Meta-analysis shows the impacts of ecological restoration on greenhouse gas emissions

Rising atmospheric greenhouse gases, driven by land-use change and ecosystem degradation, are pushing global temperatures toward critical thresholds. Ecological restoration—assisting recovery of degraded forests, grasslands, and wetlands—has been proposed as a key climate mitigation strategy, and international initiatives (e.g., UN Decade on Ecosystem Restoration, Bonn Challenge) set ambitious restoration targets. However, the net effects of restoration on the three major greenhouse gases (CO2, CH4, N2O) remain uncertain across ecosystems, restoration types, and ages, and are insufficiently represented in IPCC guidance and national inventories. This study asks how restoration affects CH4 and N2O emissions and net ecosystem CO2 exchange globally across major ecosystems, how these responses evolve with restoration age, and which factors control GHG responses, thereby informing policy and improving inventory methodologies.

Prior studies indicate forests are major carbon sinks but disturbances and land-use changes impact GHG fluxes. Afforestation can increase soil CH4 uptake (via lowered bulk density) and reduce N2O (via reduced N substrate), though some conversions (e.g., grassland to forest) may increase N2O. Grasslands are typically CH4 and CO2 sinks but N2O sources; degradation reduces CH4 uptake and alters C balance, while restoration effects on GHGs have varied by type and measure. Wetlands store disproportionate soil carbon; drainage typically decreases CH4 but increases CO2 and N2O, while restoration effects on net GHG balance are mixed, influenced by restoration type, age, hydrology, climate, and soils. Existing IPCC guidelines lack comprehensive restoration-specific emission factors and global syntheses across forest, grassland, and wetland restorations.

The authors compiled a global dataset from 253 peer-reviewed articles (Dec 1999–Jun 2023), including paired restored–control comparisons (679 cases) and chronosequence observations (1289 data points with restoration age). Inclusion criteria were field studies with paired controls or chronosequences, ≥3 replicates per treatment, at least one full year or growing season of measurements, and reporting at least one GHG variable. Extracted variables included CH4, N2O, CO2 flux components (NEE, GPP, ER), site climate (MAT, MAP), restoration age, and soil properties (water table depth, soil temperature, WFPS, Eh, bulk density, pH, SOC, TN, NH4+, NO3−). Fluxes were typically measured by static chambers and eddy covariance; NEE = GPP − ER (positive indicates net CO2 source). Means, SD, and sample sizes were extracted; when SD was missing (mainly eddy fluxes), SD was estimated as one-tenth of the mean. Data from figures were digitized where needed. Meta-analysis used Hedges’ d (RRd) as the effect size (suitable for variables with positive or negative values), with categorical random-effects models (MetaWin 3) and 95% CIs via bootstrapping (9999 iterations). Significance was inferred when CIs did not overlap zero (α = 0.05). One-way ANOVA tested differences between restored and control groups. Mixed meta-regression assessed relationships between GHG fluxes and environmental variables. GWP (t CO2-eq ha−1 yr−1) integrated CO2, CH4, and N2O using conversion factors (16/12 for CH4–C to CH4, 44/12 for CO2–C to CO2, 44/28 for N2O–N to N2O) and 100-year GWP values of 27.2 for CH4 and 273 for N2O.

