Principal component and cluster analyses to evaluate production and milk quality traits

Using multivariate analyses, this study attempted to identify important traits explaining the relationship between milk production and quality produced by Holstein cows. Monthly milk records from three commercial dairy farms located in the Agreste region of Pernambuco, Brazil, collected in the period from 2007 to 2017, were used. A total of 5,872 observations regarding milk production, milk components and somatic cell score (SCS) were analyzed using principal component analysis (PCA) and cluster analysis. According to the former analysis, the first three principal components explained 79.69% of the total variation. Total solids content contributed 29.66% of the variation in the first principal component, while lactose content contributed 49.43% of the variation in the second principal component. According to the latter analysis, three clusters differed for all characteristics (p<0.001) and cluster 2 concentrated 43.15% (2,534) of the information with lower SCS and higher lactose content and milk production. Total solids, lactose and fat were considered the most representative traits explaining the variability of the data set. The multivariate techniques used in this study proved useful in obtaining effective characteristics, with three factors considered important in explaining the relationship between Holstein cows' milk production and quality.