Fertilization-driven global and regional dynamics of phosphorus patterns under a representative fertilization...

Phosphorus (P), essential for terrestrial biogeochemical cycles and plant biochemical functions, supports agricultural production and ecosystem stability. However, global soil P imbalances threaten food security and increase environmental pollution risks, necessitating science-based fertilization guidance and P management strategies. The development of these strategies relies on a comprehensive understanding of soil P, particularly inorganic (Pi) and organic (Po) P, dynamics under fertilization. Herein, we employed machine learning to construct a predictive and analytical framework for soil Pi and Po concentrations on a global scale following fertilization. Our model identified macro-environmental and fertilization practices as key drivers of Pi and Po concentration changes, predicted a global decrease in Pi (-2.5 mg/kg) and an increase in Po (+5.5 mg/kg) under an assumed fertilization scenario. A transcontinental Pi increasing belt and high-latitude Pi accumulation areas were identified as primary hotspots, with the Great Lakes Basin exhibiting a substantial Po decline. Furthermore, ecosystems displayed heterogeneous responses to fertilization, with wetlands showing the most pronounced changes, including a 15.6% decrease in Pi and a 9.5% increase in Po. These findings provide a scientific basis for improving understanding of P cycling and may support future efforts in precision fertilization, mitigation of eutrophication risks, and sustainable P management.