9294 - Laserfiche WebLink

Home Browse Search

highly significant (0.100.05). Chi-square analysis indicated that chubs used backwaters with ~ j iigher twbidity ratings (p<0.01) (Table 5). ~~ The trends noted- above for pooled samples did not always hold for each sampling period. = In rune, only area and length were significantly different between use and non-use. Dunn this g month chubs used backwaters that were smaller than others. However, only one sampling trip was made in June. July samples showed significant differences in percent cover, main channel temperature, depth and turbidity. Backwaters used by chubs in August showed patterns of habitat differences similar to the pooled analyses. Backwaters used by chubs during August were even more turbid than in other months. In September, however, chubs used backwaters with significantly smaller area and lower water temperature, although width, depth, volume and turbidity were still greater. Attempts to develop discriminant function statistics for predicting chub use of backwater habitats proved futile. Variables important in descriptive equations for one month were not always important in other months or for pooled data. Pooled samples could not be discriminated at any better than 4% accuracy or 35% classification error rate. Model accuracy increased to between 8% and 20% far monthly samples, but classification error rates were between 30% and 42%. High classification error rates were driven by discriminant models' inability to correctly classify chub use backwaters. Depth and temperature variables drove discriminant functions for pooled data. Various combinations of depth, temperature, length and cover were important in discriminant functions for monthly sampling periods. Colorado squawfish -- Analyses for Colorado squawfish use indicated some selection of backwater habitats by type of formation (Table 6}. Colorado squawfish showed greater use of scour backwaters and lower use of shoreline eddies than expected for September (p<0.05) and 18