Improving the ecological relevance of metals risk assessment

F. De Laender

Typically, the effects occurring in mesocosm food webs are complex and hard to understand. Pattern oriented modelling (POM, Grimm et al. 2005) is a useful approach in this context. Models should be able to predict multiple patterns in the data and not be limited to predicting only one pattern. For exam-ple, a food web model that accurately predicts the observed negative effect of copper on cladocerans but fails to predict the observed increase in phytoplankton, is not a good model. A first step in this ap-proach is thus identification and definition of patterns. Examples of expected patterns in mesocosm datasets include which groups are positively or negatively affected by metal exposure, whether recovery or adaptation occurs and if ecosystem functions such as primary production are affected. Next, model performance is evaluated against these patterns: can the model replicate all observed patterns and if not, can the model be adjusted to correct for this?