OOS 7-2 - Capturing the effects of minor environmental change on a stream fish population: Comparing models differing in the complexity of individual behavior

Monday, August 6, 2012: 1:50 PM
B110, Oregon Convention Center
Bret C. Harvey, Pacific Southwest Research Station, U.S. Forest Service, Arcata, CA and Steven F. Railsback, Department of Mathematics, Humboldt State University, Arcata, CA
Background/Question/Methods

Individual-based models (IBMs) hold promise for predicting the ecological consequences of human use of ecosystem goods and services, in part because of their ability to incorporate individual behaviors critical to population-level responses to environmental change. However, this begs the question: What behaviors are critical? We addressed this question in our efforts to model the consequences of a minor water diversion (approximately 20% of dry-season streamflow) on a stream trout population, by applying two models incorporating different levels of behavioral complexity. The two models used similar general approaches to habitat selection by individual fish (fish sought to maximize their probability of survival over a time horizon), but in one model fish were limited to day-time feeding and night-time activity was ignored (Daily Model), while in the second fish made the decision to feed or hide each day and night (Diel Model).

Results/Conclusions

Using levels of food availability and predation risk determined by calibrating models of an upstream control reach to observed fish density and growth over 4 years, the Daily and Diel models exhibited similar abilities to predict the differences in fish biomass between the upstream control and downstream diversion reaches, although the Daily Model appeared less sensitive to streamflow than the Diel Model. Simulation of the reach below the diversion using the flow regime of the upstream control suggested that differences between the reaches in fish biomass were partially attributable to differences in habitat structure; the more complex model estimated a slightly larger habitat-structure effect. In this case, applying IBMs differing in the number of adaptive behaviors included strengthened our understanding of, and ability to predict, the effects of modest environmental change.