Multivariate regression relationships between ocean conditions and early marine survival of coho salmon (<i>Oncorhynchus kisutch</i>)

Abstract

Washington State hatcheries have released coded wire tagged coho salmon (Oncorhynchus kisutch) since the early 1970s and have recorded data on the number and ages of adult returns. From these data, we isolated marine survival from release to age 2 using a multinomial model incorporating the multivariate structure of the age-at-return data. Only hatcheries on or near the coast were used in the analysis in order to isolate marine effects. Using models to explore linear and curvilinear relationships, early marine survival was regressed against environmental covariates of sea surface temperature (SST), coastal upwelling, and the date of spring transitions. With hatcheries serving as replicates, curvilinear relationships were found between survival and both June SST and the date of spring transition. Maximum survival occurred for a June SST of 13.45°C and for a spring transition date around April 2. The regression results support the idea of optimal environmental conditions for coho salmon survival and may help fisheries managers better understand the dynamics of early marine survival of salmonids.