Using remote sensing to monitor behaviour and energy expenditure in livestock

Using a machine learning technique, measures of animal were posture found to be important in the accurate classification of adult and lamb behaviours. More behaviours (eight for adults and 15 for lambs) were identified with over 75% accuracy than any other previously reported study on sheep using tri- axial accelerometers.<br><br>Resting metabolic rate (3.67 ± 0.38 mL O2 kg'1 • min'1) and the thermoneutral zone of sheep was estimated via controlled indirect calorimetry experiments.

Measures of oxygen consumption via indirect calorimetry were then used to calibrate the use of the vectorial sum of dynamic acceleration, (VEDBA) as a proxy for energy expenditure. Sheep were trained to walk on a treadmill within a respirometry chamber at various speeds, incline and decline angles. Similar to previous studies, a positive linear relationship was described between VEDBA and oxygen consumption for sheep walking at various speeds on a level treadmill (rs = 0.88, p = 0.007).

However, VEDBA was not an accurate proxy for energy expenditure when incline or decline angle increased which could potentially result in under or overestimations of energy expenditure by 6 - 46%. Within this study, we suggest the measurement of animal body angle as a proxy for terrain angle, and while this measurement correlated with slope of the treadmill (X2 = 4.949, df= 1 ,p = 0.026, R2 =0.724), it did not explain any of the variation observed in energy expenditure.

Therefore, information with regards to the environment through which an animal is moving is necessary for accurate estimates of energy expenditure.<br><br>Tri-axial accelerometers were then applied to free-ranging male and female sheep during the mating season. Using information gathered from the previous studies, detailed time-energy budgets of sheep (2880 hours (133GB)) were constructed.<br><br>Along with this, measurements of animal body mass were taken to monitor animal energy balance.

During the study period, body mass did not significantly change for either sex (p = 0.06). Sex differences in activity budgets throughout the mating period were identified as females spent more time grazing than males, for whom the most common behaviours were standing and resting. The estimated energy expenditure of males was higher than that of females, and the estimated daily energy expenditure varied over the mating season for both sexes.<br><br>Quantification of the rate at which animals ingest food is a fundamental part of animal ecology and can provide information as to the energy obtained by an animal.

Inter-mandibular angle sensors (IMASENs), were deployed to three domestic herbivore species, cattle (Bos taurus), sheep (Ovis aries) and pygmy goats (Capra aegagrits hircus) to quantify feeding behaviour when eating different foodstuffs. Simultaneous video recording and deployment of devices revealed that IMASENs can accurately detect jaw movements of feeding herbivores, consuming various food items. As well as this, chewing and biting could be distinguished based on the rate of jaw movements.

Differences in rate of jaw movements were identified between species (cattle slower than sheep and goats) and between some food items (concentrate pellets were consumed faster than other food items).<br><br>Finally, the findings of the four data chapters are discussed in relation to each other and further interpreted as how these results and future applications of the remote sensing technologies included in this body of work may aid animal production systems to improve farm management and/or food provision efficiency in order to improve animal health, welfare and production.