Combining multi-surface and biotoxicity models to predict cadmium bioavailability and accumulation in a soil collembolan.

The link between internal metal concentrations in soil animals and external metal concentrations is a critical issue in soil ecotoxicity tests and involves metal transfer from solid-liquid interfaces to target soil animals. Soil cadmium (Cd) pollution is a major concern, the bioavailability of Cd to the model soil collembolan Folsomia candida was therefore determined in naturally Cd-contaminated soils by chemical extraction methods and mechanism-based multi-surface models (MSMs). Three combined models were also developed by combining MSMs with a free ion activity model (MSMs-FIAM), a modified biotic ligand model (MSMs-rBLM), and a Gouy-Chapman-Stern model (MSMs-GCSM) to predict Cd bioaccumulation in F. candida in a mechanistic way. MSMs gave better prediction results for Cd bioavailability to F. candida (determination coefficient, R2 = 0.667) than other chemical extraction methods (0.01 M CaCl2, 0.43 M HNO3, soil solution, DGT, and soil total Cd). MSMs calculated dissolved Cd is an effective indicator of bioavailable Cd to F. candida and allowed prediction under a wide range of soil properties. The combined model MSMs-rBLM more successfully predicted Cd bioaccumulation in F. candida (R2 = 0.793, root-mean-square error (RMSE) = 0.172, mean absolute percentage error (MAPE) = 15.4 %) than other combined models and linked the soil-liquid interfaces to the surface of the target soil animal. MSMs-rBLM model may be a new tool for the prediction of Cd ecological risks and bioaccumulation in soils.