When Brazil established the Xingu Indigenous Park in 1961, the reserve was far from modern civilization, nestled deep in the southern reaches of the vast Amazon forest. When I first went to live with the Kuikuro, one of the reserve’s principal indigenous groups, in 1992, the park’s boundaries were still largely hidden in thick forest, little more than lines on a map. Today the park is surrounded by a patchwork of farmland, its borders often marked by a wall of trees. For many outsiders, this towering green threshold is a portal, like the massive gates of Jurassic Park, between the present—the dynamic modern world of soy fields, irrigation systems and 18-wheelers—and the past, a timeless world of primordial nature and society.

Long before taking center stage in the world’s environmental crisis as the giant green jewel of global ecology, the Amazon held a special place in the Western imagination. Mere mention of its name conjures images of dripping, vegetation-choked jungles; cryptic, colorful and often dangerous wildlife; endlessly convoluted river networks; and Stone Age tribes. To Westerners, Amazonian peoples are quintessential simple societies, small groups that merely make do with what nature provides. They have complex knowledge about the natural world but lack the hallmarks of civilization: centralized government, urban settlements and economic production beyond subsistence. In 1690 John Locke famously proclaimed, “In the beginning all the World was America.” More than three centuries later the Amazon still grips the popular imagination as nature at its purest, home to native peoples who, in the words of Rolling Stone editor Sean Woods in October 2007, preserve “a way of life unchanged since the dawn of time.”

Looks can be deceiving. Hidden under the forest canopy are the remnants of a complex pre-Columbian society. Working with the Kuikuro, I have excavated a network of ancestral towns, villages and roads that once supported a population perhaps 20 times its present size. Huge swaths of forest have grown over the ancient settlements, gardens, fields and orchards, which fell into disuse when epidemics brought by European explorers and colonists decimated the native peoples. The region’s rich biodiversity reflects past human intervention. By developing a mix of land uses, soil-enrichment techniques and long crop rotation cycles, the ancestors of the Kuikuro thrived in the Amazon despite its infertile natural soils. Their accomplishments could inform efforts to reconcile the environmental and development goals of this region and other parts of the Amazon.

“Nature Folk”
The most famous person to go looking for lost civilizations in the southern Amazon was Percy Harrison Fawcett. The British adventurer scoured what he called the “uncharted jungles” for an ancient city, Atlantis in the Amazon, replete with stone pyramids, cobbled streets and alphabetic writing. His tales inspired Conan Doyle’s The Lost World and perhaps the Indiana Jones movies. David Grann’s gripping recent book, The Lost City of Z, retraced Fawcett’s path before his disappearance in the Xingu in 1925.

Actually, five German expeditions had already visited the Xinguano people and lands. In 1894 Karl von den Steinen’s book Unter den Naturvölkern Zentral Brasiliens, which described the earliest expeditions, became an instant classic in the fledgling discipline of anthropology. The book set the tone for 20th-century studies of Amazonian peoples as small, isolated groups living in a delicate balance with the tropical forest: “nature folk.” Later anthropologists often viewed the forest environment as uniformly inimical to agriculture; the soil’s poor fertility seemed to preclude large settlements or dense regional populations. By this reasoning, the Amazon of the past must have looked much like the Amazon in recent times.

But this view began to erode in the 1970s as scholars revisited early European accounts of the region, which talked not of small tribes but of dense populations. As Charles Mann’s best-selling book 1491 has eloquently described, the Americas were heavily populated on the eve of the European landings, and the Amazon was no exception. Gaspar de Carvajal, the missionary who chronicled the first Spanish expedition down the river, noted fortified towns, broad, well-kept roads and large numbers of people. Carvajal wrote on June 25, 1542:

We went among some islands which we thought uninhabited, but after we got to be in among them, so numerous were the settlements which came into sight … that we grieved ... and, when they saw us, there came out to meet us on the river over two hundred pirogues [canoes], that each one carries twenty or thirty Indians and some forty ... they were colorfully decorated with various emblems, and they had with them many trumpets and drums ... and on land a marvelous thing to see were the squadron formations that were in the villages, all playing instruments and dancing about, manifesting great joy upon seeing that we were passing beyond their villages.

Archaeological research in several areas along the Amazon River, such as Marajó Island at the mouth of the river and sites near the modern cities of Santarém and Manaus, has confirmed these accounts. These societies interacted in far-flung systems of trade. Less is known about the southern peripheries of the Amazon, but recent work in Llanos de Mojos in lowland Bolivia and in the Brazilian state of Acre suggests that they, too, supported complex societies. In 1720 Brazilian frontiersman António Pires de Campos described a densely settled landscape in the headwaters of the Tapajós River, just west of the Xingu:

These people exist in such vast quantity, that it is not possible to count their settlements or villages, [and] many times in one day’s march one passes ten or twelve villages, and in each one of them there are ten to thirty houses, and in these houses there are some that are thirty to forty paces across ... even their roads they make very straight and wide, and they keep them so clean that one finds not even a fallen leaf....

