Plant trait turnover and microbe networks drive forest multifunctionality across landscapes

In the species-rich forests of southwestern China, the fate of an ecosystem’s many simultaneous jobs—storing carbon, cycling nutrients, holding water, decomposing litter, and growing biomass—does not hinge on how many species live in any one place. Instead, it hinges on how differently those species are put together from plot to plot, and on how the web of connections between plants and soil microbes is rewired across the landscape. That is the central finding of a new study of 24 subtropical forest plots in Yunnan Province, published in the journal Ecological Processes, which quantifies for the first time how plant functional trait turnover and plant–microbe network turnover jointly dominate the spatial variation of ecosystem multifunctionality at landscape scales.

The research team, led by Ming-Hui Wang and colleagues working in Simao District of Pu’er City, established twenty-four 30-by-30-meter survey plots in a monsoon evergreen broad-leaved forest region, with more than 100 meters separating neighboring plots. The area, sitting between 1,310 and 1,650 meters in elevation with roughly 1,490 millimeters of annual rainfall concentrated in the May-to-October monsoon, contains both pine-dominated secondary forests and the Castanopsis-rich climax broadleaved forest. In each plot the researchers inventoried every woody plant with a stem diameter of at least one centimeter, measured nine plant functional traits ranging from leaf chemistry to seed mass, sampled soils from the top 20 centimeters, and characterized bacterial and fungal communities alongside a suite of soil nutrient and enzyme measurements.

The target of the analysis was not multifunctionality itself but its spatial turnover, which the authors call β-EMF. Drawing on sixteen indicators spanning five categories of ecosystem function—soil nutrient cycling, carbon sequestration, water conservation, decomposition, and plant productivity—the team applied a recently developed framework based on Hill-Chao numbers. Unlike simpler averaging approaches, this method treats each function as a distinct attribute, corrects for correlations among functions, and can decompose regional-scale multifunctionality into local and between-site components. The measured β-EMF values across the plots ranged from 1.0 to 1.5, with a mean of 1.13, indicating substantial spatial variation in how evenly multiple functions are distributed across the landscape.

To find what drives that variation, the researchers turned to Multiple Regression on Distance Matrices, a technique suited to the pairwise distance data generated by comparing all 276 possible plot pairs. In these single-predictor models, plant functional β-diversity emerged as the strongest single explanation, accounting for 54.3 percent of the variation in β-EMF. Plant species β-diversity and phylogenetic β-diversity followed, explaining 43.3 and 41.4 percent respectively. Notably, both components of a novel variable—the β-diversity of the plant–microbe association network—also showed strong effects: the species replacement component explained 40.8 percent and the connection reconstruction component 36.1 percent. Bacterial β-diversity, soil pH distance, and geographical distance each explained roughly a third of the variation, while topographic distance, fungal β-diversity, and soil water content distance showed no statistically significant relationship.

A stepwise multivariate version of the same regression narrowed the field to six significant drivers. Plant functional β-diversity and the species replacement component of the plant–microbe network tied for the largest relative contributions, at 21.9 and 21.7 percent of the explained variation, with the network’s connection reconstruction component close behind at 19.7 percent. Together these three predictors accounted for 62.3 percent of the total explained variation in the final model. Plant species β-diversity contributed 14.8 percent, soil pH distance 11.7 percent, and bacterial β-diversity 10.2 percent. The pattern is striking: two of the three dominant forces are not the diversity of any single group of organisms but the turnover of the cross-kingdom network linking plants to their soil partners.

Constructing that network itself required a careful statistical approach. The team calculated Spearman correlations between every plant species and every soil microbial functional group across the 24 plots, retaining only associations with a correlation strength of at least 0.4 and a false-discovery-rate-adjusted significance threshold, to build a regional meta-network. Each plot’s local network was then weighted by the Hellinger-transformed abundances of the connected plant and microbial nodes. When comparing plot pairs, total network dissimilarity was split into two mechanistically distinct parts: species turnover, reflecting the physical replacement of plants or microbes between sites, and connection reconstruction, reflecting changes in the strength or existence of interactions between species shared by both sites. This distinction matters because previous studies had only captured broad covariation between aboveground and belowground communities, without being able to tell whether function changes because species swap places or because the same species interact differently.

