Potential role of biologgers to automate detection of lame ewes and lambs

Lameness is an important health, welfare and economic problem in sheep flocks and early treatment is key to controlling lameness. Biologging technology provides high-resolution, continuous data that offers a novel opportunity to detect lameness either directly or by identifying behavioural changes; either option would facilitate more rapid treatment of lame sheep than visual observation. Here, the role of biologging data to identify lame sheep through behavioural changes within and between sheep is investigated. Accelerometers and proximity sensors were fitted to a flock of 50 Poll Dorset ewes rearing 32 single and 36 twin lambs, in Devon, UK in October 2019. Accelerometers were used to identify standing time and classify behaviour into four states for ewes (inactive, ruminating, grazing, walking) and three for lambs (inactive, sucking, moving). Principal components analysis reduced these behaviours to two components, ‘feeding’ and ‘inactive’ for ewes, and ‘inactive’ and ‘feeding’ for lambs. A visual locomotion score of each sheep was used each day to assess lameness. Complete records from sensors and locomotion observations were obtained for 513 days of ewe-activity and 720 days of lamb-activity (40 ewes, 26 single-raised and 28 twin-raised lambs). Linear mixed effects models were used to assess the effect of lameness adjusted for covariates age, litter size, social behaviour, environment and climate on standing time and the principal components. Lame ewes stood less, spent less time grazing and were more inactive than non-lame ewes. Lame lambs also stood less and were more inactive than non-lame lambs. Lambs with severely lame dams were also more inactive than those with non-lame dams. In conclusion, it is possible to identify behavioural differences between lame and non-lame ewes and lambs which could help enable automated early warning of lameness and consequently early treatment of lameness, and improved sheep welfare.