Ecological connectivity models commonly derive resistance surfaces from habitat suitability; however, it remains unclear how two-dimensional (2D) versus three-dimensional (3D) representations of landscape propagate through this modeling pipeline, particularly in urban environments. This study evaluated the sensitivity of habitat suitability and connectivity outputs derived from these models to the inclusion of 3D landscape structure for two bird species with contrasting habitat associations, the Western Bluebird (Sialia mexicana) and Acorn Woodpecker (Melanerpes formicivorus), in Los Angeles, California, USA. Species distribution models were developed in Maxent using 2D variables and compared to models incorporating 3D variables derived from LiDAR, with connectivity modeled using Omniscape under identical parameterization. Across both species, 2D and 3D Maxent models exhibited similar performance and high niche overlap, indicating limited sensitivity of overall habitat suitability patterns to the inclusion of 3D variables. Likewise, connectivity outputs were highly correlated with high-connectivity values consistently concentrated in protected open spaces, mirroring the patterns seen in the Maxent models. However, the inclusion of 3D variables altered variable importance in the models and produced discernible shifts in spatial configuration of some suitability and high connectivity values, with species-specific differences. The results suggest that even when overall habitat suitability and downstream connectivity outputs are similar, the inclusion of 3D variables still produced discernible differences in the spatial configuration of high connectivity values, having implications for conservation management decisions. More broadly, these results underscore the importance of evaluating how upstream modeling choices related to the dimensional characterization of landscape propagate through connectivity workflows.
Rogers et al. (Thu,) studied this question.