Data supporting the publication: Chronic toxicity of metal-organic mixtures to Daphnia magna: To what extent can Concentration Addition and Independent Action predict effects?

The dataset contains the underlying raw data files to each mixture series (1-3) tested and published. Additionally, the R scripts for the data analysis are provided with one script per dataset and an additional dataset for the analysis of MIF values. To predict the toxicity of chemical mixtures the Concentration Addition (CA) model is widely established in chemical risk assessment (RA). The CA model assumes that all compounds in a mixture have the same Mode of Action (MoA). For mixtures with different MoA, such as metals and OMPs, the Independent Action (IA) model was established. Here, we tested which reference model, CA or IA, describes the toxicity of metal-organic mixtures best. Binary mixtures of one metal (Cd, Cu, Ni, or Zn) and one OMP (fenazaquin, fluoranthene, methomyl, or triclosan) were tested in chronic Daphnia magna reproduction tests. Test combinations were based on a previously published prioritization study on freshwater monitoring data to ensure the relevance of the mixtures. Deviations between observed and predicted effects were quantified with the Mixture Interaction Factor (MIF), dividing the observed effect concentration (EC) by the predicted EC expressed as Sum of Toxic Units (∑TU). The results show that for metal organic mixtures CA tends to overestimate toxicity (median MIF: 1.42) and IA tends to underestimate toxicity (median MIF: 0.84). IA tend to predict mixture toxicity more accurately than CA, especially when concentrations were expressed as ∑TU relative to the EC50(median MIF, IA: 0.94, CA: 1.48). In the RA of metal-organic mixtures, the CA model can serve as first tier due to its higher conservatism. In a higher-tier RA, IA could be used due to its greater accuracy in predicting effects if suitable data are available. Alternatively, to compensate for possible effect underestimations by IA or effect overestimations by CA, correction factors could be implemented to CA or IA based mixture assessments, based on the reported MIF values.