Development of data-driven models for the assessment of macroinvertebrates in rivers in Flanders

The Flemish Environment Agency (VMM) has been gathering water quality andbiological data in more than 1000 sites per year since 1989. These data have been used todevelop predictive models for macroinvertebrates based on data-driven methods (regressiontrees). These models relate the river status to the Multimetric Macroinvertebrate Index Flanders(MMIF), which is a score system developed to report in the context of the European WaterFramework Directive. The trees have been developed in the R software, and severaloptimisations have been made by altering the dataset (variable and record selections). Modelswere evaluated based on mathematical criteria, ecological insight and user convenience (clarity,simplicity and coupling-potential with water quality models of the VMM). The study is a firstattempt to construct a set of models that can be used by the VMM to evaluate the ecologicalbenefits of their river management plans.