ABSTRACT
During past decades, biomonitors were deployed in lakes and rivers to rapidly detect hazardous chemicals by measuring the endpoints of a single aquatic species at defined short intervals. Most biomonitors, however, are only capable of indicating a departure from baseline water conditions without identifying the cause. In order to provide a more comprehensive assessment, a biomonitoring system which features a library of stereotyped responses of multiple aquatic species in various water conditions is proposed. A preliminary library was constructed by characterizing the behavioural and physiological responses of Daphnia magna, Hyalella azteca, Lumbriculus variegatus, and Pseudokirchneriella subcapitata to various concentrations of atrazine and tributyltin. By employing multivariate statistical tools such as principal component analysis (PCA) and discriminant analysis, this library (which contained responses after 6h of exposure to contaminants) was used as a template to classify and to model other sets of earlier measurements at 2 and 4h, resulting in an accuracy of 73 and 97%, respectively. These findings demonstrated the potential capability of the proposed early-warning biomonitoring system to provide real-time water quality assessment and early-warning contaminant detection.
