A PROBABILISTIC MODEL FOR DERIVING SOIL QUALITY CRITERIA-BASED ON SECONDARY POISONING OF TOP PREDATORS .1. MODEL DESCRIPTION AND UNCERTAINTY ANALYSIS

In previous studies, the risk of toxicant accumulation in food chains was used to calculate quality criteria for surface water and soil. A s imple algorithm was used to calculate maximum permissable concentratio ns [MPC = no-observed-effect concentration/bioconcentration factor(NOE C/BCF)]. These studies were limited to simple food chains. This study presents a method to calculate MPCs for more complex food webs of pred ators. The previous method is expanded. First, toxicity data (NOECs) f or several compounds were corrected for differences between laboratory animals and animals in the wild. Second, for each compound, it was as sumed these NOECs were a sample of a log-logistic distribution of mamm alian and avian NOECs. Third, bioaccumulation factors (BAFs) for major food items of predators were collected and were assumed to derive fro m different log-logistic distributions of BAFs. Fourth, MPCs for each compound were calculated using Monte Carlo sampling from NOEC and BAF distributions. An uncertainty analysis for cadmium was performed to id entify the most uncertain parameters of the model. Model analysis indi cated that most of the prediction uncertainty of the model can be ascr ibed to uncertainty of species sensitivity as expressed by NOECs. A ve ry small proportion of model uncertainty is contributed by BAFs from f ood webs. Correction factors for the conversion of NOECs from laborato ry conditions to the field have some influence on the final value of M PC5, but the total prediction uncertainty of the MPC is quite large. I t is concluded that the uncertainty in species sensitivity is quire la rge. To avoid unethical toxicity testing with mammalian or avian preda tors, it cannot be avoided to use this uncertainty in the method propo sed to calculate MPC distributions. The fifth percentile of the MPC is suggested as a safe value for top predators. (C) 1996 Academic Press, Inc.