Soil properties like organic matter (OM) content show great variation, making it hard to predict the fate and effects of a chemical in different soils. We therefore addressed the question: can we remove the complexity of the soil matrix and yet accurately predict soil toxicity from porewater exposures? Folsomia candida was exposed to imidacloprid in natural (LUFA 2.2 [4.02% OM], Grassland [12.6% OM]) and artificial soils (OECD 5 [6.61% OM], OECD 10 [10.8% OM]), in pore water extracted from spiked LUFA 2.2 soil and in water. Toxicity decreased with increasing OM content except for Grassland soil, which had the highest OM content but the lowest clay content, suggesting a role of clay minerals in the binding of imidacloprid. Distribution coefficients for imidacloprid based on toxicity (Toxicity-Kd) were derived by comparing effect concentrations in LUFA 2.2 soil and in water. Using these Toxicity-Kds to recalculate soil LC50s/EC50s to porewater concentrations, the differences in LC50/EC50s almost disappeared. The recalculated porewater LC50s did not differ by more than a factor of 0.55-1.43 from the LC50 obtained upon water exposure. This similarity suggests that the toxicity in the soil is dependent on porewater concentrations and can be obtained from water exposure. The porewater test and the corresponding "pore-water extrapolation concept" developed in this study may be used to predict the toxicity of chemicals in the soil and extrapolate among different soils.
Keywords: Bioavailability; Collembola; Hazard/risk assessment; Imidacloprid; Organic matter; Soil ecotoxicology.