The research team focused on the La Esperanza Forest, a private reserve in El Carmen canton in Manabí province, within the Tumbes-Chocó-Magdalena biodiversity hotspot. This bioregion stretches from southern Panama along the Pacific coasts of Colombia and Ecuador into northwestern Peru, and it ranks as the ninth most biodiverse hotspot on the planet, holding roughly 11,000 plant species, about 3 percent of all known plants, on less than 0.2 percent of Earth’s land surface. Yet more than 85 percent of the Ecuadorian Chocó’s original habitat has been lost in recent decades, driven by livestock, agriculture, forest fires, and urban expansion. What remains is a fragmented archipelago of isolated rainforest patches, of which La Esperanza is a rare example that scientists can actually access and study.

The reserve itself is remarkably rich for its size. Spanning just 15.22 hectares, it shelters a canopy 20 to 30 meters tall with emergent giants such as Eschweilera awaensis and palms like Attalea colenda and Oenocarpus bataua. A preliminary botanical census documented around 200 plant species in a single hectare, many still unidentified, alongside 105 bird species, 13 amphibians, 10 reptiles, and 10 large mammal species. Of the reserve’s area, about 11.22 hectares are mature forest, while the remaining 4 hectares were once farmland and pasture. Since 2022, that disturbed section has been undergoing agrochemical-free restoration, with 118 native tree species planted between 2022 and 2024, livestock excluded, invasive species removed annually, and supplementary irrigation provided during dry months. No soil amendments or chemical inputs have been applied since restoration began.

To capture the microbial signature of each land-use condition, the researchers sampled three habitat types: the remnant reference forest, the early-stage restored parcel, and an adjacent degraded cattle pasture that had been fumigated before 2017 with atrazine and the systemic herbicide Tordon 101, then maintained under grazing and manual slashing ever since. Fieldwork took place over three days in late September 2024. Within each habitat, the team established two 10 by 10 meter plots, subdivided each into a grid of 25 two-meter subplots, and collected pooled soil aliquots from the corners of eight dispersed subplots per plot. After two forest samples failed to yield viable DNA, the final dataset comprised 46 soil samples from the top 15 centimeters of the soil profile, preserved on dry ice and stored at minus 20 degrees Celsius until analysis. Soil DNA sequencing was authorized by the Ecuadorian government under a genetic resource access contract, and the samples are now permanently archived in the National Bank of Genetic Resources at INABIO.

The molecular approach relied on 16S rRNA gene metabarcoding, a technique that reads a diagnostic stretch of bacterial DNA to identify which microbes are present and in what relative proportions. The team used Oxford Nanopore long-read sequencing on a GridION device, amplifying nearly the full-length 16S gene with primers 27F and 1492R, then filtering reads to keep only sequences between 800 and 2,000 base pairs. Taxonomic classification was performed with the Kraken2 k-mer algorithm against the NCBI 16S/18S reference database, followed by Bayesian abundance re-estimation with Bracken at the genus level. The result was a catalogue of 1,208 distinct bacterial genera across all sampled habitats, with rarefaction curves confirming that sequencing depth was sufficient to capture the taxonomic diversity present in each sample.

What emerged first was a picture of surprising continuity. A core set of dominant genera, including Bacillus, Haliangium, and Paenibacillus, persisted across all three habitats with stable relative abundances, suggesting that the fundamental biological scaffolding of these soils survives even under intense aboveground disturbance. But the researchers caution that the persistence of these generalist taxa should not be mistaken for evidence of ecological health. In fact, these dominant genera tended to be more abundant in the degraded and restored habitats, which may partly explain the descriptive reduction in community evenness observed there compared to the forest.

When the team analyzed the overall structure of the bacterial communities using compositional data analysis, a fundamentally different picture emerged. Because sequencing data are inherently compositional, meaning that the abundance of one taxon is only meaningful relative to all others, the researchers applied a centered log-ratio transformation through the ALDEx2 package, which uses Monte Carlo sampling from a Dirichlet distribution to estimate true relative abundances and technical variation. They then computed Aitchison distances, a mathematically robust metric for compositional data, and ran a permutational multivariate analysis of variance aggregated at the plot level to account for the nested spatial design. The omnibus test did not reach traditional statistical significance, a consequence of the highly conservative model with only six experimental units, but the effect size was striking: habitat condition explained over 66 percent of the total structural variance in the soil communities. Pairwise comparisons reinforced this pattern, with large effect sizes separating forest from restored soils and forest from degraded soils, while the restored and degraded plots clustered much closer together.

