Soil microbial communities and fungal assembly under horizontal spatial configurations in a Larix kaempferi plantation.

Horizontal spatial configuration has emerged as a promising approach for improving plantation structure and ecosystem functions, yet its long-term belowground ecological consequences remain poorly understood. We evaluated how five horizontal configurations of large-diameter trees were associated with variation in soil nutrients, microbial community composition and assembly, and microbial co-occurrence networks over 8 years in a Larix kaempferi plantation. Five configurations were compared: dumbbell-type (R1), torch-type (R2), and 1:2 mixed (RR) random structural units, lower-layer thinning (LT), and an unmanaged control (CK). A total of 15 plots, comprising three plots per configuration, were sampled after 5 years (in 2021) and again after 8 years (in 2024). Spatial configuration showed limited and context-dependent associations with below ground responses. Total potassium differed among configurations in 2021 but converged by 2024, whereas total nitrogen was higher under R2 in 2024; however this difference was not supported by significant treatment × year interaction. Within years PERMANOVA and coverage-standardized fungal diversity analyses also detected no identifiable configuration effect given the current replication level. Fungal tNST and neutral-model outputs provided descriptive evidence but did not establish a configuration-driven temporal assembly transition. Annual pooled networks retained few robust phylum-level associations. Across 20 exploratory configuration-specific phylum networks, every 2024 network contained fewer edges than its 2021 counterpart. Network density patterns varied among domains and configurations. These topology comparisons are hypothesis-generating and do not establish network stability, ecological interaction, or configuration superiority. Overall, horizontal configuration associated with limited and context-dependent belowground signals at the scales examined. More independent plots and standardized cross-year sequence processing are required before configuration-specific network patterns can inform management decisions.