Impact of Different Breeding Strategies on the Accuracy of Genomic Selection in a Population of Pacific White Shrimp (Penaeus Vannamei): A Simulation Study

Abstract

The prediction accuracy of estimated breeding values using traditional best linear unbiased prediction (PBLUP) and single-step genomic BLUP (ssGBLUP) methods was investigated in two breeding populations of Pacific white shrimp (Penaeus vannamei) using simulated data with varying levels of marker density. Additionally, the impact of the number of generations with phenotypic and genotypic data on the prediction accuracy was also investigated. The simulated breeding populations consisted of either 25 families with 40 animals each or 50 families with 20 animals each. Each population consisted of 20 non-overlapping generations, with 1000 animals per generation. Only the last ten generations were genotyped, and phenotypes and pedigrees were available for all the animals. Phenotypes from the 20th generation were masked in data analysis to mimic the selection of young animals without phenotypes. A trait with a heritability of 0.1 was simulated, and 100% of the genetic variance was explained by 1000 QTLs evenly distributed across chromosomes and randomly located across the genome. Genomic information consisted of 44 chromosomes that mimicked the genome structure of P. vannamei with 6K, 11K, 22K, 45K, and 90K markers evenly distributed across the 44 chromosomes. Results indicated that expanding family sizes with fewer family numbers improved the prediction accuracy. In all scenarios, ssGBLUP outperformed PBLUP in terms of the prediction accuracy. The more generations of data in which genotypes and phenotypes were included in the ssGBLUP approach, the better the accuracy of the predictions. However, genotyping for more than six generations did not improve the prediction accuracy. Additionally, the maximum accuracy was achieved with 45K SNPs in both scenarios, with no substantial improvement in accuracy by increasing the marker density to 90K. Overall, more accurate predictions were obtained with ssGBLUP with more genotyped animals per family and could be a suitable method for genomic selection  in P. vannamei.

Keywords: single-step GBLUPMixed Model EquationsGenetic EvaluationBreeding