Probability maps of anthropogenic impacts affecting ecological status in European rivers

Abstract

Understanding how anthropogenic pressures affect river ecological status is pivotal to designing effective management strategies. Knowledge on river aquatic habitats status in Europe has increased tremendously since the introduction of the European Union Water Framework Directive, yet heterogeneities in mandatory monitoring and reporting still limit identification of patterns at continental scale. Concurrently, several model and data-based indicators of anthropogenic pressures to freshwater that cover the continent consistently have been developed. The objective of this work was to create European maps of the probability of occurrence of river conditions, namely failure to achieve good ecological status, or to be affected by specific pervasive impacts. To this end, we applied logistic regression methods to model the river conditions as functions of continental-scale water pressure indicators. The prediction capacity of the models varied with river condition: the probability to fail achieving good ecological status, and occurrence of nutrient and organic pollution were rather well predicted; conversely, chemical (other than nutrient and organic) pollution and alteration of habitats due to hydrological or morphological changes were poorly predicted. The most important indicators explaining river conditions were the shares of agricultural and artificial land, mean annual net abstractions, share of pollution loads from point sources, and the share of upstream river length uninterrupted by barriers. The probability of failing to achieve good ecological status was estimated to be high (>60%) for 36% of the considered river network of about 1.6 M km. Occurrence of impact of nutrient pollution was estimated high (>60%) in 26% of river length and that of organic pollution 20%. The maps are built upon information reported at country level pursuant EU legal obligations, as well as indicators generated from European scale models and data: both sources are affected by epistemic uncertainty. In particular, reported ...