An Ancient Walled Town
When I ventured to Brazil in the early 1990s to study the deep history of the Xingu, lost cities were the furthest things from my mind. I had read Steinen but had barely heard of Fawcett. Although much of the vast Amazon basin was archaeological terra incognito, it was unlikely that ethnographers, much less local Xinguanos, had missed a large monolithic center towering over the tropical forests.

Nevertheless, signs of something more elaborate than present-day settlements were all around. Robert Carneiro of the American Museum of Natural History in New York City, who lived with the Kuikuro in the 1950s, had suggested that their settled way of life and productive agricultural and fishing economy could support communities 1,000 to 2,000 strong—several times the contemporary population of a few hundred. He also cited evidence that indeed it once had: a prehistoric site (designated “X11” in our archaeological survey) that was surrounded by extensive ditches. The Villas Boas brothers—Brazilian indigenistas who were nominated for a Nobel Peace Prize for their part in creating the Xingu park—had reported such earthworks near many villages.

In January 1993, soon after I arrived in the Kuikuro village, the principal hereditary chief, Afukaka, took me to one of the ditches at a site (X6) they call Nokugu, named for the jaguar spirit being thought to live there. We passed local men who were raising a huge fish weir across the Angahuku River, which was already swelling from the seasonal rains. The ditch, which runs over two kilometers, was two to three meters deep and more than 10 meters wide. Even though I had expected to find an archaeological landscape different from today’s, the scale of these ancient communities and their constructions surprised me. Kuikuro research assistants and I spent the following months mapping it and other earthworks at the 45-hectare site.

Since that time, our team has studied numerous other sites in the area, hacking more than 30 kilometers of line-of-sight transects through the forest to map, examine and excavate the sites. Many Kuikuro helped in one way or another, and some became well versed in archaeology.

At the end of 1993, Afukaka and I went back to Nokugu so I could tell him what I had learned. We followed the contour of the site’s outer ditch and stopped at an earthen bridge, where a major road we had uncovered passed over it. I pointed down the arrow-straight ancient dirt road, which was 10 to 20 meters wide and led to another ancient site, Heulugihïtï (X13), about five kilometers away. We crossed the bridge and entered Nokugu.

The road, defined by low earthen curbs, widened to 40 meters—the size of a modern four-lane highway. After a couple of hundred meters, we passed over the inner ditch and stopped to look at our recently finished excavation trench, where we had found a funnel-shaped footing for a tree-trunk palisade. Afukaka told me a story of palisaded villages and raids in his people’s distant past.

As we moved farther into the ancient town, we passed through patches of forest, scrub and open areas that now cover the site—the footprints of diverse past activities. We emerged into a grassy glade of towering palms marking the former plaza. I slowly spun and pointed along the perfectly circular edge of the plaza, marked by a meter-high mound. The tall palms, I told him, had colonized the plaza centuries ago from compost gardens in domestic areas.

Leaving the plaza to explore the surrounding neighborhoods, we came across large refuse middens that closely resembled the one behind Afukaka’s own house. They were filled with broken pots that he noted were exactly like those his wives used to process and cook manioc, down to minute details. On a later visit, when we were excavating a pre-Columbian house, the chief bent down in the central kitchen area, popped out a big hunk of pottery, and corroborated my sense that the daily life of the ancient society was much like today’s. “You’re right!” Afukaka exclaimed. “Look here, a pot sup­port”—an undagi, as the Kuikuro call it, used to cook manioc.

These connections are what make the Xinguano sites so fascinating. They are among the few pre-Columbian settlements in the Amazon where archaeological evidence can be linked directly to present-day customs. Elsewhere, the indigenous culture was completely wiped out, or the archaeological record is spotty. The ancient walled town I showed Afukaka was much like his current village, with its central plaza and radial roads, only it was 10 times larger.

From House to Polity
“Palatial” is not the word that usually comes to mind to describe a pole-and-thatch house. Most Westerners think “hut.” But the house that the Kuikuro were building for the chief when I arrived in 1993 was massive: well over 1,000 square meters. It is hard to imagine that a house built like a giant, overturned basket without stone, mortar or nails could get any bigger. Even the average Xinguano house, at 250 square meters, is as big as the average American home.

What makes the chief’s house stand out is not just size but also its position, located on the southern point of the central circular plaza. As one enters the village along the formal entry road, high-ranking families live to the right (south) and left (north). The arrangement reproduces, on a larger scale, the layout of an individual house, whose highest-ranking occupant hangs his hammock to the right, along the long axis of the house. The entry road runs approximately east-west; in the chief’s house, his hammock is oriented in the same direction. When a chief dies, he is also laid to rest in a hammock with his head to the west.

