Soils sequester two times less carbon than hoped

Climate models needed a hero. Soil restoration seemed perfect for the role. In March 2026, a study published in MDPI Sustainability closed that chapter: across 1.2 billion hectares of degraded land, the real potential for organic carbon sequestration is 38.5 gigatonnes of carbon — two to three times lower than the estimates circulating for a decade. For grasslands alone, the figure drops to 3.4 GtC. The lever exists, but it is narrower than advertised.

This revision does not render restoration useless. It requires reformulating honestly what it can do, and what it cannot do. These are two different questions, and confusing them costs decisions.

The essentials

  • The potential for organic soil carbon sequestration across 1,223 Mha of degraded land is estimated at 38.5 GtC, of which 3.4 GtC for grasslands, according to a study published in MDPI Sustainability (March 2026); a separate study published in Nature Geoscience (2025) estimates a global potential of 96.9 Gt C (biomass and SOC combined) by 2100
  • Previous models overestimated this potential by assuming a return to pre-agricultural carbon stocks, an assumption that empirical data invalidates
  • Soil restoration retains proven value for biodiversity, agricultural yields, and hydrological resilience, but can no longer be counted as a first-rank climate solution
  • The gap between promises and reality opens a debate on budgetary and political trade-offs in climate solution portfolios

How models overestimated soil potential

For ten years, a simple equation circulated through climate reports: agricultural soils and degraded lands had lost between 50 and 70% of their original organic carbon under the impact of plowing, monocultures, and deforestation. In theory, it was sufficient to restore these lands to reconstitute the stocks. Some projections valued this potential at 100 GtC or more over several decades.

The problem lay in the initial assumption: the models assumed a return to pre-agricultural carbon stocks, that is, to levels of organic carbon that existed before agriculture transformed these landscapes. It is a convenient reference, but it is largely fictional. The climatic, hydrological, and biological conditions that allowed these high stocks no longer exist in the same configurations. Degraded soils do not recharge like filling a reservoir.

The study published in MDPI Sustainability in March 2026, titled Empirically Based Estimates of Soil Organic Carbon Gains After Ecosystem Restoration and Their Global Climate Benefits, applied a different methodology: empirical data collected from actual restoration sites, compared to undegraded soils in equivalent contexts. The result is a potential of 38.5 GtC across 1,223 million hectares, a downward revision of 60 to 70% from the high estimates that fed official strategies. For restored grasslands, the gap is even more pronounced: the 3.4 GtC measured are well below what large-scale models projected.

This type of revision is not rare in carbon science. Tropical forests, long presented as quasi-unlimited sinks, have been subject to similar reassessments since studies showed that some zones are close to saturation or emit more than they absorb due to repeated droughts, according to research conducted in the Brazilian Amazon. The precision of measurements increases, and hopes recede.

38.5 GtC in perspective: large and insufficient at once

To calibrate this figure, a few reference points. Global CO₂ emissions reached approximately 37.4 gigatonnes per year in 2023, according to data from the International Energy Agency. The 38.5 GtC of soil organic carbon represents, once converted to CO₂, approximately 141 Gt CO₂ — roughly three to four years of current global emissions, spread over several decades of restoration. In actual annual flux, the most optimistic projections bank on 1 to 2 GtC per year if large-scale restoration is actually implemented, which would require unprecedented political and financial mobilization.

This is not negligible. But in a carbon budget compatible with 1.5°C or 2°C of warming, where every gigatonne counts, the gap with previous projections has direct consequences for how governments and companies account for their carbon neutrality commitments.

The problem is not only scientific. It is financial and political. The voluntary carbon market, valued at approximately 1.4 to 2.5 billion dollars in 2024-2025, has seen the development of soil and grassland restoration projects — though they represent only a minority fraction, with the market remaining dominated by forestry projects. Large companies have incorporated these credits into their carbon footprints to offset their emissions. If the real potential is two to three times lower than initial assumptions, some of these offsets rest on carbon that will never be sequestered.

What restoration actually does well

Reducing soil restoration to its climate disappointment would be an error symmetrical to overvaluing it. The documented benefits exist, they are simply different from what was promised.

Restoration improves the physical and biological structure of degraded soils. Healthier soils retain water better, reduce runoff, and decrease erosion risks. In agricultural areas, this translates into more stable yields, especially during drought episodes. Available estimates of soil degradation costs in sub-Saharan Africa are substantial — sources affiliated with IFPRI mention agricultural productivity losses on the order of 68 to 78 billion dollars per year. Restoring these lands, even if the effects on atmospheric carbon are modest, represents a defensible economic investment.

