Modeling the Survival of Chinook Salmon Smolts Outmigrating Through the Lower Sacramento River System

To study the factors associated with the freshwater mortality of outmigrating chinook salmon, releases of tagged juvenile salmon were made at multiple locations in the Sacramento River each spring between the years 1979 and 1995. A midwater trawl located downstream of the release sites caught salmon soon after release and, 1 to 4 years later, samples taken from the catches of marine fisheries recovered other tagged fish. An extended quasi-likelihood model was fit to both the freshwater and the marine recoveries. A ridge parameter was included to stabilize the parameter estimates and to improve predictive ability. Overdispersion was due, at least in part, to heterogeneity in the trawl’s capture efficiency, as well as to the complex aggregation of marine recoveries. Different dispersion parameters were used for the river and ocean recoveries because of the additional sources of variation experienced by ocean recoveries relative to river recoveries. Interpretation of estimated coefficients was delicate, given the correlation between some of the covariates, the biases introduced by the ridge parameter, and possible confounding factors. With these caveats in mind, we found the most influential covariate to be the temperature of the water into which the fish were released, with increasing temperatures having a negative association with recoveries. Three covariates were of particular interest to the biologists and water managers: water flow, position of a water diversion gate (open or closed) separating the mainstem from the central delta, and relative fraction of water exported for irrigation and urban consumption. The effects of flow were slightly positive but were confounded by salinity levels. The effect of the water diversion gate being open was to lower apparent survival for fish released above the gate, but apparent survival increased for fish released in the central delta into which the water was diverted. There was evidence that increasing the export-to-inflow ratio lowered survival, but the effect was slight and not statistically significant.