115-12 Linked Genetic Markers and Mixed Stock Analysis: When to Care and What to Do?

Christopher Habicht , Genetics Laboratory, Division of Commercial Fisheries, Alaska Department of Fish and Game, Anchorage, AK
Jim Jasper , Alaska Department of Fish and Game, Gene Conservation Laboratory, Anchorage, AK
Tyler H. Dann , Alaska Department of Fish and Game, Gene Conservation Laboratory, Anchorage, AK
Andrew W. Barclay , Alaska Department of Fish and Game, Gene Conservation Laboratory, Anchorage, AK
Lisa W. Seeb , University of Washington, Seattle, WA
William D. Templin , Alaska Department of Fish and Game, Gene Conservation Laboratory, Anchorage, AK
An increasing number of genetic markers are being developed for use in genetic studies, particularly single nucleotide polymorphisms (SNPs) for use in mixed stock analyses (MSA) of Pacific salmon (Oncorhynchus spp.).  Along with greater power to resolve stocks in such analyses, the availability of more SNPs creates a greater likelihood of encountering the non-random association of alleles among markers, a condition known as linkage disequilibrium (LD).  Markers in LD, especially those that are physically linked, may cause overly optimistic confidence intervals around estimates of stock proportions.  On the other hand, excluding all markers that hint at LD may lead to unnecessary loss of power.  As part of the effort to better understand the harvest of chum (O. keta) and sockeye (O. nerka) salmon in western Alaska, we have genotyped approximately 70,000 chum and sockeye salmon from spawning populations for 96 SNPs.  Here we present the method we developed to identify and handle linked loci in the building of baselines for use in MSA for both species.   Sets of SNPs potentially in LD were first identified in the entire dataset using both overall probabilities and the rate of occurrence.  Following this, we investigated the increase in information available from the association as opposed to using a single SNP from the linked set.  Genotypes from SNP sets in LD that were retained due to increased information were combined and used in MSA as a single phenotype.