Mengxue Qu, Yinglong Chen, Hao Ren, Bin Zhao, Baizhao Ren, Jiwang Zhang, Hongzhang Wang, Peng Liu
Fertilization and rhizosphere effects co-regulate soil N cycling and bacterial succession, with mechanistic understanding being key to improving crop yield and N transformation in intensive farmland. Based on a 16-year field experiment, six fertilization treatments with urea (U) and manure (M) alone or in combination (CK, U200, U100, M200, M100, and U100M100; subscript numbers denote nitrogen application rates in kg N ha-1) were established to clarify the responses of bacterial communities in two components (bulk and rhizosphere) and their associations with N dynamics and maize yield. Relative to U200, U100M100 increased grain yield by 5.6%-6.5% (2023-2024), and improved N uptake and nitrogen-use efficiency, while reducing total N loss by 44% and N footprint by 45% (average 2024-2025). Manure substitution enriched copiotrophic phyla (Proteobacteria and Bacteroidota), particularly under manure-only treatments. U100M100 lowered network complexity and increased negative interactions, while decreasing N-loss-associated ASVs. Moreover, U100M100 raised the functional potential of genes regulating nitrification, dissimilatory nitrate reduction to ammonium and N fixation in bulk soil, while promoting complete denitrification in the rhizosphere. Enriched dominant phyla and divergent N-cycling gene profiles were closely associated with N losses. Nitrogen-cycling genes and ecological modules exhibited the strongest standardized total effects on plant N uptake and maize yield. Mantel tests and PLS-PM R2 comparisons further confirmed that rhizosphere microbial communities were more strongly associated with these agronomic variables than bulk soil communities. In summary, the U100M100 treatment may establish a synergistic and efficient N-cycling system, in which the rhizosphere functions as an N loss mitigation module while bulk soil serves as an N supply module. This spatial functional divergence may result from reshaped microbial communities, reconstructed network topology and altered N-cycling functional potential.