Validating Connectivity Models: A Synthesis

Connectivity models are frequently used by researchers and conservation practitioners to support land management and conservation planning, but the performance of these models often goes untested. We conducted a non-systematic review of published studies that validated connectivity model outputs. The purposes of our review were: (1) to describe the context in which validation efforts have been undertaken; (2) to create a typology of validation approaches to aid researchers in selecting suitable approaches for their own studies; and (3) to outline current best practices and highlight areas for improvement in connectivity model validation.

The number of connectivity modeling studies including validation published per year has generally increased over time, but the proportion of connectivity modeling studies including validation remains low. Most of the reviewed studies validated cost-distance or circuit theory models of functional connectivity for mammalian focal species and used GPS telemetry or species occurrence data for validation. We identified 11 distinct validation approaches, but half of the reviewed studies relied on an approach that compared modeled connectivity values at validation locations versus reference locations. Two applications that warrant greater attention are validation of structural connectivity models and testing transferability of connectivity models across space, time, species, and movement processes.

We conclude that validation approaches are well developed and applicable to a broad range of connectivity models. However, adoption of these approaches remains limited, and validation has yet to become a standard component of connectivity modeling studies.