Investigating lactation curve characteristics of dairy cows

The main goal of this study was to analysis of the main characteristics (latent variables) of lactation curve in dairy cows using multivariate Principal Component Analysis. This work used primary database from the milk production of Holstein cows (n = 238 heads) in the PJSC “Pedigree farm ‘Stepnoy’ Kamianka-Dniprovska Raion of Zaporizhzhia Oblast over a 4-yr period (2014-2017). Recording is done with an interval of 30 days for 10 test-days (TD1-TD10), i.e., TD1 is milk production recorded on milking day 30th, TD2 is day 60th, TD3 is day 90th, etc., and 305-day milk yield records (Y305) were used also. High significant correlations were found between daily milk yields for certain test-days. The Principal Component Analysis performed on the variance-correlation matrix of TD1-TD10 records are able to explain about 90.33 % of the total variance. The first principal component (PC1) explained 66.32 % of the total variance and was highly-positively correlated with TD2-TD10 records. Thus, PC1 were defined as “total milk production”. The second principal component (PC2) explained 19.06 % of the total variance and was highly-positively correlated with TD1-TD2 records and highly-negatively correlated with TD9-TD10 records.  Thus, PC2 were defined as “lactation curve persistency”. Finally, the third principal component (PC3) explained 4.95 % of the total variance and was highly-positively correlated with TD1 and TD10 records and highly-negatively correlated with TD4-TD5 records. Thus, PC3 were defined as “lactation curve type”. The use of a multivariate method (namely, the PCA) for the analysis of lactation curve characteristics based on monthly test-day records gave very close results of the analysis of milk productivity in different groups of domestic animals (cattle, goats and sheep). In all cases, the first principal component (PC1) described the absolute level of milk productivity during lactation, and the second principal component (PC2) described the persistency of the lactation curve. Significant influence on the PC1-PC3 factor scores was revealed to the greatest extent for such non-genetic factors as age of cow (in lactations), year and month of calving. Of the genetic factors, the greatest influence on the shape of the lactation curve was not so much the differences between the bull lines (Bell, Valiant, Elevation, Starbuck and Chief), but differences between individual bulls within some lines.