Laserfiche WebLink
also differed significantly among reaches, with greatest overall <br />va, as at Confluence (F = 4.19; P < 0.01; SAS, 1985). <br />P-. lation estimates adjusted for fishing effort.---Differences <br />i:-. rmalized population estimates could result from increased <br />e::- --t. To test our estimates with fishing effort fixed, we first <br />::ted two specifications: (a) that slopes of the between-camp <br />regressions of population estimate vs effort were homogeneous <br />(i.e., regression lines parallel; see Somers and Jackson, 1993), <br />and (b) th-t interaction between fishing effort and population <br />estimates was nonsignificant. Based upon a priori statistical <br />contrasts, estimated populations in Confluence and Salt Canyon <br />reaches were similar, but each was significantly larger than at <br />Powell Canyon reach, irrespective of fishing effort (Table 1--1). <br />Population estimates for each reach and for the entire river.--- <br />Three-dimensional plots of adjusted population estimates by reach <br />are presented in Figure 1--3. Powell Canyon exhibited lower <br />e,-•timates than either Confluence or Salt Canyon, particularly in <br />1991 (Fig. 1--3a). In 1992 (Fig. 1--3b), increased activity at <br />confl%enc- during early March was reflected in elevated estimates <br />at `'owell in late March-April, followed by elevated estimates at <br />Canyon April-t}_ourh-June. Estimates at confluence again <br />increased in Ar-ii-June. Raw population estimates by reach per <br />sampling periou, and estimates normalized by river km, are in <br />Appendix 1--1. Raw population estimates per sampling period for <br />the entire LCR are presented in Appendix 1--2. This appendix also <br />contains a second population estimate for the entire river, <br />derived by summing estimates calculated over reaches for each <br />sampling period (as recorded in Appendix 1--1). An ANOVA <br />comparing these two estimates for the entire LCR (i.e., monthly <br />vs summed by month over reaches) was nonsignificant (F = 1.15; df <br />= 1,36; P > 0.7; Proc GLM, SAS, 1985). Both estimates are plotted <br />in Figure 1--4. In 1991, highest estimates were recorded for <br />early August (3157 vs 5390), while lowest were for December (745 <br />vs 1285) (Fig. 1--4). In 1992, highest estimates were for April <br />(5555 vs 5683), while lowest were for August (635 vs 408). A <br />December sampling trip in 1992 was cancelled due to inclement <br />weather. Both techniques indicated elevated population estimates <br />from early March through June of 1992 (Fig. 1--4). Also, both <br />techniques demonstrated an upswing in estimated population size <br />in autumn of both years. The average estimate summed over reaches <br />was larger (but not significantly so) than that calculated by <br />month (2993 vs 2434; N = 19; Sidak's multiple range test; SAS, <br />1985). <br />Five best-fitting population estimates were retained from <br />analysis of a CH-matrix for the entire 19 month study (Table 1-- <br />2). The highest criterion (0.61) was Pollock and Otto's estimator <br />(Mbh), which assumes that capture probabilities vary by individual <br />animal and by behavioral response to capture (i.e., behavior and <br />5