Factors Causing Observed Trends
Selected ecosystem metrics associated with SEAK adult pink harvest over the 16-year SECM time series are shown in Figure 2 below. The ranges of values below each metric are color-coded, with the highest values in green, intermediate values in yellow, and the lowest values in red. Metrics to the right of the response variable column for SEAK pink harvest are ordered by declining correlation and significance (increasing “P-value” = declining significance); the corresponding correlation coefficient “r” and “P-value” are shown below each metric. Note that in addition to CPUE, four other variables are significantly correlated with harvest (Peak migration month, North Pacific Index (NPI), %pink in June-July trawl hauls, and the ADF&G Escapement Index) and suggest an intermediate pink harvest in 2013. Additionally, this matrix shows that anomalously low (red: 2000, 2006, 2008, 2012) or high (green: 1999, 2001, 2005, 2011) return years always flag 3-5 ecosystem indicators of the respective color signal in each row. For the 2013 forecast, however, no “red” ecosystem indicators were flagged. The Icy Strait temperature index (ISTI) shown in the last column is not significantly correlated with harvest, but is an important secondary parameter to explain the error in the CPUE and harvest regression model. More details about the SECM pink salmon forecasts are available on the SECM pink salmon forecast web page.
Figure 2. Matrix of ecosystem metrics considered for pink salmon forecasting, data sources include: the Alaska Department of Fish and Game (A. Piston), NOAA (SECM/Auke Creek-J. Joyce), and Climate & Global Dynamics (J. Hurrell, http://www.cgd.ucar.edu/cas/jhurrell/indices.data.html).
Additional evidence from SECM research and other biological or ecosystem indicators suggests a strong pink salmon harvest in SEAK of 53.8 M fish in 2013. The strongest indicator for this favorable forecast is the 2012 peak juvenile pink salmon CPUE, which was the fourth highest on record. Other ecosystem indicators in 2012 that were significantly correlated (P < 0.05) with SEAK pink salmon harvest (1998-2012) were 1) a favorable July month of peak seaward migration; 2) a high North Pacific Index (NPI = 16.7); and 3) a high average percentage of pink salmon (40%) caught among juveniles in June-July trawl hauls. Less favorable ecosystem indicators were a below average ADF&G escapement index for the pink salmon parent year (2011) in SEAK and a below average wild fry production in Auke Creek (2012). An additional indicator favoring a good harvest in 2013 was the ocean catch rates of juvenile pink salmon from a research survey downstream from the SECM project, the Gulf of Alaska Integrated Research Project (GOAIRP) conducted offshore of Baranof and Chichagof Islands both west and south of Icy Strait. Compared to the SECM surveys, pink salmon catch data from this project may better represent southern and coastal SEAK pink salmon stocks, and higher juvenile pink catches in 2012 than in 2011 suggest a higher harvest of these stocks in 2013 than in 2012.
Given the ecosystem conditions and SECM metrics sampled in 2012, the two best SECM forecast models for the 2013 SEAK pink salmon harvest are shown below in Table 1. Each forecast model value has an 80% bootstrap confidence interval shown in parentheses. The 2-parameter model is the best fit predictor for the relationship of the 16-year time series of SECM data parameters with subsequent SEAK pink salmon harvests from 1998 to 2012, based on the R2 and AICc.
Table 1. The two best SECM pink salmon forecast models for the 2013 SEAK harvest.