Unravelling biological biotypes for growth, visual score and reproductive traits in Nellore cattle via principal component analysis

Abstract

Principal component analysis (PCA) is used to summarize important information from multivariate data in sets of new variables named principal components (PCs). In animal breeding, these new composite variables can be used to study the associations among multiple traits using the magnitude and direction of the PCA coefficients (in the eigenvectors) for each trait. Phenotypic data from 355 524 Nellore animals were used to estimate genetic parameters and explore the relationship among growth (weaning and post-weaning weight gain), visual score (weaning and yearling conformation, finishing precocity and muscling) and reproductive (scrotal circumference) traits using PCA. Genetic parameters were estimated by multi-trait analysis using a mixed linear animal model. The eigen-decomposition of the additive genetic (co)variance matrix (AT matrix) obtained using multi-trait analysis were used to calculate the PCs. In addition, PCA using the (co)variance matrix of the breeding values (EBVs) from single- and multi-trait analyses were investigated for comparison purposes. The direct heritability estimates for the weaning and yearling traits ranged from 0.17 (birth-to-weaning weight gain and conformation) to 0.21 (finishing precocity) and from 0.18 (weaning-to-yearling weight gain) to 0.46 (scrotal circumference), respectively. Genetic correlations estimated among all analyzed traits were positive (favorable) ranging from 0.15 (conformation at weaning and scrotal circumference) to 0.96 (finishing precocity and muscling at weaning). The first three PCs from multi-trait analysis using the eigen-decomposition of the AT matrix, explained 87.11% of the total additive genetic variance for the traits. The first PC (PC1) had negative and relatively similar coefficients for all traits, the second PC (PC2) contrasted the animals with early or late biotype, and the third PC (PC3) characterized a contrast between weaning and yearling traits. Our findings suggest that the PCA could be explored in breeding programs to select Nellore cattle to tailor selection towards specific PC, targeting, for instance, faster growth and precocious biotype.