7860 - Laserfiche WebLink

Home Browse Search

18 COPEIA, 1996, NO. 1 DOUGLAS ell camp fished upriver from 1.3-7.0 km. Salt camp personnel fished from 8.0-14.9 km. Data collection.-Fishes were captured during 19 6-14 day trips at approximately monthly in- tervals from July 1991 to Dec. 1992. Hoop nets (0.76 or 1.2 m dia., 2.4 or 3.0 m length, four- or six-hoop, single- or double-throat) were de- ployed in all available habitat types of sufficient depth (i.e., > 0.4 m). Trammel nets (7.6-45.7 m length, 1.8 m depth, 1.3-3.8 cm inner and 30 cm outer meshes) were set routinely in the confluence. Fishing effort for a particular trip was recorded as number of net-hours per camp. All captured fishes were identified, measured (TL to nearest mm) and weighed (nearest g). Native species were examined for tags, mark- ings, secondary sexual characteristics, ripeness, and general health and condition. Those great- er than 150 mm TL (= adults) were injected with passive integrated transponder (PIT) tags (see Prentice et aI., 1990) and released near points of capture. Nonnative fishes were scanned for presence of PIT tags (a result of consuming tagged native fishes) then sacrificed and either dissected immediately or preserved for later study. Analytical protocol.-One-way ANOV A (Proc GLM; SAS, 1985) was used to compare total fishing effort and captures of adult G. C)Pha by reach and year. To determine movements dur- ing 1992 (which represented a full year of sam- pling), adult chubs were grouped by reach and season (winter = Dec., Jan., Feb.; spring = March, April, May; summer = June,July, Aug.; and autumn = Sept., Oct., Nov.). Numbers of G. C)pha tagged/recaptured in a given reach during a given trip were condensed into a cap- ture history (CH) matrix (Burnham et aI., 1987; Lebreton et aI., 1991). Fifty-seven matrices were derived (three camps over 19 trips). Closed population estimates.-Population esti- mates were generated from each CH-matrix un- der assumption that the three stream reaches contained closed populations. This was appro- priate given the brief sampling period at each camp (see Otis et aI., 1978) and the fact that only adults were censused. Closure was tested by examining numbers of individuals tagged within one reach then recaptured within a sec- ond reach during the same trip. Nine different closed-population estimates were derived from each CH-matrix using an updated (30 Dec. 1991) version of the computer program CAPTURE (G. C. White, D. R. Anderson, K. P. Burnham, and D. L. Otis, Los Alamos Natl. Lab., 1982, unpubl.). Models and assumptions are ex- plained in Otis et al. (1978), Pollock et a1. (1990), and Nichols (1992). The single best-fitting pop- ulation model, as indicated by goodness-of-fit tests and comparisons between competing mod- els, was retained. In this first analysis, popula- tion estimates were made relative to one anoth- er by dividing each by length of reach (in km). ANCOV A (Proc GLM; SAS Institute, Inc., 1985) then contrasted relative population esti- mates by reach, using fishing effort as a covar- iate. In a second analysis, tag/recaptures were evaluated for the entire LCR (rather than by reach). Here, 19 CH-matrices were generated, one for each month of study. Again, the single best-fitting population model was retained. ANOV A was used to test the 19 estimates against those summed by reach for each month. The hypothesis under test is that monthly estimates are not significantly different from those summed by month over reaches. Finally, a third analysis collapsed all tag/re- captures into a single CH-matrix (i.e., each col- umn of the CH-matrix represented a single month). Here, five best-fitting estimates were retained. However, assumptions of closure may be violated in this analysis by movements of G. c)pha into/from the mainstem Colorado River over the 19-month study interval and by re- cruitment of juvenile chubs into the adult pop- ulation. Thus, although this analysis is a logical culmination of population estimates by reach, by month summed over reach, and solely by month, results are heuristic rather than prac- tical. 2000 A 1000 Fig. 3. (A) Three-di and S = Salt) during Ju = Confluence; P = Po" geneous (i.e., regress ers and Jackson, 199 between fishing effor was nonsignificant. 1 in Table 1. Based uF trasts, estimated pop fluence and Salt Car cally similar, but eat than at Powell Can) fishing effort (Table Population estimates l lation estimates, star er /upper 95% confi( by reach and month timates normalized t sional plots of these Monthly population deviations and lowe limits are presented contains a summati( over reaches (as reo ANOV A comparing vs summed by mont 2) was nonsignificant 0.7; Proc GLM, SA~ plot of monthly vs su estimates is provided est estimates were ree vs 5390; Appendix 2 Dee. (745 vs 1285). were for April (555: (interestingly enoug 408). A Dec. sampli celled due to incler niques indicated ele' RESULTS Fishing effort and unadjusted population esti- mates.-Fishing effort differed significantly among reaches (F = 6.40; P < 0.0035; Proc GLM; SAS Institute, Ine., 1985), with effort at Salt Canyon greater than that at Confluence (Sidak's multiple range test; SAS Institute, Ine., 1985). However, efforts at Salt and Powell Can- yon reaches were statistically similar. Popula- tion estimates (normalized by river km) also dif- fered significantly among reaches, with greatest overall values at Confluence (F = 4.19; P < 0.01; SAS Institute, Inc., 1985). Analysis oj covariance.-Differences in normal- ized population estimates could result from in- creased effort. To test estimates with fishing effort fixed, we first evaluated two specifica- tions: (1) that slopes of the between-camp re- gressions of population vs effort were homo-