Process-based simulation of soybean rust in Brazil: A DSSAT-coupled approach

Soybeans are one of the most important crops worldwide, and a comprehensive understanding of the yield-limiting factors is critical to strengthening farmers’ capacity to adapt across different environments. Among the biotic factors, Asian soybean rust (ASR), caused by Phakopsora pachyrhizi, is one of the most damaging foliar fungal diseases, capable of causing severe yield losses. Models able to predict yield losses due to diseases are fundamental to support decision-making and risk analysis. However, research on the development of crop models integrating plant disease dynamics is still limited. In this study, we adapted a generic epidemiological model to predict ASR dynamics by coupling it to the CROPGRO-Soybean model within DSSAT. Simulated disease progress curves were compared with observations from epidemics in two major soybean-producing regions of Brazil. A two-step calibration was used, guided by parameter sensitivity, identified using the Morris method, and Monte Carlo simulations. First, 15 biological parameters were calibrated, followed by the tuning of two inoculum-specific parameters for each epidemic curve. Leave-one-out cross-validation was then applied. Two representative sets of biological parameters were obtained for each region, with mean R² values of 0.91 and 0.93 and RMSE of 7.55% and 7.18% for the Southern and Central regions, respectively. Cross-validation confirmed robustness, with Willmott’s concordance indices (d/d1), Nash–Sutcliffe efficiencies (EF), and RMSE values of 0.95/0.82, 0.81, and 11.89% for the Southern region and 0.98/0.90, 0.93, and 8.38% for the Central region. Results suggest that the coupled model provides a robust framework for representing ASR dynamics under diverse conditions, underscoring the need for regional parameterizations to account for location-specific epidemic drivers. In addition to supporting decision-support systems and prospective climate change analyses, this study demonstrates the adaptability of the coupled-DSSAT epidemiological module in creating opportunities for integration with other crop models and for applications across different pathosystems and cropping systems.