Toxicity of aryl- and benzylhalides to Daphnia magna and classification of their mode of action based on quantitative structure-activity relationship

The acute toxicity of aryl- and benzylhalides to Daphnia magna was investigated to test the validity of existing classification schemes for chemicals by mode of action, mainly based on fish studies, and the applicability of predictive quantitative structure-activity relationship (QSAR) models. Halobenzenes and halotoluenes are generally agreed to be unambiguous baseline toxicants (class I) with the major exception of the benzylic structures, which are reactive in fish tests (class III). Eighty-nine percent of the arylhalides tested in this study match a log P-ow-dependent QSAR, including fluorinated, chlorinated, brominated, and iodinated derivatives, thereby confirming the validity of the baseline models also for variously halogenated compounds (other than only-chloro compounds). The toxicities of the benzylhalides relative to baseline QSARs clearly indicate that these compounds belong to two classes by mode of action, i.e., they either act as narcotic toxicants (class I) or reveal excess toxicity due to unspecific reactivity (class III). On some occasions, the assignment to the two classes in D. magna deviates from the structural rules derived from fish, i.e., iodinated compounds as well as alpha,alpha-Cl-2-toluenes lack reactive excess toxicity but behave as nonpolar nonspecific toxicants. The QSARs derived during this study reveal lower slopes and higher intercepts than typical baseline models and, together with the analysis of mixture toxicity studies, behavioral studies, and critical body burden, advocate the hypothesis that there are several different ways to produce baseline toxicity. Most halobenzenes and halotoluenes are actually baseline chemicals with some extra reactivity and as such form a subgroup, whose limits still have to be defined. Different primary sites of action could explain why the chemicals are discriminated by different classification systems, but still they must have some rate-limiting interaction in common (e.g., lipid diffusion) as they fit the same log P-ow-dependent baseline QSAR.