Do predictions from Species Sensitivity Distributions match with field data

Abstract Species Sensitivity Distribution (SSD) is a statistical model that can be used to predict effects of contaminants on biological communities, but only few comparisons of this model with field studies have been conducted so far. In the present study we used measured pesticides concentrations from streams in Germany, France, and Finland, and we used SSD to calculate msPAF (multiple substance potentially affected fraction) values based on maximum toxic stress at localities. We compared these SSD-based predictions with the actual effects on stream invertebrates quantified by the SPEAR pesticides bioindicator. The results show that the msPAFs correlated well with the bioindicator, however, the generally accepted SSD threshold msPAF of 0.05 (5% of species are predicted to be affected) severely underestimated the observed effects (msPAF values causing significant effects are 2–1000-times lower). These results demonstrate that validation with field data is required to define the appropriate thresholds for SSD predictions.