Supplementary descriptions for poster CowCalf Contact – Effects on cow traffic in an automatic milking system

Senast ändrad: 24 juni 2024


The interest in keeping dairy calves with their dams for a prolonged period after calving (CowCalf Contact (CCC)) is increasing among European farmers. Although CCC can be implemented in many different dairy systems (Eriksson et al., 2022), it is still uncommon to keep the calves in “milking robots” or automatic milking systems (AMS).

For AMS to work satisfactory, it needs to be easy for the cows to move between different resources (feed, water, resting area, milking robot). The movement of cows between different areas within the pen is called cow traffic. If the cow traffic does not work well in a pen, this will restrict the cows’ access to important resources and affect their time budgets.

In AMS with controlled cow traffic, as in our study, selection gates directs the cows either to milking or to other resources depending on time since last milking. The aims of our study was to explore if housing CCC calves in part of the pen would affect how willing their dams were to pass a selection gate located by the waiting area by the milking robot, depending on if the gate steered them to the waiting area or to the resting area where the calves were housed.

Material and methods

Gate registrations from four research batches of CCC cow and calf pairs, housed in the same free-stall AMS between August 2019 and June 2021 was explored. Three of four batches also included conventional cows (CONV) kept in the same AMS, but without physical calf contact. The CONV cows were separated from their own calves within 24 h after calving, before leaving the calving box. For all batches, the CCC cows could access their calves between 12-24 hours per day during the study period. Depending on batch, the set-up of the pen, the contact period, and periods with pasture access differed. For the results presented here, only periods when the animals were housed indoors were included in the analyses.

Two periods were compared:

1)    Before separation of the CCC calves. For this analysis, the data contained 68 CCC cows with in total 2931 cow-days of gate registrations and 49 CONV cows with 2326 cow-days of registrations.

2)    After separation of the CCC calves. Due to only one batch being housed in the same AMS after separation of the calves, the data for this analysis contained 19 CCC cows (588 cow-days) and 17 CONV cows (554 cow-days).



Översikt över vänteyta fotograferad ovanifrån. Foto.

Functionality of the selection gate

Upon entering the gate area, the ear tag of the cow is read by sensors located just before the small gate in the lower right corner of the picture. Before the small gate opens, the selection gate (red star) is adjusted to steer the cow to the correct area depending on if the cow has milking permission or not. This cow has milking permission, so the selection gate has swung open to direct her into the waiting area. If the cow would not have had milking permission, she would instead have been directed to her resting area (CCC: straight ahead, CONV: to the right). The cow can only leave the waiting area through the milking robot (yellow star).

When the small gate opens, the cow can choose to (1) enter the waiting area directly, (2) remain standing in the gate area while the small gate opens and closes repeatedly, (3) leave the gate area and remain in the feed alley. On this particular occasion, the small gate opened and closed 261 times before the cow entered the waiting area.


Validation of gate data

To evaluate the reliability of the gate registrations, 109 passages through the selection gate of 5 CCC dams with easily identifyable coat patterns were assessed from video recordings by one observer. The passages were selected based the number of times the selection gate opened and closed while the focal cow did not pass (DNP), based on the gate registrations. Thirty passages were selected for DNP = 0, 19 for DNP = 1-5, 15 for DNP = 6-10, 16 for DNP = 11-20, 16 for DNP = 21-30 and 13 for DNP = 30+.

For the 109 evaluated passages, the gate registrations and observer disagreed on 7 passages due to incorrect registrations by the gate. The disagreements were due to (1) another cow being registered as the focal cow when passing the gate, resulting in the focal cow no longer being directed to the originally intended area (n = 1); (2) the gate did not physically opened but still registered that it repeatedly opened and closed (n = 2); (3) focal cow incorrectly registered as passing the gate, breaking the evaluated passage into two separate events in the gate data but not in the video assessment (n = 4). For the four last passages, the combined number of DNP in the two split gate events corresponded to what was registered during video assessment.

Additional time in the feed alley

The amount of additional time spent in the feed alley was calculated by the difference in time between when the gate first opened and the time when the focal cow passed. As such, the passages that occurred directly after the gate opened for the first time contributed 0 seconds to additional time, even if passage through the gate was not instantaneous. Additional time in seconds was then scaled to hours. The amount of additional time in the feed alley were summed by cow and period, and diveded with the number of days with gate registrations that each cow had in each period, to obtain average daily additional time in the feed alley for each cow. Due to substantial right scew of the data, additional time is reported as median (interquartile range).


During the pre-separation period, CCC cows did not directly pass the selection gate to the milking robot on 55% of occasions, but instead retracted their heads when the gate opened and then tried to enter the gate a median 3 (range 1-261) more times before passing. However, there were large variability between CCC cows, with individual cows not passing the gate directly on 7-91% of occasions. For CONV, the corresponding proportion was 25%, with a median 3 (1-85) DNP. Also CONV cows showed large variability, with individual animals not passing the gate directly on 0-78% of occasions.

When cows instead were directed to the resting area, CCC cows did not pass directly on 5% of occasions (individual cows 0-17%), with a median 1 (range 1-90) DNP before passing. For CONV cows the values were 13% of occasions (individual cows 1-35%) with 1 (1-79) DNP.

After separation of the CCC calves, CCC cows directed to the milking robot did not pass directly on 27% of occasions (individual cows 14-55%), with a median 2 (range 1-36) DNP. Corresponding values for CONV were 19% of occasions (individual cows 7-63%) and median 2 (1-55) DNP.

When directed to the resting area both groups behaved similarily with CCC cows not passing directly on 12% of occasions (individual cows 0-42%) and performing median 2 (range 1-39) DNP before passing, while CONV cows did not pass directly on 14% of occasions (individual cows 2-32%) with median 1 (1-29) DNP.



Stapeldiagram med proportion passager via vänteytan mot antal DNP. Grå staplar för kokalv. Blå för konventionella.Bild.

For passages with many DNP, it was much more common that the CCC cows were directed to the waiting area than the CONV cows before the CCC calves were separated. For example, for passages with 30+ DNP (representing 135 occasions for CCC and 17 occasions for CONV), the gate directed CCC cows to the milking robot 95% and CONV cows 42% of the occasions. This finding suggests that the CCC cows were more unwilling to enter the waiting area.



During video validation of the gate registrations it was discovered that passages with many tries did not necessarily affect cow traffic for other animals. Even though the data currently is very limited, it was clear that some cows stayed in the gate area for long periods if they chose not to pass the gate (obstructing the gate for other cows), while other CCC cows left the gate area and returned at a later time (leaving the gate available for other cows). As the focal cows behaved similar between days, some CCC cows likely affect the time budgets of other animals to a higher degree.

Discussion and conclusions

The differences we found between CCC and CONV before separation of the CCC calves, and the fact that these differences gradually decreased after separation, suggest that the presence of the calves affected their dams willingness to pass the selection gate, at least in the study set up we used.

It is possible that our findings are due to the same selection gate sometimes directing the cows to the waiting area and sometimes to the calves, creating occasions when the “verdicts” of the selection gate are different from what the cows want to achieve. Other AMS solutions, e.g. having two different selection gates providing access to the waiting area and the calf area, or only allowing access to the calves if the cows first pass the waiting area, may reduce the negative effects of CCC on cow traffic in AMS.


Eriksson, H. et al. (2022). Strategies for keeping dairy cows and calves together – a cross-sectional survey study. animal 16: 100624.