Spatial habitat use of young-of-the-year Atlantic salmon (Salmo salar) in response to changing stream discharge and population density : testing the instream flow model concept in a controlled experiment

Abstract

Many rivers are affected by man-induced regulations of stream-flow. The effects of these on the instream biota have been studied widely and it is generally accepted that assessment tools for the management of regulated rivers are of vital importance. In particular predictive instream habitat models like the Physical Habitat Simulation Model (PHABSIM) have become popular for this purpose with users world-wide. These models predict discharge-related changes in instream habitat availability for target species by modelling the hydraulic geometry of the river reach on the one and the microhabitat preferences of the species on the other side. Despite their popularity, validation studies for this approach have met many difficulties which are mostly related to the biological part of the model, the habitat preference curves. A review of these studies undertaken here reveals that very little information has been published on two main assumptions of the models: 1. The habitat preference of a species is independent of stream discharge 2. The habitat preference of a species is independent of the species population density

Most validation studies are undertaken in field situations. As such they frequently have had problems relating to sampling the microhabitat use consistently, a lack of experimental control and variations in other secondary variables. It was thus decided to conduct controlled experiments in a large indoor flume. Young-of-year Atlantic salmon (Salmo salar) caught in a nearby stream were stocked into pool-riffle sequences, landscaped within a natural substrate, in observation areas of 3.6 metres length and 1 metre width. Microhabitat use of fish was recorded at three different discharges within a 15-fold discharge variation.

It was found that the mean column velocity preference of the juvenile salmon, calculated by the standard method, varied largely, mainly due to a shift of preference for low water velocities. "Weighted usable area" (WUA) calculations, the final output of instream habitat models, varied up to two-fold due to these differences in preference. Habitat preference also varied with population density. Fish preferred the riffle habitat at low population density and the pool habitat during high population density. Fish used higher mean column velocities during low population density. There are hence fundamental problems related to the approach of using density functions as preference indices as is commonly done for building habitat preference curves, because of a bias for habitat availability. A new approach using Geographic Information Systems (GIS) is taken by comparing the microhabitat conditions fish experience at their chosen positions with the conditions fish would have experienced had they maintained the positions used at the other discharges. It was found that fish adjusted their positions towards significantly different microhabitats between low flow positions and the positions at the other two flows.

This research demonstrates how microhabitat use and preference of wild-caught juvenile salmon varied with discharge and population density in a large near-natural flume. The error introduced by these variations to instream habitat model predictions was large. It confirms that habitat preference curves built as density functions on empirical fish observation data are bound to misrepresent the overall habitat requirements of a species life stage which in the case of juvenile salmon appeared wide and flexible over the range of discharge.