Ordering chemical compounds by their chronic toxicity to aquatic species. A principal component analysis

This report is part of a project studying the variation in the sensitivity of species to toxicant's. Patterns in the chronic toxicity of compounds to aquatic species are studied using a multivariate statistical technique called principal componant analysis. The matrix consists of 15 aquatic species and 22 compounds. The species belong to different taxonomic classes: algae, plants, protozoans, crustaceans, fishes and amphibians. The organic chemicals belong to several toxicological classes: nonpolar narcotics, polar narcotics, reactive compounds and compounds with a specific mode of action. Three heavy metals, a halogen and a detergent are also included. The data was obtained from the EPA-database Aquire and additional literature surveys. The results show that: The major part of matrix variation is determined by the overall toxicity of compounds and not by differences between species. The compounds can be ordered unambiguously by their average chronic toxicity. This is valid for all species considered and similar to results found earlier for acute toxicity data. No clear patterns in the sensitivity of the species are observed. Comparison with acute toxicity data shows that the ratio of acute to chronic toxicity is similar for all compounds and has a mean value of 4. For the acute to chronic ratio a 95% confidence interval stretching from 1 to 50 has been calculated for the 13 compounds in this study where both acute and chronic toxicity data were available. Mean toxicities of chemicals with a nonspecific mode of action show a strong relation with their octanol-water partition coefficient (Kow) and chemicals with more specific modes of action are more toxic than predicted by their (Kow)