A distance-based structural equation model then mapped the causal pathways among all these factors. The researchers compared two competing architectures—one driven upward from plants, one driven downward from microbes—and although both explained the same 56.8 percent of the variation in β-EMF, the plant-driven model fit the data far better. Its verdict was unambiguous: plant β-diversity and plant–microbe network β-diversity are the only two variables with significant direct positive effects on multifunctionality turnover. Soil microbial β-diversity, despite its strong bidirectional feedback with plant diversity, has no direct effect; it acts only by reshaping the cross-boundary network. And soil pH distance, the one abiotic factor that mattered, works entirely through a cascade: it filters which plant functional strategies occur where, and that plant turnover in turn drives microbial replacement and network reorganization before any signal reaches multifunctionality.

The authors interpret this buffering of abiotic effects as a hallmark of complex forests. In simpler systems such as alpine meadows or arid early-successional sites, soil pH can directly control multifunctionality by governing nutrient solubility and mineral weathering. In these species-rich subtropical forests, by contrast, slight chemical differences between plots appear to be absorbed by the vegetation and its vast symbiotic microbial networks before they can act directly on soil geochemistry. The researchers also note that the failure of soil pH to directly predict microbial community turnover contradicts the classic view of pH as the primary controller of soil microbial structure, and they suggest that at their mesoscale of roughly 10 by 10 kilometers, the pH pattern is itself strongly regulated by canopy cover, litter chemistry, and root exudates. Phylogenetic β-diversity, meanwhile, failed to predict multifunctionality, likely because key functional traits show evolutionary instability or convergent evolution, making evolutionary distance a poor proxy for functional difference here.

For forest managers, the results carry concrete implications. Because functional trait turnover matters more than species richness alone, restoration projects should aim to combine species with contrasting strategies—pairing slow-growing, nutrient-conservative trees with fast-growing, nutrient-acquiring ones—rather than simply maximizing the number of species planted. Because soil pH operates through plant succession, maintaining healthy pH gradients, for example with biochar, lime, or adjusted litter composition, may prevent the functional homogenization that nitrogen-driven acidification could otherwise cause. And because the plant–microbe network acts as a flexible buffering mechanism, protecting corridors and understory microhabitats that sustain belowground connectivity may be as important as protecting the trees themselves.

The team is candid about the limits of the work. With only 24 plots, even the dimensionality-reduction and 999-permutation procedures used to stabilize the models cannot fully remove the risk of overfitting, so the directional pathways should be generalized to other regions with caution, and the study lacks the long-term monitoring needed to capture responses to climate extremes across generations. Still, by extending the biodiversity–ecosystem function relationship from the familiar local scale to the landscape scale, and by showing that the rewiring of plant–microbe interactions can outweigh the diversity of either trophic level alone, the study offers a mechanistic blueprint for designing multifunctional forest landscapes in an era of accelerating environmental change.

Subject of Research: Spatial turnover of ecosystem multifunctionality and its biotic and abiotic drivers in subtropical forests

Article Title: Plant functional trait and plant–microbe network β-diversity dominate spatial turnover of ecosystem multifunctionality in subtropical forests

Article References: Wang, M.-H., Su, J.-R., Liu, W.-D., Li, S.-F., Huang, X.-B., & Shang, R.-G. (2026). Plant functional trait and plant–microbe network β-diversity dominate spatial turnover of ecosystem multifunctionality in subtropical forests. Ecological Processes, 15(1), Article 93. https://doi.org/10.1186/s13717-026-00747-9

Image Credits: AI Generated

DOI: 10.1186/s13717-026-00747-9

Keywords: ecosystem multifunctionality, beta-diversity, plant functional traits, plant–microbe networks, subtropical forests, soil microbes, soil pH, biodiversity–ecosystem function, Hill-Chao numbers, structural equation modeling, forest management, Yunnan China