Hidden Soil Worlds: DNA Reveals How Forest and Grassland Microbes Differ Side by Side

In a well-preserved corner of the Brazilian Atlantic Forest, two ecosystems sit almost within arm’s reach of each other: a dense Native Forest crowned by the iconic Araucaria angustifolia conifer, and an open Highland Grassland shaped by cold nights, strong solar radiation and a vast underground root network. A new study published in Metabarcoding and Metagenomics has now taken a molecular census of the topsoil in both habitats, using environmental DNA metabarcoding to profile bacteria, archaea, unicellular eukaryotes and soil animals. The results show that even neighbouring, equally protected habitats can host strikingly different soil communities, and that the forest consistently supports higher microbial richness and diversity than the grassland beside it.

The research was carried out in the Private Natural Heritage Reserve RPPN PROMATA, in São Francisco de Paula, in Rio Grande do Sul, Brazil’s southernmost state, at roughly 900 metres above sea level. The team sampled five randomly chosen points in each habitat, all within 150 metres of one another and no more than four kilometres apart at the furthest distance, on a single day in November 2022. After carefully removing leaf litter, they collected two grams of topsoil from the upper five centimetres at each point, kept the samples on ice, and transported them to the laboratory within six hours to minimise shifts in community composition before DNA extraction.

The molecular work relied on high-throughput sequencing on an Illumina MiSeq platform, with separate primer sets targeting the 16S rRNA gene for bacteria and archaea and the 18S rRNA gene for eukaryotes. Raw sequences were quality filtered and denoised into Amplicon Sequence Variants, or ASVs, using the DADA2 pipeline, and taxonomy was assigned against the SILVA and PR2 reference databases with a confidence threshold of at least 80 percent at the family level. In total, the study identified 247 distinct eukaryotic ASVs, 4,201 bacterial ASVs spanning 30 phyla, and 130 archaeal ASVs, providing one of the most detailed multi-domain snapshots of soil life in this highly endemic region.

The taxonomic portraits of the two habitats diverged most clearly among the eukaryotes. In the forest soil, the phylum Cercozoa dominated the unicellular eukaryotic community, with the group Filosa-Sarcomonadea reaching 55.1 percent relative abundance, whereas in the grassland the parasitic phylum Apicomplexa took over, with the class Gregarinomorphea accounting for 83.9 percent of sequences. Among soil animals, annelid worms dominated most forest sites at 57.1 percent mean relative abundance, while nematodes were far more prevalent in grassland, representing 46.7 percent of identified sequences there compared with just 7.55 percent in the forest. Tardigrades appeared exclusively at one grassland site, and ciliates were entirely absent from grassland samples.

In contrast, the prokaryotic communities looked remarkably similar across the habitat boundary. Bacterial communities in both ecosystems were dominated by the phyla Pseudomonadota, Acidobacteriota and Actinomycetota, which together made up 79.08 percent of relative abundance, with the class Alphaproteobacteria reaching 29.7 percent in forest and 37.5 percent in grassland. Archaeal communities were overwhelmingly dominated by the phylum Thermoproteota, whose class Nitrososphaeria reached 85.1 percent in forest and 87.3 percent in grassland. Notably, the ratio between Alphaproteobacteria and Acidobacteriae, often used as an indicator of soil health, was about 1.40 in forest and 1.93 in grassland, values above one that point to healthy soils in both habitats, likely a benefit of long-term protection from disturbance.

Diversity statistics reinforced the forest’s edge. Using incidence-based rarefaction and Hill numbers, the researchers found that richness and both the Shannon and Simpson diversity indices were significantly higher in the Native Forest across all four organism groups, with non-overlapping 95 percent confidence intervals. The gap was widest for bacteria, whose estimated richness in forest soil was almost three times that of the grassland, the greatest disparity observed among the studied groups. Non-metric multidimensional scaling based on Jaccard distances revealed clear spatial separation between the habitats for unicellular eukaryotes, bacteria and archaea, and a significant, though graphically less tractable, separation for metazoans, confirming high beta diversity between the two systems.

One of the study’s most technically interesting contributions is its use of zeta diversity, a multi-site framework that tracks how many species are shared across increasing numbers of sampling sites, avoiding the bias toward rare species that can distort single beta diversity indices. Unicellular eukaryotes started with roughly 35 shared species and then declined steeply, indicating communities dominated by rare, site-specific taxa. Metazoans began with about 15 shared species and followed a similar abrupt decline, with an exponential regression providing the best fit and suggesting a core of specialised species. Bacteria, by contrast, started with around 650 shared species and declined only moderately in a long-tailed pattern best described by a power law, consistent with their habitat-generalist nature. Archaea began with about 17 shared species and showed steep early declines, hinting at communities of rare or highly specialised, possibly endemic organisms.

Co-occurrence network analysis among eukaryotes revealed two well-structured clusters: one linking rotifers, flatworms and Sagenista, and another connecting nematodes, the class Elardia, Coccidiomorphea and Cercozoa. The significant association between Coccidiomorphea, a group within Apicomplexa, and Nematoda is particularly intriguing, since both may infect overlapping categories of host organisms, although the authors caution that such patterns should be interpreted as exploratory and may reflect shared responses to local micro-environmental conditions rather than direct ecological interactions.

The findings carry broader implications for conservation. Primary forests are increasingly recognised as biodiversity reservoirs whose long-term genetic and functional value cannot be quickly replicated by recovering or secondary forests, and this study adds soil microbes and soil animals to that argument. At the same time, the grassland, often described as an underground or dwarf rainforest because of its massive belowground biomass and evolutionary antiquity, made a substantial contribution to regional diversity despite its comparatively lower within-habitat heterogeneity. The clear separation of communities between the two habitats points to environmental filtering driven by vegetation structure, soil physicochemical properties and microclimatic conditions as the prevailing mechanism shaping these assemblages.

The authors are careful to note the limits of their design. With five replicates per habitat and a single sampling date, the observed patterns should be read as indicative rather than exhaustive, and additional sampling could uncover further rare taxa and refine diversity estimates in such heterogeneous soils. Still, the work delivers a foundational characterisation of two adjacent, well-preserved habitats and demonstrates the value of pairing multi-domain metabarcoding with zeta diversity analysis to capture fine-scale community differentiation. Future studies with larger sample sizes, broader sequencing coverage and longitudinal data will be needed to determine whether these community shifts are driven by localised environmental filters, stochastic drift, or a combination of both, but the message is already clear: protecting a landscape means protecting not just its trees and grasses, but the intricate, largely unseen soil worlds beneath them.

Subject of Research: Soil microbial and metazoan community composition in adjacent preserved Atlantic Forest and Highland Grassland habitats assessed by eDNA metabarcoding

Article Title: Soil microbial community composition in two distinct preserved habitats

Article References: Huber, C. S. R., Huber, L. K., & Utz, L. R. P. (2026). Soil microbial community composition in two distinct preserved habitats. Metabarcoding and Metagenomics, 10, Article e183786. https://doi.org/10.3897/mbmg.10.183786

Image Credits: AI Generated

DOI: 10.3897/mbmg.10.183786

Keywords: soil biodiversity, eDNA metabarcoding, Atlantic Forest, Highland Grassland, microbiome, zeta diversity, bacteria, archaea, Cercozoa, Apicomplexa, Nematoda, conservation