Computational monitoring of response behaviors of aquatic invertebrates after the treatments of toxic chemicals

Recently response behaviors of indicator specimens have garnered a special attention regarding on-line monitoring of toxic chemicals in aquatic ecosystems. Behavioral monitoring at the individual level can fill the gap between large-scale (e.g., community diversity) and small-scale (e.g., molecular analyses) measurements in ecological risk assesment and could continuously monitor water quality in an appropriate time scale under economical and pro-ecological measurement conditions. The two-dimensional movement tracks of individual specimens of aquatic invertebrates including chironomid larvae, Chironomus riparius, daphnia (Daphnia magna), and oligochaetes (Limnodrilus variegatus) were analyzed by computational methods such as fractal dimension and artificial neural networks after the specimens were exposed to toxic chemicals (e.g., diazinon, copper). Parameters including speed, acceleration and stop number were analyzed before and after the treatments and were used for statistically characterizing the effects of toxic chemicals on response behaviors of indicator specimens. Fractal dimensions were obtained from the movement positions and appeared to decrease consistently after the treatments. Subsequently, the parameter in the movement tracks were trained with the self-organizing map (SOM). The sequences in deformation and abnormal body movements were accordingly detected to elucidate differences in response behaviors of indicator specimens before and after the treatments.