Objective The problems of resource waste and environmental pollution caused by long-term excessive phosphorus application have become increasingly serious. The aim was to determine a reasonable phosphorus fertilizer input rate for the development of green agriculture.

Method This study was based on long-term field experiments conducted on black soil in Gongzhuling (continuous maize cropping, single-cropping system per year) and manural loessial soil in Yangling (winter wheat–summer maize rotation, double-cropping system per year) (both initiated in 1990). Four typical fertilization treatments were selected: no fertilizer (CK), chemical nitrogen, phosphorus and potassium fertilizers (NPK), NPK plus straw return (NPKS), and NPK combined with manure (NPKM). The corresponding P (P₂O₅) application rates in the different treatments of the two sites were 82.5 and 188.2, 105.3 and 200.0, and 153.4 and 348.1 kg hm⁻², respectively. First, the SPACSYS (Soil-Plant-Atmosphere Continuous System) model was calibrated and validated using long-term observations (crop yield, soil organic carbon stock, total nitrogen stock, and soil available phosphorus stock). Second, based on the current soil available P contents at the two sites and their differences from the agronomic thresholds (Gongzhuling: 13.3 mg kg⁻¹; Yangling: 17.2 mg kg⁻¹) and environmental thresholds (Gongzhuling: 51.6 mg kg⁻¹; Yangling: 40.0 mg kg⁻¹), P fertilizer reduction scenarios were established for the period 2020 - 2060. Since the soil available P contents of all fertilization treatments at the Yangling site had either stabilized below the environmental threshold or shown a rapid declining trend, no reduction in P application rate was required. In contrast, at the Gongzhuling site, the soil available P content under the NPK and NPKS treatments continued to increase and approached the environmental threshold, necessitating moderate P fertilizer reduction, and the soil available P content under the NPKM treatment far exceeded the environmental threshold, requiring substantial P fertilizer reduction. Therefore, P fertilizer reduction scenarios were set only for the Gongzhuling site. Based on the historical crop P uptake, the P application rates for the NPK and NPKS treatments at the Gongzhuling site were set at 80% - 120% of crop P uptake (equivalent to P reduction of 2.8% - 36.1% for NPK and 30.4% - 54.3% for NPKS), while that for the NPKM treatment was set at 20% - 80% of crop P uptake (equivalent to P reduction of 46.5% - 86.6%). By simulating the long-term changes in crop yield, soil available P stocks, and surface P runoff losses under different P fertilizer reduction scenarios, an appropriate P fertilizer reduction strategy was proposed.

Result The SPACSYS model satisfactorily simulated the long-term dynamics of crop yield, soil organic carbon stock, total nitrogen stock, and available phosphorus stock under different fertilization treatments at the two experimental sites. During 2020 - 2060, under the P fertilizer reduction scenarios of 2.8% - 36.1% for NPK, 30.4% - 54.3% for NPKS, and 46.5% - 86.6% for NPKM relative to conventional P application rates at the Gongzhuling site, soil available P stocks significantly decreased by 8.0% - 26.3%, 8.4% - 24.8%, and 51.9% - 76.0%, respectively. However, as the soil available P stocks remained above the agronomic threshold in all cases, long-term maize yield was maintained without significant reduction. Under the NPKM treatment, surface P runoff loss significantly decreased by 37.7% - 82.5% compared with the conventional P application rates across different reduction scenarios.

Conclusion By comprehensively balancing P input, crop yield and P loss, the recommended P application rates for NPK, NPKS, and NPKM treatments at Gongzhuling are 52.7, 48.1, and 20.6 kg P₂O₅ hm⁻², corresponding to reductions of 36.1%, 54.3%, and 86.6% from conventional levels, respectively. This scheme can simultaneously achieve the synergistic goal of maintaining high maize yield while reducing phosphorus loss.