上海石化加大流延膜专用料产品研发力度

Introduction .
Global temperature is approaching a threshold that will have irreversible consequences for the future of our Earth, mainly due to the increasing concentrations of atmospheric greenhouse gases (GHG) such as carbon dioxide (CO 2 ), methane (CH 4 ) and nitrous oxide (N 2 O)1. Over the last 220 years, global CO 2 concentrations increased from 283 to 419 parts per million (ppm), CH 4 increased from 750 to 1925 parts per billion (ppb), and N 2 O increased from 273 to 336 ppb2. Land-use change and ecosystem degradation have caused massive anthropogenic emissions of GHG and altered natural ecological ecosystems from net sinks to net sources3 , 4. Restoring the degraded ecosystems and converting lands back to healthy ecosystems has been proposed as a vital strategy for stabilizing the Earth’s climate5. To limit global warming below the 2°C threshold, there is an urgent need to reduce atmospheric GHG concentrations by restoring degraded ecosystems such as forests, grasslands and wetlands6. Ecological restoration is the process of assisting the recovery of an ecosystem that has been degraded, damaged, or destroyed (Society for Ecological Restoration and Policy Working Group 2002). The United Nations (UN) has declared 2021-2030 as the ‘UN Decade on Ecosystem Restoration’ and calls on countries to meet commitments to restore one billion hectares of land. The Bonn Challenge and the New York Declaration on Forests have established ambitious targets to restore 350 million hectares of forests worldwide by 20305. Thus, systematically understanding the impacts of ecological restoration on GHG emissions is imperative for making better restoration policies and improving the Intergovernmental Panel on Climate Change (IPCC) guidance of GHG inventories.
Forests occupy approximately 30% of the global land surface and play a crucial role in regulating the global carbon (C) cycle and reducing global warming7 , 8 , 9. A recent estimation reported that global forests maintained a net C sink of 7.6 Gt CO 2 e yr 1 , reflecting a balance between gross C removals (15.6 Gt CO 2 e yr 1 ) and gross emissions (8.1 Gt CO 2 e yr 1 ) from deforestation and other disturbances, e.g., clear-cut, fire, windthrows, insects, etc10. Afforestation and reforestation could change biomass accumulation and alter soil biogeochemical, physical and hydrological properties, thereby affecting the GHG fluxes9 , 11 , 12 , 13. Previous work found that converting croplands to forests increased CH 4 uptake due to the decreased soil bulk density, and afforestation decreased N 2 O emissions due to the reduced nitrogen (N) substrate availability11. The conversion of grasslands to forests might decrease CH 4 emissions but increase N 2 O emissions12. Although many studies showed that afforestation could enhance the CO 2 sink function of ecosystems10 , 14, some studies observed that forest lands continued to act as a CO 2 source even after several years of afforestation15. These diverse results suggest that the magnitude and direction of GHG dynamics driven by forest restoration are highly uncertain and could be affected by multiple factors, including ecosystem types, restoration ways, and restoration age12 , 16 , 17. It is undoubtedly necessary to explore the general patterns and the major controlling factors of GHG emissions in the restored forests.
Grassland ecosystems constitute approximately 40% of the terrestrial biosphere18, and natural grasslands are usually identified as efficient sinks of atmospheric CH 4 and CO 2 19, but sources of N 2 O20. Grassland degradation leads to changes in soil nutrient content, soil moisture, and plant composition, which influences the pattern of GHG emissions21 , 22. It has been found that grassland degradation might decrease CH 4 uptake by 40%23. However, whether grassland restoration can reduce GHG emissions is still inconclusive19 , 24. Previous work reported that grassland restoration increased C accumulation and enhanced CH 4 uptake24, but some studies found that grassland restoration might stimulate N 2 O and CO 2 emissions and shift grassland from a C sink to a C source19. Furthermore, the effects of grassland types, restoration measures, and restoration age on GHG emissions in the restored grasslands at a global scale are still unclear.
