Need for Replicated Electrophoretic Analyses in Multiagency Genetic Stock Identification (GSI) Programs: Examples from a Pink Salmon (Oncorhynchus gorbuscha) GSI Fisheries Study

Genetic stock identification (GSI) studies usually assume that electrophoretic data from different laboratories are accurate, precise, and comparable. We compared the genetic scores from replicate (identical) collections of pink salmon (Oncorhynchus gorbuscha) tissues that were analyzed in a "blind" manner by three different laboratories; scores at 15 genetic systems (18 loci) were compared for 1248 fish. The scores from laboratories A and B consistently agreed. Their average discrepancy rate across all loci was 0.21%, which was approximately 18 times lower than the discrepancy rate of laboratory C (3.86%). Laboratory C also reported data with higher numbers of discrepancies (P < 0.01) in their scores of variant genotypes (non-*100/100) than the other two laboratories. Stock composition estimates using maximum likelihood analysis of the data from laboratories A and B differed by approximately 1% by stock group, while those of laboratory C differed by an average of 8% by stock group from the estimates of the other two laboratories. Our findings demonstrate that different laboratories can, but do not always, produce results that are consistent. Because many GSI programs in North America rely on data contributed by several laboratories, we recommend that testing of replicate samples be incorporated into all multiagency GSI studies.