Using Bio-logging to Improve Sheep Health and Performance

英国养羊业拥有超过1600万只种羊，每年为英国经济贡献约4.659亿英镑。健康问题使该行业损失了数百万英镑，并造成了重大的福利问题。两大相互关联的挑战制约着该行业并威胁着未来的粮食安全：i）需要改善动物健康和福利; ii）提高产量的压力。目前的绵羊育种计划通常侧重于易于测量的生产措施，如胴体性状。越来越多的证据表明，行为特征，如气质，孕产妇护理，攻击和亲社会行为可以共同预测生产（如生长率）和健康（如免疫功能）。然而，迄今为止，在商业羊群中测量行为的实际困难意味着行为特征的遗传基础在很大程度上仍然未知。目前迫切需要开发新的方法来监测绵羊的行为，并量化行为、健康和产量之间的关系。该项目与英国的行为测量设备制造商Activinsights合作，将使用动物传感器和机器学习开发新的方法来自动分类绵羊的行为和健康。我们将使用加速度计自动跟踪行为（例如放牧，步行），GPS设备来确定行为发生的位置，并使用接近标签来记录社交接触的模式。我们将联合收割机将这些多个生物记录数据流与动物健康和生产（例如生长和怀孕率）的直接观察结合在一个基于事件的数据平台中。利用这个数据平台，我们将使用机器学习来数字化测量健康和行为特征。这项研究将与Centurion合作完成，Centurion是一个拥有数千只绵羊超过25年生产数据的无角多塞特育种者组织。与这个系谱品种群的合作将使我们能够确定功能性状变异的遗传基础，并预测基因型在不同农场管理条件下的表现。该项目是“农业和粮食安全”和"健康生物科学“BBSRC研究重点的核心，并与BBSRC的一些优先事项直接一致。它将采取数据驱动的方法，并通过与国家和国际合作伙伴合作，制定新的动物健康和行为措施，以改善管理动物的福利。通过与研究的最终用户合作，该项目有望可持续地提高农业生产。

The UK sheep industry has more than 16M breeding sheep and contributes an estimated £465.9M p.a to the UK economy. Health problems cost the industry millions of pounds and cause significant welfare issues. Two major interlinked challenges constrain the industry and threaten future food security: i) a need to improve animal health and welfare; and ii) pressure to improve production. Current breeding programmes in sheep generally focus on easily measured production measures such as carcass traits. There is mounting evidence that behavioural traits such as temperament, maternal care, aggression and prosocial behaviour can jointly predict both production (e.g. growth rates) and health (e.g. immune function). To date, however, the practical difficulties of measuring behaviour in commercial flocks means that the genetic basis of behavioural traits remains largely unknown. There is an urgent need for the development of new methods to monitor sheep behaviour and to quantify the relationships between behaviour, health and production.In partnership with Activinsights, a UK-based manufacturer of devices for the measurement of behaviour, this project will use animal-borne sensors and machine learning to develop new methods to automatically classify sheep behaviour and health. We will use accelerometers to automatically track behaviour (e.g. grazing, walking), GPS devices to determine where the behaviour occurs and proximity tags to record patterns of social contact. We will combine these multiple streams of bio-logging data with direct observations of animal health and production (e.g. growth and pregnancy rates) in a single event-based data platform. Using this data platform we will use machine learning to digitally measure health and behavioural traits. The research will be done in collaboration with Centurion, a group of Poll Dorset breeders with over 25 years of production data on thousands of sheep. Working with this pedigree breed group will allow us to determine the genetic basis of variation in functional traits and predict the performance of genotypes under different farm management conditions.This project is central to the 'Agriculture and Food Security' and 'Bioscience for Health' BBSRC research priorities and is directly aligned with a number of the BBSRC priorities. It will take a data-driven approach and by working with national and international partners will develop new measures of animal health and behaviour to improve the welfare of managed animals. By collaboratively working with the end-users of the research this project promises to sustainably enhance agricultural production.