A general model for salmon run reconstruction that accounts for interception and differences in availability to harvest

Abstract

Understanding population-specific spawner–recruit relationships is necessary for sustainable salmon management. Where multiple populations are harvested together, run reconstruction methods partition mixed-stock catches and allocate recruits back to their populations of origin. Traditional run reconstruction methods often use age composition data to inform catch partitioning. However age-only methods do not account for stock-specific differences in the availability of fish to harvest within fishing areas or the incidental harvest of nontarget stocks in nearby fishing areas. Advances in molecular genetic techniques permit genetic stock identification (GSI) of both contemporary and historical catch samples. We present a statistical model for salmon run reconstruction that utilizes both age composition and GSI data to estimate differences in the availability of stocks within, and interception rates among, terminal fisheries. When applied to the commercial sockeye salmon (Oncorhynchus nerka) fishery in Bristol Bay, Alaska, new estimates of population productivity differed from those generated using previous age-only methods by 0.1%–155.1%, with stock-specific mean absolute percent differences of 9.7%–38.7% across years, underscoring the value of genetic data for run reconstruction. With more accurate run reconstruction methods, spawner–recruit relationships can be identified more precisely, thus providing more accurate management targets for salmon fisheries.