Wetlands are considered to be one of the most efficient ecosystems for sequestrating CO 2 from the atmosphere25, mainly because inundation creates anaerobic conditions that prevent the decomposition of dead plant material and restore sequestered C in soil26 , 27. Despite covering only 5–8% of the Earth’s landscape, global wetlands store 20–30% of soil C on the Earth and thereby play an important role in the global C cycle28. In general, wetland drainage and degradation decrease CH 4 emissions but enhance CO 2 and N 2 O emissions to the atmosphere, converting the wetlands from C sinks into sources29 , 30. However, the impacts of restoration on wetland GHG and the driving factors remain controversial31 , 32 , 33. Previous work reported that wetland restoration could shift the ecosystems into net GHG sources32 , 34 , 35or net sinks36. The inconsistent results are probably attributed to the wetland restoration types, restoration age, climate, water table depth, and soil properties32 , 37. Since wetland restoration generally decreases CO 2 emissions but increases CH 4 emissions32 , 34 , 36, the overall effects of wetland restoration on the global warming potentials (GWP) considering three major GHGs (i.e., CO 2 , CH 4 , and N 2 O) are not well understood.
Despite numerous studies investigating the effects of ecological restoration on the emission of individual or a few GHGs at the plot or regional level38, the general pattern of the impacts of ecological restoration on the three major GHGs at a global scale has not yet been analyzed. Furthermore, there is currently a lack of comprehensive global assessments for the three major ecosystems (i.e., forests, grasslands, and wetlands) which are crucial for the global GHG budget and the ‘UN Decade on Ecosystem Restoration’6 , 7 , 8 , 9. In addition, detailed data on the responses of GHG to ecological restoration are lacking in the IPCC reports, the IPCC Guidelines for National Greenhouse Gas Inventories, and the Good Practice Guidance for Land Use, Land-Use Change and Forestry. To fill these knowledge gaps, we compiled a global dataset from 253 peer-reviewed articles and conducted a meta-analysis to assess the effects of ecological restoration on GHG emissions (Fig.  1 ). Our specific objectives were to (1) quantify the impacts of ecological restoration on CH 4 and N 2 O emissions and net ecosystem CO 2 exchange (NEE) in forest, grassland and wetland ecosystems, (2) explore the patterns of GHG emissions with restoration age, and (3) determine the key factors influencing the response of GHG emissions to ecological restoration.
Fig. 1: Global distribution of the study sites for this meta-analysis. The free continental data of the world map was sourced from Natural Earth, supported by the North American Cartographic Information Society ( https://www.naturalearthdata.com/ ). ArcGIS Desktop 10.8 (Esri, West Redlands, CA, USA) was employed for mapping the distribution of the study sites. Source data are provided as a Source Data file.
Full size image
In this work, we show that forest and grassland restoration increases CH 4 uptake, mainly due to the changes in soil properties. Conversely, wetland restoration increases CH 4 emissions, primarily attributed to elevated water table depth. Forest and grassland restoration has no significant effect on N 2 O emissions, while wetland restoration reduces N 2 O emissions. Overall, forest, grassland and wetland restoration enhances C sink, reduces the global warming potentials, and can serve as strategies for mitigating GHG.
Results .
Effects of ecological restoration on CH 4 emissions .
Overall, forest and grassland restoration significantly decreased CH 4 emissions, and the weighted response ratios ( RRd ) of CH 4 emissions were 2.3 (95% CI: 2.9 to 1.6) and 1.6 (95% CI: 2.4 to 0.8) under forest and grassland restoration, respectively (Fig.  2a ). Compared with the paired control ecosystems, forest and grassland restoration averagely increased CH 4 uptake from 1.0 to 1.9kgC ha 1 year 1 (by 90.0%) and 2.6 to 3.4kgC ha 1 year 1 (by 30.8%), respectively (Fig.  3b, c ). Among the types of grassland restoration, temperate steppe & meadow and desert steppe increased CH 4 uptake from 2.6 to 3.8kgC ha 1 year 1 (by 46.2%) and 7.7 to 11.4kgC ha 1 year 1 (by 48.4%), respectively (Fig.  3c ). Wetland restoration significantly increased CH 4 emissions by 544.4% ( RRd : 2.9; 95% CI: 2.43.4; P < 0.05) (Fig. 2a ). The average CH 4 emissions increased from 23.4 kg C ha 1 year 1 to 150.8 kg C ha 1 year 1 after wetland restoration (Fig. 3a ). Among the types of wetland restoration, the conversion of grasslands to wetlands showed the largest increase in CH 4 emissions, with an average increase from 61.2 kg C ha 1 year 1 (in paired control) to 284.8 kg C ha 1 (in restored wetlands) (Fig. 3a ). In contrast, there was no significant change in CH 4 emissions when aquaculture ponds were converted to wetlands ( RRd : 1.7; 95% CI: 3.8 to 0.5) and mangroves were restored ( RRd : 1.1; 95% CI: 0.4 to 2.5) (Fig. 2a ).
