|
5G - Species sensitivities distribution (TUP/214) Comparing uncertainty on Ecotoxicological estimator. Payet, Jerome1, Pennington, David1, Jolliet, Olivier1, 1 GECOS-ENAC Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland ABSTRACT- The translation of the results of toxicity tests into an Ecotoxicity Indicator for relative comparison applications can be done following several methods. Older methods are based on the most sensitive species, meeting the requirements of regulatory guidelines related to the identification of a (policy) threshold of hazard for each chemical in the environment. The estimation can also be based on an SSD curve (Species Sensitivity Distribution). The retained measure for the indicator can be the HC50 (50% of species affected - the median) or, for example, the HC5 (5th percentile - sometimes the basis of the PNEC). Similarly, for chronic impact indicators, the toxicity measures underlying SSDs are based on NOECs - providing a distribution of NOECs - for which very little data exists without extrapolation for most chemicals. Using the example of the three indicators, the study underlines the importance in considering the variability of the indicator for the understanding of the uncertainty. The sample variance and the standard error on the mean are calculated in order to assess the reliability of the indicator. Results indicate that variance increase with a factor of 20 between the HC50 and the HC5, while the HC5 is not necessarily more relevant. Furthermore, HC5 value is highly dependent on the biodiversity considered and an assessment based on laboratory species can underestimate the fraction of affected species of a factor 7. Concerning the Results indicates that non-parametric approaches enable reduced confidence interval. Uncertainty on the HC50 indicates that a statistical estimator based on non-parametric method narrows the confidence interval both for the variance and the standard deviation. Key words: Ecosystem, Environmental Risk Assessment, Uncertainty, SSD |
