Genetics Provide a Forty-Five Year Retrospective of Sockeye Salmon Harvest Compositions in Bristol Bay, Alaska

The Bristol Bay sockeye salmon fishery is an intensively managed fishery that supports the largest sockeye salmon harvests in the world.  The fishery is divided into five discrete fishing districts (Egegik, Naknek-Kvichak, Nushagak, Togiak, and Ugashik) that are designed as terminal fisheries which target stocks destined for major local drainages.  The harvest of some non-local stocks from out-of-district drainages is inevitable.  However, the magnitude of this harvest is uncertain.  Estimates of stock-specific recruitment are essential to the development of biologically sustainable escapement goals designed to ensure the long-term stability of the Bristol Bay stock complex.  Accurate estimates of harvest composition are particularly important when determining stock specific recruitment in Bristol Bay, as sockeye salmon stocks can be exploited at rates up to 80%.  We used archived sockeye salmon scales from district harvests that were genotyped at 45 single nucleotide polymorphism (SNP) markers to estimate stock composition of harvests from a selection of 48 district/year strata dating from 1964 to 2005. Our specific objectives were to (i) estimate historical stock-specific harvests within Bristol Bay fishing districts, (ii) evaluate stock-specific vulnerabilities within fishing districts, and (iii) compare genetics-based harvest estimates to harvest estimates generated using the traditional age composition harvest allocation method. Genetics-based estimates showed that the majority of the sockeye salmon harvested in each district/year stratum were local stocks originating from watersheds within their respective districts, and harvests of sockeye salmon from non-local stocks typically originated from watersheds in adjacent districts. Assuming the genetics-based estimates are correct, our results demonstrate that the assumptions made under the traditional age composition allocation method can lead to large harvest misallocations for some stocks, with overestimates of total run as high as 164% for the Igushik stock and underestimates as high as 72% for the Alagnak stock. Our results suggest that traditional harvest allocation methods can be improved by incorporating genetics-based estimates, thereby improving run reconstruction models by identifying differences in vulnerability among stocks in mixed-stock fisheries.  Historical harvest composition estimates will improve accuracy in brood tables ultimately providing for continued sustainability and economic viability that will benefit local communities and the fishing industry.