Fig. 2: Effects of ecological restoration on CH 4 (a), N 2 O (b), and NEE (c) fluxes across the different wetland, forest and grassland restoration categories. The overall effect size was calculated with a categorical random effects model. Values are means ±95% CIs of the weighted response ratios ( RRd ) between the paired control ecosystems and restored ecosystems. If the 95% CI value does not overlap with zero at the α =0.05 level, the response is considered significant. The asterisks indicate significant positive or negative effects. Numbers next to the y-axis indicate sample sizes ( n ). Due to the small sample size for the paired restored-control measurements for the NEE in forests, the effects of forest restoration on the NEE were not tested by RRd . DG to W, drained grassland to wetland; DF to W, drained forest to wetland; DC to wetland, drained cropland to wetland, AQ to Wetland, aquaculture to wetland; NEE net ecosystem CO 2 exchange, TGM temperate steppe & meadow, AGM alpine steppe & meadow, DS desert steppe, AG artificial grassland. Source data are provided as a Source Data file.
Full size image
Fig. 3: Box plots of CH 4 and N 2 O fluxes in the restored wetlands and their paired control ecosystems (a and d), the restored forests and their paired control ecosystems (b and e), the restored grasslands and their paired control ecosystems (c and f). Every two adjacent boxes represent the paired control-restored measurements. The paired control ecosystems are prefixed with ‘P’. TGM, temperate steppe & meadow; AGM, alpine steppe & meadow. Box boundaries represent the 75th and 25th percentiles, whisker caps represent the 95th and 5th percentiles, and circle points represent outliers. Diamond points and solid lines inside the boxes represent means and medians, respectively. Asterisks () denote significance at p < 0.05, as determined by using a two-sided, independent samples t test. No adjustments were made for multiple comparisons. Numbers next to the x-axis indicate sample sizes (n). Exact p -values and Source data are provided as a Source Data file.
Full size image
Effects of ecological restoration on N 2 O emissions .
Overall, forest ( RRd : 0.4; 95% CI: 1.3 to 0.4) restoration did not affect N 2 O emissions, while grassland and wetland restoration reduced N 2 O emissions by 21.7% ( RRd : 0.7; 95% CI: 1.4 to 0.1) and 68.6% ( RRd : 2.9; 95% CI: 3.9 to 1.9; P < 0.05) (Fig. 2b ), respectively. When considering the types of forest restoration, the conversion of croplands to forests averagely decreased N 2 O emissions from 3.7 kg N ha 1 year 1 (in paired control) to 1.4 kg N ha 1 year 1 (in restored forests) ( RRd : 3.3; 95% CI: 4.7 to 1.9; P < 0.05) (Fig. 3e ). Among the types of wetland restoration (Fig. 2b ), the conversion of grasslands to wetlands averagely reduced N 2 O emissions from 5.2 kg N ha 1 year 1 (in paired control) to 2.6 kg N ha 1 year 1 (in restored wetlands), and the conversion of croplands to wetlands averagely decreased N 2 O emissions from 17.0 kg N ha 1 year 1 to 2.3 kg N ha 1 year 1 . Compared with the paired control ecosystems, peatland restoration averagely reduced N 2 O emissions from 2.2 kg N ha 1 year 1 to 0.5 kg N ha 1 year 1 . However, floodplains restoration did not significantly affect N 2 O emissions (Fig. 2b ). Among the measures of grassland restoration, the conversion of croplands to grasslands averagely decreased N 2 O emissions from 2.3 kg N ha 1 year 1 to 0.7 kg N ha 1 year 1 (Supplementary Fig. S1b ). Similarly, prairie restoration reduced N 2 O emissions from 4.8 kg N ha 1 year 1 to 0.1 kg N ha 1 year 1 ( RRd : 10.9; 95% CI: 14.7 to 7.1; P < 0.05 (Fig. 3f ).
Effects of ecological restoration on CO 2 fluxes and GWP .
