Advantages and Challenges of Genetic Stock Identification in Fish Stocks with Low Genetic Resolution

• Published in: Transactions of the American Fisheries Society
• Main Authors: Araujo, H. Andres, Candy, John R., Beacham, Terry D., White, Bruce, Wallace, Colin
• Format: Article in Journal/Newspaper
• Language: English
• Published: Wiley 2014
• Subjects: Fraser River; Oncorhynchus gorbuscha; Pink salmon
• Online Access: http://dx.doi.org/10.1080/00028487.2013.855258; https://afspubs.onlinelibrary.wiley.com/doi/pdf/10.1080/00028487.2013.855258

Abstract Genetic stock identification (GSI) is widely applied to mixed‐stock fisheries for many commercially exploited species. However, the accuracy of GSI depends on the level of differentiation among stocks. To evaluate our ability to estimate contributions in mixed‐stock fisheries of Pink Salmon Oncorhynchus gorbuscha , a species with limited population genetic differentiation, we analyzed 46 odd‐year Pink Salmon stocks belonging to a baseline of genotypes from southern British Columbia, the Fraser River, and Puget Sound. Samples were obtained without replacement from the baseline (known mixtures), and 16 microsatellite loci were used for analysis with two software packages (cBayes and ONCOR) to evaluate the accuracy of using this marker set to identify the correct region, subregion, and spawning site. The correct subregion was identified for Pink Salmon from southern British Columbia and Puget Sound. However, incorrect assignments were observed for the Fraser River subregions and the stock‐specific estimates. In addition, we used simulated baselines with the average genetic differentiation index F ST ranging from 0.0007 to 0.04 (the range of F ST values observed in Pink Salmon stocks) to identify biases in the GSI software programs. The results suggested that stock‐level genetic identification is subject to significant biases (>15%) when the average F ST among baseline stocks is less than 0.01. ONCOR was more accurate than cBayes in identifying the correct stock at small mean F ST values (<0.01), but there was no significant difference between the software packages at larger F ST values. Our results can help to improve GSI methods and to identify their limitations, especially for stocks with low genetic separation.