The scientific requirement for accurate and precise estimates of juvenile salmon abundance is well explained. Snorkeling is widely used as a juvenile salmonid census technique, especially in areas with listed species, because it does not involve handling individuals. However, in many cases there is no basis for estimating the degree to which snorkeling underestimates the actual number of fish present (it will likely always be an underestimate). This proposal outlines a study that will facilitate statistical models that allow snorkeling estimates to be corrected to provide more precise and accurate population censuses. The approach to resolving the uncertainty of the estimates appears sound. The ability to more accurately census juvenile salmonid populations is critical to status and trend monitoring, as well as estimating restoration effectiveness. This project has the potential to significantly improve monitoring accuracy by providing tools to correct snorkel estimates.

Table 1 provides a very nice summary of the uses of juvenile abundance data in management. The proposal describes its general relevance to other projects that involve snorkel estimates (there are apparently 17) and also the major monitoring efforts such as CSMEP, the NOAA Fisheries Pilot projects, and INPMEP.

Methods were thoroughly explained, especially the techniques used to construct the statistical models. This project will use ten-fold cross validation to evaluate model accuracy. It was nice to read a proposal that provided an adequate description of product quality. The sampling plan and analysis was excellent. The sequence of decision-making on the statistical analysis is the appropriate way to proceed in these circumstances. The presentation of the sampling, analysis, and decision-making is the best among other comparable systemwide proposals.

With regard to the effects of water clarity on snorkel enumeration, why not just use a turbidimeter instead of the secchi-disk method? It might be a bit less subjective.

Many of the habitat measurements described on pages 10-11 were not related explicitly to the goals of the proposal. How will this information factor into model development?