The complex nature, variable composition and limited aqueous solubility of crude oil and related petroleum products poses challenges for evaluating the aquatic toxicity of these substances. While considerable research has been performed to understand the effects of physically and chemically dispersed oils to aquatic organisms, differences in design and conduct of aquatic toxicity studies often makes results impossible to compare and correctly apply in decision-making. A review of past approaches for generating and characterizing oil exposures in toxicity tests in the absence and presence of chemical dispersants based on the commonly used water accommodated fraction (WAF) procedure is discussed. Differences in the multicomponent dissolution behavior and resulting exposure concentrations of dissolved hydrocarbons in WAFs obtained from different test procedures based on nominal oil loadings or WAF dilutions are illustrated using several case studies. To provide a consistent framework for interpreting toxicity results, dissolved hydrocarbon exposures are coupled to an additive toxicity model. This framework is applied to toxicity tests involving different experimental designs. Two important design considerations that influence observed toxicity are the use of open or closed test systems and constant or declining exposure conditions. A key consideration shown to impact dissolved hydrocarbon exposures in studies that rely on WAF dilutions is the carry over and dissolution of oil droplets. This process can complicate toxicity relationships obtained with predicted or measured oil exposures. Recent advances in passive sampling methods that allow accurate quantification of dissolved hydrocarbon exposure in WAF tests are discussed. Based on insights from this analysis, guidance is provided for standardizing exposure characterization and applying a mechanistic framework in the interpretation of future oil toxicity studies. Adoption of these recommendations would serve to increase comparability across studies, improve the utility of such data in hazard and risk assessments and avoid unsupported conclusions that could misguide rational spill response.
