The Forest Was a City

For most of the twentieth century, the consensus about tropical rainforest was straightforward and, it seemed, well founded.

The soils are poor. Nutrients in such systems are held almost entirely in the living biomass rather than in the ground; clear the canopy and the exposed earth is exhausted within a few seasons of cultivation. Disease pressure is high, stone for building is often absent, and the vegetation reclaims any clearing with remarkable speed. Rainforest, the textbooks explained, is a counterfeit paradise — overwhelmingly productive and almost useless to agriculture. It could support mobile groups at low density, and monumental centres at a few favoured locations, and nothing between the two.

The evidence appeared to agree. Where archaeologists surveyed, they found ceremonial architecture surrounded by forest. The temples were extraordinary and the countryside was empty.

There was an obvious problem with that evidence, and it was recognised for decades without being solvable. Surveying tropical forest on foot is brutally slow. A team may clear and map a few square kilometres in a season, at considerable cost, and any earthwork more than a metre or two high is invisible from ten metres away. Investigators went where they already knew something existed, because walking transects at random through closed-canopy forest is an almost perfect way to waste a career. The resulting map showed the distribution of archaeologists rather than the distribution of settlement, and everybody involved knew it.

---

The instrument that changed this was not developed for archaeology.

Airborne laser scanning works by emitting hundreds of thousands of pulses per second towards the ground and timing their return. Over forest, most pulses strike leaves and branches at various heights. A small proportion — sometimes only a few per cent — find a gap in the canopy, reach the actual ground, and come back.

The essential move is to discard almost everything. Software separates the returns and keeps only the lowest in each small area, from which it builds a model of the bare earth. The forest is not seen through; it is filtered out, statistically, and the surface it has concealed for a thousand years appears.

What emerges, rendered with artificial low-angle light to throw relief into shadow, is not a photograph. It is a map of ground shape at a resolution of tens of centimetres, and on that map the difference between a natural slope and a collapsed platform is often perfectly obvious to a trained eye and entirely invisible to a person standing on it.

---

The first large survey of this kind over the central lowlands of the Maya region covered some two thousand square kilometres in a matter of weeks.

The results were not an improvement on the existing picture. They were incompatible with it.

Over sixty thousand previously unrecorded structures were identified in that area alone: house platforms, terraces, reservoirs, quarries. Raised causeways ran for tens of kilometres between centres, several wide enough for processional use and elevated above seasonal flooding. There were extensive defensive works — ditches, ramparts, fortified hilltops — in a region long characterised as having engaged in limited, largely ceremonial conflict. And most consequentially, the agricultural infrastructure was everywhere: terraced hillsides, drained wetland fields in gridded patterns, water management at a scale that implied centuries of maintained investment.

Population estimates for the lowlands had run, on the older evidence, to perhaps one or two million. The revised figures based on structure counts run to seven million and above, in a landscape that had been described, in textbooks still in print, as sparsely inhabited forest.

Similar surveys elsewhere produced the same shape of result. Around a celebrated temple complex in South-East Asia, previously understood as a ceremonial centre in a rural setting, scanning revealed a vast low-density urban landscape extending far beyond the monuments — a grid of occupation mounds, roads, canals and embankments covering hundreds of square kilometres, one of the largest pre-industrial settlements known. In the southern Amazon, in a region of seasonally flooded savannah, surveys identified monumental earthen platforms up to twenty metres high, surrounded by concentric ditches and linked by straight raised causeways running for kilometres between sites.

---

The reframing this forced is larger than any individual site, and it has two parts.

The first is that low population density is not the same as low complexity. These were cities, but they were not cities of the compact, walled, densely packed kind that the European and Near Eastern record made the default. They were dispersed: household compounds with their own gardens and orchards, spread across a managed landscape, integrated by roads and water infrastructure and centred on monumental cores. Such settlements can support very large populations across very large areas, and — critically — they leave almost no trace that conventional survey detects, because their fabric is earth, wood and thatch distributed thinly rather than stone concentrated in one place.

The second is about the forest itself. If the terraces, reservoirs and field systems are real, then the vegetation standing on them is not primeval. It is regrowth over an abandoned agricultural landscape — and the composition of that regrowth carries the evidence: concentrations of useful species far above what unmanaged distribution would produce, and patches of unusually dark, fertile, charcoal-rich soil that are demonstrably anthropogenic and remain more productive than the surrounding ground centuries after the people who made them departed.

The forest was not an obstacle that a few societies overcame. In substantial part, the forest is what those societies left behind.

---

Two cautions belong here, and practitioners insist on both.

Scanning detects shape, not date, function or occupancy. A platform on a map is a hypothesis. Whether it held a dwelling, whether contemporaneous structures were occupied simultaneously, and what any of it was for are questions that only excavation answers, and excavation remains as slow as it ever was. The population figures depend on conversion assumptions that are openly contested within the field.

And the word discovery requires care. Many of these landscapes were never lost to the people living in and around them, who could frequently identify mounds and causeways and had incorporated them into local knowledge long before any aircraft flew a grid overhead. What was discovered was not the sites. It was that the discipline had been mapping its own reach and calling the result a settlement pattern.

Which is, in the end, the more useful lesson. The forest had not hidden anything. We had simply been looking with an instrument that could not see, and had mistaken the silence for an answer.