Forecasting Pink Salmon Abundance in Southeast Alaska From Juvenile Salmon Abundance and Associated Environmental Parameters

November 27, 2009

The Southeast Alaska Coastal Monitoring (SECM) project has been sampling juvenile salmon (Oncorhynchus spp.) and associated environmental parameters in northern Southeast Alaska (SEAK) since 1997 to better understand effects of environmental change on salmon production. A pragmatic application of this effort is to forecast the abundance of adult salmon returns in subsequent years. Since 2004, juvenile peak salmon catch per unit effort (CPUE) from SECM, modified by other environmental parameters as appropriate, has been used to forecast harvest of adult pink salmon (O. gorbuscha) in SEAK. The forecast of 16.1 million fish for 2008 was within 2% of the actual harvest of 15.9 million fish. This represents the fourth forecast over the period 2004-2008 which was within 11% of the actual harvest. In 2006, however, the harvest was substantially different from the forecast. Although a simple CPUE model did indicate a downturn in harvest for 2006, the forecast was nonetheless 200% higher than the actual 2006 harvest. These results show that the CPUE information has great utility for forecasting year class strength of SEAK pink salmon, but additional environmental data are needed to avoid “misses” such as the 2006 return. Since 2007, the forecast model was developed using stepwise multiple regression, jackknife hindcast analysis, and bootstrap confidence intervals. A four-parameter model was selected as the “best” forecast model for 2009. Juvenile pink salmon CPUE in northern SEAK accounted for 82% of the variability in annual harvest of SEAK pink salmon over the 1997-2008 time period. The amount of variability explained was improved to 99% when the May 20-m integrated sea water temperatures and mixed-layer depths (from the SECM strait habitat) and the El Niño Southern Oscillation (ENSO) Index were included in the four-parameter multiple regression model. The 2009 forecast from the four-parameter model, using data collected in 2008, is 44.4 million fish, with an 80% bootstrap confidence interval of 37-52 million fish. Juvenile pink salmon peak CPUE collected in southern SEAK from 2005-2007 was highly correlated (r = 0.99) with the peak CPUE from northern SEAK in those years, and was not correlated (r < 0.01) with the residuals from the forecast model. Because the pattern of juvenile abundance was similar for the two areas, no additional variation in the harvest was explained by including the southern region data. However, that time series includes only three years of data; more years may provide additional information on regional variation in pink salmon year-class strength, especially for years when the two areas have distinctly different environmental conditions.