The Amazon's 'Lost Cities' Were Never Lost. People Were Living On Them.

On 11 January 2024, the journal Science published Two thousand years of garden urbanism in the Upper Amazon — a lidar survey of Ecuador's Upano Valley led by Stéphen Rostain of the CNRS.

What the survey found is remarkable and not in dispute:

• More than 6,000 rectangular earthen platforms — houses and communal buildings — laid out in a geometric pattern
• 15 urban centres, linked by a sophisticated road network
• Terraced agricultural fields, interwoven with river drainages
• Roughly 600 km² mapped, from a lidar unit mounted on an aircraft's belly
• Built around 500 BCE, occupied for centuries, with the population moving away between 300 and 600 CE

This is urbanism, in the Amazon, contemporaneous with the Roman Republic, at a scale comparable to Maya systems recently mapped in Mexico and Guatemala. It demolishes a long-standing assumption that the Amazon basin was too poor in soils to support dense, organised populations.

The science is excellent. Now the other part.

What the paper said, and what the coverage said

The Science paper is measured. It describes "garden urbanism" — a specific, technical claim about low-density settlement interwoven with agriculture.

More than 19 international outlets, including CNN, the BBC and the Guardian, reported that archaeologists had found an "ancient Rome of the Americas."

The same pattern had already played out with lidar work in the Bolivian Amazon. The scientific paper described "low-density urbanism." Coverage from Smithsonian Magazine to Axios announced that lost cities had been found in the jungle.

Two problems with that.

First: archaeologists had documented hundreds of those sites years earlier. The lidar refined and extended existing knowledge. It did not stumble on something nobody knew about.

Second — and this is the one that matters — local communities had long been aware of the earth mounds.

They were not lost. People knew where they were. Some people were living on them.

The word "discovery" is not neutral

A peer-reviewed analysis in Advances in Archaeological Practice — Lidar and Lost Cities — examined how lidar findings are presented in news media. Its finding is blunt: changes in tone, emphasis and word choice skew archaeological results to the point that the coverage becomes difficult to distinguish from pseudoscientific posts.

That is an unusually sharp thing for a methods journal to say about the BBC.

And the critique goes past accuracy into consequence. Writing in The Conversation, researchers make the point directly about La Mosquitia in Honduras — the region behind a decade of "lost city of the monkey god" coverage. The Miskitu people have long lived there. They have always known about the archaeological sites in their ancestral homelands.

The argument is that narratives of discovery erase Indigenous presence, knowledge and agency — while enabling dispossession.

Follow the logic, because it is not merely a complaint about tone. Lost implies nobody's. Discovered implies discovered-by. A place framed as empty until outsiders found it is a place with a weaker claim attached to the people already there. That framing has a long and specific history in the Americas, and it has never been only rhetorical.

As we reported in Indigenous lands and Amazon deforestation, recognised Indigenous territories are the best-preserved land in the Amazon — 1.2% of native vegetation lost over 35 years. Recognition is the mechanism. A narrative that quietly unrecognises who was already there is not a neutral storytelling choice.

What lidar actually is, and what it isn't

Lidar fires laser pulses from an aircraft and times their return. Enough pulses find gaps in the canopy that software can strip the vegetation and model the bare ground. Structures invisible from the air — and nearly invisible on foot, in dense forest — appear as clean geometry.

It is genuinely transformative. It is also frequently oversold.

An anthropologist quoted on the Maya lidar coverage made the essential distinction: lidar data have tremendous potential to revise population and farming estimates, but those are potential results, not actual findings. Understanding waits on the hard work — the ground survey, the excavation, the dating.

Lidar tells you something is there. It does not tell you what it was, who built it, when, or why. Those answers come from years of unglamorous fieldwork that no outlet will headline.

A map is a hypothesis, not a conclusion. The same trap as GBIF occurrence records, which we have written about in our own data notes: a dataset that shows you where people looked is not the same as a dataset that shows you what is there.

What the honest version sounds like

Nothing here diminishes the Upano work. The correct framing is arguably more impressive than the hype:

Ancient Amazonians built a networked landscape of 6,000 platforms and 15 centres, farmed it for roughly a thousand years, and left. Their descendants know the ground. Lidar let outsiders finally see the shape of what they built.

That is a better story than "lost city." It has real people in it, at both ends.

Why this is on AEDO1

We are a documentary platform. Documentary storytelling runs on discovery — the reveal, the thing nobody knew, the camera pushing through the leaves.

This is the article where we say plainly that the frame is often a lie, and that we intend to be held to it. If we ever tell you we found something in a place where people live, ask us who showed us.