Further Development and Refinement of Interspecies Correlation Estimation Models for Current-Use Dispersants.

Abstract

Limited species diversity in aquatic toxicity data for most current-use dispersants leads to uncertainties in hazard assessments, which impacts the broader discussion on dispersant use. Sufficient toxicity data are available for a re-evaluation of previously developed dispersant-interspecies correlation estimation (ICE) models. These models increase species diversity because toxicity predictions for untested species are made from the known toxicity for surrogates. Data were used to refine 4 and develop 25 new dispersant-ICE models. Most of the new models are for species not included in the >2000 existing ICE models, and contain a higher species diversity than the original dispersant-ICE models (19 vs 7 species). Dispersant-ICE models predict toxicity with reasonably accuracy: predictions were within 3-fold of observed values (new models: 70% of 132 predictions; refined models: 88% of 83 predictions), and species sensitivity distributions developed with ICE-predicted data only were in most cases not statistically significantly different from those developed with empirical data (83% of 23 paired comparisons). Examples of the practical application of dispersant-ICE models, including laboratory-to-field comparisons within the context of operational dispersant application rates, are also presented. The significance of these results is that dispersant-ICE models could fill gaps in species diversity, and thus help to address concerns about species sensitivities related to the use of dispersants. Environ Toxicol Chem 2019;38:1682-1691. © 2019 SETAC.