Indirect Estimation of (Co)Variance Functions for Test-Day Yields During First and Second Lactations in the United States

(Co)variance components for milk, fat, and protein yields during first and second lactations were estimated from test-day data from 23,029 Holstein cows from 37 herds in Pennsylvania and Wisconsin using a multitrait test-day model. Canonical transformation was used with an expectation-maximization algorithm. To allow description of (co)variances within and across yield traits and parities, four lactation stages of 75 d were defined6 for each parity, and the test day nearest the center of each interval was used. Prior to analysis, data were adjusted for lactation curves within lactation stages using all records from all available cows. Data from cows with missing values were excluded to allow a canonical transformation to be used for estimation of (co)variance matrices. Data from 9110 cows were available for canonical analysis of lactations with test days in all lactation stages. (Co)variance functions were used to describe (co)variance structure within and across yield trait and parity. (Co)variance components of biological functions (305-d yield, persistency defined as difference between yields on d 280 and 60, and maturity rate defined as difference between second- and first-lactation yields) were developed from (co)variance functions. Heritabilities ranged from 0.09 to 0.22 for test-day yields, from 0.21 to 0.23 for 305-d yields, from 0.03 to 0.11 for persistencies, and from 0.05 to 0.07 for maturity rates. Phenotypic correlations between first- and second-lactation persistencies were low, but genetic correlations were high. Genetic correlations with maturity rate ranged from 0.11 to 0.61 for 305-d yields and persistencies.