Joint survival modelling for multiple species exposed to toxicants

In environmental risk assessment (ERA), the multitude of compounds and taxa demands cross-species extrapolation to cover the variability in sensitivity to toxicants. However, only the impact of a single compound to a single species is addressed by the general unified threshold model of survival (GUTS). The reduced GUTS is the recommended model to analyse lethal toxic effects in regulatory aquatic ERA. GUTS considers toxicokinetics and toxicodynamics. Two toxicodynamic approaches are considered: Stochastic death (SD) assumes that survival decreaseswith an increasing internalized amount of the toxicant. Individual tolerance (IT) assumes that individuals vary in their tolerance to toxic exposure. Existing theory suggests that the product of the threshold zw and killing rate bw (both SDtoxicodynamic parameters) are constant across species or compounds if receptors and target sites are shared. We extend that theory and show that the shape parameter ss of the loglogistic threshold distribution in IT is also constant. To verify the predicted relationships, we conducted three tests using toxicity studies for eight arthropods exposed to the insecticide flupyradifurone. We confirmed previous verifications of the relation- between SD parameters, and the newly established relation for the IT parameter ss. We enhanced GUTS to jointly model survival for multiple species with shared receptors and pathways by incorporating the relations among toxicodynamic parameters described above. The joint GUTS exploits the shared parameter relations and therefore constrains parameter uncertainty for each of the separate species. Particularly for IT, the jointGUTS more precisely predicted risk to the separate species than the standard single species GUTS under environmentally realistic exposure. We suggest that joint GUTS modelling can improve crossspecies extrapolation in regulatory ERA by increasing the reliability of risk estimates and reducing animal testing. Furthermore, the shared toxicodynamic response provides potential to reduce complexity of ecosystem models.