W5 PM Toxicity of Complex Mixtures
Wednesday, 16 November 2005: 1:50 PM - 5:30 PM in 321-323

(RED-1117-826580) Mechanistic fate and effects model to predict the aquatic toxicity of complex petroleum products in laboratory tests.

Redman, A¹, McGrath, J¹, Parkerton, T², Di Toro, D^{1, 3}, ¹ HydroQual, Inc.² ExxonMobil³ University of Delaware

ABSTRACT- A modified narcosis target lipid model (NTLM) was used to predict the toxicity of water accommodated fractions (WAFs) prepared from eight gas oils and three kerosenes to three different species, an algae (Selenastrum capricornutum), a fish (Oncorhynchus mykiss) and a daphnid (Daphnia magna). Petroleum products are complex substances that are comprised of mixtures of hydrocarbons that upon dissolution into the aqueous phase are expected to act via narcosis. The NTLM has been modified by expressing aquatic toxicity of petroleum products based on membrane-water partitioning (K_mw) rather than octanol-water partitioning (K_ow). This is required because it has been shown that K_ow is not a suitable descriptor for the partitioning that occurs across an organism membrane for higher molecular weight compounds with log(K_ow) greater than approximately 5.5. At a log(K_ow) greater than 5.5, the log(K_mw )-log(K_ow) relationships deviates from linearity. Aquatic toxicity data were obtained using a lethal loading test procedure in which WAFs were prepared at different product loadings. The compositions of a given petroleum product were based on reported boiling point ranges and other available compositional information (e.g. % aromatic and aliphatic hydrocarbons). A three-phase (oil, water, air) fate model was developed to compute the WAF composition of aromatic and aliphatic hydrocarbons in each boiling point range based on product composition and loading. The model accounts for the change in the volume of the product phase that occurs during WAF preparation, which varies depending on product loading as well as volatilization into the headspace (of sealed systems). Toxicity calculations using solely log(K_ow) resulted in calculated median lethal loadings (LL50) an order of magnitude or more lower than observed LL50s. Toxicity predictions were greatly improved by using K_mw instead of K_ow for compounds with log(K_ow) above 5.5 highlighting the importance of adopting K_mw as the preferred descriptor for assessing narcotic effects of high K_ow hydrocarbons.

Key words: petroleum product, toxicity, narcosis, target lipid model