51 Opportunities to Apply and Learn from Deep Phenotyping in Dairy Cattle

Abstract

Abstract Data from automated systems including sensors, cameras and high-throughput assays, such as milk testing, generate huge amounts of valuable data for research and on-farm applications. The potential to use data collected from automated systems in real-time for predictive analytics or as indicator traits for genetic selection is an exciting opportunity for the dairy industry to address difficult phenotypes. While high frequency recording of data provides high-resolution views of animal behavior, it also creates logistical challenges for determining what data are most informative, when or if to use repeated measures, and how to integrate these repeated measurements with static or less frequently measured traits of interest. Measuring the precision and accuracy of data from automated systems is also a continued challenge. As feed is the largest cost on commercial dairy farms and intake is difficult to measure, our lab is investigating the use of different wearable and stationary sensors to monitor cow and environmental level measurements to determine their utility as potential indicators of variation in feed intake. Associations of sensor measures with feed intake and relationships with health events have been identified. Ongoing research is evaluating the ability to predict trait phenotypes with various sensor data. Beyond the potential to learning more about trait relationships and underlaying genetics of behavior, there is also the potential to develop precision management tools from sensors for producers. Integrating automated systems measurements with research data (e.g. omics) may help to fuel new discoveries and tool development. Future advances in sensing technologies will require transdisciplinary research to develop new types of sensor measures and data analytics to identify hidden information within the data that may lead to new actionable applications on farm.