8163 - Laserfiche WebLink

Home Browse Search

960 OSMUNDSON AND BURNHAM TABLE I.-Number of Colorado squawfish lengths available for size-frequency comparisons by year and sub- reach (lengths from 1974-1976 could not be partitioned by year). Recaptures within a given year were not includ- ed. Year 1974- Subreach 1976 1979 1982 1991 1992 1993 1994 1995 Lower reach Entire 4 37' 32' 82' 66' o 5 Lower 0 33' Middle 66' 7 Upper 0 4 Upper reach 9 21b 8b l6c l3c 23d l3b 32b 40c 37c 23d 23b 21 b ]7c 21 C a Numbers used in the analysis. b Data for 1991-1992 were pooled within subreach and used in the analysis. C Data for 1993-1994 were pooled within subreach and used in the analysis. d Middle and upper subreaches were pooled within year and used in the analysis. pled population. The RECAP program was also used to compute standard errors and confidence intervals on these estimated averages by use of a nonparametric bootstrap method. Size-frequency analyses.-To detect recent changes in age structure, size-frequency data from the Colorado River were compared with data re- ported from earlier studies (Seethaler 1978; U.S. Fish and Wildlife Service [USFWS], Grand Junc- tion, Colorado, unpublished field office data files). Because capture rates differed by subreach in the upper reach, tests for differences in size distri- . butions among the three subreaches (combined 1991-1994 data) were conducted by using the Kol- mogorov-Smirnov two-sample test (K-S test). A minimum sample of about 30 different fish was assumed necessary to develop a meaningful length-frequency histogram. Data sets were avail- able from 1974 to 1995; however, sufficient data were not available for many years, and data from 2 to 3 years were combined when necessary to obtain an adequate sample (Table 1). The K-S tests were used to identify changes in size distributions among sampling periods. Test of proportions was used to identify changes in proportions of partic- ular size-groups between sampling periods. At- tempts were made to make sample regions and gear types consistent in data comparisons between time periods. To look for evidence of recent recruitment in the upper reach, we compared pooled 1991- 1992 data with pooled 1993-1994 data for each of the three subreaches and compared middle sub- o reach data from these years with data collected there in 1995 by the USFWS. Catch rates.-Standard linear regression (SAS Institute 1985) was applied to numbers of fish cap- tured during each occasion to test for a temporal trend in fish abundance. Fits of CJS models using Pr were used as an aid in interpreting these results. Catch per unit effort (CPUE) of adults (mean num- ber of fish caught per net set) in the upper reach was also compared among years (1991-1994) as an additional means to assess temporal trends in population size. Kruskal-Wallace nonparametric analysis of variance (ANOV A) was used to test for differences in CPUE among years. Reproductive success.-Samples of age-O and larval fish were collected (1986 through 1994) to monitor annual reproductive success of Colorado squawfish and distribution of young in the middle and upper subreaches (Grand Valley) of the upper reach. This study area was subdivided into 183.2- km sections with one sample of larvae collected weekly from each section during early to mid-July through August. Samples were collected from backwater, embayment, or shoreline habitats with a 0.6-m-wide, 0.5-mm-mesh, hand seine. Sam- pling locations within each section changed week- ly, and although sampling effort was not constant among sites (5-15 min of seining per site), average effort per site was assumed constant among years. Age-O Colorado squawfish were sampled once yearly in late September or early October by sein- ing backwaters with a 4.6-m-wide, 3-mm-mesh, beach seine. The middle and upper subreaches were subdivided into 8-km sections and two same ples were collected from each of two backwaters within each section. Samples of both larval and age-O fish were pre- served in 10% formalin and sent to the Larval Fish Laboratory at Colorado State University for iden- tification. Because of the non normal distribution of fish captures, CPUE was calculated as a geo- metric mean. Kruskal-Wallis one-way ANOVA (K-W ANOVA) and Kruskal-Wallis multip1e- comparison z-value tests were used to test for dif- ferences in CPUE among years. o 53' o Results Adult Survival Rate, Abundance, and Recruitment Estimates Upper reach.-The two best models (lowest AIC values), differing by only 0.034 AIC units, provided annual survival rate estimates (<l>) of 0.845 and 0.860. The best model as determined by