9724 - Laserfiche WebLink

Home Browse Search

1.0 LITTLE COLORADO RIVER HUMPBACK CHUB 237 ~ 0.8 ~ 0.6 = o ';: 0.4 o c. e 0.2 0.. Cl:!. ...fb.. .:.. c I;l~ . .:p 0$ .. .... 0'. aD [J. .. ~.\; . oJ 0 ,"" . " .. " . . . " . " , , 0.0 ,of ,~' ,~'" ,0,0,"> ,~... ,0,0,<-' ~o,'" ,~" ,0,0,0" ,o,~ ",#' -f'''' ",<S>'" Year FIGURE L-Observed monthly mark rates of adult (age-4 and older) humpback chub captured in the Little Colorado River (diamonds) and the Little Colorado River inflow reach of the Colorado River (squares). on the model parameter of interest. We again evaluated model fit using AlCc (Table 2). The annual ASMR model.-We used the ASMR model (Coggins et al. 2006) to estimate the age-specific mortality rate, the age- and time-specific capture probabilities, adult abundance, and recruit (age-2) abundance. Like the traditional Jolly-Seber type models, the annual ASMR model uses data aggregated within years and ignores multiple within-year recapture information. Three formulations (ASMR 1-3) of the overall model structure were used to explore alter- natives in estimating time- and age-specific capture probabilities. We used a Markov chain-Monte Carlo algorithm to assess parameter uncertainty for each model formulation (Coggins et al. 2006). Multistate movement models.-The ASMR and traditional Jolly-Seber models do not explicitly represent seasonal spawning movements between the LCR and the LCR inflow reach. Because sampling has historically been concentrated in the LCR (there were over 12,500 recapture events in the LCR versus about 1,100 in the LCR inflow reach), it is possible that the capture probabilities calCulated from spatially aggre- gated data could lead to misleading population estimates (e.g., underestimates of the numbers of adult fish that are seasonally resident in the LCR inflow reach). To evaluate this possibility, we developed a spatially explicit model for individual recapture histories (Appendix), treating fish as being in two possible location states (the LCR or the LCR inflow reach) in each sampling month. The binomial likeli- hood function for this model was combined with the Poisson likelihood function for unmarked fish from the ASMR model (Coggins et al. 2006) to provide an overall model for the dynamics of both marked and unmarked fish. Our goal was to estimate age-specific mortality rates, monthly movement rates to and from the LCR, monthly capture probabilities in both locations, and the abundance of llIIl11aIked fish over time by maximizing this combined likelihood function. ", We also evaluated multistate movement models (Brownie et al. 1993; Hestbeck et al. 1991) in MARK that allow animals to move between strata and provide a probability of transitioning between strata. Results High mark rates (>80%) were achieved in the mid- 1990s (Figure 1). Examination of the monthly mark rates among recapture locations (LCR versus LCR inflow reach) indicates that fish mix and move between the two areas, as there are similar patterns in overall mark rate with large changes in sampling intensity. Index-Based Assessments Long-term catch rate indices from the standardized hoop and trammel net sampling show 2-3-fold declines in catch rate from the late 1980s to the present (Figure 2). Although all monthly trammel net samples from the LCR inflow reach for 1990-2003 are presented, only samples from 1990-1993 and 200 1 represent robust 0.3 2 ~ 0.25 ~ 0.2 ~ 0.15 2: 0.1 a. ~ 0.05 o ~~#~~#~~~~~~~~~#\$ ~ J51-I99mmTL --$->200mmTL Year : 1.2 ~ \ B Ll 0.8 Z 0.6 ~ 0.4 E E f- . . " " ~.:~. -.".. ._~~. - _. h . . . '. . . . 0.2 o ~rf,> ~o,' ~o,'" ~o," ,~... ~o,<-' ,~'" ~o," ,~o" ~~ ",#' ",<S>' ",<S>'" -f''''' ",# . . " . Year FIGURE 2.-AnnuaI mean humpback chub caleh rate using hoop nets (top panel; caleh}h, with 95% confidence intervals) in the lower 1,200 m of the Little Colorado River (LCR) and mean monthly humpback chub (1L >200 rom) catch rate using trammel nets (bonom panel; catch .h-1.1oo m-I) in the Little Colorado River inflow reach (LCR inflow reach) of the Colorado River. In the bottom panel, diamonds represent the extensive sampling that took place in 1990--1993 and 200t and that was well distributed throughout the LCR inflow reach. Squares represent the limited sampling that took place in 1994-2000 and 2002-2003 and that, in some years, was concentrated near the confluence of the LCR. Though both regression lines (dashed, 1990--2003; solid, 1990--1993 and 2001) suggest a declining trend in abundance, the slope of neither regression is significantly different from zero (P = 0.14 for both regressions).