Using conditional tree forests and life history traits to assess specific risks of stream degradation under multiple pressure scenario

Abstract

The full accomplishment of the European Water Framework Directive objectives has required from EU members three successive steps: (i) the evaluation of their water body ecological status, (ii) the risk assessment of different anthropogenic pressure categories and (iii) the implementation of appropriate management and restoration programs. We aimed at designing an innovative retrospective ecological risk assessment (ERA) tool working for most of the French wadeable rivers in a context of multiple anthropogenic pressures (step ii). This tool, including conditional tree forest (CTF) models, was built on combinations of benthic macroinvertebrate trait-based metrics for each of sixteen anthropogenic pressure categories. For eleven pressure categories, CTF models have given good impairment risk assessment (i.e. AUC≥0.70), even at moderate risk level and in a multi-pressure context. The four other models have provided poorer but promising results (AUC=0.67±0.02). Identifying the potential weight of individual anthropogenic pressures that lead to biotic assemblage impairment in streams under multiple pressure scenario, is a key step for managers to implement appropriate stream restoration programs. Simultaneously considering the whole complexity of bio-ecological adaptations within biotic assemblages subjected to human pressures provides a functional diagnostic tool both (i) ecologically relevant and (ii) efficient for ERA.

Keywords: Benthic macroinvertebrates; Bio/ecological traits; Ecological risk assessment; European Water Framework Directive; Functional diagnostic tool; Random forest.