Predicting Earthworm Toxicity in Untested Soils with Toxicokinetic Modelling: Concept and Validation

Earthworm risk assessment for chemicals is normally based on toxicity tests performed in artificial test soils with either 5% or 10% organic matter. Soil characteristics, in particular organic matter content and pH, influence the bioavailability of chemicals, and one would therefore expect different levels of toxicity in natural soils. Predicting toxicity in different soils is a major challenge. Here we demonstrate a novel approach for predicting effect concentrations in untested soils by using the results of a toxicity test in a standard test soil together with a previously published empirical model for uptake of chemicals into earthworms. The model predicts the uptake and elimination rate constants of a one-compartment toxicokinetic model based on earthworm species properties (lipid content and specific surface area), topological polar surface area of the molecule, and the organic matter content and pH of the soil. The accuracy of the currently proposed model for predicting toxic effect concentrations (EC50 and LC50) was evaluated against an independent dataset, including 145 measured effect concentrations in non-standard soils, covering 30 synthetic organic compounds and 5 earthworm species. The model showed a high accuracy with 90% of predictions within a factor of 3 of observations. We show the current bias in European risk assessment related to differences in organic matter content between standardized test soils and common agricultural soils and demonstrate how application of the new approach removes that bias. An example with two fungicides illustrates how the model could be applied to increase the environmental realism of the risk assessment.