Nitrogen mineralization characteristics, drivers, and prediction in paddy soils of representative regions in the Yangtze River Delta and Northeast China

Soil mineralized nitrogen (N) constitutes a significant source of plant-available N in paddy systems, and improving its rapid assessment is critical for guiding fertilizer decisions in China's major rice-producing regions. Yet, the spatial variability of N mineralization (Nm) dynamics, and the indicators that best represent long-term Nm across subregions remain insufficiently understood. This study analyzed 36 and 24 paddy soil samples from the Yangtze River Delta (YRD) and Northeast China (NC), respectively, to compare regional Nm behaviors and identify mechanistic and operational predictors. Long-term Nm potential (N0) was substantially higher in the YRD soils, exceeding that of the NC by 29.4%, reflecting regional contrasts in substrate lability. Using anaerobic incubations and first-order kinetic modeling, the shortest representative incubation duration was defined as the threshold at which short-term Nm (STNm) reliably reflected long-term Nm (LTNm). It was found that this threshold differed markedly between regions (7–14 d, R2 = 0.57–0.75), reflecting contrasts in substrate lability and organic matter stabilization. Rapid chemical extractions demonstrated strong potential for estimating long-term Nm. In the YRD, UV-absorbing dissolved organic N fractions (HW260) were closely associated with Nm potential. In contrast, in the NC, hydrolysable hot-water–extractable N (HWN) provided a more mechanistically relevant proxy. Across sites, pH exerted the most decisive influence on LTNm, while other factors varied in their interrelationships. Integrating rapid chemical indices with soil properties significantly improved long-term Nm prediction. The optimal models combined TN, pH, and HW260 in the YRD (R2 = 0.63) and CEC, pH, clay content, and 14-d STNm in the NC (R2 = 0.75), demonstrating high predictive accuracy and clear geographic differentiation. Collectively, this study revealed distinct Nm behaviors among regions, and developed site-specific models by combining short-term incubation and rapid extraction methods, emphasizing the need for region-tailored prediction frameworks for accurate Nm assessment in paddy soils.