BIODIVERSITY: A NEW INDEX USING PREDICTIVE MODELS.

S. Linke, R.H. Norris, N.C. Sims, and J.L. Coysh. CRC for Freshwater Ecology, University of Canberra, Belconnen, ACT, 2601 AUSTRALIA

Most existing biodiversity measures do not consider the habitat features at a site, which results in an assessment without a reference and can lead to over- or underestimation of biodiversity. Predictive models, based on habitat features, provide insight into the biological potential of a river. These models use an assessment of the presence of taxa with a high chance of occurrence to indicate the health at a site. A reference for biodiversity may be set by the summing the probabilities of rare taxa. Comparing this sum to the actual number of rare taxa gives a quantitative measure of the conservation value at a site. A site with many more rare taxa than expected may be considered biodiverse. In a trial run in the Sydney water supply catchments, NSW, Australia, using macroinvertebrates, we identified different categories of sites: healthy sites with many rare taxa; healthy sites that are less biodiverse; sites in poor health with few, if any rare taxa and; sites failing the health component that had many more rare taxa than expected. The first and last categories indicate either a change in composition, or the potential for a biodiversity 'hotspots'.