The research, led by Yingyi Zhou, Dongliang Li, and colleagues at the Hainan Academy of Agricultural Sciences, took advantage of a naturally regenerated tea plantation inside Hainan Tropical Rainforest National Park in Wuzhishan City. The region receives between 1,300 and 1,800 millimeters of rain per year at a mean annual temperature of roughly 25 degrees Celsius, and its lateritic red soils are typically acidic under tropical conditions. Because the plantation had been unmanaged for over twenty years, the researchers could treat habitat variation as a natural experiment rather than a manipulated treatment. They identified three ecological categories within the same landscape: open sites with low canopy cover, slope-shaded sites with moderate vegetation, and rainforest-shaded sites beneath closed canopy, each represented by three replicate plots of 20 by 20 meters, separated by at least 50 meters to ensure spatial independence.

Sampling followed a rigorous protocol designed to capture the true rhizosphere, the narrow zone of soil where root exudates and microbial activity converge. Within each plot, seven healthy tea plants were selected along an S-shaped path, fine roots were excavated from the top 20 centimeters of soil, and the tightly adhering soil was collected by gentle brushing. Soils from the seven plants were pooled into a single composite sample per plot, yielding nine biological replicates across the study. Each sample was split, with one portion analyzed for a full suite of physicochemical properties including pH, organic matter by Walkley-Black oxidation, total and available nitrogen, phosphorus, and potassium, cation exchange capacity, and humus carbon, and the other flash-frozen for DNA extraction and shotgun metagenomic sequencing.

The sequencing effort was substantial. Using an Illumina NovaSeq 6000 platform, the team generated between 4.84 and 6.15 gigabases of clean sequence per sample, averaging 18.16 million read pairs each and totaling 49.03 gigabases across the study. Quality-filtered reads were assembled de novo with MEGAHIT, genes were predicted with Prodigal, and a non-redundant gene catalog was clustered at 95 percent identity. Taxonomic annotation used Kraken2 with the PlusPF database, while functional annotation relied on DIAMOND alignment against the KEGG database to assign KEGG Orthologs, the standard units for inferring potential microbial functions from metagenomes. Good’s coverage exceeded 96 percent in every sample, indicating that the sequencing depth was sufficient to capture the majority of the community.

The soil chemistry results revealed a striking edaphic gradient across the three habitats. All soils were acidic, but rainforest-shaded soils were significantly less acidic, with a higher pH than open and slope-shaded sites at a statistical threshold of P ≤ 0.001. The rainforest-shaded soils also contained significantly more organic matter than open sites, along with higher total and available potassium and the highest available phosphorus, while slope-shaded soils recorded the lowest available phosphorus. Intriguingly, nitrogen-related indices, including total nitrogen and alkali-hydrolyzable nitrogen, as well as cation exchange capacity and humus carbon, remained statistically stable across habitats. In other words, natural habitat heterogeneity shaped soil acidity, carbon accumulation, and phosphorus and potassium availability without materially altering the bulk nitrogen pools.

The microbiome data told an equally nuanced story. Alpha diversity, the within-sample richness and evenness measured by Shannon, Simpson, ACE, and related indices, did not differ significantly among habitats, with even Chao1 showing only a marginal trend at P = 0.051. Yet beta diversity, which measures how community composition differs between samples, revealed a significant overall habitat effect. PERMANOVA based on Bray-Curtis dissimilarities returned F = 7.206 with R² = 0.706 at the phylum level and F = 3.864 with R² = 0.563 at the genus level, both significant. The principal compositional gradient separated the rainforest-shaded communities from those of the open and slope-shaded sites, which were comparatively similar to each other. This dissociation between stable diversity and shifting composition is ecologically meaningful: the habitats appear to filter different taxa into place without changing the total number of species a given soil can support.

