Monitoring individual drinking behaviour and the social hierarchy in dairy cows using electronic drinkers

Sufficient access to clean drinking water is essential for the production and welfare of dairy cows. However, dairy farms vary greatly in how they provide water to cows. Electronic drinkers are now available to monitor cows’ water intake on individual level and can be used to better define their water need. The objective of this study was to validate the detection of agonistic replacements when competing for water access with a new electronic drinker and to explore the ability of using these interactions to identify the social hierarchy within a group of cows. A group of 22 Holstein mid-lactating cows were housed in a free-stall barn equipped with 6 electronic drinkers, 22 individual feeders and 7 cameras. For 9 days over the 18 days of the trial, agonistic interactions were identified via video. We used an algorithm relying on visit start times of subsequent cows at the electronic drinkers to automatically detect agonistic replacements. We determined 4 social hierarchies using the normalized David's score (NormDS) based on 1) agonistic interactions video observed in areas outside the drinkers, 2) agonistic interactions video observed at the drinkers, 3) replacements at the drinkers obtained from video, and 4) replacements at the drinkers detected by the algorithm. We found a weak correlation between the hierarchy recorded outside the drinkers and the one based on agonistic interactions at the drinkers. The hierarchy based on agonistic interactions at the drinkers was strongly correlated with the one relying on video-based replacements but showed a weaker association with the hierarchy relying on replacements detected by the algorithm. To explore the influence of dominance on drinker use, we defined dominant (n = 6), mid-dominant (n = 6), mid-subordinate (n = 4) and subordinate (n = 6) cows according to NormDS quartiles in the video-based replacement hierarchy. Subordinate cows drank less than the other social groups at the peak consumption of the whole group at 1800 h but drank more than the most dominant cows 2hours later. Subordinate cows preferentially used the most isolated drinker of the barn, furthest from the feeding area, and they drank less from drinkers in higher density areas. Detecting replacements at the drinker can be useful in studies aimed to analyze the effect of social groups on individual drinking behaviour. However, we suggest that for the automatic detection of replacements at the drinker, technology that allows for the recording of non-drinking visits is necessary.