Estimates of Survival:

Methods

Survival is estimated from release to first detection site, and between series of dams, by the Cormack-Jolly-Seber (CJS) release-recapture method outlined in American Fisheries Society Monograph 5, Design and analysis methods for fish survival experiments based on release-recapture, by K.P. Burnham, D.R. Anderson, G.C. White, C. Brownie, and K.H. Pollock, 1987. This methodology is used to estimate survivals both to and between the dams in the hydro system possessing PIT tag detection capabilities, along with an estimate of collection efficiency at these dams. The CJS method is based on mark release-recapture theory in which the subsequent detection histories on a known number of marked fish re-released at a particular dam is used to estimate the number of fish that past that particular dam alive but undetected. The software program MARK (White and Burnham 1999) was used to perform the survival estimates with the “identity “design matrix and “identity” link function set. The program MARK provides estimates of survival between the tailraces of each detection site. Generating extended multi-dam reach survival estimates requires taking the product of a set of these shorter reach estimates. The associated variance for the extended reach estimate is computed using formulas for propagation of error in products of non-independent estimates (Meyer 1975). Extended reach survival estimates with associated 95% confidence intervals are obtained for each species, and release location and period of interest.

Sets of survival estimates are computed each year for various river reaches. In the Snake River basin, estimates of survival are made from key hatcheries to John Day Dam tailrace, from SMP traps to Lower Monumental Dam tailrace, and from Lower Granite Dam tailrace to McNary Dam tailrace. In the Columbia River basin, estimates of survival are made from key hatcheries to McNary Dam tailrace, from Rock Island Dam (release site) to McNary Dam tailrace, and from McNary Dam tailrace to John Day Dam tailrace, and specifically for yearling chinook from McNary Dam tailrace to Bonneville Dam tailrace utilizing the NMFS trawl in the lower Columbia River as the final detection site. The goal is to have at least 600 PIT tagged smolts released (or detected and re-released at the starting site) in each group for which survival estimates are desired. Generally, release period of a week are attempted, but in some instances release periods of up to 15 consecutive days was required in order to try to achieve the target release size. Detailed results for the individual release (or detected and released) groups of interest are present in Appendix F.

For SMP traps and key dams, the 2002 survival data is summarized to annual averages for comparison to recent past years. A single seasonal average survival estimates is obtained for those PIT tagged groups released over time that do not differ significantly. To determine any significant differences occurred within a year, a test of whether the “between group” variance component was significantly greater than zero (Burnham 1987 et al., Chapter 4). This is a chi-square test equal to [empirical variance of mean survival*(1-degrees of freedom)]/ [theoretical variance of mean survival]. In cases where the chi-square test was not significant at the 95% confidence level, then the average was computed for the season, along with the average theoretical variance. In cases where the chi-square test was significant, then the season was split into periods showing the different survival levels.