Developing the DSSAT-CERES-Millet Model for Dynamic Simulation of Grain Protein and Starch Accumulation in Foxtail Millet (Setaria italica) Under Varying Irrigation and Nitrogen Regimes

Foxtail millet (<i>Setaria italica</i>), vital in northern China, has its quality and taste influenced by starch and protein. Existing models do not simulate the accumulation of these components during growth. To address this, we enhanced the DSSAT-CERES-Millet model (referred to as DSSAT) by integrating two newly developed modules: the protein simulation module and the starch simulation module. The protein simulation module uses a nitrogen-to-protein conversion coefficient to determine grain protein accumulation based on grain nitrogen accumulation simulated by the DSSAT model. In the starch simulation module, the carbon source supply (carbohydrates) received by millet grains is calculated based on the simulated aboveground and vegetative dry matter by the DSSAT model, and starch synthesis is modeled using the Michaelis-Menten equation to convert carbohydrates into starch within the grains. The integrated model demonstrates good performance in simulating grain protein and starch accumulation, with NRMSE (normalized root mean square error) values of 3.06-26.22% and 4.06-26.88%, respectively. It also accurately simulates grain amylopectin and amylose accumulation at maturity, achieving an NRMSE of less than 14%. The enhanced DSSAT-CERES-Millet model can provide guidance for optimizing irrigation and nitrogen management to enhance the protein and starch quality of millet grains.