GENOMIC PREDICTION AND GENOME-WISE ASSOCIATIONS IN THE PACIFIC WHITELEG SHRIMP Litopenaeus vannamei.

Reprogen - Animal Bioscience, Faculty of Veterinary Science, University of Sydney, Camden NSW, Australia. E-mail address: mehar.khatkar@sydney.edu.au <mailto:mehar.khatkar@sydney.edu.au>.

^b Centre for Sustainable Tropical Fisheries & Aquaculture, and the School of Marine and Tropical Biology, James Cook University, Townsville QLD, Australia.

^c Global Gen, Jalan Raya Narogong Km 14, Desa Cikiwul Bantar Gebang Bekasi, 17310 Indonesia.

The Pacific white-leg shrimp (Litopenaeus vannamei) is an important aquaculture species. Traditional genetic improvement programs for L. vannamei rely on family selection to improve growth and disease resistance traits. Recent advances in high-throughput genotyping and analytical methods can help in simplify breeding schemes and increase genetic gain particularly for complex or difficult to measure traits.

We generated a resource database for L. vannamei by genotyping a total of 1,934 samples with 8,967 genome-wide SNPs. These include 1,134 female and 123 male parents of 416 families, along with 677 nauplii pools. Following SNP QC, 5,893 SNPs were used for analysis.  Genetic diversity analysis indicated no evidence of inbreeding or a reduction in genetic diversity over time within this population.

A linkage map was constructed using 344 individuals from the largest 13 families. In total, 4,390 SNPs were mapped to 45 linkage groups that span a total of 4559.0 cM and cover an estimated 97.89% of the L. vannamei genome. The average interval, excluding intervals of 0 cM, was 2.67 cM indicating the most dense coverage to-date of SNP markers across the genome.

Genome-wide association analyses were conducted using mean family allele frequencies (nauplii DNA pools) of 5,893 SNPs and the family mean (of 416 families) of  two growth traits, two survival rate traits, one larval rearing survival rate  and three disease challenge traits. Few clusters of SNPs with P-value < 0.001 were detected, however, after correcting for false discovery cut-off no region of major effect remained significant.

Genomic selection uses information from all SNP to derive Direct Genetic Values (DGV). The equations are based on a "training set" and validated in an independent "test set". The data on a growth trait on 416 families were analysed for this analysis. DGV was estimated using best linear unbiased prediction. The accuracy of DGV in mirror prediction (randomly dividing families in training and test set) for growth was "high" (0.63-0.69). Other traits such as disease resistance were also investigated and expressed lower accuracy. This project demonstrates that genomic selection has potential application with moderate number of SNPs, family average phenotypic records, and based on family DNA pool frequency data for commercially important traits.