Mastitis remains the most frequent health issue affecting dairy cattle, with an estimated 25% of U.S. dairy cows experiencing at least one case of clinical mastitis annually. This disorder is also one of the costliest diseases globally, with an average cost per case of approximately $444. In addition to the direct costs that are more apparent to farmers (such as treatments, diagnostics, discarded milk and veterinary services), there are numerous indirect losses, including reduced future milk production, lowered milk quality and premature culling. Consequently, novel tools that enhance the efficiency of mastitis control programs would greatly benefit the dairy industry.
Automated milking systems were first commercially introduced in the Netherlands in 1992, with the first system in the U.S. installed in Wisconsin in 2000. Since then, these systems have expanded worldwide, offering farmers the opportunity to record multiple variables in individual cows, allowing for continuous monitoring of performance and specific behaviors during the milking process.
One area of growing interest is the potential for behavioral changes during milking to detect specific health disorders. Briefly, milking behavior is defined as a group of actions expressed by cows and captured by automated milking systems. Examples of these behaviors include milking duration, teat not found, incomplete milking and milking unit kick-off, among others. The most important question is whether these behaviors are altered in sick cows to a degree that allows the prompt detection of affected cows, enabling dairy producers to manage these cases in a timely manner, avoiding larger losses and improving animal welfare.
Although Holsteins clearly remain the dominant dairy breed in the U.S., Jersey cows and their crosses with Holsteins are a growing population, particularly in the Southwest region. According to the Council on Dairy Cattle Breeding, in 2020, Holstein and Jersey cows made up 79.9% and 7.9% of the U.S. dairy herd, respectively, while crossbreeds constituted 11.8% of the cow population. Moreover, although herds using a single breed remain more representative of the U.S. dairy industry, this shift has led some farms to house multiple breeds, as well as crossbred animals.
Overall, it is recognized that these breeds differ in performance, behavior and susceptibility to disease. Due to differences in milk yield and conformation, Holstein cows are more susceptible to developing clinical diseases such as metabolic disorders, mastitis and lameness. Additionally, Holstein cows are more likely to exhibit reproductive inefficiencies. On the other hand, Jerseys are more prone to oral behaviors such as tongue rolling and flicking, especially in cows kept under pasture-based systems. Furthermore, Jersey cows have smaller teats than Holsteins, which may influence their milkability.
The effect of clinical mastitis on the behavior of dairy cows has been documented. These deviations include variable lying times, increased kicking during milking and extended milking durations. However, there is a need for research on the impact of clinical mastitis on cows of different breeds, particularly regarding how discomfort and pain from mastitis affect their milking behavior.
To explore these differences, our research group conducted a study from 2023 to 2024 at a large organic multibreed dairy equipped with an automated milking system. The goal was to describe and compare the milking behavior of cows from different breeds during the week before and after the diagnosis of clinical mastitis. Unlike traditional voluntary automated milking systems, this dairy utilizes a batch milking design, where cows are brought in twice per day for milking. This system allows cows the opportunity to choose which robot unit they prefer.
A total of 2,224 cows were included in the study, corresponding to 55% Jerseys, 35% Holsteins and 10% crossbred animals. In total, 204 clinical mastitis cases were identified, with 71 cases in Holsteins, 113 in Jerseys and 20 in crossbred cows. For each case, 10 control cows were selected, matched by breed, parity, days in milk and the absence of other clinical diseases. These controls were used to make comparisons at five distinct time points: -7 days, -4 days, day 0 and 4- and 7-days post-diagnosis.
Our findings indicated that Jersey and crossbred cows affected by clinical mastitis had longer milking times compared to healthy cows. In Jerseys, these extended milking times began as early as four days before diagnosis and persisted for the next seven days. Notably, Holstein cows with clinical mastitis did not show significant changes in milking time compared to healthy animals. A potential explanation for this finding is presented in previous studies suggesting that the distress caused by moderate and severe clinical mastitis can impair milkability (the ability to produce milk quickly, regularly and completely), potentially prolonging milking times due to disruptions in the milk ejection reflex caused by reduced oxytocin secretion.
Regarding other milking behaviors, we found that, compared to healthy herdmates, Jersey cows affected by clinical mastitis presented a greater frequency of “teat not found” events, where one or more teats are not milked by the system. No such differences were observed in Holsteins or crossbred cows. The smaller teats of Jerseys and the variability in udder conformation could play a role in milking unit attachment efficiency, especially if Jerseys are more sensitive to mastitis-related discomfort than Holsteins.
We also observed that, regardless of breed, cows with clinical mastitis exhibited a greater frequency of “incomplete milking,” where at least 50% of the expected milk yield was not delivered. This behavior was evident four days earlier in Jerseys than in other breeds. This event is relevant, as incomplete milking can lead to increased intramammary pressure, compromising further milk production. Additionally, incomplete milking may result in milk leakage between milkings, increasing cow discomfort and disrupting the milking routine.
Interestingly, despite expectations, we did not observe any differences in “milking unit kick-off” frequency, an event where the milking unit is abruptly detached, between cows affected by clinical mastitis and healthy cows across all breeds. This suggests that, although cows affected by clinical mastitis may be more reactive to mechanical stimuli due to potential pain or discomfort, the degree of inflammation might not be sufficient to cause the expression of this behavior.
In summary, Jersey and crossbred cows affected by clinical mastitis exhibited longer milking times, a higher incidence of teat not found events and more incomplete milking compared to healthy cows. However, no such differences were observed in Holsteins, suggesting they may be more resilient to mastitis-related discomfort. Data from automated milking systems provides valuable insights into disease prevention and management. Further research is needed to enhance our understanding of how to effectively utilize this technology to improve both the health and productivity of dairy cows.
Pablo Pinedo is a professor in the department of animal sciences at Colorado State University. Pablo Muñoz Boettcher is a graduate research assistant in the Ph.D. program in the department of animal sciences at Colorado State University. Juan Velez is chief agricultural officer of Aurora Organic Dairy.
