The relations between ‘standard’ fluvial habitat variables and turbulent flow at multiple scales in morphological units of a gravel‐bed river

Abstract

AbstractFluvial fish habitat is often characterized by highly turbulent flow conditions. Several laboratory experiments suggest that unpredictable turbulent fluctuations can increase the swimming energy costs of fish. At the scale of fish habitat models, it can be hypothesized that turbulence can be captured by the combined effects of the standard habitat variables: depth, velocity and substrate. However, recent studies conducted at the reach scale suggest that turbulent properties are more controlled by the large‐scale bed morphology than by individual roughness elements. In this study, we investigate the spatial structure of turbulent flow and the potential relationships between ‘standard’ habitat variables and turbulent flow properties in pools and riffles of a shallow gravel‐bed river. The study explores these relations at multiple spatial scales. Mean turbulent properties and turbulent flow structures statistics were computed from 1932 near bed velocity time series sampled with acoustic Doppler velocimeters on a regular grid in four morphological units (two pools and two riffles) presenting a gradient of complexity. We used a novel multivariate variation partitioning analysis involving principal coordinates of neighbour matrices (PCNM) to partition turbulent flow properties into six significant spatial scales (VF: 0.35, F: 0.75, M: 1.25, L: 2, XL: 2.5 and XXL: 3 m). Between 45 and 70% of the variance of the turbulent flow properties were explained by the spatial PCNM. In the four units, turbulent properties exhibited a spatial dependence across the entire range of scales. However, the proportion of variation explained by the larger‐scaled PCNMs was higher in the most homogeneous units. In general, the spatial dependence of turbulent flow was lower in the riffles than in the pools, where the mean flow velocity was slower. The capacity of ‘standard’ fish habitat variables to explain turbulent properties was relatively low, especially in the smaller scales, but varied greatly between the units. From a practical point of view, this level of complexity suggests that turbulence should be considered as a ‘distinct’ ecological variable within the range of spatial scales included in this study. Further research should attempt to link the spatial scales of turbulent flow variability to benthic organism patchiness and fish habitat use. Copyright © 2009 John Wiley & Sons, Ltd.