Interactive Metal Mixture Toxicity to Daphnia magna Populations as an Emergent Property in a Dynamic Energy Budget Individual-Based Model.

Environmental risk assessment of metal mixtures is challenging due to the large number of possible mixtures and interactions. Mixture toxicity data cannot realistically be generated for all relevant scenarios. Therefore, methods for prediction of mixture toxicity from single-metal toxicity data are needed. We tested how well toxicity of Cu-Ni-Zn mixtures to Daphnia magna populations can be predicted based on the Dynamic Energy Budget theory with an individual-based model (DEB-IBM), assuming non-interactivity of metals on the physiological level. We exposed D. magna populations to Cu, Ni, and Zn and their mixture at a fixed concentration ratio. We calibrated the DEB-IBM with single-metal data and generated blind predictions of mixture toxicity (population size over time), with account for uncertainty. We compared the predictive performance of the DEB-IBM with respect to mixture effects on population density and population growth rates with that of two reference models applied on the population level, independent action and concentration addition. Our inferred physiological modes of action (pMoA) differed from literature-reported pMoAs, raising the question of whether this is a result of different model selection approaches, intraspecific variability, or whether different pMoAs might actually drive toxicity in a population context. Observed mixture effects were concentration- and endpoint-dependent. The independent action was overall more accurate than the concentration addition but concentration addition-predicted effects on population growth rate were slightly better. The DEB-IBM most accurately predicted effects on 6-week density, including antagonistic effects at high concentrations, which emerged from non-interactivity at the physiological level. Mixture effects on initial population growth rate appear to be more difficult to predict. To explain why model accuracy is endpoint-dependent, relationships between individual-level and population-level endpoints should be illuminated. Environ Toxicol Chem 2021;40:3034-3048. © 2021 SETAC.