Milking-time test indicators reveal links between vacuum dynamics, milk-flow patterns, and milk-quality traits in...

Variations in milking vacuum affect mechanical stress on teat tissue, with consequences for udder health and milk quality. This study evaluated key milking-time test indicators in commercial Holstein herds, and their contemporaneous associations with milk yield, composition, and somatic cell score (SCS). Data comprised 4,246 test-day records from 3,739 cows in 214 Italian herds equipped with conventional milking parlors. Vacuum was continuously recorded at the short milk tube (SMT) and mouthpiece chamber (MPC) to characterize machine-on time, milk let-down (MLD), overmilking, positive and negative cyclic vacuum fluctuations (CVF_max and CVF_min, respectively). A random forest model identified the milking-time test, herd and animal characteristics most strongly associated with variation in SCS. The most relevant milking-time test indicators were included in repeated mixed models to quantify their effects on milk yield and composition. Similar mixed model approaches were used to determine how herd and animal characteristics influenced milking-time test indicators. Overmilking, SMT vacuum during the milking phase were the variables most strongly associated with SCS variation. SCS significantly increased when overmilking exceeded 30 s and peaked beyond 60 s (2.60 vs. 2.95, respectively), while lactose and fat content were significantly lower in the ≥ 60 s class (4.78% and 3.81%, respectively), milk yield per session was unaffected by overmilking. SMT vacuum influenced milk yield and composition: SCS were significantly highest at 37.3-38.0 kPa (3.13) and lowest at 39.3-39.8 kPa (2.57), whereas lactose was lowest between 36.7 and 37.3 kPa (4.76%) and greatest between 38.7 and 40.4 kPa. Greater CVF_min and CVF_max were significantly associated with reduced fat content once fluctuations exceeded ∼3.50 kPa, likely reflecting altered milk-air flow dynamics. Large-size herds (>329 lactating cows) had significantly shorter machine-on time (328 vs. 338 s) and overmilking (47.6 vs. 59.2 s), and higher milk flow (3.07 vs. 2.73 kg/min), consistent with the presence of higher-genetics merit animals, typically associated with greater milk production potential in larger operations. However, large size herds showed significantly longer MLD (34.9 vs. 24.8 s), likely reflecting reduced pre-stimulation in high-throughput routines. Similar trends were observed in high-yielding herds (>38 kg/day per cow) that had significant shorter machine-on time (328 vs. 341 s), lower relative overmilking (15.0% vs. 18.2%), longer relative MLD (9.70% vs. 8.51%), and higher milk flow (3.17 vs. 2.71 kg/min) compared with low-yielding herds (<34 kg/day per cow). Finally, MPC vacuum significantly declined as parity increased, from 21.8 kPa in primiparous cows to 12.8 kPa in cows ≥ 4 parities, consistent with progressive teat canal widening. This study provides a field-based characterization of milking-time test indicators and their associations with milk yield and composition in Italian Holstein herds. Milking-time test indicators related to overmilking and vacuum dynamics were associated with variation in milk SCS. At herd level, different phases of milking, including MLD, main milking phase, and overmilking, differed according to herd size and productivity, while at animal level parity order mainly affected vacuum-based parameters. Overall, milking-time test indicators captured biological and mechanical variation in milking conditions which could be used to detect inefficient milking patterns under field conditions.