An ecosystem model of the lower Po river for use in ecological risk assessment of xenobiotics

Ecological modelling has the potential to increase the realism of chemical risk assessment for better informed risk management and decision making meeting the protection goals and requirements of the EU's chemicals- and water-related regulations. We developed a food web model of the final lowland section of the longest Italian river (Po) to assess the importance of ecological interactions in setting protective thresholds for river ecosystems exposed to chemicals discharged via wastewater. An integrated 14 compartment model was setup using AQUATOX 3.1, providing a dynamic, quantitative representation of the main functional groups in the food web. To our knowledge, this is the first time that the functioning of the Po ecosystem is quantitatively described. The model was calibrated against observations of biomass density of riverine biota, as typically available for a large lowland river in Europe. The role of ecological interactions on the response of the modelled organisms to chemical exposure was tested on realistic and hypothetical exposure scenarios for two compounds contained in home and personal care products: the anionic surfactant linear alkylbenzene sulphonate and the antimicrobial triclosan. At realistic exposure concentrations the model showed no significant effect compared to control simulations. At hypothetical higher exposure, effects resulting from complex ecosystem-scale interactions emerge. Depending on the organism's position in the food web, indirect effects due to ecological interactions can either amplify or mitigate the effect of direct toxicity. Model results indicate that organisms’ responses to chemicals in real ecosystems is poorly correlated to their direct toxicity (i.e. measured by L/ECX values) for a range of simulated exposure, including concentrations equal to the reported PNEC values. AQUATOX is a useful tool to investigate the relative importance of direct toxicity and ecological interactions, but at this stage it is difficult to use it for prospective chemical risk assessment, given the underlying model uncertainties and the practical limitations of field scale evaluations. An improvement of the quantitative monitoring of feeding preferences and abundance over time of the most representative species within each functional group would be of great help to refine the model parameterisation and calibration. Nonetheless, the development of river ecosystem model scenarios is a stepping stone towards the incorporation of ecological modelling in risk assessment. When extensively tested on different scenarios AQUATOX provides a useful platform, which can be linked to mechanistic effect models as long as this component can be evaluated in controlled settings (i.e. laboratory or mesocosm scale).