A comparison of metrics predicting landscape connectivity for a highly interactive species along an urban gradient in Colorado, USA

Many organisms persist in fragmented habitat where movement between patches is essential for long-term demographic and genetic stability. In the absence of direct observation of movement, connectivity or isolation metrics are useful to characterize potential patch-level connectivity. However, multiple metrics exist at varying levels of complexity, and empirical data on species distribution are rarely used to compare performance of metrics. We compared 12 connectivity metrics of varying degrees of complexity to determine which metric best predicts the distribution of prairie dog colonies along an urban gradient of 385 isolated habitat patches in Denver, Colorado, USA. We found that a modified version of the incidence function model including area-weighting of patches and a cost-weighted distance surface best predicted occupancy, where we assumed roads were fairly impermeable to movement, and low-lying drainages provided dispersal corridors. We also found this result to be robust to a range of cost weight parameters. Our results suggest that metrics should incorporate both patch area and the composition of the surrounding matrix. These results provide guidance for improved landscape habitat modeling in fragmented landscapes and can help identify target habitat for conservation and management of prairie dogs in urban systems.