Using information theory to select spatial scales for species-habitat responses with camera traps

Widespread anthropogenic landscape change, particularly from energy development, has fundamentally reshaped ecosystems, and understanding species responses remains a central ecological challenge. Remote camera traps are widely used to estimate mammal abundance and distribution, but inferring species-habitat relationships from these data is complicated by the spatial scale at which landscape features are measured. We apply foundational concepts of landscape ecology to explore whether predictable patterns in scale emerge within a broad mammal community inhabiting a highly developed landscape. Using camera trap data from 11 mammal species and a multi-scale, information-theoretic modeling approach, we investigated: (1) how best-supported spatial scales differ between anthropogenic and natural features; (2) species-scale patterns when modeling similar landscape attributes separately versus together; and (3) evidence for patterns of species-scale relationships based on body size and trophic level. We found that best-supported spatial scales spanned the entire range of sizes considered, from 250 to 5000 m. Best-supported scales differed between anthropogenic disturbances and natural landcover model sets for most species, but without a predictable direction. Similarly, optimal scales varied across species but showed no consistent relationship with species traits such as body size or trophic level. Species' best-supported spatial scales did not converge on a characteristic scale more frequently when similar landscape attributes were considered in individual models versus combined into a global model, nor did scale domains emerge consistently for global models. Our findings emphasize the complex and context-dependent nature of species-landscape interactions and underscore the importance of evaluating spatial scale in ecological analyses. Explicit consideration of scale, including evaluating a broad range of spatial extents, and, where appropriate, implementing multi-scale approaches can improve ecological inference and interpretation. Transparent reporting of scale decisions and critical evaluation of how scale influences results are essential for robust assessments of species responses to landscape change.