Influence of stream temperature and human disturbance on prespawn mortality of Chinook Salmon in the Willamette River basin

Abstract Objective Premature mortality of adult female Chinook Salmon Oncorhynchus tshawytscha is a major barrier to population recovery. The Willamette River basin, Oregon, typifies the problems that are faced by fishery managers in the Pacific Northwest (USA). Adult salmon are trapped and transported upstream of dams to access historical spawning grounds, but annual rates of prespawn mortality (PSM) are high (often >40%) and may limit the recovery of natural populations. The purpose of this study was to identify potential factors related to PSM of female Chinook Salmon that are outplanted above dams and incorporate them into a modeling framework to facilitate adaptive management of outplanting operations. Methods We evaluated PSM in Fall Creek of the Willamette River basin prior to transport facility improvements in summer and fall of 2010–2017 and postimprovement during 2020–2021. We estimated PSM and conducted exploratory analyses to identify possible nontransport sources of stress that may contribute to the observed high PSM rates. Candidate factors included long-term elevated temperature exposure, elevated temperature exposure below the trap, total number of outplanted fish, and monthly human disturbance of outplanted fish. We then developed and fit three models, each representing a hypothesis of a factor influencing PSM, incorporated them into a single alternative decision model, and conducted sensitivity analyses. Result Prespawn mortality averaged 0.66 (ranging from 0.37 to 0.94) over the study period. According to the simulation results, the top two management actions were to exclude human activities—swimming and fishing—from Fall Creek in July and August. Conclusion Expected PSM rates were predicted to be 0.38 when human activity was excluded in July and 0.37 for August. Sensitivity analyses indicated that the most influential decision model component was the choice of the alternative model.