Animal husbandry in the 21st century: Application of ecological theory and precision technology to inform understanding of modern grazing systems

Ruminant animals comprise the greatest proportion of herbivores around the world, provide essential ecosystem services and human consumable protein by consuming grass and human inedible dietary fiber. Herbivory pressure alters plant communities and species diversity, effectively making grazing animals ecosystem engineers in dynamic ecosystems. Development of advanced computer processing power coupled with biometric and ecosystem sensors may be employed in the internet of things framework to create an integrated information system designed to inform understanding of grazing system function and animal energy balance. Towards this end, I utilized Bos indicus / Bos taurus crossbred steers (n = 20) across two study sites each in consecutive calendar years and fitted them with GPS and accelerometer collar systems. Steers were grazed in improved grass pastures containing Tall Fescue (Festuca arundinacea) and Bermudagrass (Cyanodon dactylon). Forage samples were collected in a 20-m grid pattern at 35-day intervals to test nutritional composition, and NDVI maps were created using remotely sensed data collected using a UAV mounted camera system. In the first chapter, I utilize the movement ecology framework to investigate metabolic theory and animal behavior on energy budgets, then explore available technology to utilize in an integrative information system. In Chapter 2, I tested preprocessing and behavior collection methods used to train a machine learning randomforest classification model to predict animal behavior using triaxial accelerometers. Landscape functional scale and optimal sampling density is the primary focus of Chapter 3, where I explored the complex relationship between sampling regime, interpolation strategy, and landscape complexity, demonstrating that sampling density is a product of desired accuracy and landscape complexity. Finally, I focused on animal growth in Chapter 4, demonstrating the functionality of a walk-over-weigh system, and identified robust regression as the most accurate smoothing method to identify and remove spurious animal weights