Overall, wetland restoration significantly reduced NEE by 138.8% ( RRd : 3.2; 95% CI: 3.8 to 2.5; P < 0.05) (Figs. 2 c and 4a , Table 1 ). Compared with the paired control ecosystems, the conversion of grasslands to wetlands averagely reduced NEE from 231.9 g C m 2 year 1 to 219.5 g C m 2 year 1 , and the conversion of aquaculture to wetlands averagely reduced NEE from 41.9 g C m 2 year 1 to 151.5 g C m 2 year 1 (Supplementary Table S1 ). Bogs restoration averagely reduced NEE from 159.2 to 35.8 g C m 2 year 1 (Fig. 4a ). The conversion of grasslands to wetlands decreased gross primary productivity (GPP) and ecosystem respiration (ER), while bogs restoration increased GPP and ER (Supplementary Fig. S2 ). The floodplains and mangrove restoration showed no significant effect on GPP and ER (Supplementary Fig. S2 ).
Fig. 4: Box plots of annual CO 2 fluxes (NEE, GPP and ER) the restored wetlands and their paired control ecosystems (a–c), the restored forests and their paired control ecosystems (d–f), and the restored grasslands and their paired control ecosystems (g–i). Every two adjacent boxes represent the paired control-restored measurements. The paired control ecosystems are prefixed with ‘P’. Box boundaries represent the 75th and 25th percentiles, whisker caps represent the 95th and 5th percentiles, and circle points represent outliers. Diamond points and solid lines inside the boxes represent means and medians, respectively. Asterisks () denote significance at P < 0.05, as determined by using a two-sided, independent samples t test. No adjustments were made for multiple comparisons. Numbers next to the x-axis indicate sample sizes (n). NEE net ecosystem CO 2 exchange, GPP, gross primary productivity, ER ecosystem respiration, TGM temperate steppe & meadow, AGM alpine steppe & meadow. Due to the small sample size for the paired restored-control measurements for the NEE, GPP, and ER in forests, the effects of forest restoration on CO 2 fluxes were not tested by the t-test. Exact p -values and Source data are provided as a Source Data file.
Full size image
Table 1 Changes in comprehensive C budget and GWP when converting the paired control ecosystems to the restored ecosystems (Mean±SE)Full size table
Overall, grassland restoration decreased NEE by 146.9% ( RRd : 4.7; 95% CI: 5.8 to 3.5; P < 0.05) (Figs. 2 c, 4h and Supplementary Fig. S1c ). Compared with the paired control ecosystems, grassland restoration by grazing exclusion averagely decreased NEE from 245.3 g C m 2 year 1 to 703.0 g C m 2 year 1 , grassland restoration by reducing grazing density averagely reduced NEE from 587.9 g C m 2 year 1 to 1460.1 g C m 2 year 1 , and the conversion of cropland to grassland averagely reduced NEE from 10.3 g C m 2 year 1 to 75.8 g C m 2 year 1 (Supplementary Fig. S1c , Table S2 ). Grassland restoration increased GPP and ER (Supplementary Fig. S2 ).
Due to the small sample size for the paired restored-control measurements for the NEE, GPP, and ER in forests, the effects of forest restoration on CO 2 fluxes were not tested by the RRd and t-test (Figs.  2 c, 4d–f ). Based on the restoration chronosequence sub-dataset, the NEE in restored forests decreased first and then tended to be stable and showed a negative exponential relationship with afforestation age, while the GPP and ER showed a positive exponential relationship with afforestation age (Fig.  5e ; P < 0.001). Similarly, the NEE was negatively and exponentially correlated with reforestation age and time since restoration after disturbance (Fig. 5f, g ; P < 0.001).
Fig. 5: Relationships of the response ratios ( RRd ) of CH 4 and N 2 O fluxes, and the annual CO 2 fluxes with restoration age.a , b relationships between the response ratios of wetland CH 4 ( a ) and N 2 O ( b ) fluxes and restoration age. c Relationship between the response ratios of forest and grassland CH 4 fluxes and restoration age. d – f relationships between the CO 2 fluxes in wetland ( d ) and forest ( e and f ) and restoration age. g relationships between the fluxes of NEE and restoration age. h , i relationships between GPP/ER and afforestation time ( h ), and restoration age after forest disturbance ( i ). Linear and nonlinear regression were used and the error bands surrounding the regression lines represent the 95% confidence interval of the correlation. The n is the number of observations. Exact p -values and Source data are provided as a Source Data file.
Full size image
On average, the C budget (CO 2 and CH 4 ) was 295.5, 506.5 and 53.4gC m 2 year 1 for forest, grassland and wetland restoration, respectively, indicating the capacity of enhanced C sink in the restored ecosystems (Table  1 ). On average, forest, grassland and wetland restoration decreased the GWP by 327.7%, 157.7% and 62.0% compared with their paired control ecosystems, respectively (Table  1 ).
