Validation of the narcosis target lipid model for petroleum products: gasoline as a case study.

The narcosis target lipid model (NTLM) was used to predict the toxicity of water-accommodated fractions (WAFs) of six gasoline blending streams to algae (Pseudokirchnereilla subcapitata, formerly Selenastrum capricornutum), juvenile rainbow trout (Oncorhynchus mykiss), and water flea (Daphnia magna). Gasolines are comprised of hydrocarbons that on dissolution into the aqueous phase are expected to act via narcosis. Aquatic toxicity data were obtained using a lethal-loading test in which WAFs were prepared using different gasoline loadings. The compositions of the gasolines were determined by analysis of C3 to C13 hydrocarbons grouped in classes of n-alkanes, iso-alkanes, aromatics, cyclic alkanes, and olefins. A model was developed to compute the concentrations of hydrocarbon blocks in WAFs based on gasoline composition and loading. The model accounts for the volume change of the gasoline, which varies depending on loading and volatilization loss. The predicted aqueous composition of WAFs compared favorably to measurements, and the predicted aqueous concentrations of WAFs were used in the NTLM to predict the aquatic toxicity of the gasolines. For each gasoline loading and species, total toxic units (TUs) were computed with an assumption of additivity. The acute toxicity of gasolines was predicted to within a factor of two for algae and daphnids. Predicted TUs overestimated toxicity to trout because of experimental factors that were not considered in the model. This analysis demonstrates the importance of aliphatic hydrocarbon loss to headspace during WAF preparation and the contribution of both aromatic and aliphatic hydrocarbons test to the toxicity of gasolines in closed systems and loss of aliphatics to headspace during WAF preparation. Model calculations indicate that satisfactory toxicity predictions can be achieved by describing gasoline composition using a limited number of aromatic and aliphatic hydrocarbon blocks with different octanol-water partition coefficients.