Predictive models for photoinduced acute toxicity of polycyclic aromatic hydrocarbons to Daphnia magna, strauss (cladocera, crustacea)

Abstract

The photodynamic, actue toxicities of 20 polycyclic aromatic hydrocarbons (PAHs) to Daphnia magna were predicted by photophysical and physiochemical parameters. The photophysical parameters considered were lowest singlet energy, lowest triplet energy, singlet‐triplet splitting energy and phosphorescence lifetime. The physiochemical parameters were first‐ and second‐order connectivity indices and log P values. D. magna were exposed to aqueous solutions of each PAH such that equimolar concentrations in D. magna were achieved. The organisms and the PAHs were then exposed to 120 μW/cm² UV‐A and 25 μW/cm² UV‐B light. Mortality times (min) were recorded and the results reported as median lethal time (LT50). Potency factors (π) were calculated and used to rank the PAHs in terms of relative photodynamic toxicity. Some statistically relevant correlations between individual physical parameters and toxicity were observed. Linear, multiple regression models were poor predictors of photoinduced PAH toxicity. A curve‐linear model was developed to predict photoinduced toxicity from triplet energy. Goodness‐of‐fit chi‐square tests were performed and demonstrated that triplet energy was an effective predictor both of observed LT50 and of LT50 values adjusted to a constant PAH concentration. Toxicity data were also analyzed using discriminant functional analysis. A stepwise, canonical correlation parameter selection method separated the PAHs into three toxic categories by using triplet energy and phosphorescence lifetime as variables. This model, which classifies PAHs as very toxic, moderately toxic or nontoxic, was 100% accurate when the model developed from one set of PAHs was tested with a different set.