Changes of CH 4 and N 2 O emissions and NEE with restoration age .
Given the critical impact of restoration age on GHG emissions in restored ecosystems, the patterns of CH 4 and N 2 O emissions and NEE with restoration age were first explored. Overall, the restoration age had a significant effect on CH 4 emissions (Fig.  5c and Supplementary Fig.  S3b ). The RRd of CH 4 emissions in the restored forests was negatively correlated with restoration age (i.e., years since restoration) (Fig.  5c ). The soil CH 4 uptake showed no response to the afforestation age within 10 years, while soil CH 4 uptake increased with afforestation age for longer time intervals (Supplementary Fig.  S3b ). The RRd of CH 4 emissions in the restored wetlands was exponentially and positively correlated with restoration age and achieved a relatively stable value in about 10 years since restoration (Fig.  5a ). The RRd of N 2 O emissions in the wetland was negatively correlated with restoration age (Fig.  5b ).
Restoration age was an important factor influencing CO 2 fluxes (Fig.  5 ). The GPP/ER showed low values (<1 or ≈1) in the early years following afforestation and restoration from fire and clear-cutting (Fig. 5g, h ). The GPP/RE became greater than 1 (i.e., NEE < 0) by 4 years, 6 years, 13 years, and 8 years after restoration for the afforestation sites, clear-cutting sites, post-fire sites, and all disturbances sites, respectively (Fig. 5 ). The GPP/ER ratio varied with afforestation age and time since restoration after disturbance, and had an asymptote of 1.19 and 1.09, respectively (Fig. 5g, h ; P < 0.001). The NEE in the wetlands was exponentially and negatively correlated with restoration age and the switchover time from net CO 2 sources to net CO 2 sinks was estimated to be approximately 4 years (Fig. 5d ).
Factors influencing the response of CH 4 , N 2 O, and NEE to ecological restoration .
Forest restoration significantly increased concentrations of soil organic C (SOC), NH 4+ -N and dissolved organic C (DOC), but reduced soil temperature, soil water-filled pore space (WFPS), soil moisture, and pH ( P < 0.05; Fig. 6b ). Grassland restoration remarkably increased soil WFPS, soil moisture, vegetation coverage, and grassland aboveground and belowground biomass, but reduced soil bulk density (BD) and soil NO 3 -N concentrations ( P < 0.05; Fig. 6c ). Wetland restoration significantly increased water table depth, soil SOC and total N (TN), but decreased soil BD, soil redox potential (Eh), pH and NH 4+ -N concentrations ( P < 0.05; Fig. 6a ).
Fig. 6: Effects of ecological restoration on soil properties in wetland (a), forest (b), and grassland (c) ecosystems. The overall effect size was calculated with a categorical random effects model. Values are meant ±95% CIs of the weighted response ratios ( RRd ) between the paired control ecosystems and restored ecosystems. If the 95% CI value does not overlap with zero, the response is considered significant. The asterisks indicate significant positive or negative effects. SOC soil organic carbon, TN total N, C: N carbon/nitrogen ratio, Eh soil redox potential, BD bulk density, SM soil moisture, WFPS water-filled pore space, ST soil temperature, WT water table depth, VC vegetation coverage, TB total biomass, AB aboveground biomass, BB belowground biomass; NH 4+ ammonium, NO 3 nitrate. Source data are provided as a Source Data file.
Full size image
The RRd of CH 4 emissions in the restored forests was positively correlated with the RRd of soil WFPS ( P < 0.01; Fig. 7b ). The RRd of CH 4 emissions in the restored forests and grasslands was negatively correlated with the RRd of BD (Fig. 7a ). Afforestation decreased CH 4 emissions regardless of the tree types (i.e., coniferous and deciduous forest) ( P < 0.01; Supplementary Fig. S3c ). The RRd of N 2 O emissions in the forests and grasslands were positively correlated with the RRd of soil NH 4+ -N and NO 3 -N concentrations (Fig. 7d, e ), and the RRd of N 2 O emissions in the forests was negatively correlated with the RRd of soil pH ( P < 0.01; Fig. 7f ).
