Northern Prairie Wildlife Research Center
The results of this analysis help to explain the highly restricted distribution patterns of many species and reveal the limiting factors affecting fishes of the Red River basin. Large differences were noted in the way species of the same family, and even of the same genera, responded to the physiochemical variables. Within each group (Figures 15-21), there were some species that preferred waters with levels of the environmental variables at or near the basin averages for those variables (the centroid of the plots), whereas other species preferred waters with levels either much below or much above averages for the basin. The northern hogsucker and greater redhorse (Figure 15), pugnose shiner and weed shiner (Figure 16), central stoneroller (Figure 17), yellow bullhead (Figure 18), rainbow darter, least darter, and logperch (Figure 20), and the bowfin (Figure 21) all have preferred habitats within the basin characterized by high discharges, low flow variability, and low conductivity, residue, and hardness. In contrast, there are several species which have widespread distribution patterns in the basin (the white sucker, common shiner, and others). These species have occurred in reaches which had waters which were average in their levels of discharge, flow variability, and chemical composition. These habitat "generalists" are likely much more capable of adapting to and surviving any changing environmental conditions at reaches throughout the basin than species which are more specialized in their habitat requirements.
No other studies of midwestern fishes have been conducted which have correlated stream fish assemblages with physiochemical variables, so comparisons can only be made with studies from other geographic regions. Stevenson (1974) conducted one of the first multivariate analyses of fish distribution over a large geographic region. The study included four water-quality variables (total dissolved solids, chloride, sulfate, and hardness) which were found to correspond with the distribution of five fish species in Oklahoma. In Arkansas, where natural environmental gradients in 101 drainages in the state were well-defined from northwest to southeast, Matthews and Robison (1988) conducted an analysis of native stream fishes. Multiple regression analysis determined that fish distributions were significantly correlated with 11 geographic, meteorological, and local factors. Taylor et al. (1993) studied the relationship between fish species and environmental factors in the upper Red River basin, southwestern Oklahoma. Assemblages were predicted along measured environmental gradients; and it was determined that conductivity, stream size, alkalinity, woody debris, and water clarity were the most important variables. Matthews et al. (1992) determined that the spatial distributions of fishes in Arkansas were significantly related to overall patterns in water quality. Total dissolved solids, total hardness, chloride, conductivity, ammonia, dissolved oxygen, pH, alkalinity, and magnesium had high loadings on the first two PCA axes of their analysis. Their results emphasized, in agreement with Hawkes et al. (1986) and the present study, that "environmental variables likely operate in concert with each other as a multivariate system and not as univariate variables." While individual water quality variables, such as pH or salinity, alone have been shown to restrict the distributions of fishes (Echelle et al. 1972, Koel 1993, Koel and Peterka 1995), it is unlikely that any one water-quality variable alone limits the distribution of fishes in the Red River basin.
A potential criticism of this gradient analysis may be that its scale is too large. Information regarding micro habitat use of fishes was not utilized; rather, fishes were grouped on a broad geographic scale by ecoregion. Studies of both types are valid. While working on the distribution of the red shiner Notropis lutrensis, Matthews (1985) stated that "to fully understand the use of habitats by a widespread species, there is clearly a need both for micro habitat information at the local scale and broad information on typical habitats on a geographic scale." Two comparable studies were conducted by Strayler and Ralley (1993), who studied micro habitat use by a stream-dwelling unionid assemblage, and Strayler (1993), who studied the macro habitat use of the same group. Predictive power of micro habitat variables (water depth, current speed, bottom roughness, and others) was so low that the adequacy of the traditional micro habitat approach to unionacean ecology was questioned (Strayler and Ralley 1993). In contrast, macro habitat descriptors (stream size, hydrologic variability, calcium concentration, and others) were highly effective predictors of mussel distributions (Strayler 1993). It is the second approach that was taken in the present study. Fish species were directly related to several physical and chemical variables in large stream reaches from throughout the entire Red River basin. Undoubtably, studies of micro habitat use of fishes, such as that by Aadland (1993), will complement the results of this research.
The present study of species and environment would not have been feasible if all data had to be personally collected. The importance of this study was that it indicated the relationships between fish assemblages (even individual fish species) and trends in water quality and flow variables within an extremely large geographic region, the Red River basin, primarily using published information.