Rainforest-to-pasture conversion stimulates soil methanogenesis across the Brazilian Amazon

The study addresses how conversion of Amazonian primary rainforest to cattle pasture affects the soil methane (CH4) cycle, which is governed by methanogenic archaea (CH4 producers) and methanotrophic bacteria (CH4 consumers). Tropical deforestation contributes substantially to greenhouse gas emissions (e.g., ~1 Pg C yr−1 globally, with 78% of Brazil’s total greenhouse gas emissions attributed to land-use change). Methane has 86× the 20-year global warming potential of CO2, so shifts in soil microbial processes that control net CH4 flux are climatically important. Prior in situ flux studies generally show mature rainforest soils consume CH4 while pastures emit CH4, but mechanisms at the level of active microbial communities remain unresolved. Previous genomic surveys in Brazilian Amazon soils reported mixed responses of methanotroph and methanogen functional markers to land-use change. The authors hypothesized that increased soil methane production observed in cattle pastures is caused by a decrease in active methanotrophy. They aimed to determine how the composition and functions of the active CH4-cycling community and associated metabolic pathways vary across land-use types (primary forest, pasture, secondary forest) and across two distant regions of the Brazilian Amazon (Pará and Rondônia).

The introduction synthesizes several strands of prior research. Long-term measurements in Amazonia show primary rainforest soils typically act as CH4 sinks, whereas pastures tend to be CH4 sources. Molecular studies using functional gene markers found variable responses to land-use change: some reported decreases in methanotrophy markers (pmoA, mmoX) in pastures without changes in methanogenesis (mcrA), while others found decreased Type II methanotroph pmoA and increased mcrA in pastures. Broader microbial ecology indicates Type II methanotrophs dominate under high CH4, low O2, and N/Cu limitations, though they can also persist at low CH4 due to multiple pMMO isoenzymes. Anaerobic methane oxidation is established in wetlands but is only potentially present in upland soils. The literature also notes challenges with metatranscriptomics and metaproteomics in soils due to low target mRNA counts, extraction issues, and database limitations, motivating the use of stable-isotope probing to target actively growing CH4-cycling taxa.

Study sites and sampling: Intact soil cores (5 cm diameter × 10 cm depth) were collected from Tapajós National Forest and adjacent areas (Pará; June 2016) and from Fazenda Nova Vida and adjacent areas (Rondônia; April 2017). For each location, soils were sampled from two primary rainforest sites, one cattle pasture, and one secondary rainforest, along 100–200 m transects with five equidistant points. Eighteen cores were collected per land-use type. Stable-isotope probing (SIP): Intact cores (~200 g) were incubated in gas-tight glass jars in the dark at 25 °C for ~7 months, reflecting low gas exchange in intact cores relative to homogenized soil. Substrates and dosing: every 2 weeks, either 25 mL 13C-CO2 (3% headspace), 25 mL 13C-CH4 (3% headspace), or 1 mL 13C-sodium acetate (NaAOC; 1 mM final, applied to the top) was added. To maintain an oxic headspace in CH4 incubations, 1 mL sterile water was added weekly. Jar pressure was periodically released prior to substrate injection. Methane production/consumption was monitored by gas chromatography. Incubation duration targeted incorporation of ~100 µM 13C g−1 soil (within recommended 5–500 µM 13C g−1). Post-incubation, cores were sectioned into five 2 cm segments and frozen (−20 °C). Experimental design: For each location and set of four sites (two PF, one pasture, one SF), five cores per site were incubated with a 13C substrate and one core with a 12C control. This yielded six cores per site per substrate; across three substrates, 72 cores per location were processed. DNA extraction and fractionation: DNA was first extracted from 0.25 g soil from all segments of two 13C cores to identify the segment with highest methanogen or methanotroph abundance (qPCR of mcrA and pmoA). From the identified segment, DNA was extracted from 4 g soil of three 13C cores and the 12C control (DNeasy PowerMax). Five micrograms of DNA were separated by CsCl density gradient ultracentrifugation, fractionated into 12 equal volumes, precipitated (with linear acrylamide carrier), and quantified. Fractions containing 13C-labeled DNA were identified by qPCR of marker genes compared to 12C controls. To control for GC content effects on density, both light and heavy fractions from 12C controls were sequenced. Pooling: light fractions (~1–5) and heavy fractions (~6–12) were pooled prior to sequencing. qPCR assays: pmoA (primers A189f/mb661r) and mcrA (mlas/mcra-rev). Sequencing: Amplicon sequencing targeted 16S rRNA, pmoA, and mcrA with custom dual-indexed primers; pooled heavy fractions for three 13C samples per site/substrate and both heavy/light fractions for all 12C controls were sequenced on Illumina MiSeq (paired-end 300 bp). Additional 16S rRNA sequences from fresh field soils were obtained to assess incubation effects. Metagenomes from heavy fractions of two 13C samples per site/substrate and all 12C controls were sequenced on an Illumina HiSeq4000 (two flow lanes per location). Soil physicochemical analyses followed established protocols. Bioinformatics and statistics: Amplicons were processed with DADA2 in QIIME2. Metagenomes were annotated in MG-RAST using GenBank (taxonomic) and SEED Subsystem (functional) databases; the SEED category “Methanogenesis strays” was included. Community composition differences were tested using Bray–Curtis dissimilarities with Adonis (PERMANOVA) and Permdisp (dispersion) in vegan (R). For incubation impact assessment, 16S reads from 12C control light+heavy fractions were combined to represent the total community, rarefied to 20,000 annotations, and compared to fresh soil communities. Active fraction definition and normalization: Metagenomic annotations were rarefied and normalized to corresponding 12C controls; features lower in 13C than 12C were set to zero. Annotations significantly higher in 13C versus 12C (p<0.05, Fisher’s exact test in STAMP) were considered active (anabolic 13C incorporation). Between land-use comparisons within substrates used ANOVA with Tukey–Kramer post hoc in STAMP, with verification that features were active. Soil chemistry used ANOVA with Tukey–Kramer; correlations used Pearson tests. Data availability: MG-RAST projects mgp88468 and mgp86794; SIP amplicon data on figshare (DOIs provided).

