Performance and Validation of an On-Line Fish Ventilatory Early Warning System

Abstract

A continuous, automated biological early warning system for rapidly detecting acutely toxic wastewater was developed and tested over a two-year period at an ammunition manufacturing plant. The computerized system simultaneously tracked and analyzed the ventilatory frequency of control and exposed bluegill (Lepomis macrochirus), generating either “warnings” or “alarms” depending on the toxicity of the exposure water. Standard acute flow-through toxicity tests (96 h) were concurrently performed on the wastewater to validate the results of the ventilatory system. Standard water quality and chemical monitoring were also performed during the study to interpret the results of the early warning system. Several preliminary trials indicated that a data assessment interval of 10 min and a sampling rate of 60 Hz for each fish was generally sufficient for detecting acutely toxic conditions. Alarms (statistically significant changes in ventilatory rate over time) were usually generated within 30 min of exposure to effluent concentrations approximating the 96-h median lethal concentration (LC50) value. Time to alarm was inversely related to the toxicity (percentage of the LC50) of the effluent; longer times to alarm occurred with decreasing effluent toxicity (R2 = -0.88). The system was capable of detecting 10% of the 96-h LC50 concentration within 24 h. Less toxic effluent concentrations were not reliably detected with the present system. Water quality and chemical monitoring data generally contributed little information in this study. A general lack of analytical sensitivity may have contributed to this finding. It was thought that better analytical sensitivity would have yielded stronger correlations between the chemical composition of the effluent and toxicity effects. Regression analyses indicated that 35% of the variability in the toxicity tests and ventilatory data was explained by the available chemical data. Two general problems were evident in this system. First, ventilatory rate alone was occasionally an insufficient parameter for detecting acute toxicity. In several runs, ventilatory amplitude and waveform patterns were more informative response variables than frequency. Second, due to the insufficient information generated at times by frequency data alone, false alarms or lack of alarms were evident, reducing the reliability of the system. Results of this study confirm that fish ventilatory systems can serve as useful early warning tools for monitoring wastewaters and that all facets of ventilatory behavior need to be monitored to ensure system reliability. This study also shows that validation of early warning systems, through traditional toxicity tests and chemical monitoring, is crucial to the development of sound approaches for rapidly detecting toxics in surface waters.