14 An agent-based modeling approach for balancing feed efficiency and environmental footprint of growing-finishing pig production systems.

Abstract The United States (US) pork industry is a significant economic sector, contributing $62.7 billion annually to the gross national product. However, this economic success comes with substantial environmental challenges. Life cycle assessment indicates that nearly 81% of the potential impacts of global warming from agriculture are associated with livestock production. Specifically, pork supply chains account for about 9% of total greenhouse gas (GHG) emissions from livestock, with swine feed production being the second-largest source of emissions in the US. This research employs an agent-based model (ABM), developed in NetLogo, to thoroughly assess the environmental footprint of various swine feeding strategies and optimize nutrient utilization by individual pigs, ensuring an overall minimal environmental impact. Unlike previous studies that relied on average feed intake and growth rates, which masked inherent variability and uncertainty, the proposed model simulates individual pigs’ dynamic feed intake, metabolism, and growth processes. The stochastic nature of our ABM allows us to generate statistical distributions that explicitly account for natural biological variations among growing-finishing pig populations. All agents were designed with variations in key attributes such as body weight gain, starting and finishing weights, sex, and feed ingredient variety to more accurately reflect the inherent complexity of pig development and the variability within a herd. The model estimates critical nutritional parameters, including daily feed intake, and requirements for protein, amino acids, nitrogen (N), calcium, phosphorus (P), and other essential nutrients. This study emphasizes five key environmental indicators: GHG emissions, including carbon footprint, N and P excretion, water consumption, and land use per market hog. The ABM calculates these environmental impacts and animal performance for different feeding strategies across growing-finishing pig populations with various herd sizes, sexes, and geographic distributions. We explore the effects of feeding different diets on these environmental indicators, including varying proportions of corn, soybean meal, distillers’ dried grains with solubles, and synthetic amino acids. The model also accounts for regional heterogeneity by simulating swine production across three production regions in the US, examining how optimizing nutritional strategies tailored to individual pig variation affects regional environmental footprints. This spatial analysis with our ABM concludes which regionally specific feed optimization strategies are crucial for minimizing environmental impact. The ABM was calibrated and validated using data from existing literature. The results offer detailed insights into the trade-offs between different feeding strategies and their environmental consequences, providing recommendations for enhancing the sustainability of swine production. Our follow-up studies will incorporate optimization algorithms to identify the best nutritional availability and utilization strategies for individual pig requirements, minimizing environmental footprint while ensuring sufficient nutrition. Scaling up the industrial adoption of our model will aid in developing more environmentally responsible and economically viable swine farming practices.