USING GENOMIC INFORMATION TO OPTIMIZE GENETIC IMPROVEMENT IN CHANNEL CATFISH

Catfish production is the largest aquaculture segment in the US and recent research has been conducted to improve production efficiency. Since 2006 selection has been based on traditional BLUP evaluations. With the availability of genomic information for catfish in early 2017, the first objective of this study was to investigate the feasibility of using genomic selection in US catfish. The second objective was to identify major SNP associated with harvest weight and residual carcass weight.

Phenotypes were available for harvest weight (n=27,160) and residual carcass weight (n=6020), and the number of fish in the pedigree was 36,365. After quality control, genotypes on 54,837 SNPs were available for 2911 animals. Genomic and pedigree predictions were calculated in a 5-fold cross validation approach, using single-trait models. Single-step genomic BLUP (ssGBLUP) was the method of choice for genomic predictions. Ability to predict breeding values was calculated as the correlation between adjusted phenotypes based on complete data and EBV or genomic EBV (GEBV) for each one of the folds. Inflation was assessed as the regression coefficient (b1) of adjusted phenotype on (G)EBV. For the association study, the GEBV were back solved to SNP effects and the percentage of variance explained by each SNP was calculated as SNP effect squared.  

Predictive ability for both traits increased 8 points when genomic information was used, compared to traditional evaluations (Table 1). Genomic information also helped to reduce inflation (Table 1), especially for residual carcass weight (b1=0.97). The proportion of variance explained by windows of 20 SNP was at maximum 2.2% for harvest weight and 3.3% for residual carcass weight. Both traits appear to be polygenic with no major SNP. Using genomic information is beneficial in catfish selection because of higher predictive abilities and it also allows estimation of the Mendelian sampling, helping to identify individuals within families, which is not possible with only pedigree information.