Multivariate factor analysis of detailed milk fatty acid profile: Effects of dairy system, feeding, herd, parity, and stage of lactation

We investigated the potential of using multivariate factor analysis to extract metabolic information from data on the quantity and quality of milk produced under different management systems. We collected data from individual milk samples taken from 1,158 Brown Swiss cows farmed in 85 traditional or modern herds in Trento Province (Italy). Factor analysis was carried out on 47 individual fatty acids, milk yield, and 5 compositional milk traits (fat, protein, casein, and lactose contents, somatic cell score). According to a previous study on multivariate factor analysis, a variable was considered to be associated with a specific factor if the absolute value of its correlation with the factor was ≥0.60. The extracted factors were representative of the following 12 groups of fatty acids or functions: de novo fatty acids, branched fatty acid-milk yield, biohydrogenation, long-chain fatty acids, desaturation, short-chain fatty acids, milk protein and fat contents, odd fatty acids, conjugated linoleic acids, linoleic acid, udder health, and vaccelenic acid. Only 5 fatty acids showed small correlations with these groups. Factor analysis suggested the existence of differences in the metabolic pathways for de novo short- and medium-chain fatty acids and Δ(9)-desaturase products. An ANOVA of factor scores highlighted significant effects of the dairy farming system (traditional or modern), season, herd/date, parity, and days in milk. Factor behavior across levels of fixed factors was consistent with current knowledge. For example, compared with cows farmed in modern herds, those in traditional herds had higher scores for branched fatty acids, which were inversely associated with milk yield; primiparous cows had lower scores than older cows for de novo fatty acids, probably due to a larger contribution of lipids mobilized from body depots on milk fat yield. The statistical approach allowed us to reduce a large number of variables to a few latent factors with biological meaning and able to represent groups of fatty acids with a common origin and function. Multivariate factor analysis would therefore be a valuable tool for studying the influence of different production environments and individual animal factors on milk fatty acid composition, and for developing nutritional strategies able to manipulate the milk fatty acid profile according to consumer demand.