7860 - Laserfiche WebLink

Home Browse Search

DOUGLAS AND MARSH-ESTIMA TESjMOVEMENTS OF GILA CYPHA 19 2000 A 1000 2000 1000 FJan. Ma~fj) A~.ar.( May June Jut Aug. Sap. Oct. Nov. Fig. 3. (A) Three-dimensional plot of population estimates by reach (where C = Confluence; P = Powell; and 5 = Salt) during July-Dec. 1991. (B) Three-dimensional plot of population estimates by reach (where C = Confluence; P = Powell; and S = Salt) during Jan.-Nov. 1992. geneous (i.e., regression lines parallel; see Som- ers and Jackson, 1993), and (2) that interaction between fishing effort and population estimates was nonsignificant. The resulting ANCOV A is in Table 1. Based upon a priori statistical con- trasts, estimated populations within both Con- fluence and Salt Canyon reaches were statisti- cally similar, but each was significantly larger than at Powell Canyon reach, irrespective of fishing effort (Table 1). Population estimates by reach and river.-Popu- lation estimates, standard deviations, and low- er jupper 95% confidence limits are presented by reach and month in Appendix 1, as are es- timates normalized by river km. Three-dimen- sional plots of these estimates are in Figure 3. Monthly population estimates, with standard deviations and lower jupper 95% confidence limits are presented in Appendix 2, which also contains a summation of estimates by month over reaches (as recorded in Appendix I). An ANOV A comparing these estimates (monthly vs summed by month over reaches; Appendix 2) was nonsignificant (F = 1.15; df = 1,36; P > 0.7; Proc GLM, SAS Institute, Inc., 1985). A plot of monthly vs summed monthly population estimates is provided in Figure 4. In 1991, high- est estimates were recorded for early Aug. (3 I 57 vs 5390; Appendix 2), whereas lowest were for Dee. (745 vs 1285). In 1992, highest estimates were for April (5555 vs 5683), whereas lowest (interestingly enough) were for Aug. (635 vs 408). A Dec. sampling trip in 1992 was can- celled due to inclement weather. Both tech- nigues indicated elevated population estimates from early March through June of 1992 (Fig. 4). Both years demonstrated an upswing in es- timated population size in autumn. Average monthly estimate summed over reaches was larger (but not significantly so) than that cal- culated by month (2993 vs 2434; n = 19; Sidak's multiple range test; SAS Institute, Inc., 1985). Five best-fitting population estimates were re- tained from analysis of a CH-matrix that in- cluded all 19 months of the study (Table 2). The highest criterion (0.61) was Pollock and Otto's estimator (Mbh), which assumes that cap- ture probabilities vary by individual animal and by behavioral response to capture (i.e., behavior and heterogeneity effects; Otis et aI., 1978:40- TABLE 1. POPULATION ESTIMATES (NUMBER/RIVER KILOMETER) OF ADULT Gila C)'pha (> 150 mm TL) IN THREE REACHES OF THE LITTLE COLORADO RIVER (CONFLUENCE, POWELL, SALT) FROM JULY 1991 THROUGH DECEMBER 1992. Estimates were adjusted before analysis for length of reach (in km). Logl. fish- ing effort was used as ANCOV A covariate. Diagonal elements represent average least-squares population estimates (adjusted for log.. fishing effort) and have been converted from logl. values. Upper triangular cells represent F-values for pairwise a priori contrasts. Confluence Powell Salt Confluence Powell Salt 6.2' 110< 0.2b 4.3' 222 263 'P < 0.016. b P > 0.657, < F = 4.34; P < 0.019; df - 3,48. . P < 0.044.