The large role of declining atmospheric sulfate deposition and rising CO2 concentrations in stimulating future wetland CH₄ emissions

Atmospheric methane (CH₄) is a potent greenhouse gas responsible for substantial warming since preindustrial times. Natural wetlands are the largest individual natural source, contributing an estimated 120 to 210 Tg CH₄ annually (2000–2019), representing 20 to 40% of the global methane budget. Isotopic evidence (negative δ¹³CCH₄ shifts) and recent exceptional atmospheric CH₄ growth (notably 2020–2021) indicate increased microbial emissions, especially wetland feedbacks, as a major driver. Field and observational studies highlight the sensitivity of wetland CH₄ emissions to meteorological changes (temperature and precipitation) and to biogeochemical factors such as atmospheric sulfate deposition (suppressive) and CO₂ enrichment (potentially stimulative). However, most projections have focused on meteorology alone or have only partially explored CO₂ fertilization, and Integrated Assessment Models do not directly account for future wetland CH₄ changes. The research aims to quantify future wetland methane emissions under multiple climate and socioeconomic scenarios by jointly assessing meteorological feedbacks and biogeochemical feedbacks from changing sulfate deposition and rising CO₂, thereby addressing a critical gap in projections and mitigation planning.

Prior modeling studies consistently project that warming and hydrological changes increase wetland CH₄ emissions, but with large spread due to uncertainties in parameterizing methane processes and wetland dynamics. End-of-century increases range widely: 40–270 Tg a⁻¹ in high-forcing scenarios (e.g., RCP8.5) and 14–50 Tg a⁻¹ in low-CO₂ scenarios (e.g., RCP2.6). Biogeochemical perturbations have been less comprehensively included. Field experiments show sulfate deposition can suppress wetland CH₄ emissions by up to ~40% via stimulation of sulfate-reducing bacteria that outcompete methanogens. Earlier projections assumed increasing sulfur emissions would strengthen suppression, potentially offsetting warming-induced increases, but anticipated 21st-century sulfur emission declines necessitate re-evaluation. Nitrogen deposition may stimulate methane by enhancing carbon substrates and methanogenic activity, though results are mixed and confounded by fertilizer runoff. CO₂ effects are ambiguous in process models: elevated CO₂ can increase substrate supply but may also enhance CH₄ oxidation via plant-mediated oxygen transport. Paleoclimate correlations between δ¹³CCH₄ and CO₂ suggest an important CO₂ fertilization role, especially for C₃ plants, yet most models use simplified mechanisms. This study synthesizes a broader set of field experiments to better constrain sulfate and CO₂ effects globally and integrates them with data-driven meteorological sensitivities.

The study develops a data-driven framework to project global wetland CH₄ emissions (1900–2100) by combining: (1) empirical meteorological sensitivities derived from bottom-up and top-down inventories; (2) biogeochemical response functions from meta-analyses of field experiments for sulfate suppression and CO₂ fertilization; and (3) external forcings from CMIP6 and AerChemMIP ensembles. - Baseline emissions: Compiled 135 estimates (2000–2019) from 100 publications, including 31 bottom-up inventories (e.g., WetCHARTs v1.3.1 and 13 GCP models) and 8 top-down inversions, yielding 174 ± 45 Tg CH₄ a⁻¹ (90% CI) with similar magnitudes from bottom-up (173 ± 55 Tg a⁻¹) and top-down (176 ± 29 Tg a⁻¹) approaches. - Meteorological sensitivity model: Constructed grid-scale empirical regressions linking monthly wetland CH₄ emissions to local temperature (T) and precipitation (P) and to regional climate patterns (SVD-based modes within ~1000 km). Two forms were fitted per grid cell: E = a₁T + a₂P + β₁S₁ + β₂S₂ and log(E) = a₁T + a₂log(P) + β₁S₁ + β₂S₂, selecting the higher R². Including regional modes improved explained variance to 70–90% (20–30% higher than local-only). The model was trained on 31 bottom-up inventories and 8 inversions and validated against FLUXNET-CH₄ site data and the UpCH₄ upscaling product (good agreement in non-tropical regions in totals, seasonality, and Q10 behavior). - Biogeochemical response functions: • Sulfate suppression: Synthesized 30 field and controlled experiments; fitted a Michaelis–Menten-type relationship g(x) = a x/(b + x) with a = −42.8% maximum suppression and b = 13.0 kg S ha⁻¹ a⁻¹ (R² = 0.77). Suppression increases with deposition up to ~40 kg S ha⁻¹ a⁻¹ then saturates (~40%). • CO₂ fertilization: Compiled 118 experiments from 34–43 sites across wetland types (mostly mid- to high-latitudes) elevating CO₂ from ~300–400 to ~600–800 ppm. Derived global mean sensitivity of CH₄ emissions of 0.038% ppm⁻¹ (90% CI range ~0.01–0.067% ppm⁻¹) using ln response ratio (lnRR) per 100 ppm. No strong dependence on wetland type or water table was found, though nutrient status may modulate responses in controlled studies. Sensitivity tests excluded C₄-dominated wetlands in the CO₂ calculation and applied latitude-dependent attenuation north of 50°N. - Forcings and scenarios: Used CMIP6 meteorology (39 models, 423 realizations) under seven SSP/RCP combinations (ssp119, ssp126, ssp245, ssp370, ssp434, ssp460, ssp585). Applied quantile delta mapping bias correction to T and P. Sulfate deposition (1900–2100) from AerChemMIP multi-model ensemble. Gridded atmospheric CO₂ (1850–2150) from Cheng et al. (2022). Wetland area dynamics are implicitly represented via monthly meteorological drivers within the empirical model; explicit CMIP6 wetland area projections were not used to rescale. - Synthesis equation: Monthly grid-cell emissions computed as E(T,P) × g(S deposition) × h(CO₂), where g and h are relative modifiers referenced to 2000–2019 conditions. CO₂ effect applied via response ratio scaling. Projections focus on low- to mid-CO₂ scenarios in the main text; high-CO₂ scenarios (ssp370, ssp585) are shown with caveats. - Uncertainty: Monte Carlo propagation of uncertainties in biogeochemical sensitivities (bootstrap on experiments), choice of wetland inventories in training, and spread across CMIP6 climate models. Reported as 90% confidence intervals.

