Multiobjective spatial optimization of fertilizer rates enables sustainable crop production in southwest China

Achieving the dual goals of ensuring food security and environmental security is a global challenge in sustainable nutrient management. However, previous studies have neglected the spatialization of nutrient management. Here, we proposed a new approach to assess and optimize fertilizer rates by combining data-driven forecasting and machine learning methods to address spatially optimal allocation of nutrient resources. We found that the contribution of fertilizer application to crop yields in Southwest China decreased by 1–3% from 2009 to 2019. Crop nutrients were widely unbalanced, with obvious nitrogen excess, local phosphorus deficiency, and general potassium deficiency, while plains and riverbanks were hotspots of nutrient excess. Multiobjective optimization reduced the nitrogen fertilizer rate by 18% (11×104 t) with a slight increase in the crop yield, whereas the ratio of nitrogen, phosphorus, and potassium fertilizers was optimally adjusted from 1:0.38:0.33 to 1:0.51:0.42, conforming with national expectations.