Multivariate factor analysis of reproductive and productive traits in Holstein cattle

The aim of this study was to elucidate the structure of relationships between productive and reproductive traits and to identify the most informative traits regarding to phenotypic variance in Holstein cattle. Records of 5,217 Holstein (primiparous and multiparous) females born between 1993 and 2011 were used. Animals were from a herd located in the town of Descalvado/ SP, Brazil. The analyzed productive traits were 305-day cumulative milk yield (Y305), peak yield (PY) and milk production per day of calving interval (MYCI). The reproductive traits were interval between calving and first estrus (CFE), interval between calving and last service (CLS), calving interval (CI) and gestation length (GL). Multivariate factor analysis was used in the derivation of new synthetic variables that summarize the original group of studied traits. Multivariate factor analysis was performed by applying axis orthogonal rotation (rotation type VARIMAX); the analysis grouped the seven traits into four latent or common factors with an associated communality equal to 88.9% of the original total variance. The first factor grouped all the productive traits and explained 30.9% of the total variance. The second factor explained 29.5% of the total variance and was represented by the CLS and CI traits. The third factor, retained 14.3% of the variation of the dataset and was highly correlated to the GL. The fourth factor is linked with CFE and explained 14.2% of the total variance. The use of multivariate factor analysis permitted the separation of studied traits in four factors, interpreted as indicators of milk production, reproductive activity, gestation and early reproductive activity, respectively. The obtained results showed the first two factors, which included the Y305, MYCI, PY, CLS and CI traits, explained the most variability in the dataset. The GL showed independence in the phenotypic expression of the group of evaluated variables.