Canonical correlation of milk immunoglobulins, lactoferrin concentration and Dairy Herd Improvement data of Chinese Holstein cows

Abstract

In this study, canonical correlation analysis was applied to estimate the relationships of DHI data with milk Ig (IgG1, IgA, and IgM) and Lf concentrations. Specifically, we evaluated the relationships between two groups of variables: milk IgG1, IgA, IgM and Lf concentration as variables ( y) and lactation number, stage of lactation, daily milk production, milk fat, protein, lactose, milk total solids and somatic cell score (SCS) as variables ( x). The results indicated that four canonical variables were identified. The canonical correlations of the first and second pairs of canonical variables were 0.662 and 0.469 respectively, which were highly significant and accounted for 91.6% of the variability observed in the data. Stage of lactation, daily milk production, milk protein and SCS were the significant factors affecting milk Lf concentration in both canonical correlations and multiple correlation analysis, and lactation number was the significant factors affecting milk IgG1 concentration. The first standardized canonical variation combination could be regarded as a predictable measure of Lf concentration and the second as a predictor of IgG1 level. These results indicated that dairy producers could select cows for increased Ig and Lf production using DHI data directly.