Including Predation Risk in Mechanistic Habitat Assessment Models for Stream Fish

Steven F. Railsback^1,2*Bret C. Harvey³

• ¹Lang Railsback & Associates, Arcata, CA, United States
• ²Department of Mathematics, Cal Poly Humboldt, Arcata, California, United States
• ³Pacific Southwest Research Station, Forest Service (USDA), Albany, California, United States

Mechanistic habitat assessment models have long been used for stream fish, especially drift-feeding salmonids. Most of these models assess habitat value as the rate of net energy intake (growth) obtained by a fish feeding in a habitat unit. However, the fitness value of habitat and the willingness of fish to occupy it also depends on predation risk: habitat is not valuable if it offers high growth but also high risk. Methods for modeling how predation risk varies with characteristics of habitat and fish are much less developed than those for modeling net energy intake. We present approaches we use in InSTREAM, an individual-based salmonid population model, to represent how risk from several kinds of predation depend on fish characteristics (size, activity) and habitat characteristics including depth, velocity, availability of escape and concealment cover, temperature, light intensity, and turbidity. Such models of risk are by nature complex, but they can be designed and parameterized using a variety of conceptual models, literature, and field experiments. Incorporating risk in habitat assessment models also requires combining simulated growth and risk into a meaningful measure of the overall fitness value of habitat. We present a measure of expected future survival of both predation and starvation as a practical, proven measure of fitness value. Assessing habitat explicitly as a measure of future fitness provides conceptual clarity to models, for example by identifying habitat variables more meaningful than some traditional measures (e.g., distance to escape cover instead of generic cover availability) and by illuminating differences between predation by fish and by terrestrial animals. But explicitly considering fitness also highlights the conceptual limitations of habitat-only models for management decision support. In contrast, individual-based population models like InSTREAM provide a way to make meaningful and testable predictions of the effects of habitat change on fish populations.