Assembly mechanisms and successional patterns of sediment bacterial communities along a riverine wetland restoration chronosequence

Understanding the assembly mechanisms of bacterial communities in sediment is vital for assessing ecological outcomes of wetland restoration. This study examined these processes within sediment bacterial communities along a restoration chronosequence (2, 4, and 8 years) in riverine wetlands. Employing multi-seasonal sampling combined with 16S rRNA gene sequencing, null model frameworks, and co-occurrence network analyses, we observed distinct stage-dependent differences in community assembly. The 2-year wetland was characterized by a relatively high contribution of stochastic processes (notably dispersal limitation), although deterministic processes already dominated assembly. With increasing restoration age, the influence of deterministic processes, especially homogeneous selection, strengthened and remained dominant through the late stage (8 years). Community assembly exhibited significant spatiotemporal variation, with seasonal fluctuations strongly associated with this variation. Concurrently, bacterial co-occurrence networks demonstrated increased modularity, indicative of increasingly modular, functionally compartmentalized community structures. Network topology differed consistently along the restoration gradient, from simple and loosely connected to highly complex and modular configurations, reflecting ecosystem maturation. These findings reveal a pattern of increasing deterministic selection and network modularity with restoration age, thereby advancing understanding of microbial processes during wetland restoration and providing a scientific foundation for ecosystem recovery assessment.IMPORTANCEWetland restoration is widely regarded as a nature-based solution, yet the microbial recovery trajectories following restoration remain poorly characterized. Our study reveals a consistent pattern across restoration ages in sediment bacterial communities across 2-, 4-, and 8-year restored wetlands: from relatively high stochasticity alongside dominant deterministic selection, to strengthened deterministic dominance, and finally to modular, functionally compartmentalized networks. This pattern is associated with restoration duration itself, not with measured environmental factors, and occurs despite strong seasonal variation. Importantly, we show that co-occurrence network modularity and robustness can serve as quantitative, sensitive indicators of ecosystem maturation-more informative than traditional diversity indices. These findings provide a scientific basis for evaluating restoration success and suggest that allowing at least 8 years is necessary for functional stabilization. Our work offers microbial ecological criteria to guide adaptive management and engineering decisions in nature-based wetland restoration.