Insights · Land eligibility
Planting trees in grasslands and savannas: when not to
Open, treeless land looks like the obvious place for a tree-planting carbon project. Often it is the wrong place. Many of the world's grasslands and savannas are ancient ecosystems, not forests that were cut down, and the carbon rules treat converting them very differently from restoring a cleared forest. This post explains why, how to tell the difference, and where trees on open land can still work.
The mistake: treating open land as deforested land
A widely used map of restoration opportunity, published by the World Resources Institute, treated non-forest areas where the climate could support forest as deforested. In a 2015 letter in Science, Veldman and colleagues called that assumption erroneous: much of the area marked for “forest restoration” corresponds to the world's ancient grassy biomes. They drew a line that matters for every carbon project: reforestation is planting trees on deforested land; afforestation is converting land that was historically not forest into forest or plantation.
The scale of the issue is largest in Africa. Bond and colleagues (2019) note that about one million square kilometres in Africa, mostly grassy biomes, have been targeted for “restoration” by 2030 under national pledges to the AFR100 initiative, and that ancient savanna landscapes, including the Serengeti and Kruger National Park, appear on the opportunity maps as deforested and degraded.
The FAO definitions help here. Forest is land over 0.5 ha with trees over 5 m and canopy cover above 10% (or trees able to reach that). Land with 5–10% canopy, or a combined cover of shrubs, bushes and trees above 10%, is other wooded land. Much of a savanna sits below or around those thresholds by nature, not because it was degraded. See afforestation vs reforestation for what the two words mean in carbon rules.
What the carbon rules say
None of the main standards bans tree planting on open land, but each one draws the line at converting a healthy native ecosystem:
| Standard | Rule | Where |
|---|---|---|
| Verra VCS | Activities that convert an ecosystem may only be implemented in degraded ecosystems; an ARR project must show it restores a native ecosystem type represented in the same ecoregion | VCS Standard v4.7, 3.19.29 |
| Verra VM0047 | Area-based approach not applicable where clearing of pre-existing woody biomass results in degradation of native ecosystems, or on land that was managed forest in the previous 10 years | VM0047 v1.1, 4.4.1 |
| Plan Vivo | Natural ecosystems must not be converted to generate carbon benefits; the guidance names planting trees on native grasslands as conversion | PV Climate Project Requirements v5.7, 3.9.13 |
Note what VM0047 itself does not say: the word “grassland” does not appear in it. When Verra consulted on the methodology, it answered comments about planting in native ecosystems by pointing to the VCS Standard: the prohibition on clearing native ecosystems is not repeated in the methodology, but it still applies. So a project passing the methodology's 10-year forest test can still fail at validation on the standard's ecosystem rules. The forest test is covered in the 10-year rule.
What goes wrong on the ground
The rules exist because the damage is documented. Research on planting trees in grassy biomes points to three main effects:
Water
A synthesis of more than 600 observations by Jackson and colleagues (2005) found that plantations reduced streamflow by 227 mm a year on average, 52%, and that 13% of streams dried up completely for at least a year. Bond and colleagues warn that less streamflow from savanna afforestation hits dry-season water supply for local communities, and note that South Africa restricts new afforestation by law to conserve water.Fire
Africa accounts for about 70% of the world's annually burnt area. Most grassland carbon is below ground, where it survives fire. Grass fires can be managed; high-intensity plantation fires are much harder to suppress. A fire that kills the trees reverses the carbon, which is exactly the risk a buffer is there to cover.Biodiversity
Veldman and colleagues write that afforestation of grassy biomes can severely compromise hydrology and soil nutrient cycles and markedly reduce biodiversity. Savanna species depend on open, sunny ground, grazing and fire; shade removes them.
Bond and colleagues add a climate point: forests absorb more incoming sunlight than grassland, so plantations on bright, open land may warm the local climate more than their carbon cools it.
The carbon case is weaker too
Even where the rules allow it, dry open land is usually a poor carbon site. Two of our real screenings show why.
- Tropical shrubland, Africa (IPCC Tier 1)0.775 tCO2e/ha/yr
- Kenya dryland site: eligible94.7%
- Kenya dryland site: net−2,976 tCO2e
- A dryland site in the Kenyan highlands (about 1,700 ha, Acacia-Commiphora bushland): almost all of it passed the eligibility screen, with essentially no forest in 2000. But the dry climate puts it in the IPCC tropical shrubland zone, the slowest-growing zone we have screened. Over 40 years the average case came to about 69,800 tCO2e gross, and because the intake did not say how the land is used, the mapped crops and rangeland were treated as displaced: leakage of about 55,300 tCO2e. Add the 10% uncertainty deduction (about 7,000) and the 15% buffer (about 10,500), and the net screening estimate was below zero: −2,976 tCO2e. Details on the Kenya page.