Fig. 7: Relationships of the response ratios ( RRd ) of CH 4 (a–c), N 2 O (d–g), and NEE (h and i) with response ratios ( RRd ) of soil properties after wetlands, forest, and grassland ecosystems were restored. WT water table depth, BD bulk density, SM soil moisture, WFPS water-filled pore space, AB aboveground biomass, VC vegetation coverage. Linear and nonlinear regression were used and the error bands surrounding the regression lines represent the 95% confidence interval of the correlation. The n is the number of paired observations. Exact p -values and Source data are provided as a Source Data file.
Full size image
The RRd of CH 4 emissions in the restored wetlands was exponentially and positively correlated with water table depth ( P < 0.01; Fig. 5a and Fig. 7c ). The RRd of N 2 O emissions in the wetland was positively correlated with the RRd of soil NH 4+ -N concentrations (Fig. 7g ). The N 2 O emissions and NEE of the restored wetlands were negatively related to water table depth ( P < 0.001; Supplementary Fig. S4c ).
Across all restoration groups, GPP and ER were positively correlated with the temperature and precipitation ( P < 0.01; Supplementary Fig. S5 ). The RRd of CH 4 emissions in all systems was negatively correlated with the aridity index ( P < 0.01; Supplementary Fig. S6c ). When the precipitation was larger than 900 mm, the RRd of N 2 O emissions in all ecosystems was positively correlated with precipitation ( P < 0.05; Supplementary Fig. S6e ). The RRd of NEE in all ecosystems was negatively correlated with the aridity index when the aridity index was greater than 0.9 ( P < 0.05; Supplementary Fig. S6i ).
Discussion .
Forest and grassland restoration increased CH 4 uptake while wetland restoration enhanced CH 4 emissions .
We found that forest and grassland restoration significantly increased CH 4 uptake (Table  1 , Fig.  2 ), suggesting the great potential of forest and grassland ecosystem restoration in enhancing sink function for CH 4 . The conversion of croplands and grasslands to forests increased CH 4 uptake by 84.8% and 106.8% (Fig.  3 ), respectively, indicating high CH 4 removal efficiency. The RRd of CH 4 emissions linearly increased with the RRd of WFPS (Fig.  7b ), indicating that lower soil moisture and WFPS enhanced CH 4 uptake and inhibited CH 4 emissions from soil. Afforestation significantly decreased WFPS in the forest ecosystems (Fig.  6b ), mainly because trees had deeper roots and higher water demands than crops and grasses, and afforestation enhanced evapotranspiration and canopy interception of precipitation39 , 40. The decrease in soil WFPS caused by afforestation can enhance the diffusion of atmospheric O 2 and CH 4 into the soils, thereby increasing CH 4 oxidation and uptake in the afforested soils39. Soil compaction by machinery in the agricultural lands and trampling by livestock in the grasslands may increase soil bulk density and reduce soil porosity41 , 42 , 43. Our results showed that grassland restoration significantly reduced soil bulk density (Fig.  6c ), and the RRd of CH 4 emissions showed a negative relationship with the RRd of soil bulk density (Fig.  7a ), implying that the lower soil bulk density in the restored grasslands increased CH 4 diffusion from atmosphere into soils and thus increased CH 4 uptake44. Grassland restoration by reducing grazing intensity or grazing exclusion increased belowground biomass (Fig.  6c ), which may form “root holes” and improve soil aerobic conditions for diffusion of atmospheric CH 4 into the soil profiles and the growth of methanotrophs45, thereby enhancing CH 4 uptake in the restored grassland. In addition, the increase in SOC in the afforested lands (Fig.  6b ) could enhance soil macropores and the number of coarse pores46, and thus create favorable environments for methanotrophs growth and CH 4 oxidation47. Taken together, the increased CH 4 uptake in the restored forests and grasslands could be attributed to the changes in soil properties.
Forest restoration significantly increased CH 4 uptake with the increase of afforestation age (Fig.  5c and Supplementary Fig.  S3b ), which could be mainly attributed to the increased SOC and decreased soil moisture and WFPS following afforestation (Figs.  6 b and 7b )48 , 49. Bárcena et al. reported that soil SOC concentrations increased with stand age and therefore increased the abundance and activity of methane-oxidizing bacteria growth by supplying abundant substrates50, consequently resulting in an enhanced CH 4 oxidation rate with afforestation age. Gatica et al. found that, with the increase of afforestation age, soil moisture was decreased by the combined effects of increasing rainfall interception51and tree transpiration in the older forest stands39, and thus enhanced CH 4 consumption with time. Therefore, these individual observations support our results and inferences48 , 49 , 50 , 51.
