8161 - Laserfiche WebLink

Home Browse Search

......- 692 1000 900 800 - 700 E E 600 - .r::. - Q) 500 c ~ 400 (ij - ~ 300 200 100 0 0 OSMUNDSON ET AL. fffff!!fttttftttffttft! -' fttt ,. ttt ,.' ttt " ttt I ,/ tttt l.f't t t 5 10 15 20 25 30 35 Age (years) 40 45 50 55 FIGURE 3.-Estimated mean length by age for Colorado squaw fish in the Colorado River. Lengths calculated by four methods (see Tables ], 2). Bars represent :t 1 SE. For ages where lengths were calculated by adding mean increments to mean lengths of the preceding age (ages 8-55), SE was calculated assuming a linear combination of mean increments. Additional Jines are growth curves reported by others: solid, Seethaler (1978) for the Colorado River; dash, Hawkins (1992) for the Colorado River; dot-dash, Hawkins (1992) for four rivers combined. 800 mm, growth appears to slow (Table 2). Dif- ferences in mean annual growth increments among all size-classes greater than 550 mm were, how- ever, not statistically significant. Rate differences among years (1991-1995) were not present in fish 345 mm or longer (ANCOV A, F = 0.34, df = 4, 71, P = 0.85). Relationships between growth in one year and subsequent years was not indicated for fish 400 mm or longer (X2 = 0.024, df = 1, P = 0.88), i.e., most fish did not consistently grow more or less than the average. Simulated length distributions produced mean lengths by age similar to those derived by adding mean increments (Figure 3) and indicate an ex- pected range of variation by age (Figure 4a). Growth increments were log-normal within size- classes, and simulations used log-transformed growth rates. When growth was assumed constant for fish 550 mm and longer, the rate was steady, as expected, but variance in lengths for a given age was greater than when calculated increments were used (Figure 4b). Twenty simulations that used different random-number sequences pro- duced nearly identical distributions by age. Simulations indicated broad ranges of age for ~ fi,h of ,;mil" length,_ "p<'Cially roc fi,h 550 mm lond long" (Figore 5a). lodividoa!. may tak, 10-22 (mean = 15) years to reach 600 mm; 16- 30 (mean = 25) years to reach 700 mm; and 20- 40 (mean = 32) years to reach 800 mm. Similar, but even more variable, ages were indicated for a given TL when growth of fish 550 mm and longer was assumed constant (Figure 5b). Survival Tests of assumptions.-No significant differ- ences (P > 0.05) existed among years in the TL distributions of fish 550 mm or longer captured in the upper reach 1991-1994 (N = 34, 41, 49, 34, respectively). Similarly, no significant differences (P > 0.05) were present between TL distributions for those years and for 1990 (N = 15) and 1995 (N = 44) under different sampling regimes. The TL distribution of 1982 fish (N = 41) was signif- icantly different from 1992 (P = 0.028), but was not (P > 0.05) for all other years (1990-1995), suggesting essential stability during the period. Catch rates, expressed as number of Colorado squawfish 550 mm on longer per net set, were compared for the period 1991-l994 (number of sets = 139, 117, 121, 105, respectively). No sig- nificant differences existed among years (Kruskal- Wallis one-way ANOV A, X2 = 1.916, df = 3, P = 0.590).