Soil microbes have long been recognized to substantially affect the coexistence of pairwise plant species across terrestrial ecosystems. However, projecting their impacts on the coexistence of multi-species plant systems remains a pressing challenge. To address this challenge, we conducted a greenhouse experiment with 540 seedlings of five tree species in a subtropical forest in China and evaluated microbial effects on multispecies coexistence using the structural method, which quantifies how the structure of species interactions influences the likelihood of multiple species to persist. Specifically, we grew seedlings alone or with competitors in different microbial contexts and fitted individual biomass to a population dynamic model to calculate intra- and inter-specific interaction strength with and without soil microbes. We then used these interaction structures to calculate two metrics of multispecies coexistence, structural niche differences (which promote coexistence) and structural fitness differences (which drive exclusion), for all possible communities comprising two to five plant species. We found that the soil microbes generally increased both the structural niche and fitness differences across all communities, with a much stronger effect on structural fitness differences. A further examination of functional traits between plant species pairs found that trait differences are stronger predictors of structural niche differences than structural fitness differences and that soil microbes have the potential to change trait-mediated plant interactions. Our findings underscore that soil microbes strongly influence the coexistence of multispecies plant systems, and also add to the experimental evidence that the influence is more on fitness differences rather than niche differences.

We conducted the experiment in a greenhouse located at the Heishiding Natural Reserve (111°53′ E, 23°27′ N, 150–927 m elevation), Guangdong Province, south China. Our experiment included five species that vary in their relative abundance in the reserve and in mycorrhizal types: Artocarpus styracifolius (ASTY), Cryptocarya concinna (CCON), Cyclobalanopsis hui (CHUI), Castanopsis hystrix (CHYS) and Cyclobalanopsis pachyloma (CPAC) (Appendix S1: Table S1). Briefly, ASTY and CCON form root associations with arbuscular mycorrhizal fungi, and the others associate with ectomycorrhizal fungi.

In our experiment, we grew either a single individual (to measure the intrinsic growth rate) or two conspecific individuals (to measure intraspecific interaction coefficients) of each species in pots inoculated with soil from conspecific adult trees, or in pots with only sterile soil (5 species × 2 soils × 2 density levels = 20 treatments; Figure 1). Additionally, we grew each pairwise combination of species (10 combinations) to obtain the interspecific interaction coefficients. To do so, we grew each pair of species (i/j) in inoculated soil with species i’s soil microbes and separately with species j’s soil microbes (to quantify interaction coefficients in the presence of microbes), and in sterile soil (to quantify interaction coefficients in the absence of microbes), which led to a total of 30 treatments (10 species pairs × 3 soil microbial treatments = 30 treatments; Figure 1). For each treatment, we established six replicate blocks.

To generate allometric equations for estimating the initial biomass of experimental seedlings, we grew an additional 20 seedlings per species in the same greenhouse conditions in sterilized soils. In September 2020, we harvested these seedlings and measured each individual’s height, basal diameter, leaf count, and total biomass. We performed linear regression models of the total biomass with these three predictors for each species (Appendix S1: Table S2). Using this linear regression, we estimated the initial biomass of seedlings based on height, basal diameter, and leaf count measurements from the beginning of the experiment. In May 2021, after 12 months of growth, we harvested the total (above- and below-ground) biomass for each seedling dried for 48 h at 60 ℃, and weighed.

To investigate how functional traits relate to the microbially mediated plant coexistence metrics, we measured eight plant functional traits that are closely associated with plant life-history: Height (H, the shortest distance between the upper boundary of the main photosynthetic tissues on a plant and the ground level, cm), specific leaf area (SLA, the one-sided area of a fresh leaf divided by its dry mass, cm²/g), total leaf area (TLA, one-sided or projected area of all leaf from the same plant, cm²), root average diameter (RAD, averaged diameter of all roots, mm), root length (RL, total length of all roots, cm), root to shoot ratio (RS, the ratio of plant dry mass in root and shoot), root surface area (RSA, total surface area of all roots, cm²), and specific root length (SRL, length per unit dry mass of roots, cm/g).