GEOMORPHIC PREDICTORS OF FISHES IN A SOUTHERN PIEDMONT DRAINAGE BASIN

The hypothesis that geomorphic conditions relate to fish assemblages is tested in the upper Etowah River basin north of Atlanta, Georgia. The Etowah drains moderate to high relief terrain of the upper Piedmont and Blue Ridge, which exhibits a wide variety of rock types and geologic structures. This locality is considered a hot spot for biotic diversity, harboring some of the richest stream fish communities in the temperate world.

Sample sites include 31 wadeable streams on the Piedmont with sub-basins ranging from 10-130 km^2. Channel morphology, sedimentology, and basin morphometry were measured by standard field, lab, and map methods using multiple methods for certain variables (i.e., pebble counts and sieving). Fishes were collected by electroshocking and assemblage properties are objectively characterized by numerical expressions including: axis scores from nonmetric multidimensional scaling; a ratio of endemic/cosmopolitan species; and percentages of distinct fish guilds determined by principal components analysis. Geomorphic predictors of fish variables are identified by Spearman and Pearson correlation as well as by multiple linear regression.

More than 100 geomorphic variables were calculated and 51 fish species were collected with a maximum richness of 30 species at a site. The sampled streams exhibit large geomorphic variation with gradients of 0.001 to 0.01, average bed textures of sand to cobbles, and diverse channel morphologies (i.e., wide range of width/depth ratios). The best predictors (r^2 up to 0.70) include variables related to stream bed composition (texture, sediment mobility, riffles/pools), whereas bankfull morphology and basin morphometry are less important. Basin size and related variables generally are not good predictors. The results fail to support the River Continuum Concept, which predicts that stream habitat and biotic communities are strongly related to watershed size. This probably results because of the relatively small size range studied (10-130 km^2). Instead, these data strongly support the Process Domains Concept, which posits that spatially variable geomorphic conditions govern ecosystem structure and dynamics. Many predictors are easily obtained continuous data, which questions the common practice of characterizing streams with interval data.