SYMP 6-5 - Accounting for imperfect detection of species in the estimates of different biodiversity facets

Tuesday, August 8, 2017: 10:10 AM
Portland Blrm 253, Oregon Convention Center
Marta A. Jarzyna, Ecology and Evolutionary Biology, Yale University, New Haven, CT and Walter Jetz, Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT
Background/Question/Methods

Interest in, and opportunities to include functional and phylogenetic attributes of species in community ecology and biogeography are rapidly growing and seen as vital for the assessment of status and trends in biodiversity. However, the fundamental underlying evidence remains the (co-)occurrence of the biological units, such as species, in time and space and our ability to appropriately detect and quantify them. Rarely are all species or individuals detected during the survey event, and failure to account for imperfect detection might result in flawed estimates of biodiversity and its change, ultimately hindering conservation efforts. Here, we examine the implications of imperfect detection of species for functional and phylogenetic diversity (FD and PD) estimates. We explore how FD and PD might have different detectabilities than taxonomic diversity (TD) and how all three might vary differently along spatial and environmental gradients. We also extend occupancy modeling and dendrogram-based methods to address the imperfect detection of different biodiversity facets.

Results/Conclusions

Using a near-half century dataset on birds, we found that accounting for imperfect detection is critical for appropriately capturing all facets of biodiversity as well as spatial and temporal diversity trends. Estimates of taxonomic and functional diversity increased after imperfect detection of species was accounted for. Temporal biodiversity change—measured by relative % change D, dissimilarity DIS, and turnover TUR in assemblage composition—were also biased by imperfect detection, with D generally being overestimated and TUR and DIS being underestimated when imperfect detection was ignored, especially at sites with low levels of turnover and dissimilarity. Estimates of temporal change in FD and PD were biased more than the estimates of change in TD. We conclude that recognizing and accounting for potential detection-related biases in studies of biodiversity dynamics should be given careful consideration, particularly if the study spans large spatial and temporal extents and encompasses a wide range of different species with varied characteristics in terms of their trait-based and phylogenetic position.