Calibration, validation and sensitivity analysis of an ecosystem model applied to artificial streams.

There is a growing interest for the use of ecological models to improve the Ecological Risk Assessment (ERA) of toxic pollutants. However, although several models have been proposed, there is little information available on their relevance (i.e., validation through comparison with original data) and on their sensitivity to the input parameters. In this study, the Aquatox Model was used to simulate biomass dynamics of various biological compartments in artificial streams designed for measuring the effects of pollutants on aquatic communities. Calibration and validation of the model were performed using data from control streams. Multi-variate sensitivity analysis was implemented to identify those parameters that exert a prominent role in the outputs of the model. The calibrated model was able to adequately describe the dynamics of most of the simulated biological compartments of a stream. Using data from other streams, it was shown that between-streams natural variability was a source of discrepancy between observed and simulated data. Sensitivity analysis showed that the model was highly sensitive to the parameters related to the temperature limitation, maximum rate of photosynthesis of producers and consumption by consumers. This strongly suggests that particular attention should be devoted to the estimation of these parameters if this model has to be used for ERA of toxicants in aquatic ecosystems.