We examined microhabitat use among fishes in a 37-m section of Coweeta Creek, North Carolina. Numerical abundances of species changed substantially during the 17-month study period. Microhabitat availability, however, did not change markedly during our investigation. Seasonal principal component analyses of microhabitat availability and fish microhabitat use data extracted two main patterns of non-random microhabitat use. Clinostomus funduloides, Notropis eoccogenis and Semotilus atromaculatm were significantly over-represented in deep areas with low to intermediate velocities and depositional substrata. Campostoma oligolepis, Coitus bairdi, Etheostoma blennioides, Rhinichthys cataractae and Salmo gairdneri all occurred in intermediate to deep microhabitats with moderate to high velocities and erosional substrata. Five of seven species exhibited seasonal variation in microhabitat utilization, whereas six species displayed size-related variation in use. Size-related variation was probably ontogenetic. We attributed most seasonal changes in microhabitat use to variations in microhabitat availability.

We used canonical analysis of discriminants to identify factors maximizing interspecific differences in microhabitat use. This analysis indicated that species could be assigned to either a benthic or a water column guild. Species within a guild generally could not be differentiated statistically, whereas members of different guilds were readily separable. These patterns persisted throughout the study, despite changes in numerical abundances of assemblage members. There was no evidence of either exploitation or interference competition for microhabitat, consequently it is unlikely that spatial resources were limiting during our study.