Investigating a major assumption of predictive instream habitat models: is water velocity preference of juvenile Atlantic salmon independent of discharge?

Abstract

The mean column velocity preference of juvenile Atlantic salmon Salmo salar (LF 30–55 mm) was investigated by observing their spatial pattern of habitat use in a laboratory flume while varying discharge (Q) over a 18-fold range (Q=2·6–46·8l s-1). Based on 341 fish observations at three discharges (Q=2·6, 15·0 and 46·8l s-1), three separate velocity preference curves were developed using standard procedures. The mean column velocities measured at 0·6 depth for the fish positions at the set low, medium and high discharges had medians of 7, 10 and 24 cm s-1, respectively, and varied significantly between the discharges. Across the range of flows, the fish utilized mean column velocities between 0 and 56 cm s-1, but the three velocity preference curves differed. Differences between juvenile Atlantic salmon use of habitat, defined according to mean column velocities at different discharges, were greatest at the lower end of the available range of velocities (<20 cm s-1). Weighted usable area (WUA), the output of the instream flow model PHABSIM that is used to describe the available habitat at a given discharge, was calculated for the flume using the preference curves built at the three set discharges. The model was highly sensitive to differences between the three preference curves and WUA varied by up to a two-fold difference. Furthermore, habitat-discharge relationships derived from the three preference curves were very different. Predicted habitat losses across the modelled range of discharges varied by up to 150% depending upon which velocity preference curve was used in the model. Thus, the assumption that a single preference curve can be applied across a range of discharges is not valid and is likely to result in large errors when employing PHABSIM and other models that use similar principles.