- Baseline emissions: Global wetland CH₄ emissions for 2000–2019 are 174 ± 45 Tg a⁻¹ (90% CI), with bottom-up 173 ± 55 Tg a⁻¹ and top-down 176 ± 29 Tg a⁻¹. - Sulfate suppression: Empirical suppression increases with deposition up to ~40 kg S ha⁻¹ a⁻¹ and saturates near ~40% inhibition; global sulfate deposition peaked in the 1980s and declines in clean-air scenarios (ssp119), returning to ~1900 levels after ~2050, implying reduced suppression and higher CH₄ emissions. In polluted pathways (e.g., ssp370), deposition remains ~70% of 1980s levels by 2100. - CO₂ fertilization: Meta-analysis yields a mean CH₄ emission sensitivity of 0.038% ppm⁻¹ (range ~0.01–0.067% ppm⁻¹). Responses are broadly positive across wetland types; C₄-dominated systems likely less responsive. CO₂ concentrations rise from ~370 ppm (2000) to ~390 ppm (2100) in ssp119 (after an overshoot to ~440 ppm mid-century) and to ~1150 ppm in ssp585. - End-of-century increases relative to 2000 (171 Tg a⁻¹): • Low-CO₂ scenarios: ssp119 +20 Tg a⁻¹; ssp126 +34 Tg a⁻¹ by 2100. Biogeochemical feedbacks explain 43% (ssp119) and 38% (ssp126) of increases—CO₂ enrichment contributing 9% and 17%, and declining sulfate suppression 35% and 21%, respectively. Sulfate-driven increases are 6.9 ± 1.8 Tg a⁻¹ (ssp119) and 7.0 ± 1.8 Tg a⁻¹ (ssp126). Meteorology-only increase in ssp126 is 21 ± 4 Tg a⁻¹. • Mid-CO₂ scenarios: ssp434 +44 Tg a⁻¹; ssp245 +62 Tg a⁻¹; ssp460 +85 Tg a⁻¹ by 2100. Meteorological feedback contributes ~31–56 Tg a⁻¹; CO₂ fertilization adds ~8–27 Tg a⁻¹; sulfate effects are modest (~1–5 Tg a⁻¹). Excluding C₄ wetlands lowers CO₂ contribution by ~3.1 Tg a⁻¹ (ssp245) and ~4.4 Tg a⁻¹ (ssp460) (~16%). Biogeochemical feedbacks contribute ~30–40% of total increases in these mid-CO₂ scenarios. • High-CO₂ scenarios (ssp370, ssp585): Increases of ~60–110% by 2100 with very high uncertainty due to extrapolated CO₂ effects, unmodeled permafrost CH₄, and vegetation adaptation; biogeochemical share estimated at ~40–50%. - Policy-relevant impacts: Along 1.5° and 2°C pathways (ssp119 and ssp126), wetland CH₄ increases of ~20–34 Tg a⁻¹ by 2100 reduce the allowable anthropogenic methane emissions by ~8–15%. From 2020–2050, wetland emissions increase by ~14 Tg a⁻¹ (1.5°) and ~19 Tg a⁻¹ (2°C), equivalent to 8–15% of the required anthropogenic CH₄ mitigation (180 and 130 Tg a⁻¹, respectively). For the Global Methane Pledge (2020–2030 reduction target ~100 Tg a⁻¹), wetlands add ~10 Tg a⁻¹ in ssp119, implying an extra ~11 ± 4% mitigation effort. - Comparative magnitude: Projected wetland increases in low-CO₂ scenarios are 2.2–3.8× global oceanic CH₄ (~9 Tg a⁻¹), 1.6–2.7× U.S. oil/gas (~12.5 Tg a⁻¹), and comparable to China’s coal (~21 Tg a⁻¹) or global coal (~33 Tg a⁻¹) emissions.

