Agrostology and the trillion-tree fight

Agrostology, the branch of botany devoted to grasses, sounds like the sleepiest corner of systematic botany. It is not. Its subject is the family Poaceae, the plants under your feet on every continent, from the tropics to the fringe of Antarctica.[9] Over the last decade the science of grasses widened into the science of grassy biomes, and there it collided head-on with the most fashionable idea in climate policy: plant a trillion trees. Grassland scientists say much of the land being targeted for tree-planting is not degraded forest at all but ancient grassland and savanna — ecosystems millions of years old that tree-planting would destroy. The fight has run through Science and Nature since 2019, pits grass ecologists against a celebrated tree-restoration lab, and turns on a deceptively simple question: how do you tell an old grassland from a bald patch of former forest?

01The field in brief

Agrostology proper is the taxonomy, morphology, and physiology of grasses. Poaceae is one of the largest plant families (~11,000 species) and, by land area and calories, the most consequential — wheat, rice, maize, sugarcane, bamboo, and the turf of every pasture are grasses. The field dates itself to a 1708 Swiss taxonomic treatise and spent two centuries sorting grasses by the architecture of the flowering spikelet.[9] Molecular phylogenetics has since rebuilt the family tree, but a few pieces of grass biology matter for the fight, so define them first.

The first is C4 photosynthesis. Many grasses of warm, open, seasonally dry regions run a carbon-concentrating pathway, called C4, that outperforms the ordinary C3 pathway in heat and bright light. C4 grasses are what make tropical savannas productive, and highly flammable when they dry out.

The second is fire, which in grassy biomes is a system input rather than an accident. Fire is not damage here; it is the process that keeps the system open. Africa's C4 grasslands and savannas account for the majority of the planet's annually burned area.[7] Grasses recover within weeks because most of their biomass, and most of their carbon, sits belowground, in roots and storage organs a fire cannot reach.

The third concept is the load-bearing one: old-growth grassland. An ancient, undisturbed grassland is rich in forbs (non-grass herbaceous plants) with deep, slow-growing underground storage organs, the bulbs, tubers, and woody rootstocks that take decades to centuries to build and cannot regrow once plowed or shaded out.[7] A secondary grassland, an abandoned field or a cleared forest, looks superficially similar from a satellite but lacks that hidden architecture. Telling the two apart from space is nearly impossible, and that difficulty is the crux of the whole dispute.

02The fight

It broke open on 5 July 2019. A team led by Jean-François Bastin at Thomas Crowther's lab at ETH Zürich published a Science paper, "The global tree restoration potential," estimating that Earth had room for roughly 0.9 billion hectares of additional tree canopy, which could store on the order of 205 gigatonnes of carbon.[1] The paper's own framing called tree restoration the most effective climate solution then available. It became one of the most-covered ecology papers in memory and fed directly into trillion-tree pledges, the Bonn Challenge, and a wave of corporate offset planting.

The grassland ecologists read the same map and saw a catastrophe in the making. Their objection was that much of the "restorable" land Bastin's model identified was not deforested land at all. It was savanna and grassland, open by nature and by fire, that the model had misread as tree-shaped holes waiting to be filled. In October 2019 Science ran a barrage of formal Technical Comments. One, led by Joseph Veldman with more than forty co-signatories (a roll call of the field including William Bond, Catherine Parr, Caroline Lehmann, Nicola Stevens, and Giselda Durigan), argued the carbon figure was roughly five times too large.[2] It faulted the model on three grounds: it inflated soil-carbon gains, it ignored that new trees at high latitudes darken the surface and can warm the climate by lowering albedo, and, the core charge, it counted afforestation of grasslands and savannas as "restoration."

A second comment, from carbon-cycle physicists led by Pierre Friedlingstein at Exeter, attacked the number on pure carbon-accounting grounds: it was inconsistent with how the global carbon cycle actually responds to emissions.[3] A third, from forest scientists including Simon Lewis, piled on. Bastin's group replied, conceding the albedo and evapotranspiration points as real but beyond their study's scope, and standing by the carbon estimate.[4] Veldman's re-analysis, using literature values, put the plausible figure near 107 GtC; a follow-up in Annals of Forest Science using biome-specific data landed between about 72 and 76 GtC — under 40% of the original.[2]

What makes this a live fight and not a settled one is that it kept escalating and then, unexpectedly, partly converged. The grassland camp coined "biome awareness disparity" for the reflex that treats trees as good and open land as empty, and pressed the point institutionally. In 2024, Parr, te Beest, and Stevens argued in Science that conflating reforestation with restoration is widespread and doing real ecological harm;[5] the same year, a Frontiers paper flatly titled "Rangeland afforestation is not a natural climate solution" extended the critique to the world's rangelands. Then in 2025 came the surprise: a Nature Communications paper, "Addressing critiques refines global estimates of reforestation potential," co-authored by Veldman and by tree-restoration scientists including The Nature Conservancy's Susan Cook-Patton, adopted the grassland exclusions and shrank the global reforestation estimate to roughly 195 million hectares — an area about the size of Mexico, and a fraction of the 2019 vision.[6] A 2026 Nature Ecology & Evolution paper then re-mapped the world's grasslands themselves, finding they had been chronically misclassified.[8] Bastin, for his part, has continued to defend wide-net mapping, arguing that in some shifting-climate regions afforestation is a defensible practical option and that restrictive forest definitions wrongly exclude drylands.[10]

03What the fight reveals

This is not a both-sides dispute on the factual core, and it would be dishonest to frame it as one. On the specific claims the grassland ecologists raised, that the 2019 number was far too high, that albedo matters, and that planting trees on ancient grassland destroys irreplaceable biodiversity and can release more carbon than it stores, the evidence has moved decisively their way, to the point that the tree-mapping community has published revised, much smaller estimates alongside them.[6] That convergence is the strongest evidence of who was right.

What the dispute exposes about the field is subtler. Agrostology's central problem is that its most valuable object, the old-growth grassland, is defined by what you can't see from a satellite: the belowground forb architecture that takes centuries to assemble. A discipline whose ground truth is literally underground is structurally vulnerable to being overruled by whoever controls the remote-sensing map. The 2019 paper won the news cycle precisely because a global map is legible and a soil core is not. The grassland scientists' real achievement was forcing a slower, field-based kind of evidence back into a debate that satellites had seemed to settle.

One thing remains genuinely unresolved: there is still no cheap, scalable way to distinguish an ancient grassland from a secondary one at global scale. Until there is, every trillion-tree map will keep making the same error, and the fight will keep having to be re-won region by region.