An Analysis of Differential Delayed Mortality Experienced by Stream-Type Chinook Salmon of the Snake River

Mass transportation of juvenile fish in the lower Snake River was initiated in the late 1970's in an effort to reduce mortality of salmon and steelhead during downstream migration. Fish are transported in barges and trucks to below BON, thereby circumventing direct mortality due to passage through the hydroelectric projects and reservoirs. Measurement of the efficacy of smolt transportation has taken the form of studies of "T/C" (transport/control) ratios. These mark recapture studies measured the smolt-to-adult return rates (SARs) of test fish, which were transported, and control fish which were returned to the river. These studies estimated the relative effectiveness of transportation to improve survival rates of fish from the site where they were collected as juveniles back to (usually) the same site when they returned as adults. Included in this T/C ratio is any differential mortality from the collection point to the end of the hydrosystem (to BON tailrace), as well as any differential mortality from below BON to the adult recapture site(s). Although fish generally appear to survive reasonably well while in the trucks and barges, it is harder to gauge how well transported fish survive below BON, after they are released and continue their life cycle in the estuary and ocean. NMFS suggests there may be partial support for delaying a decision to breach the lower Snake River hydroelectric dams because 'D' estimates, using "improved methods provided by PIT-tag technology," appears to be high for the recent past (A-Fish). Based on these estimates, NMFS further suggest that "ongoing experiments by NMFS are likely to resolve the uncertainty regarding differential delayed transportation mortality in 5 to 10 years." Alternatively, the Plan for Analyzing and Testing Hypotheses (PATH) analyses include a larger set of T/C studies and stock recruitment data that suggests 'D' is low, which lends support to breaching of the four Snake River dams as the most robust hydro action for recovery of Snake River salmon and steelhead (Marmorek et al. 1998). In this paper, we evaluate the NMFS conclusion that 'D' is now much higher than previously thought (A-Fish), demonstrate the sensitivity of estimates of 'D' to the numerous assumptions required to make an estimate of 'D', clarify and discuss the evidence for and against various interpretations of these assumptions, and discuss the possibility of improving estimates of 'D' in the future. We note that 'D' is not a measurement. Instead, it is an indirect estimate from data and requires numerous assumptions, with many different possible interpretations. In our analysis we evaluate the effect of these different assumptions on 'D' estimates: 1) including and excluding different control and transport groups; 2) using different techniques to expand reach survival rate estimates from a shorter experimental reach to the entire migration corridor; 3) using different approaches to weight cohort reach survival rate estimates to produce seasonal estimates; 4) using different approaches to summarize experimental groups on a daily or on a weekly basis for wild fish only or wild and hatchery fish combined; and finally 5) using different approaches of pooling or averaging estimates across years. In addition, the effects of using alternative tools (passage models) to estimate reach survival rates on 'D' value estimates were evaluated.