Application of multiple index development approaches to benthic invertebrate data from the Virginian Biogeographic Province, USA

Previous work had indicated that the Virginian Province Index did not perform well in a smaller estuarine complex. While it was hoped that the existing Chesapeake Bay Benthic Index of Biotic Integrity, with its greater number of metrics and habitat separation would be more adaptable, this index also did not perform well outside of Chesapeake Bay. In this study we assembled additional metrics and applied different methods of index compilation to explore the indices relative strengths and weaknesses. Three different approaches were utilized - two multimetric indices (Chesapeake Bay IBI and the Mebane IBI) and a statistical logistic regression technique. The data were subdivided by habitat (salinity and grain size), and indices compiled using the same initial group of benthic metrics. Each approach was examined for its classification accuracy for both reference and impaired sites for the entire Virginian Province. The Chesapeake Bay IBI approach did not perform well in this study. In contrast, another multimetric approach, the Mebane IBI approach, performed well, as did the statistical logistic regression approach. Both techniques have promise for index development and could be useful in applying a biological condition gradient to estuaries.