We collected data on the behaviour of dairy cows in barns, clinical signs of diseases as well as events that may stress or agitate the cows. A Real-Time Locating System gives the position of individual cows every second. The position of the cow is determined by triangulation based on radio waves emitted by a tag fixed on each cow neck collar and captured by antennas in the barn. The cow’s activity is inferred from its position: ‘eating’ if the cow is positioned at the feeding table, ‘resting’ if the cow is in a resting area (typically cubicles), else ‘in alleys’. We aggregated this information to get the time spent in each activity per hour. We also calculated the activity level of the cow for each hour of the day by attributing a weight to the time spent in each activity. For each cow and day, we collected information on health events or other events that may affect behaviour. There were 11 types of events. Six events were linked to health: lameness; mastitis; LPS (i.e. administration of lipopolysaccharide (LPS) in the mammary gland, an experimental treatment to induce udder inflammation); subacute ruminal acidosis; other diseases (such as colic, diarrhoea, ketosis, milk fever or other infectious diseases); and accidents (such as retained placenta or vaginal laceration). Two events were linked to reproduction: oestrus and calving. Three events were stress events: animal mixing, disturbance (i.e. mild intervention on animals such as late feeding, alarm test) and marginal management changes (ration changes, fill bed). In addition, a Boolean sums up whether this hour was considered as normal or not. Data contain four datasets. It consists of univariate time series. Each time series corresponds to the hourly activity level of a cow. Datasets 1 and 2 are from the INRAE Herbipôle experimental farm and include data from experiments; datasets 3 and 4 are from commercial farms. They contain data on respectively 28, 28, 30 and 300 cows monitored for 6 months, 2 months, 40 days and one year. The data can be used to study the links between health, reproduction events and stress on the one hand and cow behaviour on the other hand. More specifically, it can be used to build and test tools for an earlier detection of health and disturbances, with a view to inform caretakers so that corrective actions can be rapidly put in place.
Read full abstract