Nitrous oxide (N₂O) emissions from paddy soils significantly contribute to global warming; however, the regulatory mechanisms of microbial denitrification remain poorly understood. This study investigated the biotic and abiotic drivers of N₂O production and reduction across seven paddy soils spanning China’s major rice-growing regions, using integrated qPCR, incubation experiments, and multivariate analyses. Results demonstrated niche partitioning among denitrifying microorganisms. Pearson correlation analysis demonstrated significant positive correlations between potential N₂O production rates and the abundances of denitrification genes (nirS, nirK, and fungal nirK), as well as between N₂O reduction rates and nosZ gene abundances (both clade I and II). Key soil chemical properties, including pH, total carbon (TC), and NH₄⁺-N content, showed significant relationships with both potential N₂O production rates and reduction rates. Furthermore, random forest analysis identified nirS, fungal nirK, TC, and pH as key predictors of N₂O production, while nosZ (clade I and II), TC, and pH governed N₂O reduction. Structural equation modeling revealed that nirS-type bacteria predominantly drove N₂O production, whereas nosZ II-encoded microorganisms primarily mediated N₂O reduction. Moreover, TC exhibited direct positive effects on both processes, while pH indirectly influenced N₂O production by regulating nirS abundance and affected reduction via nosZ Ⅱ modulation. These findings provide a mechanistic framework for mitigating agricultural denitrification-derived N₂O emissions through a targeted management of soil carbon and pH conditions to optimize complete denitrification.

Moisture, salinity and aeration in saline–alkali soil are three critical factors affecting the biotic and abiotic environment. A three-factorial experiment including two tillage measures (ridge and flat tillage, denoted as R and F, respectively), three drip irrigation levels (soil wetting proportion of 40, 55 and 70%, denoted as P₁, P₂ and P₃) and the presence or absence of air injection (AI) were investigated to determine their effects on soil moisture, salinity, aeration and sunflower growth and yield. Field trials were conducted in the Hetao irrigation district of Inner Mongolia in 2021 and 2022. Results showed that R increased daily average topsoil (0–20 cm depth) temperature by 0~4.7 °C, water-filled pore space (WFPS) by 4.3~9.1% and redox potential (Eh) by 16.7~31.6% compared to F. P₂ reduced the Eh of topsoil by 50.4% and 55.1% respectively under R and F. Under the same P, the effect of different tillage methods (R and F) on salt accumulation was not notable. AI increased topsoil temperature under R (0.1~2.7 °C) and F (0~2.2 °C) and increased salt accumulation in the topsoil. Compared with other treatments, the yield of sunflower increased by 10~36% and 12~37% respectively under the conditions of P₃R and P₂FAI. The net profit of P₃R treatment was 3421–3551 USD ha⁻¹, which was 12.0–71.5% higher than the other treatments. Furthermore, random forest analysis revealed that air injection, salinity, tillage method, and Eh were the primary determinants of sunflower yield and quality, while WFPS and temperature were of secondary importance. These findings also suggest that, in heavily saline–alkali soils, sunflower yield can be effectively enhanced either by adopting flat tillage with air injection or by using ridge tillage without air injection.

To address the challenges of cumulative pseudo-label noise and substantial phenological differences across regions in cropping structure mapping, this study develops a phenological knowledge-guided weakly supervised semantic segmentation framework, termed PhenoStruct-WSF. The framework first constructs typical phenological curves for major annual cropping patterns and uses TWDTW to measure the phenological similarity between samples and templates. A two-class margin indicator, ∆_i, is then proposed, and phenological confidence is derived through probability calibration, after which high-confidence samples are selected as anchors under posterior-FDR control. Subsequently, the anchors are embedded into a structure-aware weakly supervised segmentation network, PhenoStruct-Net, forming a two-stage training strategy of “fidelity first, expansion later”, in which robust pseudo-label expansion is achieved through phenological confidence gating and adaptive weighting. Experiments were conducted in three representative study areas, namely Handan in Hebei Province, Jingzhou in Hubei Province, and Siping in Jilin Province. The anchor samples selected using posterior-FDR achieved accuracies above 94% in all three study areas. PhenoStruct-WSF consistently achieved overall accuracy (OA), mean intersection over union (mIoU), and macro F1-score (mF1) above 89%, 81%, and 89%, respectively, across the three regions, with an average OA improvement of 4.52 percentage points over the comparison methods. Sample sensitivity experiments indicate that the proposed method maintains a stable trend of accuracy improvement under different training sample sizes, demonstrating good adaptability to limited samples. The ablation experiments further show that phenological confidence anchors, posterior-FDR control, and the confidence-weighting mechanism can effectively suppress pseudo-label noise propagation and improve the robustness of crop mapping in complex agricultural landscapes.