Laserfiche WebLink
268 M. J. JENNINGS, L. S. FORE AND J. R. KARR <br />Table II. Statistical significance of reservoir rank concordance <br />based on individual metric values and RFAI scores. Kendall's <br />coefficient of concordance (Zar, 1984) was used to evaluate the <br />agreement among annual (1989-1992) rank orders of reser- <br />voirs. Significant concordance indicates similar rankings among <br />years <br />Metric Inflow Transition Forebay <br />TAXA NS NS NS <br />TSUN NS NS <br />TSUK * NS <br />TINT * NS <br />PTOL <br />POMN * NS <br />PINV NS * NS <br />TMIG NS <br />TLIT <br />FHAI NS NS NS <br />ABUN NS NS <br />RFAI NS NS NS <br />*p < 0.05 <br />inflow zones in four reservoirs (Kentucky, Nickajack, Chickamauga and Watts Bar). Each sample consisted <br />of the standard 10 electrofishing runs plus 10 additional runs to improve our understanding of variability <br />among runs and accumulation of new species beyond 10 runs. The 10 runs closest to the dam from October <br />samples were used to calculate annual RFAI scores for these four inflow sites because these 10 runs most <br />closely approximated the location of samples taken in other years. <br />The precision of any estimate made from a sample depends both on the method by which the estimate is <br />calculated from the sample data and on the plan of sampling (Cochran, 1977). To evaluate the within-year <br />variation in RFAI scores, the index was calculated for each monthly sample with the 10 runs closest to the <br />dam. Bootstrap sampling from these 10 runs was used to estimate the variance of each monthly sample. <br />These variance estimates allow comparison of the four samples at a site, providing an evaluation of measure- <br />ment error (or precision) of RFAI. <br />A similar resampling algorithm was used to evaluate the sampling error caused by slight differences in <br />sampling locations. In this instance, resampling was conducted at the level of the run, rather than at the <br />Table III. Variance estimates of RFAI scores generated from boot- <br />strap analysis with nine metrics. Data presented are averages of <br />four years, except Melton Hill Reservoir, which was sampled over <br />two years <br />Reservoir Inflows Transitions Forebays <br />Kentucky 4.5 3.0 3.2 <br />Pickwick 3.9 5.3 3.6 <br />Wilson 4.0 NA 7.9 <br />Wheeler 6.0 6.7 5.1 <br />Guntersville 1.8 2.6 2.9 <br />Nickajack 4.0 NA 1.2 <br />Chickamauga 2.0 1.5 0.4 <br />Watts Bar 6.9 2.7 3.0 <br />Ft. Loudoun 4.6 2.4 8.4 <br />Melton Hill 6.3 7.9 4.0