PARENT SESSION

2A - Mixture Toxicity
Hall 6
8:30 AM - 12:30 PM, Monday, 28 April 2003
Chair: Hermens, J.¹, ¹
Co-chair: Toy, R.², Backhaus, T.³, ^{2 3}

(MO6/7) Ecotoxicity QSAR Development for Complex Substances: A Case Study for Alcohol Ethoxylates.

Boeije, Geert¹, Marshall, Stuart², Cano, Manuel³, Louallen, Jeff³, ¹ Procter & Gamble, Brussels, Belgium² Unilever, Bedford, UK³ Shell, Houston, USA

ABSTRACT- One way to conduct an aquatic risk assessment for a complex substance, is to assess and add the toxic contributions of its individual components. For linear-type alcohol ethoxylates (AE), a large ecotoxicity dataset is available, mainly for commercial mixtures. QSARs are needed to interpolate these data to individual homologues, enabling the toxicity of other mixtures (e.g. an environmental ′fingerprint′) to be assessed. For the existing AE QSARs, the observed mixture ecotoxicities were assigned to the average structures of the tested materials. However, Daphnia magna acute toxicity tests for binary mixtures showed that mixtures are more toxic than corresponding average structures. These observations can be explained by the nonlinear relationship of structure with toxicity (e.g. toxicity typically increases exponentially with alkyl chain-length). Based on this understanding, a new QSAR fitting technique for complex substances was developed, which interprets the actual mixture toxicity, in terms of the entire homologue distribution rather than the average structure. This technique was applied to re-fit existing QSARs for acute and chronic invertebrate, acute fish, and mesocosm toxicities, and to develop a new QSAR for chronic fish toxicity. The quality of fit and accuracy for the new QSARs was found to be at least as good as for the existing ones, despite the higher mathematical complexity. A comparison between the new and existing chronic Daphnia magna QSARs showed that the new QSAR generally predicts less toxicity, although for the most toxic homologues it predicts slightly increased toxicity. Across all AE homologues (C_9-18 and EO_0-20), it was found that the average ratio between the old and new QSAR predictions is slightly less than 2.

Key words: ecotoxicity qsar, alcohol ethoxylates, complex substances, environmental fingerprint