Nonstationary effects of ocean temperature on Pacific salmon productivity

We tested the hypothesis that ocean temperature effects on productivity for northeast Pacific pink (Oncorhynchus gorbuscha), sockeye (Oncorhynchus nerka), and chum salmon (Oncorhynchus keta) changed after 1988-1989, coincident with a decline in Aleutian Low variance. Nonstationary temperature effects were tested with three different analytical methods (correlation, mixed-effects models, and variable coefficient generalized additive models) applied to spawner-recruit time series from 86 wild runs between Puget Sound and the northern Bering Sea. All three methods supported the hypothesis, with evidence for change in temperature effects that was strongest in the Gulf of Alaska, British Columbia, and Washington and weakest in the Bering Sea. Productivity for all three species showed generally positive responses to ocean temperature in Alaska before 1988-1989, but generally neutral responses after 1988-1989. British Columbia and Washington salmon showed either neutral responses to temperature (pink), negative responses that weakened after 1988-1989 (sockeye), or a switch from neutral to negative responses (chum). We conclude that the inverse response of Alaskan and more southern salmon populations to temperature variability is a time-dependent phenomenon.