- A savanna pasture in the Colombian Llanos (about 1,190 ha): 91% eligible and high rainfall, so the growth rates are high. But the whole planting area was treated as displaced cattle pasture: leakage of about 209,200 tCO2e against 372,500 gross. See the Colombia page.
Grazing is the common thread. Open land in the tropics is often someone's rangeland, and moving the animals moves the emissions. That is what leakage measures.
How to tell degraded land from native grassland
This is the hard question, and a forest history alone cannot answer it. A cleared forest that never regrew and an ancient savanna look the same in a satellite record from 2000 onwards: both show no forest. The evidence that separates them:
- The ecoregion. Is the site in a forest ecoregion, or a savanna, grassland or shrubland one? Under the VCS Standard, the project has to restore a native ecosystem type of its own ecoregion, which can be shown with remote sensing, aerial imagery, modelling or literature.
- The longer history. Older imagery, maps, records and local knowledge of when and why trees disappeared, if they ever were there.
- The vegetation today. An intact grass layer with native grasses, fire-adapted trees and grassland species points to a native system. Exotic grasses, cropland, eroded bare ground and stumps point to degradation.
- Fire and grazing. Regular fire and wild or domestic grazers are part of how a savanna works. A plan that depends on stopping both is a plan to convert it.
Where trees on open land can work
None of this means open land is off limits. Trees on open land are easiest to defend in three situations:
- Farmland and pasture that replaced forest or woodland, where the planting restores the native vegetation of the ecoregion.
- Forest-savanna mosaics, where the gaps between forest patches qualify and the patches themselves do not. Our screening in the west of DR Congo found 58% of the site eligible on that pattern (DR Congo page).
- Agroforestry and silvopasture that keep the land in use with scattered trees. VM0047's census-based approach is built for this: direct planting only, the pre-project land use maintained, at most 50 planting units per hectare. Keeping the farming also keeps leakage down. See agroforestry carbon credits.
Start by finding out what the land was. An eligibility check maps the forest history and exclusions on your boundary, so you know before design whether the rest of the evidence is worth gathering.
Frequently asked questions
Only if the land is a degraded ecosystem that the planting restores, not a healthy native grassland or savanna. Under the VCS Standard v4.7, activities that convert an ecosystem may only take place in degraded ecosystems, and an ARR project must show it restores a native ecosystem type found in the same ecoregion. Plan Vivo prohibits converting natural ecosystems and names planting trees on native grassland as an example of conversion.
Many grasslands and savannas are ancient ecosystems, not cleared forest. Researchers have shown that planting them with trees can reduce biodiversity, lower streamflow and change fire behaviour. A global synthesis of more than 600 observations found that plantations reduced streamflow by 227 mm a year on average, 52%, and 13% of streams dried completely for at least a year.
No. Having no forest in the last 10 years is necessary but not sufficient. A satellite forest history cannot tell an ancient savanna from a cleared forest that never regrew: both show no forest. The project still has to show the land is degraded and that the planting restores the native ecosystem type of its ecoregion.
On farmland and pasture that replaced forest or woodland, in forest-savanna mosaics where the planting restores former forest patches, and in agroforestry or silvopasture that keeps the land in production with scattered trees. In each case the evidence that the site was degraded, and what its native vegetation is, decides it.
Sources
- Bond, Stevens, Midgley and Lehmann (2019), The trouble with trees: afforestation plans for Africa, Trends in Ecology & Evolution 34(11), doi:10.1016/j.tree.2019.08.003 (accepted manuscript)
- Veldman et al. (2015), Tyranny of trees in grassy biomes, Science 347(6221):484, doi:10.1126/science.347.6221.484-c
- Jackson et al. (2005), Trading water for carbon with biological carbon sequestration, Science 310:1944–1947 (abstract)
- Verra: VCS Standard v4.7, Section 3.19 (biodiversity and ecosystems)
- Verra: VM0047 Afforestation, Reforestation and Revegetation, v1.1 (14 May 2025), Sections 4.3 and 4.4
- Plan Vivo: PV Climate Project Requirements, v5.7 (30 April 2026), Section 3.9