Autotetraploid Crassostrea gigas plays an important role in C. gigas artificial breeding because of their ability to stably produce triploid C. gigas with a fast growth rate, delicious meat, and abundant nutrients. There is an urgent need for autotetraploid C. gigas germplasm preservation and genetic improvement methods to ensure the sustainable development of the autotetraploid population. Recently, genomic selection (GS) has been widely applied in autotetraploid plants, significantly shortening the breeding cycle and promoting genetic improvement, but it has not been applied to autotetraploid C. gigas. This study is the first to apply GS methods for genomic prediction of five growth traits (e.g., shell length, shell height, shell width, total weight, and soft tissue weight) in autotetraploid C. gigas real dataset using whole-genome resequencing. Compared to diploid C. gigas, the optimal genotyping of whole-genome resequencing and the SNP encoding model of autotetraploid C. gigas are more complex. Thus, we investigated the SNP calling abilities of two alignment methods (e.g., BWA and Bowtie 2) and two genotyping methods (e.g., GATK and Freebayes) for whole-genome resequencing. Then, we also evaluated the ability of four potential SNP encoding models (e.g., additive, diploidized additive, simplex dominant, and duplex dominant) to estimate the heritability and genomic prediction accuracy of the growth traits in autotetraploid C. gigas, finding that the additive model was the most reliable SNP encoding model. Lastly, four linear models (e.g., Bayes A, Bayes B, Bayes C, and GBLUP), as well as one nonlinear model (e.g., RKHS), were tested to estimate the accuracy of breeding values of growth traits in autotetraploid C. gigas. Overall, GBLUP outperformed all the other four methods on C. gigas growth traits. The impact of varied SNP densities on genomic prediction accuracy was investigated as well. Our results will provide evidence that supports germplasm protection and molecular breeding in autotetraploid C. gigas.
Read full abstract