Scale-dependence of fish habitat associations in a North American river system

Determining the appropriate measurement scale to assess habitat variables is critical for ecologists assessing biological or ecological conditions. Depth, velocity, substrate, woody debris and other fish cover variables occur on both reach and microhabitat scales, and fish habitat associations with these variables may be scale-dependent. The aim of this work was to better understand the importance of scale for fish-habitat associations with these variables in a framework consistent with environmental filtering and to test the hypothesis that habitat variable importance is scale-dependent. I used prepositioned areal electrofishing in wadeable streams of the Delaware River basin to evaluate the associations of fish with the same variables summarised on different reach and microhabitat scales. The importance of scale for fish-habitat associations was assessed using two approaches that approximate an environmental filtering framework: variance partitioning with (1) ordination and (2) generalised linear mixed models. Variables on both the reach and microhabitat scales explained a significant fraction of the total variation in fish community composition (p < 0.05). Variation decomposition of reach- and microhabitat-scale effects revealed 20.2% and 2.0% of all variation were due uniquely to reach and microhabitat scales, respectively. Measures of coarseness, embeddedness, amount of riffle and areal coverage of five fish cover variables were significant explanatory variables of community composition at the reach scale only (p < 0.05). Velocity and mesohabitat (amount or presence of riffle) were the only two habitat features that were significant explanatory variables of fish community composition at both the reach and microhabitat scales (p < 0.05). Individual models of species occurrence revealed similar patterns as seen with analyses of community composition. For many fishes, habitat features quantified at the reach scale were more explanatory than at the microhabitat scale. Longnose dace (Rhinichthys cataractae) were more dependent upon microhabitat variables than reach-scale variables, relative to other fishes. Mean velocity at the reach scale was the most important explanatory variable for explaining fish community composition and indicated support for the concept of environmental filtering at the reach and microhabitat scales. Few studies of fish occurrence have incorporated a study design and analytical framework that approximates the hierarchical nature of habitat. This study identifies important scales and predictors, demonstrates the importance of a multiscale approach, and provides support for the environmental filtering concept at the reach and microhabitat scales. These findings will allow ecologists to better account for scale-dependent habitat associations and justify the use of fish habitat associations on reach and microhabitat scales for assessing biotic integrity, restoration and conservation of fishes.