7206 - Laserfiche WebLink

Home Browse Search

POPULATION SIZE OF RAZORBACK SUCKER 69 data. In this model, all fish have the same capture probability in a given year (t), and past capture history of a fish does not influence its subsequent capture probability (White et al. 1982; the Peter- sen-Lincoln estimator for t = 2 and the Schnabel estimator for t > 2 approximate the maximum likelihood estimator used for model M, in CAP- TURE). Program selection of model M, was ex- pected because sampling effort varied each year, we did not believe capture altered fish behavior in subsequent years, and we know of no apparent biological reason why each fish would have a unique probability of capture. Results and Discussion the data, and selects the "best" model. The eight models represent all possible combinations of three factors affecting capture probabilities: time, be- havioral response to capture, and innate hetero- geneity (i.e., variations among individuals in cap- ture probabilities; White et al. 1982). We considered each year as a separate sampling occasion in the data matrix. Each fish was tagged with a uniquely numbered Carlin dangler tag (Wy- doski and Emery 1983) so that capture history of individuals would be known. Daia from 1983 were not used as part of the population estimation data matrix because sampling effort was greatly cur- tailed that year due to flood conditions, and no razorback suckers were caught in the upper Green River. Recaptures of fish originally tagged by other workers from point samples were considered as original captures in the data matrix. Fish removed from the population were not included in the data matrix. Model selection. -The model-selection proce- dure of the CAPTURE program identified "M„" a model allowing time specific changes in capture probability, as the most appropriate model for our Population Estimation Razorback suckers were found in the upper (km 282-555) and lower (lan 0-211) Green River. None were captured in Gray and lower Desolation can- yons (km 211-282). Altogether, 423 adult razor- back suckers were captured, of which 69 were re- captures. These fish ranged from 426 to 617 mm total length (Figure 2). Catch per unit effort (CPUE; fish/hour of electrofishing) averaged 0.49 for the upper and 0.05 for the lower Green River (Table 1). The razorback sucker population (N) in the up- per Green River was estimated at 948 fish (SE, 97; 95% confidence interval, 758-1,138). The yearly probability of capture (P) varied from 0.02 to 0.09 (Table 2). The coefficient of variation of N (100 SE/mean) was 10.2%, indicating good precision. White et al. (1982) suggested that reliable scientific studies require a coefficient of variation of less than 20%. A way to further increase the reliability of the population estimate would be to increase the sampling effort in future years so that P also increases (White et al. 1982). The number of captures of razorback suckers in the lower Green River was so small (13 fish from 1980 to 1988) that we felt it impractical to estimate that population. Tyus (1987) indicated that ra- zorback sucker density in spring samples from the lower Green River was about one order of mag- nitude less than that in the upper Green River. Our catch-effort data (Table 1) also suggested low numbers of fish. We believe the population in the lower Green River is so small it is relict. Further evidence of razorback sucker scarcity in the Green River basin was reported by Tyus (1987) for 1979-1986; the numbers of razorback suckers in the total catch and taken by spring electrofishing (when they are most vulnerable to sampling) were FIGURE 1.-Location of Green River sampling areas for razorback sucker, 1980-1987.