Assessment of circadian rhythm of activity combined with random regression model as a novel approach to monitoring sheep in an extensive system

Sensor-based technologies are becoming increasingly available and can be used to automatically gather long-term data about animal behaviour. With this information, it is possible to assess the circadian rhythm of activity and monitor its response to internal and external factors. Identifying irregularities in this rhythm may indicate animal health and welfare issues. The aim of this study was to collect sensor-based general activity and investigate circadian rhythm of this activity to identify the changes due to weather influences that act on these parameters throughout the year; to identify the differences between individuals; and to assess links between general activity and circadian rhythm of activity with sheep body weight change. In total, 29 Scottish Blackface ewes of different ages and body condition scores were used. The animals were monitored for four consecutive weeks in each of four seasonal periods, in extensive systems on Scottish upland pastures, and without human handling during study periods. Accelerometer-integrated collars were fitted to the animals to collect motion index continuously every minute. These data were used to calculate the percentage of cyclic behaviour that was harmonic/synchronized with the environment (over 24 h period), as Degree of Functional Coupling (DFC). The DFC was shown within rolling seven-day periods. Low DFCs indicate low synchronization. Weather data were collected daily. Random regression models were used to assess between-individual variation. During the winter period, the level of the DFC for the activity of nineteen ewes lowered in response to a period of high level of precipitation combined with low winter temperatures. However, four ewes exhibited a lower level of variation in the DFC values, showing that there were differences between individuals in regard to their response to the precipitation level. The overall mean of the DFC for the general activity was highest in autumn (95.4%, P < 0.001), however, it did not differ between summer and spring (respectively 90.2% and 88.1%, P > 0.05), but was significantly lower during the winter (81.7%, P < 0.001) compared with summer and autumn. Over the spring and summer, variation in DFC was a good estimator of body weight gain. It was concluded that the assessment of circadian rhythms of general activity using the DFC-parameter allows a better understanding of sheep responses to weather influences, compared to the evaluation of general activity alone. The random regression model method was effective in identifying animals that deviated positively or negatively from population responses.