Spatial divergence of virus-bacteria interaction networks in upstream and downstream sediments of the Three Gorges Dam and their association with sediment ecological multifunctionality

The Three Gorges Dam (TGD) has substantially altered the hydrodynamic and sediment-transport regime of the Yangtze River, yet how the resulting environmental change reshapes sediment microbiome organisation and ecosystem multifunctionality remains poorly understood, particularly whether cross-domain virus-bacteria interaction networks, beyond diversity alone, are the key structure linking environmental gradients to function. We sampled sediments at five sites along the upstream-downstream gradient of the TGD (n = 50) and combined physicochemical analyses with internal-standard-based absolute quantitative metagenomics to characterise the abundance, diversity and co-occurrence network structure of viral and bacterial communities and their links to multifunctionality. Downstream sediments were markedly depleted in dissolved organic carbon and cation exchange capacity (both down by approximately 50%), with synchronous declines of approximately 60% in viral and bacterial absolute abundances. Directional dependence analysis revealed asymmetric coupling (q = 0.57 vs. 0.39), indicating that shifts in bacterial community structure constrained viral community variation. Downstream networks lost 61.42% of edges, driven mainly by preferential stripping of an environmentally sensitive dynamic component, whereas a conserved component persistent across all sites formed a stable backbone. Bacterial diversity, conserved viral diversity and network interactions jointly explained 59.60% of the spatial variation in multifunctionality, and structural equation modelling showed that network interactions exerted a stronger direct effect than diversity, with diversity acting mainly indirectly through the network. These findings identify the structural divergence between dynamic and conserved network components as a key mediator linking dam-induced environmental gradients to sediment multifunctionality, and provide a testable network-level framework for assessing the ecological effects of large hydraulic projects on riverine sediments.