Revision of biological indices for aquatic systems: A ridge-regression solution

Abstract

Biological indices are used worldwide as tools for assessing the ecological status of waters, for example as recommended in the implementation of the European Union Water Framework Directive. The biological index is usually calculated on the basis of taxonomic composition and abundance information, and pre-defined environmental sensitivity values of each taxon. However, the extensive expert- or lab-based process of defining taxon-based sensitivity values makes it a challenge for the biological index to be calibrated, revised or localised. To address this challenge, this paper proposes a ridge-regression method to adjust efficiently the sensitivity values for each taxon on the basis of existing expert-judgement-based values and calibrate the biological indices from real environmental stressor information. The macroinvertebrate Average Score Per Taxon (ASPT) index, which is calculated with the Biological Monitoring Working Party (BMWP) sensitivity values, is selected as an example. Macroinvertebrate and physicochemical data were collected from a total of 107 sampling sites in the Chishui River basin over two years. By testing the proposed approach, this study localises the European-originated BMWP sensitivity values and improves the performance of the ASPT index in a river basin in China. The statistical analysis demonstrates that the ridge-regression strategy is an efficient approach to calibrate and revise biological indices. The calibrations based on different observation groups of macroinvertebrates show a good consistency. The research also identifies some under- and over- estimated taxon-specific sensitivity values in the original BMWP scheme, which agree with the results from some previous calibration studies. The ridge-regression based calibration approach is a promising tool with high flexibility for biologically-centred water quality assessment. The approach can be used for reappraising the BMWP values, and for revising biological indicators that have similar structures with the ASPT index. It can also be used to efficiently assign locally-appropriate ecological attributes and traits in new regions or for new taxa, or even to construct new stressor-specific biological indicators.