Feeding modes in stream salmonid population models: is drift feeding the whole story?

Abstract

Drift-feeding models are essential components of broader models that link stream habitat to salmonid populations and community dynamics. But is an additional feeding mode needed for understanding and predicting salmonid population responses to streamflow and other environmental factors? We addressed this question by applying two versions of the individual-based model inSTREAM to a field experiment in which streamflow was varied in experimental units that each contained a stream pool and the adjacent upstream riffle. The two model versions differed only in the feeding options available to fish. Both versions of inSTREAM included drift feeding; one also included a search feeding mode to represent feeding in which food availability is largely independent of streamflow, such as feeding from the benthos, or feeding from the water column or the water’s surface in low water velocities. We compared the abilities of the two model versions to fit the observed distributions of growth by individual rainbow trout (Oncorhynchus mykiss) in the field experiment. The version giving fish the daily choice between drift or search feeding better fit observations than the version in which fish fed only on drift. Values for drift and search food availability from calibration to the individual mass changes of fish in experimental units with unaltered streamflow yielded realistic distributions of individual growth when applied to experimental units in which streamflow was reduced by 80 %. These results correspond with empirical studies that show search feeding can be an important alternative to drift feeding for salmonids in some settings, and indicate that relatively simple formulations of both processes in individual-based population models can be useful in predicting the effects of environmental alterations on fish populations.