Preseason Forecasts of Sockeye Salmon (Oncorhynchus nerka): Comparison of Methods and Economic Considerations

Abstract

We compared the ability of two methods to forecast observed abundances of sockeye salmon (Oncorhynchus nerka) in Bristol Bay, Alaska, stocks from 1956 through 1980. We determined that the commonly used linear forecasting method, based on untransformed abundances of sibling age classes, overestimated sockeye abundances more frequently than did a nonlinear method. Furthermore, the nonlinear method produced mean forecasts that were closer to actual abundances of returns than forecasts from the linear method. We therefore recommend use of the nonlinear method. We also quantified significant economic implications of incorrect preseason forecasts for Bristol Bay sockeye. Deviations of actual escapements from the target are correlated with errors in the preseason forecast. Thus, large errors in preseason forecasts can influence potential long-term harvests and resulting economic benefits. In addition, annual changes in prices of set and drift gill net permits for fishermen were correlated with forecasted annual change in the abundance of sockeye recruits. This result is consistent with the hypothesis that preseason forecasts of sockeye salmon returns affect expectations by fishermen of economic rents (related to profits). Overestimates of abundance of recruits by current methods may thus subsequently lead to overinvestment by fishermen seeking to increase their share of expected rents.