The primary purpose of this research was to advance our present understanding of the behavioral responses of fish to toxicant gradients. The swimming patterns of individual bluegills, Lepomis macrochirus Rafinesque, were monitored for 5 h before and after exposure to either of two sublethal gradients of ammonium chloride. The exposure took place in a laminar-flow tank, which produced an abrupt toxicant gradient across the middle of a square observation area. Fish behavior was quantified using a television-computer monitor that (a) recorded the position of the fish as a pair of coordinates at 4-s intervals and (b) calculated behavioral parameters from the coordinates at 10-min intervals. The parameter used to measure the activity level of the fish was the distance traveled during each 10-min period. The turning behavior of each fish was assessed by defining each movement as a right or left turn, a straight path, or a reversal, and by storing the angular size of each turn. Spatial selection was measured by analyzing the time spent by each fish in each side of the tank or by counting the number of coordinates recorded in each of three lateral tank areas.

Nominal ammonium chloride gradients of 0.5 mg of ammonia-nitrogen per litre (approximately 4.0 µg/litre of unionized ammonia) produced (a) a temporary exploratory response in which the fish positioned themselves in or near the gradient, (b) a slight increase in activity, and (c) a slight increase in turn sizes. Higher concentration gradients of 5.0 mg of ammonia-nitrogen per litre (approximately 40 µg/litre of unionized ammonia) usually produced a brief drop in activity followed by variable preference-avoidance responses, which were probably related to the amount of physiological stress incurred by each fish. The behavioral data obtained from both exposed and control fish were characterized by a high degree of variability, making statistical hypothesis testing difficult. However, the overall results of these experiments demonstrated that this methodology should prove useful both by hastening the incorporation of behavioral data into water quality criteria and by allowing more detailed analyses of preference-avoidance responses.