Factors associated with milking characteristics in dairy cows

Milking characteristics, and in particular milking duration, are a known contributor to costs in dairy production systems. Results from previous studies suggest that higher-yielding animals, on average, milk for a longer duration. Culling or selection for reduced milking duration alone may, therefore, reduce milk yield. Here, we propose 2 new traits, residual milking duration (RMD) and residual milking duration including somatic cell score (RMDS). Residual milking duration is represented by the residuals from a least squares regression of milking duration on milk yield; RMDS is represented by the residuals from a least squares regression of milking duration on both milk yield and somatic cell score [i.e., logarithm (base 10) of somatic cell count]. The mathematical properties of least squares regression ensure than the residual traits are independent from the regressor variables, or, in other words, RMDS is not correlated with either milk yield or somatic cell score. Both RMD and RMDS were defined using electronically measured individual cow milking duration from 235,036 part-day milking events from 74,607 cows from 1,075 Irish dairy herds. Twenty-four percent of the variation in milking duration was explained by the multiple regression model containing both milk yield and somatic cell score. The phenotypic standard deviation of RMD and RMDS was 102.2 and 98.2s, respectively, suggesting large variation in milking duration independent of milk yield (and somatic cell score). The correlation of RMD and RMDS with average milk flow rate, which may also be considered a measure of milking efficiency, was -0.74 and -0.75, respectively. Neither RMD nor RMDS was correlated with somatic cell score. However, average milk flow rate was correlated with milk yield (0.57) and milking duration (-0.38). Both RMD and RMDS are useful traits, which exhibit considerable variation and, therefore, can be used by farmers to identify phenotypically slower milking animals irrespective of milk yield (and somatic cell score). However, because of the lack of a correlation between RMD and somatic cell score in the sample population used in the present study, RMD and RMDS values per milking were almost identical.