Abstract
In arid sandy plantations, stand density critically regulates belowground ecosystems, yet its effects on microbial network complexity, stability, and function are not fully understood. This study examined Pinus sylvestris var. mongholica plantations along a density gradient (Very high density (VHD): 2,450 trees ha(-1), High density (HD): 1,633 trees ha(-1), Moderate density (MD): 1,067 trees ha(-1), Low density (LD): 583 trees ha(-1)) at two soil depths (0-20, 20-40 cm) in the Otingdag Sandy Land, integrating soil physicochemical, enzymatic, and microbial network analyses. Key findings were: (1) Soil organic carbon, total nitrogen, key enzyme activities [urease (URE), acid phosphatase (PHO), and nitrate reductase (NR)], and ecosystem multifunctionality showed a unimodal response, peaking at medium densities. (2) Microbial responses diverged: bacterial α-diversity changed but composition remained stable, whereas fungal composition was highly density-sensitive. Mid- to low densities promoted more complex, modular, and stable microbial networks. Mantel tests identified pH, URE, and Ammonium nitrogen (NH₄(+)-N, AN) as key drivers for bacterial phyla, and pH, Soil organic carbon (SOC), Total nitrogen (TN), PHO, and Polyphenol oxidase (PPO) for differentiating Ascomycota and Basidiomycota. (3) Random Forest regression identified microbial network stability as the top predictor of multifunctionality, surpassing diversity. Partial Least Squares Path Modeling (PLS-PM) analysis revealed that stand density enhances multifunctionality primarily by improving the soil environment, with microbial networks acting as environment-dependent regulators. This study demonstrates that moderate stand densities optimize microbial network resilience and ecosystem multifunctionality in sandy plantations, providing a novel perspective from microbial network stability.