We found that wetland restoration significantly increased annual CH 4 emissions by 5.4 times compared with the paired control ecosystems, indicating that wetland restoration enhanced the CH 4 source strength. Among the types of wetland restoration, the conversion of grasslands to wetlands showed the largest increase in CH 4 emissions, followed by the conversion of croplands to wetlands (Fig.  3a ). These results indicated that greater attention should be paid to the increased CH 4 emissions from the restored wetlands in global GHG accounting. Previous work showed that CH 4 emissions in the wetlands were mainly controlled by water table level26 , 38, nutrient status44, plant species52, and microbial activity53. Unlike the CH 4 emissions in forest and grassland ecosystems which were influenced by soil properties (Fig.  7a, b ), the RRd of wetland CH 4 emissions was positively correlated with water table depth, suggesting that the wetland restoration facilitated the production of CH 4 by creating an anaerobic environment through elevated water table levels (Figs.  6 a, 7c ). Long-time waterlogging during the restoration period reduced O 2 penetration into the sediments and thus induced a reduction in the redox potential of 1.23-fold compared with the paired control ecosystems (Fig.  6a ), which in turn stimulated methanogen growth and activity, thereby enhancing CH 4 emissions54. Bog restoration by rewetting may be beneficial to the proliferation of aerenchymatous vascular plants, and thus allow CH 4 to bypass the oxidized surface soil, consequently enhancing CH 4 emission into the atmosphere via the plant-mediated transport32. In addition, restoration of wetlands by rewetting created an anaerobic environment which may inhibit microbial activity and reduce SOC decomposition55, thus leading to a higher SOC concentration in the restored wetlands (Fig.  6a ). The higher SOC concentrations in the restored wetlands could provide more substrates for methanogen growth than in the paired control ecosystems44 , 56. Thus, elevating the water table level and increasing the substrate supply for methanogen is likely to result in higher CH 4 fluxes in the restored wetlands (Figs.  6 a, 7c , and Supplementary Fig.  S4a ).
The RRd of wetland CH 4 showed exponentially relationship with restoration age and achieved a relatively stable value in about 10 years since restoration (Fig.  5a ). Similarly, Mitsch et al. reported that wetland restoration initially stimulated CH 4 emissions but decreased over time and reached CH 4 emissions comparable to the natural wetland after 13-15 years28. The rapid response of CH 4 emissions to wetland restoration at the initial stage was mainly due to the restoration of the natural hydrology and the inundation of easily decomposable plant litters, which created an anaerobic environment and sufficient substrate for the growth of methanogens and CH 4 production53 , 57 , 58 , 59.
Diverse responses of N 2 O emissions to ecological restoration .
Our findings revealed that the conversion of agricultural lands to forests significantly decreased N 2 O emissions and the conversion of grasslands and wetlands to forests stimulated N 2 O emissions, indicating the response patterns of N 2 O emissions to forest restoration depends on the prior land-use type. The RRd of N 2 O emissions in forests was positively related to the RRd of NH 4+ and NO 3 (Fig.  7 ), indicating that the cessation of fertilization in the afforested croplands may lead to a reduction in soil N 2 O emissions compared with the fertilized croplands11. The main reasons for the increased N 2 O emissions in the forests converted from grasslands could be attributed to that afforestation in grasslands increased the concentrations of soil SOC and NH 4+ and decreased soil pH (Fig.  7 , Supplementary Fig.  S7 ). The increased SOC and NH 4+ could increase soil C and N availability and soil nitrification for N 2 O production12 , 60. Our results demonstrated that the RRd of N 2 O emissions showed a negative relationship with the RRd of soil pH (Figs.  6 b, 7f ). The reduction of soil pH may inhibit the activity of the N 2 O reductase enzyme and in turn increase N 2 O/N 2 ratios in the denitrification, consequently increasing N 2 O emissions from denitrification in the afforested soils4 , 61. Grassland restoration by conversion of cropland to grassland sharply decreased N 2 O emissions (Supplementary Fig.  S1b ), mainly due to the decreased concentrations of soil NO 3 by stopping fertilization (Figs.  6 c, 7 ). In contrast, artificial assisted restoration in the degraded grasslands by applying organic or mineral fertilizer may increase N 2 O emissions by increasing the availability of N for N 2 O production62. Thus, our results indicated that different grassland restoration measures showed distinct impacts on N 2 O emissions (Supplementary Fig.  S1b ).
Wetland restoration significantly decreased N