Biodiversity is the other concrete gain. Restored grasslands provide habitats for pollinators, birds, and insectivores whose disappearance ultimately compromises food systems. These ecosystem services are difficult to monetize, but they have real value that poorly designed carbon markets tend to ignore because they did not fit the categories.

Are these benefits sufficient to justify investments? Probably yes, provided they are evaluated honestly. The problem arises when restoration is financed and assessed exclusively through the carbon lens, which produces seductive but incorrect metrics.

The trade-offs this revision forces into being

If soil restoration captures less carbon than announced, two responses are possible. The first is to maintain the same climate objectives by increasing effort in other directions. The second is to accept a revision of trajectories, something few actors are willing to do publicly.

Solutions that do not need downward reassessment are well known. Decarbonization of electricity and heavy industry remains the main path: renewable energies and nuclear power together represent a potential for emission avoidance incomparably broader than any lever based on natural sinks. Energy efficiency in buildings, transport, and industrial processes offers real and measurable savings.

Direct air carbon capture technologies, or DAC, are often presented as an alternative. They remain expensive and at a very small scale: according to the IEA, global operational DAC capacity reached only approximately 10,000 tonnes per year in 2023. By end of 2024, the total capacity of operational facilities approaches the order of one million tonnes according to some sources, but the vast majority remains in the form of planned or under-construction projects — an infinitesimal fraction of global emissions. Their possible scaling over the next two decades could not alone compensate for a deficit of several tens of gigatonnes.

The forest remains a more robust sink than agricultural soils, provided existing primary forests are protected rather than planting new monocultures. Deforestation represents approximately 10% of annual global emissions according to the IPCC. Stopping deforestation is climatically more efficient than most restoration projects, and data on this point have not been revised downward.

What carbon markets will have to correct

The voluntary carbon market has faced a credibility crisis since 2023, when several studies showed that protected forest projects sold credits corresponding to largely overestimated avoided emissions. The revision of soil potential fits into the same dynamic: the carbon sold does not correspond to the carbon sequestered.

This situation is not a reason to condemn carbon markets. It is a reason to correct them. Several initiatives are working on standardizing methodologies for measurement, reporting, and verification. The Integrity Council for the Voluntary Carbon Market published in 2023 a set of standards that credit issuers are encouraged to adopt. Adoption remains slow and uneven, but the direction is clear.

For soil restoration projects, the correction involves replacing modeled projections with direct field measurements, repeated over time. This increases project costs and mechanically reduces the credits they can issue. Some projects will become economically unviable. Others will reorient toward biodiversity and agricultural resilience benefits, financed by payment for ecosystem services mechanisms rather than carbon markets.

This adjustment is painful for project proponents, but it is necessary for carbon markets to remain an instrument of real decarbonization rather than creative accounting.

Neither abandonment nor overestimation

The sequence leading from promise to revision is familiar. A scientific discovery identifies a lever. Project proponents and governments amplify it because it offers a practical solution. Finer data subsequently correct it, often downward. This pattern does not disqualify science, it is its normal expression.

Soil restoration falls into this sequence. It exits the category of first-rank climate solutions. It remains a justified practice for its effects on biodiversity, agricultural resilience, and ecosystem quality. These effects merit solid financing, based on what they actually are.

The real question posed by the MDPI Sustainability study is not “should we abandon soil restoration?”. It is: which levers, in the climate portfolios of states and companies, were underfunded because restoration seemed sufficient? Identifying these blind spots and reallocating resources accordingly is the work this revision makes urgent.

Sources

  1. Main study (38.5 GtC SOC) — Empirically Based Estimates of Soil Organic Carbon Gains After Ecosystem Restoration and Their Global Climate Benefits, MDPI Sustainability, March 2026 — https://www.mdpi.com/2071-1050/18/5/2516
  2. Study on ecosystem restoration (global potential 96.9 Gt C by 2100), Nature Geoscience, 2025 — https://www.nature.com/articles/s41561-025-01742-z
  3. International Energy Agency (IEA), CO₂ Emissions in 2023, Paris, 2024 — https://www.iea.org/reports/co2-emissions-in-2023
  4. IPCC, Sixth Assessment Report — Working Group III, chapter on land use and forestry, 2022
  5. Integrity Council for the Voluntary Carbon Market, Core Carbon Principles, 2023 — https://icvcm.org/core-carbon-principles/
  6. International Food Policy Research Institute (IFPRI), data on soil degradation and agricultural productivity in sub-Saharan Africa — IFPRI Discussion Paper 01811
  7. Cambridge/Science — Overestimation of REDD+ forest credits — https://www.cam.ac.uk/stories/carbon-credits-hot-air
  8. International Energy Agency (IEA), global DAC capacity — https://www.iea.org/energy-system/carbon-capture-utilisation-and-storage/direct-air-capture