Association between daily cow data and milk production in dairy herds milked with automatic milking systems

Abstract

Automatic milking systems (AMS) for dairy cows and its associated technologies can continuously monitor individual production, behavior, and physiological parameters, allowing for real-time decision making on these farms. Furthermore, the data recorded automatically in these systems allow dairy advisors to extrapolate some of the findings to cows milked with systems other than AMS. The main objective of this dissertation was to investigate cow daily behavior associated with milk production on AMS farms. Data from 47 AMS dairy farms located in Minnesota and Wisconsin in the U.S. were collected for a period of 12 months. RStudio was used for all the data management and statistical analyses. Multiparous cows with greater average rumination time and greater rumination time increase in the immediate postpartum period produced more milk at lactation peak, while the same association was absent for primiparous cows. Various behavior variables recorded during the time cows are in the AMS to be milked were associated with daily milk production, such as average daily milking interval, concentrate intake, number of failures, milking speed and time, and pre- and post-treatment time. Moreover, a quadratic relationship was found between body weight change in early lactation and 90-d milk yield, where cows with the ability to maintain or lose less than 10% of their body weight from DIM 1 presented greater total milk production over the first 90 DIM. Findings reported in this dissertation may be used to make milk production predictions and serve as benchmark data for AMS producers in the U.S.