Multiple-trait genomic evaluation for milk yield and milk quality traits using genomic and phenotypic data in buffalo in Brazil

Abstract

The use of markers distributed all long the genome may increase the accuracy of the predicted additive genetic value of young animals that are candidates to be selected as reproducers. In commercial herds, due to the cost of genotyping, only some animals are genotyped and procedures, divided in two or three steps, are done in order to include these genomic data in genetic evaluation. However, genomic evaluation may be calculated using one unified step that combines phenotypic data, pedigree and genomics. The aim of the study was to compare a multiple-trait model using only pedigree information with another using pedigree and genomic data. In this study, 9,318 lactations from 3061 buffaloes were used, 384 buffaloes were genotyped using a Illumina bovine chip (Illumina Infinium (R) bovineHD BeadChip). Seven traits were analyzed milk yield (MY), fat yield (FY), protein yield (PY), lactose yield (LY), fat percentage (F%), protein percentage (P%) and somatic cell score (SCSt). Two analyses were done: one using phenotypic and pedigree information (matrix A) and in the other using a matrix based in pedigree and genomic information (one step, matrix H). The (co) variance components were estimated using multiple-trait analysis by Bayesian inference method, applying an animal model, through Gibbs sampling. The model included the fixed effects of contemporary groups (herd-year-calving season), number of milking (2 levels), and age of buffalo at calving as (co) variable (quadratic and linear effect). The additive genetic, permanent environmental, and residual effects were included as random effects in the model. The heritability estimates using matrix A were 0.25, 0.22, 0.26, 0.17, 0.37, 0.42 and 0.26 and using matrix H were 0.25, 0.24, 0.26, 0.18, 0.38, 0.46 and 0.26 for MY, FY, PY, LY, % F, % P and SCCt, respectively. The estimates of the additive genetic effect for the traits were similar in both analyses, but the accuracy were bigger using matrix H (superior to 15% for traits studied). The heritability estimates were moderated indicating genetic gain under selection. The use of genomic information in the analyses increases the accuracy. It permits a better estimation of the additive genetic value of the animals.