Abstract
BACKGROUND: The alpine grasslands of the Qinghai-Tibetan Plateau (QTP) are critical ecosystems for regional climate regulation and biodiversity but are threatened by severe degradation. The success of ecological restoration in these fragile environments is intrinsically linked to the reassembly of belowground soil microbial communities. However, the specific successional trajectories and network-level mechanisms driving their reassembly during long-term restoration remain unclear. OBJECTIVE: This study aimed to characterize the distinct successional trajectories of bacterial and fungal communities and to elucidate the topological mechanisms driving their transition along a long-term restoration chronosequence. METHODS: Utilizing a space-for-time substitution approach on "Black Beach" degraded plots, we integrated consensus clustering with topological null model analyses to quantify community assembly processes. We evaluated the transition from stochastic to deterministic assembly by calculating Modularity Z-scores and analyzing structural changes in co-occurrence networks. RESULTS: Microbial reassembly follows a predictable, non-linear trajectory comprising an early "Chaos Stage" (structural fragmentation) and a later "Recovery Stage" (high cohesion and deterministic interactions). Bacteria and fungi exhibited decoupled temporal dynamics: bacteria displayed an oscillatory "fast-in, fast-out" pattern, whereas fungi followed a "lagged but persistent" trajectory. Network analysis revealed a shift from random associations to competitive dominance. While the Recovery Stage exhibited increased complexity, it revealed a hidden topological fragility where stability became heavily dependent on specific keystone taxa, particularly within fungal networks. CONCLUSIONS: Microbial succession is driven by a regime shift from stochastic to deterministic network assembly. The identified topological fragility constitutes a core mechanism underlying the differential stability of fungi relative to bacteria. Monitoring network vulnerability and kingdom-specific decoupling provides a more accurate indicator for evaluating long-term ecosystem restoration success than species diversity alone.