Are Ni-Cd toxicity models derived from simple bioassay applicable to natural soils? A bioassay-MSMs coupling approach

The application of terrestrial toxicity models to real soil environments is a challenge due to the heterogeneity of soil. We presented here a novel bioassay-geochem-model coupling approach to predict the bioaccumulation and toxicity of a metal mixture in different soils. Equations describing dose-accumulation and dose-toxicity re-lationships of Cd-Ni mixtures to wheat root elongation were generated using a standard sand culture bioassay. A geochemical multi-surface speciation model (MSM) with default parameters was employed to predicted the dissolved metal contents Cdis of three soils collected in China with different properties. Several toxicity models (TU/CAM + FIAM/BLM/GCSM) were screened to identify the one yielding the most accurate predictions. The results showed that metal bioaccumulation was better predicted by the MSM-Cdis integrated with the dose -accumulation equation (RMSE < 0.58) than by the detection of metal contents in soil porewater (RMSE < 0.74). The MSM predicted metal ion activities combined with the BLM+CAM model successfully predicted Cd-Ni toxicity in all three soils (RMSE = 20.7%), but the deviation was slightly larger than that obtained in mea-surements of metals in soil porewater (RMSE =17.2%). Our study demonstrates that the MSM coupled with the dose-accumulation and dose-toxicity equations derived from sand bioassays offers a promising approach in risk assessments of mixed metals in actual soils while avoiding tedious soil bioassays, which will greatly improve the practical application value of terrestrial toxic models.

Automated summary

The application of a new bioassay-geochem-model coupling approach can predict the bioaccumulation and toxicity of metal mixtures in different soils. The results show that the coupling of a geochemical model with dose-accumulation equations can better predict metal bioaccumulation, while the coupling of toxicity models with geochemical models can predict metal toxicity with good accuracy.