- CH4: Forest and grassland restoration significantly increased CH4 uptake (decreased CH4 emissions). Forest RRd = −2.3 (95% CI: −2.9 to −1.6), Grassland RRd = −1.6 (95% CI: −2.4 to −0.8). Average CH4 uptake rose from 1.0 to 1.9 kg C ha−1 yr−1 (+90.0%) in forests and from 2.6 to 3.4 kg C ha−1 yr−1 (+30.8%) in grasslands. Temperate steppe & meadow and desert steppe increased CH4 uptake by ~46–48%. - Wetlands increased CH4 emissions by 544.4% (RRd = 2.9; 95% CI: 2.4–3.4; P < 0.05), from 23.4 to 150.8 kg C ha−1 yr−1 on average; largest increase when converting grasslands to wetlands. No significant CH4 change for aquaculture→wetland and mangrove restoration categories. - N2O: Forest restoration overall had no significant effect (RRd = −0.4; 95% CI: −1.3 to 0.4). Grassland restoration reduced N2O emissions by 21.7% (RRd = −0.7; 95% CI: −1.4 to −0.1). Wetland restoration reduced N2O by 68.6% (RRd = −2.9; 95% CI: −3.9 to −1.9; P < 0.05). Specific conversions: cropland→forest 3.7 to 1.4 kg N ha−1 yr−1; grassland→wetland 5.2 to 2.6; cropland→wetland 17.0 to 2.3; peatland restoration 2.2 to 0.5; floodplain restoration not significant. - CO2 fluxes: Wetland restoration significantly reduced NEE by 138.8% (RRd = −3.2), turning systems into CO2 sinks; examples: grassland→wetland NEE from 231.9 to −219.5 g C m−2 yr−1; aquaculture→wetland 41.9 to −151.5; bogs 159.2 to −35.8. Grassland restoration reduced NEE by 146.9% (RRd = −4.7); cropland→grassland NEE from 10.3 to −75.8 g C m−2 yr−1. - Age effects: Forest NEE decreased (more negative) with restoration age; transition from net CO2 source to sink in ~3–5 years for afforestation/reforestation, and ~6 years after clear-cutting and ~13 years after fire. Wetland NEE became net CO2 sink in ~4 years. Forest CH4 uptake increased with afforestation age; wetland CH4 response to restoration age stabilized after ~10 years; wetland N2O decreased with age. - Drivers: In forests and grasslands, lower soil bulk density and lower WFPS favored CH4 uptake; N2O was positively related to NH4+ and NO3− and negatively to pH. In wetlands, elevated water table depth increased CH4 emissions and decreased NEE and N2O. GPP and ER increased with temperature and precipitation. - Carbon budget and GWP: Mean C budgets (CO2 + CH4) under restoration were −295.5 (forest), −506.5 (grassland), and −53.4 (wetland) g C m−2 yr−1, indicating enhanced C sinks. GWP decreased by 327.7% (forest), 157.7% (grassland), and 62.0% (wetland) compared to paired controls.

The meta-analysis shows consistent mitigation benefits of ecological restoration across major biomes but with ecosystem-specific trade-offs and controls. Forest and grassland restoration enhance CH4 sinks primarily via soil structural and hydrological changes (reduced bulk density and WFPS, increased SOC), and N2O responses depend on prior land use and nutrient availability. Wetland restoration elevates water tables, creating anaerobic conditions that boost CH4 production yet simultaneously lower N2O and shift CO2 balance strongly toward net uptake; overall, GWP declines due to large CO2 sequestration and reduced N2O despite higher CH4. Age-dependent trajectories reveal relatively rapid transitions to net CO2 sinks (3–5 years in forests; ~4 years in wetlands), emphasizing the importance of time since restoration in evaluating climate benefits. These findings address uncertainties in how restoration affects the full GHG suite, identify key drivers (water table, bulk density, WFPS, nutrient status, pH), and provide quantitative evidence to refine national inventories and IPCC guidance. They indicate that targeted measures—afforestation/reforestation, rewetting of drained wetlands, and grassland restoration via grazing management or cropland conversion—can deliver substantial GHG mitigation.

By synthesizing 253 studies, this work quantifies global effects of ecological restoration on CH4, N2O, and CO2 fluxes across forests, grasslands, and wetlands, highlights restoration-age trajectories, and identifies biophysical drivers controlling GHG responses. Restoration generally strengthens carbon sinks and reduces GWP: forests and grasslands increase CH4 uptake without increasing N2O overall, while wetlands become CO2 sinks and reduce N2O but increase CH4 due to higher water tables, with net GWP reductions across ecosystems. Practical implications include prioritizing afforestation/reforestation, rewetting drained wetlands (including bog restoration), returning aquaculture to wetlands, and grassland restoration via grazing exclusion or reduced intensity and cropland conversion. The study provides data and emission factors useful for updating IPCC inventories and underscores the need to account for restoration age in assessments and models. Future research should focus on microbial mechanisms (methanogens, methanotrophs, nitrifiers, denitrifiers), long-term trajectories of GHG fluxes, and improving coverage of underrepresented restoration types to reduce uncertainties.

Paired restored–control datasets for forest CO2 fluxes (NEE, GPP, ER) were limited, precluding formal tests with RRd and t-tests for forests. Considerable heterogeneity exists among ecosystem types, restoration measures, and climates, and uncertainties remain about the duration and variability of transitions to net CO2 sinks. Mechanistic understanding is constrained by scarce studies on microbial communities governing CH4 and N2O dynamics during restoration. Some restoration categories and regions are underrepresented, which may affect generalizability.