Features of 24-h behavioural patterns of dairy herds in relation to milking system, pasture access and season

Circadian rhythm and farm management shape the 24-h behavioural patterns of dairy herds. In turn, the features of 24-h behavioural patterns can be hypothesized to reflect farm management, such as the pasture access and milking system. This study aimed to quantify and compare the 24-h behavioural patterns of farms with different milking systems, pasture access and seasons, and distil features which could represent the characteristics of herd behaviour, through visualization of 24-h behavioural patterns, assessment of features and hierarchical clustering. Lying and step data per 15-min interval were available for individual cows on eight farms in 2019, which were processed into monthly 24-h herd patterns per 15-min interval. Five farms applied a conventional milking system (CMS) and pasture access, one CMS and no pasture access, one automatic milking system (AMS) and pasture access, one AMS and no pasture access. The farm with CMS and no pasture access had the most consistent monthly 24-h behavioural patterns across the whole year. After comparing the visualization of monthly 24-h behavioural patterns per herd, we engineered five 24-h features to characterize the 24-h patterns, including the average, maximum, and minimum lying time and average and maximum number of steps in 15-min intervals. Subsequent hierarchical clustering showed that the clustering of farms with different milking systems and pasture access using features of herd 24-h behavioural patterns is feasible, and it is more informative than classification based on daily behavioural features.