Predicting and Optimizing Cotton Yield and Nitrogen-Phosphorus Fertilization Using the DSSAT Model and Ecological Stoichiometry Theory
摘要
Cotton cultivation faces sustainability challenges due to excessive fertilization practices. Hence, how to rationally optimize cotton fertilization management, and to achieve the reduction of fertilizer input and the increase of benefits in cotton production, is a key problem that this paper intends to solve. In the first year, we established a control using a local recommended fertilization treatment (450 kg ha-1 of urea and 375 kg ha− 1 of monoammonium phosphate) and implemented six nitrogen-phosphorus reduction experiments to investigate the ecological stoichiometric characteristics and interactions between cotton leaves and root soil during flowering and boll-setting stage. After determining the theoretical optimal fertilization treatment based on ecological stoichiometry theory, we validated the optimization effects using machine learning combined with DSSAT simulations. In the second year, we further tested the theoretical optimal fertilization treatment in the field. Cotton leaf N: P ratios were below 13, and root soil N: P ratios were below 2.35, indicating nitrogen limitation in both. The DSSAT + XGBoost algorithm effectively modeled growth responses to various fertilization treatments. The treatment reducing nitrogen to 90% and phosphorus to 80% better aligned with the nutrient balance state predicted by stoichiometric principles compared to the local recommendation, corresponding to the maximum yield (increased by 8.3%) and economic benefits (improved by 12.1%) of cotton. Applying ecological stoichiometry theory can effectively guide fertilization strategies. Rational optimization of nutrient ratios in nitrogen-limited areas not only improves environmental sustainability but also enhances agricultural productivity through balanced element management.