Maximum Likelihood Estimation of the Proportion of Hatchery-Origin Juvenile Migrants at Lower Granite Dam Using Visible Marks, Coded-Wire Tag, and PIT-Tag Data

For the Snake River Fall Chinook salmon Oncorhynchus tshawytscha population, estimates of the proportion of hatchery-origin juvenile outmigrants are needed to assess population status. Some hatchery-origin fish released above Lower Granite Dam were given visible marks, coded-wire tags, and/or PIT-tags, and a fraction were collected in the bypass system sample tank of the dam. At certain times, a separation-by-code system selectively routed PIT-tagged fish away from the sample tank, thus biasing the sample tank count. This problem was solved by estimating a separate sample rate for PIT-tagged fish programmed not to enter the sample tank. We derived daily and seasonal maximum likelihood estimates of release-specific proportions of hatchery-origin juvenile outmigrants that use the bypass system at Lower Granite Dam, 1995-2014. Precision and accuracy in the estimates increased with the expected number of PIT-tag detections in the bypass system, and the expected numbers of visibly marked or CWT fish observed in the sample tank. During certain times of the year, small samples yielded daily estimates that were highly uncertain, inaccurate, or impossible to compute. When this occurred, we used weekly estimates to increase sample size. We discussed using a Bayesian method as an alternative to maximum likelihood.