Multi-scale assessment of functional connectivity: Landscape genetics of eastern indigo snakes in an anthropogenically fragmented landscape in central Florida.

Landscape features can strongly influence gene flow and the strength and direction of these effects may vary across spatial scales. However, few studies have evaluated methodological approaches for selecting spatial scales in landscape genetics analyses, in part because of computational challenges associated with optimizing landscape resistance surfaces (LRS). We used the federally threatened eastern indigo snake (Drymarchon couperi) in central Florida as a case study with which to compare the importance of landscape features and their scales-of-effect in influencing gene flow. We used genetic algorithms (ResistanceGA) to empirically optimize LRS using categorical land cover surfaces, multi-scale resource selection surfaces (RSS), and four combinations of landscape covariates measured at multiple spatial scales (multi-surface multi-scale LRS). We compared LRS where scale was selected using pseudo and full optimization. Multi-surface multi-scale LRS received more empirical support than LRS optimized from categorical land cover surfaces or RSS. Multi-scale LRS with scale selected using full optimization generally outperformed those with scale selected using pseudo optimization. Multi-scale LRS with large spatial scales (1200-1800 m) received the most empirical support. Our results highlight the importance of considering landscape features across multiple spatial scales in landscape genetic analyses, particularly broad scales relative to species movement potential. Different effects of scale on home range-level movements and dispersal could explain weak associations between habitat suitability and gene flow in other studies. Our results also demonstrate the importance of large tracts of undeveloped upland habitat with heterogenous vegetation communities and low urbanization for promoting indigo snake connectivity.