Network-scale spatiotemporal dynamics and drivers of nitrogen-cycling genes and their microbial hosts in a multi-tributary mountain river.

Rivers are critical regulators of the global nitrogen (N) cycle, yet their microbial nitrogen-cycling dynamics and feedbacks to environmental change remain poorly understood at the river-network scale. Here, we combined systematic field sampling at 187 water-column sites spanning two mainstems and 22 tributaries of the upper Hanjiang River with shotgun metagenomics, metatranscriptomics and 30 environmental variables representing geography, climate, hydromorphology, land use, and water quality. Our results showed that nitrogen-cycling genes (NCGs) mediating six major pathways showed clear longitudinal differentiation along the cumulative dendritic distance upstream. Compared to small-river-network (SN), Large-river-network (LN) consistently exhibited higher abundances, diversity and expression of denitrification, DNRA, ANRA, ODAS, and N fixation genes, identifying LN as convergence zones for N transport and hotspots of N removal. By contrast, nitrification genes (amoA, nxrB, hao) and their hosts were scarce, with only a transient autumn increase in LN. We recovered 1508 medium- to high-quality NCG-hosting MAGs spanning a broad phylogenetic range but strongly skewed toward a few bacterial phyla. Pseudomonadota, Actinomycetota and Bacteroidota together accounted for ∼87% of total NCG-host abundance. Assembly- and MAG-based community analyses revealed a shift from low-diversity, Pseudomonadota-dominated assemblages in SN to diverse, high-evenness communities in LN. Along the SN-LN gradient, MAG co-occurrence networks became richer, denser and more modular, and keystone MAGs were disproportionately enriched in denitrification and DNRA genes, with nitrification genes rare among hubs. Water quality emerged as the dominant driver of NCGs and their hosts, explaining up to 47.3% of the variance, while geography and land use exerted pathway-specific influences through indirect effects mediated by hydromorphology and water quality. Metatranscriptomic norB/nosZ further revealed a nitrate tipping point at ∼0.8 mg L⁻¹, above which microbial N2O emission potential increased by an order of magnitude and formed spatially confined, seasonally shifting hotspots. Collectively, our results demonstrate that riverine nitrogen-cycling functions are structured by hierarchical multi-scale controls and that N2O emission potential responds nonlinearly to nitrate loading, with LN acting as critical integrators of SN inputs. These findings highlight the need for river-network-scale N management to enhance riverine self-purification while mitigating microbial N2O feedbacks arising from human activities.