Article — "Lidar reveals a million-strong Amazonian civilization that collapsed around 850 CE": Article — "Lidar reveals a million-strong Amazonian civilization that collapsed around 850 CE": Four hundred thirty-two earthworks over 4,497 square kilometers of flight lines, only 36 of which anyone knew about beforehand.
That is the raw catch from a 2024 lidar survey over the Brazilian states of Acre and Amazonas. Nearly 400 monumental structures — geometric ditched enclosures called geoglyphs — appearing where science had recorded almost nothing.
The patch of southwestern Amazonia that produced those numbers was, until recently, the textbook example of what Martti Pärssinen of the University of Helsinki calls "an almost uninhabited pristine rainforest."
Now the same cultural area alone is estimated to hold as many earthworks as researchers previously attributed to the whole of Amazonia. The map didn't get an addendum. It got inverted.
Start with the physics, because the physics is the whole story.
Lidar works by firing laser pulses at the ground from an aircraft — here, about 30 per square meter. A dense forest canopy blocks most of them, but small gaps let some slip through, hit the terrain, and bounce back.
Enough of those returns, processed together, reconstruct a bare-earth model of the surface beneath the trees. That is how you find a ditch 9 to 13 meters wide and up to 5 meters deep under dense vegetation.
For a century, earthwork detection in this region depended on deforestation — you could only see the structures where the trees had already been cut. Lidar broke that dependency. In this survey it reportedly found roughly five times more earthworks than satellite imagery could, and about ten times more geoglyphs than the team expected.
So the technique is extraordinary at what it does. But notice precisely what it does: it detects large, geometrically regular scars carved deep into the earth. Ceremonial architecture. Monuments.
The geoglyphs, on the excavation evidence, were not where people lived. Pärssinen describes them as venues for gatherings, political decision-making, ritual dancing, feasting, ball games — the civic and sacred cathedrals of the Aquiry.
The homes were something else entirely. Long dwellings housing several families, reportedly situated in a dispersed pattern around the earthworks, per New Scientist's account of Pärssinen's work.
Dispersed wooden longhouses leave shallow, irregular traces — nothing like a 5-meter ditch. They are exactly the kind of thing that lidar, tuned to find monumental scaffolding under a canopy, does not reliably see.
This is the epistemic hinge. We are mapping a civilization by its cathedrals because its houses are, for this instrument, close to invisible.
We are mapping a civilization by its cathedrals because its houses are, for this instrument, close to invisible.
Which is why the headline number deserves a hard look. The 1.25-to-3-million-people estimate for the population peak around AD 100–300 is not a count. It is a chain of extrapolations.
Link one: the 432 earthworks found over 4,497 square kilometers are scaled up across the 183,000 square kilometers the team now assigns to the Aquiry, yielding an estimated 24,000 to 30,000 earthworks total.
Link two: the population figure rests on a labor assumption — that an average geoglyph, around 3 hectares, would have required roughly 300 people to build and maintain. Multiply structures by workers, adjust, and you arrive at millions.
Every link is arguably defensible. None is measured. And the whole edifice sits on a sample Pärssinen himself flags as "less than 3 percent of the whole Amazonia."
The scientists are unusually candid about this. Michael Heckenberger of the University of Florida calls the population claims "plausible and extremely important, even mind-blowing" — then names the missing piece precisely: "What we now need to know is what is going on around the detectable earthworks, such as occupational sites to support the high estimated population."
That is the gap in one sentence. We can see the monuments. We have not yet seen the settlements the monuments are supposed to imply.
Then there is the sharper objection. Heiko Prümers of the German Archaeological Institute, who studies ancient Amazonian settlement and was not part of the study, points out that it isn't even clear whether nearby structures were counted as separate sites or lumped together — which would move the totals around.
More fundamentally, he offers an alternative reading of the same data: the high site count "could be the result of frequent relocations by the Aquiry rather than a high population density."
Sit with that, because it dissolves the arithmetic. If a community built a geoglyph, used it, moved on, and built another, then 30,000 earthworks accumulated over more than a millennium do not mean 30,000 simultaneously occupied places. Many structures could be the sequential footprints of far fewer people.
The peak-population figure assumes the monuments are roughly contemporaneous and densely inhabited. Relocation assumes they are a time-lapse. The same lidar map is compatible with both, and lidar alone cannot tell them apart — because, again, it sees the ditches, not the households that would date and populate them.
Here is the argument I want to make: the uncertainty is not a footnote to the discovery. It is the discovery.
Because whether the true figure is 1.25 million or 3 million — or, on Prümers's reading, something more mobile and dispersed — one thing does not survive contact with the data. The "pristine wilderness" framing.
You do not carve 24,000 to 30,000 monumental earthworks, some running to 50 hectares, connected by roads, into an untouched forest. Even Prümers, the skeptic on the numbers, credits the work with correcting the perception "that the Amazon region is a region of 'untouched nature.'"
The Aquiry burned bamboo forest for maize, squash, and manioc. They cleared land for enclosures and the extensive roads leading to them. They cultivated Brazil nut and peach palm — species still economically important in the modern rainforest.
That last detail reframes everything downstream. The biodiversity we treat as baseline wilderness may partly be the residue of ancient agroforestry. The forest is, in meaningful part, an artifact.
And then, around 850 CE, after almost 1,500 years, the whole thing ended. What caused the collapse is not known. The people who engineered this landscape vanished from the record, and the trees closed over their cathedrals until the lasers arrived.
If Pärssinen is right that similar densities lie elsewhere — he reportedly suspects a precolonial Amazon of possibly tens of millions, per Scientific American — then estimates of the region's population, its soil, its forest structure, and its role in some global models would all need revisiting. That is a large "if," resting on the same 3 percent.
But the conservation implication doesn't wait for the census. If the Amazon is a landscape people built and then abandoned, protecting it stops being a matter of guarding pristine nature and becomes something harder: stewardship of an engineered inheritance.
Pärssinen puts the lesson plainly. Two thousand years ago, people shaped this land at monumental scale and apparently lived in balance with it. "We could learn something about how they managed the forest in the past."
The honest state of the science, then, is a map dense with monuments, a population figure floating on assumptions, and a live debate about whether we are looking at a crowded civilization or a mobile one. What is settled is smaller and more radical than the headline.
We have been counting a civilization by its cathedrals, because the instrument that revealed it cannot see where anyone actually lived.