Report — Matters Arising
The Saw-Tooth Forest
Amazon trees have been getting bigger. Two of the field's senior researchers argue that the plots used to measure them are too small and watched too briefly to see what usually comes next. This week the original authors answered them.
Across the Amazon, field crews return to marked patches of forest every few years and measure every tree above a set size. In 2025, a team led by Adriane Esquivel-Muelbert analyzed 188 of these plots. They averaged 1.2 hectares, a little under two soccer fields, and each was watched for thirteen years on average, some for as few as two and some for thirty.
Sourced · Esquivel-Muelbert et al., Nature Plants (2025) · plot positions drawn schematicallyOver that time the average trunk, measured by its cross-section at chest height, grew 3.3 percent per decade. Trees wider than forty centimeters became more common, up 6.6 percent per decade, and the smallest, under twenty, became rarer, down 1.2 percent. The authors named rising carbon dioxide as the most likely driver, though not necessarily the only one.
Sourced · Esquivel-Muelbert et al., Nature Plants (2025), abstract and Table 1Sebastian Leuzinger and Christian Körner replied this month in the same journal. Their first point concerns a single tree. A tree adds trunk and height slowly, all its life, until it falls in a day. Draw the carbon in one patch of forest over centuries and you get a saw-tooth: long climbs, sheer drops. They argue the plots show what any crowded mature forest does as fewer, larger survivors take up the space, a process called self-thinning, with no change in the air required.
Sourced · Leuzinger & Körner, Nature Plants (5 Nov 2026) · line is schematicTheir second point concerns the drops. The forest's giants, the trees that break through the top of the canopy, live about 150 to 300 years and stand one or two to a hectare. By their estimate, a one-hectare plot loses one about once in two hundred years. At that rate, 188 plots watched for thirteen years would see roughly a dozen giants fall. The other plots record only the climb.
Sourced · Leuzinger & Körner, Nature Plants (5 Nov 2026) (lifespan, density, 1 in 200) · Editor's arithmetic: 1 − (199/200)¹³ ≈ 6% of 188 ≈ 12How fast a forest grows does not settle how much it holds. In a 2017 essay, Körner put it as a recipe: take the carbon that comes in, subtract what goes out, and multiply what is left by how long it stays. It is the same logic as the diner earlier in this issue. The challengers point out that the plot study gave no direct count of deaths and no estimate of how long carbon stays, so the growth it measured cannot yet be turned into a lasting store.
Sourced · Körner, Science 355:130 (2017); Leuzinger & Körner 2026 · tank is schematicTrees that grow faster tend to die younger. Roel Brienen and colleagues found this in tree rings across nearly all the species and climates they studied, and argued that growth sped up by the environment brings a delayed rise in deaths that eventually cancels the gain. Körner points to plantations: wood grown fast is carbon held briefly. Nothing in a tree banks carbon for later. The carbon stays as long as the tree stands.
Sourced · Brienen et al., Nature Communications 11:4241 (2020); Körner 2017 · lines are schematicThe original authors answered in the same issue. The small-plot worry is old, they write, and was tested with simulations in 2009. Most death in the Amazon is small: nearly 99 percent of lost biomass comes from gaps smaller than a tenth of a hectare. In those simulations, short watches gave lopsided answers, and the lopsidedness went away with more watching. A hectare watched for a year is a hectare-year. The simulations found 2,260 of them enough to detect a real trend. This study has 3,532.
Sourced · Esquivel-Muelbert, Banbury Morgan … Phillips, reply, Nature Plants (5 Nov 2026); Gloor et al., Global Change Biology (2009) · blocks are schematicThey add that the plots do record big trees dying: the number of trees wider than forty centimeters fell in 28 percent of the intervals between censuses. A forest still recovering from old damage should gain fastest where its trees started smallest. These plots showed no such pattern; the relationship was flat.
Sourced · Esquivel-Muelbert, Banbury Morgan … Phillips, reply, Nature Plants (5 Nov 2026) · points are schematicHere the two sides talk past each other. The challengers' rare deaths are the giants, one or two to a hectare. The reply's big trees are anything wider than forty centimeters, about 42 to a hectare in these plots. A plot can see many of those die and never see a giant fall. Neither piece reconciles the two.
Editor's comparison · Leuzinger & Körner, Nature Plants (5 Nov 2026); Esquivel-Muelbert et al., Nature Plants (2025), Table 1 (42.5 stems ≥40 cm per hectare); Esquivel-Muelbert, Banbury Morgan … Phillips, reply, Nature Plants (5 Nov 2026)Neither side claims the growth can go on forever. The reply calls indefinite increases in tree size impossible and allows that the trade-off between growth and lifespan, along with fragmentation and climate change, may limit how long the sink lasts. The challengers accept that the trees in the plots grew. The argument that remains has two parts. One is cause: carbon dioxide, or the ordinary thinning of a mature forest. The other is measurement, and it turns on units. The challengers want single plots of fifty hectares or more, or records a century long. The reply counts hectare-years summed across a whole network. Whether many small plots watched for years can stand in for one very large plot watched for a lifetime is the open question.
The argument is about twenty years old. Simulations and rebuttals in 2008 and 2009 covered much of the same ground, and this exchange does not end it. What the two sides share is the shape of the problem. Growth arrives a little at a time and is recorded at every visit. Loss arrives all at once and is easy to miss. The question left open is not whether the trees grew, but whether anyone has watched long enough to see what follows. The phrase to search is Körner's: slow in, rapid out.
Sourced — Leuzinger & Körner, "Trees grow until they fall: why small permanent plots cannot identify an Amazon carbon sink", Nature Plants (5 Nov 2026) · Esquivel-Muelbert, Banbury Morgan … Phillips, "Reply to Leuzinger and Körner: Increasing tree size across Amazonia detected by plot-networks", Nature Plants (5 Nov 2026) · Esquivel-Muelbert et al., "Increasing tree size across Amazonia", Nature Plants 11, 2016–2025 (2025) · Körner, "A matter of tree longevity", Science 355, 130–131 (2017) · Brienen et al., Nature Communications 11, 4241 (2020) · Gloor et al., Global Change Biology (2009). Drawings are schematic; counts and rates are the papers'.