This basic corporeal calculus applies on all scales, from houses to the entire Upper Xingu basin. Ancient towns are distributed across the region and interconnected by a lattice of precisely aligned roads. When I first arrived in the area, it took weeks to map the ditches, plazas and roads using standard archaeological techniques. Beginning in 2002, we began using precise GPS, enabling us to map major earthworks in a matter of days. What we have found is an impressive degree of regional integration. The landscape planning seems almost overdetermined, with a specific place for everything. Yet it was based on the same basic principles of the current village. Main roads run east-west, secondary roads radiate out to the north and south, and smaller roads proliferate in other directions.

We mapped two hierarchical clusters of towns and villages in our study area. Each consists of a major ceremonial center and several large satellite towns in precise orientations relative to the center. These towns likely held 1,000 or more inhabitants. Smaller villages are located farther from the center. The northern cluster is centered on X13, which is not a town so much as a ritual center, rather like a fairground. Two large walled settlements lie equidistant to the north and south of X13, and two medium-size walled towns lie equidistant to the northeast and southwest. The southern cluster is slightly different. It is centered on X11, which is both a ritual center and a town, around which are medium- and small-size plaza settlements.

In land area, each cluster was more than 250 square kilometers, of which about a fifth was the built-up core area, making it roughly equivalent in size to a small modern city. Today most of the ancient landscape is overgrown, but forests in the core areas have distinctive concentrations of certain plants, animals, soils and archaeological artifacts, such as prolific ceramics. Land use was more intensive in the past, but the remains suggest that many practices were similar to those of the Kuikuro: manioc plots, small orchards of pequi fruit trees and fields of sapé grass, the preferred material for house thatch. The countryside was a patchy landscape interspersed with secondary forests that invaded fallow agricultural areas. Wetlands, which today are choked with Buriti palm, the most important industrial crop, preserve diverse evidence of fish farming, such as artificial ponds, raised causeways and weir footings. Outside the core areas was a more lightly populated green belt and even deeper forest wilderness between clusters. This forest, too, had its uses for animals, medicinal plants and certain trees, and it was considered the home of diverse forest spirits.

The areas in and around residential sites are marked by dark earth, which the Kuikuro call egepe, a highly fertile soil that has been enriched by household refuse and specialized soil-management activities, such as controlled burning of vegetation cover. People have altered soils the world over, making them darker, more loamy and richer in certain chemicals. In the Amazon these changes are particularly important for agriculture in many areas because the natural soil is so poor. In the Xingu, the dark earth is less prevalent than some areas, because local populations depend mostly on manioc and orchards, which do not require high-fertility soils.

Identification of large walled settlements over an area about the size of Vermont suggests that at least 15 clusters were spread across the Upper Xingu. But most of the region is unstudied, so the true number could have been much higher. Radiocarbon dating of our excavated sites suggests that ancestors of the Xinguanos settled the area, most likely from the west, and began to mold the forests and wetlands to their design about 1,500 years ago or before. In the centuries before Europeans first discovered the Americas, the communities were re-formed into hierarchical clusters. Records date back only to 1884, so the settlement patterns are our only way of estimating the pre-Columbian population; the scale of the clusters suggests a regional population many times larger than today, perhaps numbering 30,000 to 50,000.

Garden Cities of the Amazon
A century ago Ebenezer Howard’s Garden Cities of To-morrow proposed a model for low-density, sustainable urban growth. A forerunner of today’s green movement, Howard envisioned networked towns as an alternative to an industrial world filling with high-rise cities. Ten towns with tens of thousands of people, he suggested, could have the same functional and administrative capacity of a single megacity.

The ancient Xinguanos built such a system, a flat, green style of urbanism or proto-urbanism: an inchoate garden city. Perhaps Percy Fawcett was in the right place but looking for the wrong thing: stone cities. What the small-scale centers lacked in size and elaborate structures, they made up for in numbers and integration. Had Howard known of them, he might have devoted a passage to the “Garden Cities of Yesterday.” The common conception of the city as a dense grid of masonry buildings dates to early desert oasis civilizations such as Mesopotamia but was uncharacteristic of many other environments. Not only the Amazon’s tropical forests but also temperate forest landscapes throughout much of medieval Europe were dotted with towns and villages of similar size to those in the Xingu.

These insights are especially important today as the southern Amazon is redeveloped, this time by Western civilization. The transitional forest of the southern Amazon is being quickly converted into farmland and pasture. At the present rate, it will be reduced to 20 percent of its original size over the next decade. Much of what is left will be restricted to reserves, such as the Xingu, where indigenous people are the stewards of the remaining biodiversity. In these areas, saving tropical forests and protecting indigenous cultural heritage are, in many respects, one and the same thing.

Note: This article was originally printed with the title, "Lost Cities of the Amazon."