A Bayesian Method for Deriving Species-Sensitivity Distributions: Selecting the Best-Fit Tolerance Distributions of Taxonomic Groups

摘要

We present a Bayesian method for deriving species-sensitivity distributions (SSDs). We employed four Bayesian statistical models to consider differences in tolerance to toxic substances among different taxonomic groups. We first used a Malkov chain Monte Carlo simulation based on these models to estimate the SSD parameters. We then computed deviance information criterion values of the models and compared them in order to select the model with the best predictive ability. We applied this approach to seven substances (zinc, lead, hexavalent chromium, cadmium, nickel, short-chain chloride paraffin, and chloroform) as case examples, and then compared the derived SSDs from the selected models and a model that assumed no tolerance differences among taxonomic groups. We discuss the advantages and limitations of our approach on the basis of our results.