Landscape Influences on Stream Habitats and Biological Assemblages

<em>Abstract.</em>—Habitat data collected at three spatial scales (catchments, reaches, and sites) were used to predict individual fish species occurrences and assemblage structure at 150 sites in the Kansas River basin. Habitat measurements for the catchments and reaches of each sample site were derived from available geographic information system (GIS) data layers. Habitat measurements at the sample sites were collected at the time of fish sampling. Because habitat measurements are typically more difficult to collect as the spatial scale of sampling decreases (i.e., field measurement versus a GIS analysis), our objective was to quantify the relative increase in predictive ability as we added habitat measurements from increasingly finer spatial scales. Although the addition of site-scale habitat variables increased the predictive performance of models, the relative magnitude of these increases was small. This was largely due to the general association of species occurrences with measurements of catchment area and soil factors, both of which could be quantified with a GIS. Habitat measurements taken at different spatial scales were often correlated; however, a partial canonical correspondence analysis showed that catchment- scale habitat measurements accounted for a slightly higher percent of the variation in fish-assemblage structure across the 150 sample sites than reach- or site-scale habitat measurements. We concluded that field habitat measurements were less informative for predicting species occurrences within the Kansas River basin than catchment data. However, because of the hierarchical nature of the geomorphological processes that form stream habitats, a refined understanding of the relationship between catchment-, reach- and site-scale habitats provides a mechanistic understanding of fish–habitat relations across spatial scales.