Mapping multi-scale species distributions across the United States and determining their utility for conservation applications

Background/Question/Methods

Single-species distribution models are vital to decision-making and management that require an understanding of habitat use patterns, particularly for species of conservation concern. However, the production of distribution maps for indivdual species is hampered by data scarcity for many species, and often multi-species-, guild-, or other surrogate-based models are used in conservation practice instead. Furthermore, existing single-species maps are rarely seamless or spatially validated due to limited occurrence data. We used ensemble modeling to generate fine-scale, nationwide, seamless, and spatially validated maps for 240 bird species represented in the North American Breeding Bird Survey (BBS) to address these limitations. Models for each species span the conterminous US at three spatial resolutions (0.5, 2.5, and 5 km). We also compared correlations between functional richness patterns of stacked single-species models to rasters representing richness of 19 functional guild-level models that incorporate the same datasets to assess spatial differences in predictions. We further examined differences in the finest resolution viable for mapping single species versus guilds based on sample size.

Results/Conclusions

We successfully modeled 240 individual bird species at the coarsest spatial resolution (5 km) based on an initial 0.5 True Skill Statistic (TSS) cutoff. We generated models for fewer species at each subsequent resolution due to increasing data limitations at finer spatial scales. As expected, modeling guilds with few individuals (e.g., threatened or rare species) or individual species with a small number of observations necessitates coarser resolutions compared to modeling guilds with many individuals (e.g., forest affiliates) or individual species with many observations. We also found that stacked distribution maps and functional guild maps had high spatial agreement at 5-km resolution. Thus, stacked single-species maps can support broader-level conservation goals at some spatial scales. Our bird distribution maps can support fine-scale decision-making across the US and further illustrate the importance of spatial validation in conservation planning.