Abstract:
Algae sustain biodiversity in aquatic ecosystems by producing oxygen and recycling nutrients. In contrast to their key role in the environment, toxicity data of many organic pollutants, such as phenols, on algae, especially for marine algae, are severely limited. On the other hand, the data requirement in algal toxicity is almost impossible to be supplied through exhaustive laboratory testing considering the huge number of chemicals to be assessed. Therefore, the use of alternative methods to laboratory testing, such as the quantitative structure-activity/toxicity relationships (QSARs/QSTRs), can help reduce the data gap in algal ecotoxicity. In this study, novel toxicity data of phenols on marine alga Dunaliella tertiolecta and freshwater alga Chlorella vulgaris were generated and subjected to QSAR analysis. The phenols selected for toxicological assessment are known to elicit toxicity through different modes of toxic action including polar narcosis, respiratory uncoupling and reactive mechanisms; as such, the data set was regarded as a miniature model of industrial chemical space and provided a realistic basis upon which to explore the development of algal QSTRs. Multiple linear regression and counter propagation artificial neural network techniques were used to build internally and externally validated QSTR models. Most of the QSTRs highlighted the importance of hydrophobicity and electrophilicity related parameters among numerous descriptors. Hydrophobicity was found to underpin the toxicity of phenols to algae. On the other hand, pyrogallol, hydroquinones and catechols, which are potentially capable of being oxidized to reactive species, displayed algal toxicity in excess of that predicted by hydrophobicity. The toxicity of these reactive phenols was better described by electrophilicity parameters. The external validation of the models was also verified using a data set obtained from literature comprising the toxicity of phenols and anilines to another freshwater alga, Pseudokirchneriella subcapitata. Consequently, the developed QSTRs were shown to be applicable to data from another algal test system and at least for another class of organic compounds. Apart from the QSTRs, investigation of inter-algal and inter-species toxicity correlations between algae and other aquatic organisms such as bacteria, protozoa, daphnia and fish revealed that the response of aquatic organisms to phenols differentiated above the level of polar narcosis. As a result, for a heterogeneous set of compounds acting through different modes of toxic action, the models developed in this study can be used to predict the toxicity of untested compounds provided that the new chemicals are within the applicability domain of the respective model.
