Model-based estimates of Chinook Salmon redd abundance in a large wilderness area

ABSTRACT Objective The objectives of this study were to develop a model to estimate redd abundance for Chinook Salmon Oncorhynchus tshawytscha in eight populations within a large wilderness area in the Middle Fork Salmon River basin, Idaho, without the use of manned aerial surveys. Discontinuing aerial surveys would reduce risk to staff and allow resources to be reallocated to evaluate the assumptions of historical redd surveys. Methods We developed an integrated state-space model that combined adult abundance and genetic stock identification data at a trap at a downstream dam, with long-term ground- and aerial-based spawning-ground survey data, to estimate redd abundance. Aerial survey data were then withheld from the model for each individual year of the study, and observed total redd counts, which included aerial surveys, were then compared to model-predicted estimates. Results An average of 724 redds were observed (using ground and aerial surveys) per year over the duration of the study in the Middle Fork Salmon River basin. Model-predicted estimates were similar, where an average of 719 redds were predicted when aerial survey data were withheld from the model for 1 year at a time. The accuracy at the individual-population scale varied but were relatively unbiased, where maximum bias was 2% and maximum root mean squared error was 18 redds. Conclusions Results of this study suggest that Chinook Salmon redd abundance can be estimated relatively accurately without the use of aerial surveys, thereby increasing staff safety and more efficiently using resources.