This paper presents a new approach which was developed to find an optimal combination of candidate metrics for creating a fish-based estuarine biotic index (EBI) for defining the quality status of an estuarine area. The key idea was that a powerful index should minimise two prediction errors simultaneously: falsely declaring the status of a site as disturbed while it is not (Type I error) and the reverse, falsely declaring a disturbed site as undisturbed (Type II error). The balance between both errors is an inherent characteristic of an index and can be displayed as a curve. The area under this curve (AUC) is a measure of the misclassification rate (smaller = better). This criterion was the basis for a stepwise approach whereby in each step a metric resulting in the highest reduction of AUC was added. Five metrics were selected and the distribution of their average was the basis to derive the thresholds for the classes of the EBI. This paper presents the fish-based index (EBI) for the brackish Schelde estuary in Flanders (Belgium). The index was calibrated against fyke net data from five sites during the period from 1995 to 2004. The sites ranged in quality from moderately impacted to very disturbed, classes 3 to 5 respectively. Despite there being neither of the highest classes 1 (high) and 2 (good), the EBI presented can serve as an evaluation tool in the highly impacted situation in Flanders as it discriminates well between moderate and highly impacted sites. In addition, its definition complies with the biological status classes of the European Water Framework Directive (WFD).
EWI Biomedische wetenschappen
- vissen (Pisces)