Combining field experiments, modeling, and analytics for integrated crop-livestock system management

Integrated crop-livestock systems have been suggested as a promising means for diversifying agricultural production as an alternative to specialized cropping systems in modern agriculture. In the U.S. Corn Belt, tremendous amounts of corn residues left on the land after harvest represent a vast potential nutrition source for beef cattle producers, particularly when farms are specialized in continuous corn operations. Allowing cattle to graze corn residues after harvest has been suggested as a simple and economical approach to integrating cattle and corn operations. While residue grazing can significantly reduce winter feed cost, its effects on croplands remain unclear. For example, spatially uneven grazing may potentially lead to heterogeneous impacts on cropland, possibly resulting in adverse impacts on subsequent crop development. Furthermore, the integration usually increases system complexity, and consequently requires more complicated management. This dissertation seeks to provide insights into the development of decision support for managing integrated corn-cattle systems, through combining field experiments, data analytics, and model simulations. A three-year residue grazing trial was conducted on an integrated corn-cattle farm in central Illinois. Two grazing management practices (continuous grazing and strip grazing) were both investigated during the trial. A custom GPS tracking system was successfully developed, tested, and then implemented to monitor beef cattle movements during the residue grazing. Results of GPS accuracy evaluation showed that the mean horizontal error of the tracking system was less than 4 meters and was not significantly affected by its position on a cow’s neck. The long-term performance and reliability of the system were evaluated based on the maintenance records and the success of data collection across multiple years' tests and applications. Results showed that intermittent data loss was typical during the studies. An initial adaptation period was recommended for effectively reducing shaking and rubbing that may lead to damage to the GPS equipment or problems for the animals. Data collected during the grazing studies were investigated via a set of analytical tools to identify the spatiotemporal characteristics of cattle movements during residue grazing. Results of the movement characterization showed that individual cow movements were highly synchronized within the group. Cattle under the strip grazing treatment had averagely greater daily travel distances than those under the continuous grazing treatment. Results of the spatial analysis show that cattle had spatially heterogeneous visitations to the residue fields. The distributions of cattle locations were affected by management factors such as the locations of supplement feeders and the cross-fence settings in the strip grazing treatment. Results of the periodic pattern analysis suggested that cattle had periodic movements associated with their bedding areas likely due to the circadian rhythmicity of their behaviors. Besides mining GPS data for movement patterns, a computational approach was developed to evaluate errors in subsequent GPS data analysis caused by monitoring subset groups of cattle instead of the entire herd. Results suggested that monitoring an appropriate subset group can be sufficient to preserve information with acceptable errors for subsequent analysis. Results of grazing impact analysis showed that the mean subsequent corn yields were not affected by residue grazing under the two management treatments (i.e. continuous grazing and strip grazing) as compared with the ungrazed control at the study site. However, areas around the supplemental feeders have shown a significant decline in yield after the grazing, which was likely caused by the trampling effects of heavy cattle traffic. While experimental validation is needed in the future, it is expected that management strategies such as moving supplemental feeder during residue grazing may alleviate such negative impacts. With the field data collected during the grazing experiments, a spatially explicit agent-based model has been developed to simulate cattle movements during corn residue grazing. The model was calibrated and validated for two management scenarios (continuous grazing and strip grazing) using the experimental data. Results showed that the simulated spatial distributions of cattle during grazing were consistent with observation data. It is anticipated that, with further model refinement, this model can serve as a research tool to aid future development of decision support tools for managing integrated corn-cattle systems