Taxonomically, the dominant phyla were shared across all habitats, but their relative abundances shifted. Pseudomonadota was the most abundant phylum everywhere yet declined in the rainforest-shaded soils, where Acidobacteriota rose. At the genus level, Bradyrhizobium dominated all three habitats but was less abundant under rainforest canopy, while the aggregated pool of less common genera contributed a larger share of the community there. The authors caution that the elevated Acidobacteriota should not be read as a simple acid-loving signal, since the phylum contains ecologically diverse subgroups that may respond differently to pH, carbon, and nutrients. They also stress that because the habitats differ not only in canopy cover but in vegetation structure and topographic setting, the patterns cannot be attributed to shading alone; litter inputs, soil moisture buffering, and slope-mediated redistribution of soil particles and labile nutrients are plausible but unmeasured contributors.

Redundancy analysis tied the microbial differentiation back to soil chemistry. At the phylum level, the first two canonical axes explained 95.73 percent of the constrained variation, and at the genus level 98.62 percent, with pH, available phosphorus, organic matter, and cation exchange capacity retained after multicollinearity screening. Rainforest-shaded samples aligned along the ordination axes corresponding to higher pH, phosphorus, and organic matter, while open and slope-shaded samples occupied the opposite side. Functional analysis of the 35 most abundant KEGG Orthologs added another layer: open-site communities showed higher relative representation of genes annotated to substrate transport and environmental sensing, slope-shaded communities of genes tied to redox balance and stress responses, and rainforest-shaded communities of genes involved in genetic information processing and cellular maintenance. The authors are appropriately careful here, noting that no individual KO remained significant after global false discovery rate correction, and that metagenomes capture encoded potential rather than realized gene expression or enzyme activity.

What makes this study resonate beyond tea is its demonstration of aboveground-belowground linkage under genuinely natural conditions. Most microbiome research in tea systems has focused on intensively managed plantations, where fertilization, mulching, and tillage confound the picture. Here, canopy structure, understory vegetation, and topography alone, operating over decades without human input, were sufficient to produce coordinated variation in soil fertility, microbial composition, and the inferred functional repertoire of the rhizosphere. The findings suggest that habitat heterogeneity acts as an environmental filter, favoring different microbial lineages and different encoded capabilities depending on local soil conditions, and that pH, organic matter, and phosphorus availability are the key gradients structuring these communities.

Important caveats remain, and the authors flag them clearly. Because the three habitats were naturally occurring field categories rather than controlled treatments, the results describe ecological associations, not causal single-factor effects. Vegetation composition, litter inputs, soil moisture, enzyme activities, and nitrogen transformation rates were not directly measured, and the single-time-point nitrogen measurements may miss dynamic cycling processes. Future work combining metatranscriptomics, enzyme assays, microbial biomass estimates, and microclimate monitoring would test whether the observed metagenome-inferred patterns translate into actual ecological function. Even so, the message is compelling: in one of the world’s least disturbed tea ecosystems, the forest canopy above and the microbes below are telling the same story, written in the chemistry of the soil that connects them. Data from the study are publicly available in the NCBI Sequence Read Archive under BioProject accession PRJNA1452514, offering a valuable resource for anyone seeking to unravel how natural habitats sculpt the invisible life beneath our feet.

Subject of Research: Habitat-associated differentiation of rhizosphere soil chemistry and microbial communities in wild tea plantations

Article Title: Habitat-associated rhizosphere microbiome differentiation in wild tea soils

Article References: Zhou, Y., Li, D., Zhang, L., & Zhang, D. (2026). Habitat-associated rhizosphere microbiome differentiation in wild tea soils. Spanish Journal of Soil Science, 16, Article 17010. https://doi.org/10.3389/sjss.2026.17010

Image Credits: AI Generated

DOI: 10.3389/sjss.2026.17010

Keywords: wild tea, rhizosphere, metagenomics, soil pH, habitat heterogeneity, microbiome, edaphic gradient, Hainan, Camellia sinensis, beta diversity, KEGG Orthologs, aboveground-belowground linkage