|
5F - Probabilistic approaches (MOP/197) Practical approaches for the validation of probabilistic risk assessment techniques used to evaluate pesticides. Van den Brink, Paul1, 1 Alterra Green World Research, Wageningen, The Netherlands ABSTRACT- The risk assessment of pesticides for non-target organisms is deterministic, i.e. the Hazard Quotient is based on the ratio between exposure and a toxicity value for some defined standard test species derived in the laboratory. There is international agreement that, while deterministic approaches to risk assessment have their advantages (e.g., in terms of their computational simplicity and low data demand), they also have potentially serious limitations (e.g. many assumptions and, hence, safety factors to allow for unknown uncertainty and variability). Working groups in the US (ECOFRAM) and Europe (EUPRA) have recently recommended that risk assessments should be improved by taking better account of variability and uncertainty, e.g., by using probabilistic statistical approaches. One of the recommendations of the EUPRA workshop was that major components for probabilistic risk assessment should be validated although a comprehensive validation is probably unachievable. This paper will a) summarise possible approaches to validating major components of probabilistic approaches and b) assess the confidence that can be placed in probabilistic risk estimates. As a start the much used probabilistic risk assessment technique, species sensitivity distribution (SSD) concept will be used. Validation methods will focus on the comparison of threshold values derived by SSD with the results of microcosm and mesocosm experiments, complying with the variability in the outcome of these semi-field experiments. Also the predictions made by the artificial intelligence model PERPEST will be compared with the full SSD curve. This to describe the relation between concentration and effects on ecosystems using different tools. The paper will end with an outlook for further research. Key words: ecological risk assessment, pesticides, probabilistic risk assessment, species sensitivity distribution |
