The goal of this research was to formulate new tools for instream-flow analyses, with emphasis on protecting fish assemblages in speciose, warmer-water streams. This included habitat and fish data sets for the upper Roanoke River (URR) in southwestern Virginia, collected during the warmer, low- f low seasons of 1989-1991. Physical data were collected in small, rectangular quadrats or crosssectional transacts, whereas fish were collected by seining and electroshocking in the quadrats. Statistical analyses included uni-, bi-, and multivariate analyses.

The habitat analyses showed that mesohabitat types could be effectively defined with hydraulic, channel-roughness, and geomorphologic variables, and different habitat types showed characteristically different hydraulic dynamics across flows. The analyses also showed that physical habitat could be effectively described by 4 axes (sets of similar variables) at microhabitat scales, although greater covariation occurred at intermediate spatial scales (meso- and macrohabitat level). Habitat assessment for instream-flow analyses can thus be effectively simplified by undertaking visual assessment of several mesohabitat types and/or by measuring only a few physical variables for microhabitat analyses. The differences in hydraulic dynamics among mesohabitat types can simplify determination of changes in the availability of fish habitat across flows.

The fish analyses showed that habitat-use guilds of fish species could be effectively defined with aggregated data, i.e., across mesohabitat types or by calculation of fishes' habitat-use means for physical variables. In contrast, unaggregated data (each quadrat considered separately) gave cruder habitat-use segregation of fish species, particularly because fish species were independent of each other at smaller spatial resolutions and larger spatial extents. The 7 guilds included 4 rheophilic (riffle-oriented) and 3 limnophilic groups (pooloriented). Habitat-suitability models developed for these guilds showed that fish selected habitat variables independently, such that simple models can be developed to predict fish abundance in response to differences in habitat availability.

In sum, the analyses provided habitat-flow and fish-habitat data that can be interfaced (in future analyses) to predict changes in fish abundance and biodiversity to be expected from permanent changes in flow levels and thus habitat availability.