The main differences between methods were not in their output, but in their implementation and use of computational resources. Our findings suggest that all three currently published omnidirectional connectivity models produce similar current density maps and can accurately map landscape connectivity. We provide guidelines for selecting a method based on factors such as computational resources, study area, and project goals.