The study addresses the gap in projections by jointly quantifying meteorological and biogeochemical feedbacks on wetland CH₄ emissions. Findings show that, beyond warming and hydrological changes, declining sulfate deposition (due to clean air policies) and rising CO₂ can substantially and systematically elevate wetland emissions. In low-CO₂ pathways, the diminishing sulfate suppression contributes a large fraction of the increase, while in mid-CO₂ pathways, CO₂ fertilization is a dominant biogeochemical amplifier. These contributions are large enough to erode part of the allowable anthropogenic methane emissions, implying that climate mitigation strategies and Integrated Assessment Models should explicitly include natural wetland feedbacks. The emergent sulfate dose–response relationship and the empirically grounded CO₂ sensitivity provide mechanistic consistency with microbial competition and substrate availability and align with paleoclimate hints of CO₂–CH₄ coupling. Despite uncertainties, the results underscore that meeting 1.5°–2°C targets requires recognizing and compensating for natural wetland emission increases in policy planning.

By integrating data-driven meteorological sensitivities with empirically constrained biogeochemical feedbacks, the study demonstrates that declining atmospheric sulfate deposition and rising CO₂ will substantially stimulate future wetland CH₄ emissions. Across low- to mid-CO₂ scenarios, biogeochemical feedbacks account for ~30–45% of emission increases, with sulfate effects prominent in strong mitigation pathways and CO₂ fertilization dominating in mid-CO₂ pathways. Under 1.5°–2°C trajectories, wetland increases of ~20–34 Tg a⁻¹ by 2100 reduce the allowable space for anthropogenic methane by ~8–15%, necessitating more ambitious mitigation than currently assessed. Future work should expand field manipulation experiments across latitudes (especially tropics), better constrain CO₂ fertilization and nutrient interactions, incorporate evolving vegetation (C₃/C₄) distributions, improve representation of methanotrophy, and account for permafrost thaw and lateral nutrient/sulfate inputs, thereby reducing uncertainties, particularly in high-CO₂ scenarios.

- Experimental constraints: CO₂ fertilization estimates rely on limited field experiments (mainly mid- to high-latitudes) with CO₂ elevated to 600–800 ppm, making extrapolation to >900–1150 ppm (high-CO₂ scenarios) uncertain. The relative SD of CO₂ sensitivity (~30%) propagates large uncertainty (differences of ~30–40 Tg a⁻¹ by 2100 in some scenarios). - Sulfate dynamics: Assumes instantaneous annual response and no multi-year carryover; while supported for low deposition rates, broader persistence effects remain uncertain. Sparse direct observational detection of trends due to small interannual sulfate signals versus larger meteorological variability. - Missing processes: Limited representation of permafrost thaw emissions, ecological adaptation to warming (e.g., shifts in C₃/C₄ distribution), community changes in methanogens/methanotrophs, and incomplete understanding/parameterization of methanotrophy and wetland extent sensitivity. Nitrogen deposition and runoff impacts remain controversial and not fully integrated. - Model structural uncertainty: Empirical approach captures apparent sensitivities and may not fully disentangle confounding or future-changing interactions among drivers. CMIP6 meteorology biases and sparse tropical observations introduce regional uncertainties. - High-CO₂ scenarios: Projections in ssp370 and ssp585 entail extreme extrapolation and are subject to very high uncertainty.