Cooperative Fish and Wildlife Research Units Program: all
Education, Research and Technical Assistance for Managing Our Natural Resources

Snorkel surveys are frequently used to monitor stream-dwelling fish. Inferring total abundance from snorkel surveys is complicated by two primary factors: the snorkelers’ fish detection probability and the relative abundance of fish in habitat types where snorkel counts can and cannot be conducted. We examined these factors across three salmonid species (Oncorhynchus spp.), 4 years, and 113 location-years distributed randomly across the Oregon coast. We used Bayesian data augmentation techniques to integrate snorkel counts into the fitting of mark-recapture models in a unified analysis of all location-years for each species. We explored several mark-recapture model formulations. We developed mathematically explicit expressions that convert a new snorkel count into a probability density of abundance in a generalized habitat unit where snorkeling cannot be done. Snorkelers detected 63%, 33%, and 40% of juvenile coho, cutthroat trout, and steelhead [CR1] [FM(2] estimated by mark-recapture, respectively, but uncertainty within and among sampling units was large. Coho abundance in pools where snorkeling is possible was 3.3 times greater than abundances in fast-water units where snorkeling is not possible. Cutthroat and steelhead had more equitable abundances between the two stream types, and abundance estimates were more uncertain because of low sample size. We did not find significant evidence for density-dependent habitat selection between pool and fast-water stream units for any of the three species. The proportion of juvenile salmonids seen by snorkelers generally declined with increasing abundance for coho and cutthroat trout but not necessarily steelhead. Our quantification of uncertainty arising from using snorkel counts as a proxy for abundance will help managers make decisions that balance monitoring cost and biological risk.