Editorial NexoScience
A wetland works by drowning its own soil. Water sits over the ground, oxygen cannot get in, dead plant material stops rotting properly, and the carbon locked inside it stays in the mud instead of drifting off as carbon dioxide. That is why draining marshes and peatlands is treated as a climate offense, and why refilling them is one of the flagship moves of the UN Decade on Ecosystem Restoration. It is also, awkwardly, why restored wetlands give off so much methane.
Rewetting a drained wetland raises its methane emissions by 544%, according to researchers who pooled the results of 253 published field studies spanning more than two decades of measurements in forests, grasslands and wetlands worldwide. Across the restored sites in that dataset, methane output climbed from about 21 to 135 pounds of carbon per acre each year (23 to 151 kilograms per hectare) compared with the degraded land the restoration replaced. The analysis was published in 2024 in the journal Nature Communications.
Forest and grassland restoration does the exact opposite. Restored forests roughly doubled the amount of methane their soils pulled out of the air, and restored grasslands improved their uptake by about a third. The reason is dryness, not chemistry. Trees have deeper roots and higher water demand than the crops or grasses they replace, and their canopies intercept rain, so afforested soil holds less water. That leaves more open pore space for atmospheric oxygen and methane to diffuse downward, feeding methanotrophs, the bacteria that consume methane as food. In restored grasslands the mechanism is compaction: without livestock trampling the surface, soil bulk density falls, and the resulting root channels let more air through.
In a wetland, the water table decides everything. Raising the water level is like fitting a lid over the soil. Oxygen stops penetrating, the redox potential of the sediment drops, and methanogens, the microbes that produce methane precisely because there is no oxygen around, take over from everything else. The flooded ground also accumulates more organic carbon (decomposition has slowed down), which hands those same microbes extra fuel. Rewetted bogs add one more route: they favor plants with internal air channels running down through their tissue, which act like straws, carrying methane from the waterlogged root zone straight past the thin oxidized layer at the surface where it would otherwise be eaten. The surge does not last forever. Across the dataset, the methane increase leveled off roughly a decade after restoration.
The same flooding shuts down a different greenhouse gas. Nitrous oxide emissions fell by 68.6% at restored wetlands, while forest restoration had no significant effect on them at all. Two things drive that. Converting cropland or grazing land back into wetland cuts off fertilizer and animal waste, starving the nitrifying and denitrifying microbes of the nitrogen they work on. And a high water table lets denitrifying bacteria finish the reaction completely, reducing nitrous oxide all the way down to ordinary nitrogen gas rather than releasing it half-processed.
Added up, restoration still comes out ahead. Global warming potential, the standard accounting trick that puts all three gases on a single carbon dioxide scale, fell by 62% at restored wetlands relative to their paired control sites. That figure already includes methane’s outsize punch, since a given mass of methane traps dozens of times more heat than the same mass of carbon dioxide over a century, which is the conversion the authors used to add the gases together. Restored wetlands also flipped from being net carbon dioxide sources to net sinks in about four years. Forests and grasslands cut their warming potential by considerably more than wetlands did, which is the real shape of the finding: restoration works, but not equally, and not for free.
The comparison has limits worth stating plainly. These are field observations, not controlled experiments. Restored sites were compared against paired degraded sites nearby, or tracked across sites of different ages, so the differences are associations rather than proof that restoration alone caused them, and the design cannot rule out that restored and control plots differed in ways unrelated to the restoration itself. The authors were explicit about one gap: because too few paired measurements of carbon dioxide flux existed for forests, they did not statistically test the effect of forest restoration on those fluxes. The researchers also noted that when an original study failed to report a standard deviation, they substituted one tenth of the mean. And not every kind of restoration behaved like the average. Mangrove replanting and the conversion of aquaculture ponds back into wetland showed no significant change in methane at all.
The next questions are microbial. The practical target here is the IPCC’s own guidance for national greenhouse gas inventories, which currently assumes that rewetted organic soils emit no nitrous oxide whatsoever, a default the study attributes to a shortage of data rather than to evidence. This dataset puts that emission factor clearly above zero, and it gives inventory compilers numbers for wetland types they previously had none for. What it does not give them is the biology underneath. The researchers wrote that future investigations should prioritize the microbial mechanisms behind these shifts, because the methanogens, methanotrophs and denitrifiers doing the actual work remain the least studied part of every restored ecosystem on the list.
Source: He, T., Ding, W., Cheng, X., Cai, Y., Zhang, Y., Xia, H., Wang, X., Zhang, J., Zhang, K., & Zhang, Q. (2024). Meta-analysis shows the impacts of ecological restoration on greenhouse gas emissions. Nature Communications, 15, 2668. DOI: 10.1038/s41467-024-46991-5
