• climate;
  • coho salmon;
  • General Additive Model;
  • Oncorhynchus kisutch;
  • Oregon;
  • survival


To better understand and predict Oregon coho (Oncorhynchus kisutch) marine survival, we developed a conceptual model of processes occurring during four sequential periods: (1) winter climate prior to smolt migration from freshwater to ocean, (2) spring transition from winter downwelling to spring/summer upwelling, (3) the spring upwelling season and (4) winter ocean conditions near the end of the maturing coho's first year at sea. We then parameterized a General Additive Model (GAM) with Oregon Production Index (OPI) coho smolt-to-adult survival estimates from 1970 to 2001 and environmental data representing processes occurring during each period (presmolt winter SST, spring transition date, spring sea level, and post-smolt winter SST). The model explained a high and significant proportion of the variation in coho survival (R2 = 0.75). The model forecast of 2002 adult survival rate ranged from 4 to 8%. Our forecast was higher than predictions based on the return of precocious males (‘jacks’), and it won't be known until fall 2002 which forecast is most accurate. An advantage to our environmentally based predictive model is the potential for linkages with predictive climate models, which might allow for forecasts more than 1 year in advance. Relationships between the environmental variables in the GAM and others (such as the North Pacific Index and water column stratification) provided insight into the processes driving production in the Pacific Northwest coastal ocean. Thus, coho may be a bellwether for the coastal environment and models such as ours may apply to populations of other species in this habitat.