The effect of avoidance behavior on predicting fish passage rates through water diversion structures

Conserving fish populations within lotic ecosystems requires an understanding of the mechanisms that impede fish movement through a river. For rivers with diversions, negative fish population effects are ameliorated by barriers whose effectiveness is based on individual fish avoidance responses. Here, we evaluate two alternative avoidance models using new algorithms and videotape analyses. We tested the effects of a directed “danger-minimizing” behavior compared to “random avoidance” on predictions of fish passage rates at water diversion facilities using individual-based models. Validation data were collected using a two-dimensional analysis of fish swimming behavior in a small-scale channel with a louver diversion barrier. the danger-minimizing model adequately captured the range of swimming behavior, it underestimated swimming exertion and the time exposed to the barrier. The random avoidance model produced significantly higher passage rates compared to the danger-minimizing model, yet it also resulted in a significant rise in oxygen consumption rate. Given departures between predicted and observed swimming behaviors we included an a posteriori evaluation of adding behavioral complexity to the movement rules. These evaluations demonstrated that positive system level predictions of high passage rates can obscure significant differences in individual level energetic costs. Our study emphasizes the value of examining multiple models at the individual level when extrapolating to the population level effects of fish encountering artificial barriers.