SNP panels for differentiating advanced-generation hybrid classes in recently diverged stocks: A sensitivity analysis to inform monitoring of sockeye salmon re-stocking programs

Molecular markers have been employed for monitoring fisheries re-stocking initiatives for decades, however, as such programs mature, traditional genetic markers may be ineffective at distinguishing more advanced introgression classes (e.g. backcrosses and F2 hybrids). Genome-wide single nucleotide polymorphisms (SNPs) can enable more accurate and cost-effective monitoring, but there is a need to evaluate the number of markers relative to their information content, especially in cases where native and re-introduced stocks have only recently diverged. This is the case with the anadromous sockeye salmon (Oncorhynchus nerka) stocking program in Skaha Lake that has seen successful stock establishment since commencement in 2004, but also hybridization with resident kokanee. Genetic monitoring to date has been unable to accurately assign hybrids beyond F1, which is a management need moving forward. Here, we conducted a simulation-based study to evaluate the accuracy and power of varying subsets of the data to assign individuals to advanced-generation hybrid classes. Using empirical data at 6339 variable SNPs previously collected from the reference populations, we simulated six introgression classes ranging from F1 to B3-backcrosses, using eight different datasets [100 random, 100 highest Fst, 300 random, 300 highest Fst, 1000 random, 1000 highest Fst, 23 outliers, 6326 non-outliers] and evaluated their performance using two different assignment algorithms (LEA and NEWHYBRIDS). As expected, greater number and divergence (high Fst) of markers yielded greater assignment accuracy. NEWHYBRIDS outperformed LEA in all comparisons except for the 6326 SNP panel. Overall, the 300 highest Fst SNP panel should provide the optimal balance of resolution and cost-effectiveness for monitoring introgression trends in Skaha Lake over the coming decade. Given the need for system-specific assessment of introgressive hybridization, we recommend similar simulation-based analyses for optimizing SNP panels for a given system and management question prior to commencing genetic monitoring.