CONTEXT: Process-based models can jointly simulate crop yield and soil N2O emissions, enabling evaluation of crop production and environmental effects under field management. However, site-scale calibration remains challenging because multiple parameters and coupled crop growth-soil nitrogen processes jointly control these outputs. Multi-objective optimization can constrain multiple targets, but its application to parameter calibration in APSIM Next Generation (APSIM NG) remains limited. OBJECTIVE: This study used the Non-dominated Sorting Genetic Algorithm II (NSGA-II) to calibrate crop and soil parameters in APSIM NG, aiming to improve simulations of wheat yield and soil N2O emissions under nitrogen gradients. After validation, the optimized model was used for long-term simulations to explore responses of yield and growing-season soil N2O emissions to nitrogen gradients under climate scenarios. METHODS: Sobol sensitivity analysis identified 8 crop parameters sensitive to wheat yield and soil N2O emissions, and 5 soil nitrogen transformation parameters were selected based on model mechanisms and literature evidence. These 13 parameters were optimized using NSGA-II, with NRMSE for yield and NRMSE for soil N2O fluxes as the two objectives. Field observations from nitrogen treatments N0-N3 during the 2021-2022 calibration period on the Loess Plateau of central Gansu, China, were used for parameter optimization, while 2023-2024 data were used for independent validation. Long-term simulations used historical meteorological data (1975-2024) and future climate projections (SSP126, SSP245, and SSP585; 2026-2075). RESULTS AND CONCLUSIONS: Sensitivity analysis indicated that crop growth and phenological parameters were more influential in simulations of yield and soil N2O emissions than nitrogen concentration threshold parameters. After optimization, RMSE and NRMSE for yield decreased to 150.29 kg center dot ha(-1) and 19.38%, respectively, with NSE increasing to 0.66. For growing-season soil N2O emissions, RMSE and NRMSE decreased to 0.163 kg N center dot ha(-1) and 21.17%, respectively, with NSE increasing to 0.49. The NSGA-II-derived Pareto front indicated a trade-off between the two optimization objectives. Independent validation showed that the optimized model adequately simulated treatment-level differences in yield and growing-season soil N2O emissions. Long-term simulations indicated that, under current site conditions, yield gains diminished with increasing nitrogen input, whereas N2O emissions continued to increase. Linear mixed-effects model analysis indicated that nitrogen treatment explained more variation in outputs than climate scenarios, with climate scenarios mainly modulating response magnitude rather than changing response direction. SIGNIFICANCE: This site-specific methodological case study shows that NSGA-II can support joint calibration of crop and soil parameters in APSIM NG, providing a methodological reference for multi-objective calibration in similar modeling contexts.