• Across both regions (Pará and Rondônia), isotope-labeled methane-cycling communities shifted significantly with land use, with pastures showing increased abundance and diversity of active methanogens relative to primary and secondary forests.
• Total 16S rRNA-based community composition differed significantly by location (Rondônia vs Pará; r2≈0.118, p≤0.03), land use (r2≈0.08, p≤0.03), substrate (CH4, CO2, NaAOC; r2≈0.08, p=0.001), and their interactions.
• Within the specifically active fraction, significant location differences were detected primarily in CO2-incubated samples (Pará CO2 p≤3e-03; Rondônia CO2 p≤1.8e-02).
• Richness of active methanogens was highest in pasture soils in both regions and substrates, with a significant increase in Rondônia NaAOC incubations (Pasture vs PF p=9.6e-03; Pasture vs SF p=7.9e-03).
• In Rondônia, total active methanogens were significantly more abundant in pasture than primary forest and secondary forest in 13C-NaAOC incubations (p=1e-03 and p=3.8e-02, respectively), and higher than secondary forest in 13C-CO2 incubations (p=9e-03).
• Methanosarcina spp. dominated active methanogens across locations and substrates, consistent with their metabolic versatility (capable of both CO2-reducing and acetoclastic pathways). Independent non-SIP data from the same sites also implicated Methanosarcina spp. in methane fluxes.
• Secondary forests exhibited reduced methanogenic activity closer to primary forests, indicating potential recovery of methane sink function after forest regeneration.
• Overall, results indicate that increased methanogenesis (not decreased methanotrophy) is the main driver of enhanced net methane emissions from pasture soils following rainforest conversion.

The study set out with the hypothesis that increased methane emissions from pastures result from decreased active methanotrophy. Instead, stable-isotope probing coupled with metagenomics showed that conversion to pasture stimulates methanogenesis: active methanogens were more diverse and abundant in pastures, especially evident in Rondônia and with acetate labeling. Dominance of Methanosarcina spp., which can use multiple methanogenic pathways, suggests that substrate availability (e.g., acetate) and anoxic microsites in pasture soils may favor methanogen growth. While total community composition varies by location and substrate, the active methane-cycling fraction consistently shows pasture-associated increases in methanogens, indicating a mechanism for the observed switch from CH4 sink (primary forest) to source (pasture) in field flux studies. Secondary forests show decreased methanogenic activity similar to primary forests, suggesting that restoration can help recover methane sink function. Together, these findings refine the mechanistic understanding of how land-use change alters the soil CH4 cycle in tropical forests and underscore the importance of conserving and restoring forests to mitigate greenhouse gas emissions.

This work directly identifies active methane-cycling microorganisms across land-use types in the Brazilian Amazon using DNA-SIP and metagenomics. Rainforest-to-pasture conversion increases the abundance and richness of active methanogens, particularly Methanosarcina spp., providing a mechanistic explanation for pasture soils’ net positive CH4 flux. Secondary forests display methanogenic activity more similar to primary forests, indicating potential recovery of methane sink function with restoration. These insights inform land management and conservation strategies aimed at mitigating greenhouse gas emissions in the tropics. Future research should (1) resolve abiotic and biotic drivers (e.g., soil structure, redox dynamics, substrate availability, nutrient status) that favor methanogenesis in pastures, (2) incorporate seasonal and depth-resolved sampling to capture temporal dynamics, (3) integrate SIP-metatranscriptomics/proteomics as databases and methods improve to link activity with expression, and (4) expand the geographic and land-use gradients to test generality across the Amazon and other tropical regions.

• Long incubation duration (~7 months) of intact cores and repeated substrate additions may alter community composition relative to in situ conditions or favor taxa capable of growth under laboratory conditions.
• SIP targets actively growing organisms incorporating 13C; metabolically active but non-growing taxa may be underrepresented despite efforts to normalize to 12C controls.
• Primer-based amplicon approaches (16S rRNA, pmoA, mcrA) can introduce primer bias and have limited phylogenetic resolution.
• GC-content confounds buoyant density; although 12C light/heavy controls were sequenced to account for this, residual biases may remain.
• Only two regions were sampled, and location effects (including seasonal and soil physicochemical differences) were significant; generalizability across the entire Amazon may require broader sampling.
• The study emphasizes methanogens; constraints on methanotroph activity under different oxygen and nutrient regimes were not resolved mechanistically.