Unraveling the dynamics of tar spot epidemics in corn fields: A DATA-DRIVEN FRAMEWORK FOR disease development and Surveillance

Tar spot is a destructive foliar disease that has significantly impacted U.S. corn production since its first arrival in the mid-late 2010s. This disease is caused by the fungus Phyllachora maydis and despite advancements made to understand this pathogen, the mechanistic progression of tar spot epidemics in corn fields remains poorly characterized. This study investigates the spatiotemporal dynamics of tar spot epidemics over four growing seasons across two field-locations in northwest Indiana from the years 2021 to 2024. Leveraging high-resolution severity assessments, we developed deterministic population growth models and stochastic Markov chains for the 2021-2023 epidemics to evaluate their spatiotemporal progressions across three canopy positions (lower, middle, upper). Our results reveal three epidemiological phases: an establishment phase characterized by sporadic onset with <0.5% severity; a lag phase marked by widespread low severity between 0.5-1% across the field; and an exponential phase, marked by rapid severity increases up to 40% or more. Contrary to the conventional "bottom-up" development theory, we observed diverse infection-like patterns influenced by canopy-position, initial onset timing, and corn growth stage. Exponential growth models best described disease progression and yielded an average relative growth rate of 0.20, providing a general yet adaptable framework across canopy positions, locations, and years. We also built a combined Markov chain model from the 2021-2023 data to estimate the spatiotemporal progression of incidence for a 2024 epidemic, which yielded high accuracy and agreement for all canopy positions. These findings offer critical insights into disease surveillance and potential management, suggesting that generalized fungicide applications may be suboptimal under certain conditions. In the end, our work presents a practical framework that outlines disease development, enhances disease surveillance, supports informed and potentially more targeted management decisions, and ultimately lays the foundation for future integration of real-time detection and probabilistic modeling to further improve agroeconomic and sustainability outcomes.