Principal component analysis made the separation visually unmistakable. The first principal component, accounting for 21.7 percent of the variation, completely partitioned the forest samples away from both the restored and degraded plots. The taxa driving this separation were revealing. Forest soils were strongly associated with genera such as Dinghuibacter, Ilumatobacter, Amaricoccus, Aminobacter, and Phyllobacterium, along with two uncultured lineages from the Saprospiraceae and Nitrosomonadaceae families. Many of these are Gram-negative bacteria previously isolated from forest soils or associated with tree roots and leaves, and some species of Phyllobacterium are symbiotic nitrogen-fixers used in microbial fertilizers. In contrast, the degraded and restored soils were characterized by genera including Sedimentibacter, Methylocystis, Chungangia, Fervidicella, and notably Rhodanobacter, a genus frequently linked to denitrification in contaminated and agricultural soils. The differential abundance analysis confirmed that Rhodanobacter was significantly enriched in the degraded environment, while the restored parcel showed significant enrichment of Fictibacillus and an uncultured Nitrosomonadaceae genus, and the forest maintained significantly higher levels of an uncultured Saprospiraceae genus compared to the restored soils.

Perhaps the most ecologically revealing finding came from partitioning beta diversity into its two components: species turnover, which reflects the replacement of one set of taxa by another, and nestedness, which reflects the sequential loss of taxa from a richer community. Using the Sørensen dissimilarity index, the researchers found that turnover overwhelmingly dominated the structural differences between habitats, with the turnover fraction ranging between roughly 0.15 and 0.30 while nestedness remained below 0.10 across all comparisons. This means that agricultural conversion and early restoration did not simply strip away pieces of the original forest microbiome; instead, they replaced it with an entirely different bacterial architecture. The turnover component alone explained over 64 percent of the variance in species replacement, with large effect sizes between forest and restored plots and between forest and degraded plots, but a much smaller effect size between restored and degraded plots. The nestedness component, by contrast, showed virtually no variance explained by habitat type, confirming that subset-loss dynamics played almost no role in shaping the observed differences.

The implications for restoration practice are sobering but not discouraging. The restored parcel, despite hosting 118 native tree species and two years of careful agrochemical-free management, clustered structurally closer to the degraded pasture than to the reference forest, lacking the high-loading structural drivers that define the forest soil microbiome. This decoupling of aboveground and belowground recovery is increasingly recognized across tropical systems, with recent studies showing that microbial community composition shifts distinctly during secondary succession while overall diversity does not necessarily parallel the recovery of plant diversity, and that the structural transition to a mature, forest-like metabolic network spans decades rather than years. The researchers emphasize that their findings represent a localized proof-of-concept within a single reserve rather than a universal model for the heterogeneous Chocó bioregion, and they acknowledge that the absence of soil physicochemical measurements such as pH, organic carbon, and nutrient levels limits mechanistic interpretation. Nevertheless, the study establishes a rigorous molecular baseline for long-term monitoring at La Esperanza and makes a compelling case that remnant forest patches are not just refuges for visible biodiversity but critical reservoirs of belowground microbial diversity that early-stage restoration cannot quickly replicate. The authors propose moving beyond static baselines to longitudinal plots tracked over decades, coupling DNA sequencing with biogeochemical measurements and transitioning to shotgun metagenomics to capture the full functional dimension of ecosystem recovery. For now, the message is clear: the invisible half of the forest is far harder to rebuild than the half we can see, and protecting what remains may be the most efficient restoration strategy of all.

Subject of Research: Soil bacterial community structure across remnant forest, restored, and degraded land uses in the Ecuadorian Chocó

Article Title: Soil metabarcoding reveals the distinct microbial signature of a remnant forest patch at a local restoration initiative in the Ecuadorian Chocó

Article References: Santillán-Sarmiento, A., Carrión-Olmedo, J. C., Salinas-Tapia, J., Lojan-Cueva, P., Flores, D., Reyes-Barriga, D., Jaramillo, L., Narváez, A. E., & Jarrín-V, P. (2026). Soil metabarcoding reveals the distinct microbial signature of a remnant forest patch at a local restoration initiative in the Ecuadorian Chocó. Frontiers in Forests and Global Change, 9, Article 1892271. https://doi.org/10.3389/ffgc.2026.1892271

Image Credits: AI Generated

DOI: 10.3389/ffgc.2026.1892271

Keywords: soil microbiome, 16S rRNA metabarcoding, Ecuadorian Chocó, tropical forest restoration, bacterial communities, beta diversity, compositional data analysis, biodiversity hotspot, La Esperanza Forest, Rhodanobacter, Phyllobacterium, soil health