Potential for improving description of bovine udder health status by combined analysis of milk parameters

Abstract

The objective of this study was to assess the potential of a stepwise multivariate procedure to quantify cow-level udder health based on eight milk parameters: milk yield, protein percentage, fat percentage, lactose percentage, citrate percentage, somatic cell count (SCC), and two electrical conductivity parameters. The data were collected in one research herd and included 821 cow-level observations. In addition to milk parameters, disease recordings and bacteriology on quarter milk samples every eighth week throughout lactation were included. A multivariate mixed model was applied to the milk parameters in a healthy subset to adjust for the following systematic factors: total mixed ration (TMR) energy density, breed-line combination, parity, stage of lactation, and season. The proportion of variance accounted for by the mixed model ranged from 0.14 to 0.82 depending on milk parameter. The adjustments estimated in the healthy subset were applied to the whole dataset, including observations pertaining to nonhealthy cows. Combined description of the adjusted variation in the milk parameters was performed with a principal component analysis. The first principal component (Prin1) described 30% of the adjusted variation and was interpreted as being the main consequences of mastitis. Finally, cluster analysis based on Prin1 separated the observations into nine clusters, which were strongly associated with udder health in terms of increasing clinical and subclinical mastitis with increasing level of Prin1. It was concluded that a multivariate approach to assess udder health from milk parameters has the potential to substantially improve description of udder health.