Multi-species occurrence models to evaluate the effects of conservation and management actions

摘要

Conservation and management actions often have direct and indirect effects on a wide range of species. As such, it is important to evaluate the impacts that such actions may have on both target and non-target species within a region. Understanding how species richness and composition differ as a result of management treatments can help determine potential ecological consequences. Yet it is difficult to estimate richness because traditional sampling approaches detect species at variable rates and some species are never observed. We present a framework for assessing management actions on biodiversity using a multi-species hierarchical model that estimates individual species occurrences, while accounting for imperfect detection of species. Our model incorporates species-specific responses to management treatments and local vegetation characteristics and a hierarchical component that links species at a community-level. This allows for comprehensive inferences on the whole community or on assemblages of interest. Compared to traditional species models, occurrence estimates are improved for all species, even for those that are rarely observed, resulting in more precise estimates of species richness (including species that were unobserved during sampling). We demonstrate the utility of this approach for conservation through an analysis comparing bird communities in two geographically similar study areas: one in which white-tailed deer (Odocoileus virginianus) densities have been regulated through hunting and one in which deer densities have gone unregulated. Although our results indicate that species and assemblage richness were similar in the two study areas, point-level richness was significantly influenced by local vegetation characteristics, a result that would have been underestimated had we not accounted for variability in species detection.