New ways of measuring intake, efficiency and behaviour of grazing livestock Paul L. Greenwood A,B,D , Philip Valencia C , Leslie Overs C , David R. Paull B and Ian W. Purvis B A NSW Department of Primary Industries Beef Industry Centre, University of New England, Armidale, NSW 2351, Australia. B CSIRO Animal Food and Health Sciences, Armidale, NSW 2350, Australia. C CSIRO Computational Informatics, Pullenvale, Qld 4069, Australia. D Corresponding author. Email: paul.greenwood@dpi.nsw.gov.au Abstract. Wireless sensor networks (WSN) offer a novel method for measuring important livestock phenotypes in commercial grazing environments. This information can then be used to inform genetic parameter estimation and improve precision livestock management. Arguably, these technologies are well suited for such tasks due to their small, non-intrusive form, which does not constrain the animals from expressing the genetic drivers for traits of interest. There are many technical challenges to be met in developing WSN technologies that can function on animals in commercial grazing environments. This paper discusses the challenges of the software development required for the collection of data from multiple types of sensors, the management and analyses of the very large volumes of data, determination of which sensing modalities are sufficient and/or necessary, and the management of the constrained power source. Assuming such challenges can be met however, validation of the sensor accuracy against benchmark data for specific traits must be performed before such a sensor can be confidently adopted. To achieve this, a pasture intake research platform is being established to provide detailed estimates of pasture intake by individual animals through chemical markers and biomass disappearance, augmented with highly annotated video recordings of animal behaviours. This provides a benchmark against which any novel sensor can be validated, with a high degree of flexibility to allow experiments to be designed and conducted under continually differing environmental conditions. This paper also discusses issues underlying the need for new and novel phenotyping methods and in the establishment of the WSN and pasture intake research platforms to enable prediction of feed intake and feed efficiency of individual grazing animals. Additional keywords: alkanes, cattle, chromic oxide, goats, phenomics, sheep. Received 16 March 2014, accepted 28 June 2014, published online 19 August 2014 Introduction Improved productivity and production efficiency through both genetic improvement and precision livestock management are becoming increasingly important as resources become more constrained and environmental concerns increase in importance (Reynolds et al. 2011; Scollan et al. 2011). In Australia the major cost of livestock production is associated with the maternal-offspring unit. It has been estimated that the feed costs of the breeding female and her offspring can represent 60–70% of the total herd or flock feed costs (Bell and Greenwood 2013), and as much as 90%, when the rearing of replacement females are included. As such, this is a critical component of the input costs of the meat production enterprise; genetic variation in the amount of feed required to rear offspring and its association with production efficiency should be key targets of the breeding objectives of breeders of meat sires. To date, the focus for determining variation in efficiency of livestock, calculated as either feed conversion efficiency or net feed intake and which require estimates of feed intake of individuals, has been the evaluation of young animals in a feedlot environment where test animals are maintained in group pens and fed grain-based high energy concentrate diets ad libitum (Arthur and Herd 2005). Within pasture systems, chemical markers such as chromic oxide (Barlow et al. 1988) and n-alkanes (Dove and Mayes 2006) have been used to estimate intake, selectivity and/or digestibility of the pasture. However, marker methods have limitations, and are difficult to apply for the lengthy periods that may be needed for robust estimates of an animal’s underlying intake of pasture. The development of a practical measure of feed intake for all classes of animals maintained in a pasture-based environment remains a serious challenge (Cottle 2013) and would provide a means of estimating the heritability and genetic correlations necessary to evaluate the utility of direct and indirect selection criteria for a range of breeding objectives (Pollak et al. 2012) and for more precise livestock management. WSN offer the opportunity to develop new phenotypes for livestock measured in the commercial grazing environment (Greenwood and Bell 2014). These phenotypes would enable CSIRO PUBLISHING Animal Production Science, 2014, 54, 1796–1804 http://dx.doi.org/10.1071/AN14409 Journal compilation Ó CSIRO 2014 www.publish.csiro.au/journals/an