Developing a multimetric estuarine health index for the Swan Estuary, Western Australia: An approach for selecting fish community metrics in the absence of independent measures of ecological condition

摘要

We describe a novel, weight of evidence-based approach for selecting fish community metrics to assess estuarine health. In the absence of reliable measures of estuarine quality against which to test the sensitivity of candidate metrics, objective, multivariate statistical analyses and multimodel inference were employed to select metric subsets likely to be most sensitive to inter-annual changes in ecosystem condition. Novel pre-treatment techniques were first applied to down-weight the influence of highly erratic metrics and to minimise the effects of seasonal and spatial differences in sampling upon metric variability. A weight of evidence approach was then adopted to select those metrics which responded most consistently across multiple analyses of nearshore and offshore fish abundance data collected between 1977 and 2009. Sets of 11 and seven metrics were selected for assessing the nearshore and offshore waters of this system, respectively. Metrics represented species composition and diversity, trophic structure, life history and habitat functions and, in the case of the nearshore index, a potential indicator species. These metric sets are currently being used to construct a multimetric health index for the Swan Estuary, which is, to our knowledge, the first such tool to be developed for assessing the health of estuaries in Australia.