We examined the latent structure of 26 cheese related phenotypes in dairy cattle. Traits related to milk yield and quality (8 traits), milk protein fractions (8 traits), coagulation and curd firmness indicators (CF, 5 traits) and cheese-making phenotypes (cheese yields (%CY) and nutrient recoveries in the curd (REC), 5 traits) were analysed through multivariate factor analysis (MFA) using a varimax rotation. All phenotypes were measured in 1264 Brown Swiss cows. Ten mutual orthogonal, latent variables (factors; Fs) were obtained explaining 74% of the original variability. These Fs captured basic concepts of the cheese-making process. More precisely, the first 4 Fs, sorted by variance explained, were able to capture the underlying structure of the CY percentage (F1: %CY), the CF process with time (F2: CFt), the milk and solids yield (F3: Yield) and the presence of nitrogen (N) in the cheese (F4: Cheese N). Moreover, 4 Fs (F5: as1-β-CN, F7: κ-β-CN, F8: as2-CN and F9: as1-CN-Ph) were related to the basic milk caseins and 1 factor was associated with the α-LA whey protein (F10: α-LA). A factor describing udder health status (F6: Udder health), mainly loaded on lactose, other nitrogen compounds in the milk and SCS, was also obtained. Further, we inferred the effects of some potential sources of variation (e.g. stage of lactation and parity) including feeding and management systems. Stage of lactation had a significant effect for 7 of the 10 Fs, followed by parity of the cow (3 Fs), dairy system and feeding (3 Fs). Our work demonstrates the usefulness of MFA in reducing a large number of variables to a few latent factors with biological meaning and representing groups of traits that describe a complex process like cheese-making. Such an approach would be a valuable tool for studying the influence of different production environments and individual animal factors on protein composition and cheese-making related traits.

We studied the genetics of cheese-related latent variables (factors; Fs) for application in dairy cattle breeding. In total, 26 traits, recorded in 1264 Brown Swiss cows, were analyzed through multivariate factor analysis (MFA). Traits analyzed were descriptors of milk quality and yield (including protein fractions) and measures of coagulation, curd firmness (CF), cheese yields (%CY) and nutrient recoveries in the curd (REC). A total of 10 Fs (mutual orthogonal with a varimax rotation) were obtained. To assess the practical use of the Fs into breeding, we inferred their genetic parameters using single and bivariate animal models under a Bayesian framework. Heritability estimates (intra-herd) varied between 0.11 and 0.72 (F3: Yield and F7: κ-β-CN, respectively). The Fs underlined basic characteristics of the cheese-making process, milk components and udder health, while retaining 74% of the original variability. The first two Fs were indicators of the CY percentage (F1: %CY) and the CF process (F2: CFt), and presented similar heritability estimates: 0.268 and 0.295, respectively. The third factor was associated with the yield of milk and solids (F3: Yield) characterized by a low heritability (0.108) and the fourth with the cheese nitrogen (N) (F4: Cheese N) that conversely appeared to be characterized by a high heritability (0.618). Three Fs were associated with the proportion of the basic milk caseins on total milk protein (F5: as1-β-CN, F7: κ-β-CN, F8: as2-CN), also highly heritable (0.565, 0.723 and 0.397, respectively) and 1 factor with the phosphorylated form of the as1-CN (F9: as1-CN-Ph; 0.318). Moreover, 1 factor was linked to the whey protein α-LA (F10: α-LA; 0.147). An indicator factor of a cow’s udder health (F6: Udder health) was also obtained and showed a moderate heritability (0.204). Although the Fs were phenotypically uncorrelated, considerable additive genetic correlations existed among them, with highest values observed between F10: α-LA and F6: Udder health (−0.67) as well as between F9: as1-CN-Ph and F3: Yield (−0.60). Our results show the usefulness of MFA in dairy cattle breeding. The ability to replace a large number of variables with a few latent indicators of the same biological meaning marks MFA as a valuable tool for developing breeding strategies to improve cow’s cheese-related traits.

A total of 7492 test-day observations for mean contents of fat, protein, casein, serum protein and lactose and individual laboratory cheese yield (ILCY) were obtained, at approximately monthly intervals, from 1119 ewes belonging to eight Churra dairy flocks. The effect of various factors on these variables was examined and phenotypic correlations among all traits were estimated. Least squares analyses showed significant effects of flock test-date, stage of lactation, age of ewe, and number of lambs weaned on almost all variables. Protein content and composition were not affected by the number of lambs weaned. ILCY had an unadjusted mean (26·55 kg cheese/100 l milk) close to those reported for real cheese yield in dairy ewes and was affected similarly to the main milk components. Fat, protein, casein, and serum protein contents, and ILCY, showed a generally increasing trend as lactation progressed. These components reached a minimum at 1 month into lactation, when milk yield was highest, and increased for the remainder of the lactation. ILCY depended mainly on fat, protein and casein contents. Protein and casein contents were closely related and equally correlated with ILCY. An increase in somatic cell count (SCC) was associated with decreased milk yield and decreased lactose content.