Casarabe Culture

← Encyclopedia

Twice a year the water decides where people may stand. From roughly November to April the rains swell the rivers of the Beni and the savanna of the Llanos de Mojos drowns under a thin sheet of standing water; from May to October the same plain bakes and cracks, the channels shrink to threads, and the grass goes the color of old straw. It is a landscape that punishes both extremes and rewards whoever can hold water against the drought and shed it against the flood. Between roughly 500 and 1400 of the common era — across the centuries the Mediterranean world counts as late antiquity sliding into the medieval — a people now called the Casarabe culture answered that double demand by rebuilding the plain itself. They raised dry ground where the water wanted to spread, cut channels where it wanted to pool, and laid out across the savanna a network of settlements, embankments, and reservoirs whose scale was not even suspected until laser light was flown over the canopy in the twenty-first century.

The culture takes its name from a modern village near Trinidad, in the Bolivian department of Beni, where the first of its mounds were dug. Those mounds — the local Spanish word is lomas — had been read for over a century as the visible part of the story: low artificial hills standing above the flood, holding burials and house floors, scattered across the grassland like islands. They are not natural rises but raised ground, built up over generations of occupation and deliberate construction, and the digging into them yielded the human texture of the culture before its plan was ever seen — successive house floors, hearths and refuse layered like a calendar, and the dead interred in large ceramic funerary urns. The pottery itself, worked through a sequence of styles whose continuities and breaks span roughly the culture’s eight centuries, is the thread by which the Casarabe occupation was first defined and dated. What no one on the ground could see was the geometry that connected the mounds — the lines reaching from one to the next and the banked rings around the largest of them.

The savanna and the laser

The Llanos de Mojos is one of the great seasonally inundated lowlands of South America, a forest-and-savanna mosaic of some 120,000 square kilometers spread across northern Bolivia. Under tall grass in the dry season and under water in the wet, low relief vanishes; a bank a meter high reads as nothing to a surveyor walking it, and the forest islands that dot the plain hide their own interiors. The earthworks of Mojos had been mapped piecemeal for decades — by the Swedish archaeologist Erland Nordenskiöld, who excavated mounds near Trinidad and Loreto in 1908–9 and pulled urn burials and incised pottery from them; by the geographer William Denevan and his collaborators George Plafker and Kenneth Lee, who in the 1960s recognized raised fields, causeways, canals, and dikes ribboning the savanna; and by the landscape archaeologist Clark Erickson, who argued that the plain was a built environment, an engineered countryside rather than a wild one. But ground survey and aerial photography could only catch what stood above the grass or showed against the seasonal flood.

Airborne lidar broke the canopy. The technique fires hundreds of thousands of laser pulses downward from an aircraft and times the return of each; because some pulses slip between leaves all the way to the ground while others bounce off the canopy, the dense vegetation returns can be filtered out and discarded, leaving a model of the bare earth as if the forest and grass had been stripped away and only the soil’s true shape remained. A meter-high bank invisible to a surveyor wading through it stands out unmistakably on such a model. Flown over six blocks of Casarabe territory and reduced to bare-earth terrain models, the survey resolved what no overflight had — the full plan of the settlements, their banked perimeters, and the long straight lines reaching from one to the next. The results, published in Nature in 2022 by Heiko Prümers of the German Archaeological Institute, Carla Jaimes Betancourt of the University of Bonn, and José Iriarte and Mark Robinson of the University of Exeter, described what the authors called a tropical low-density urbanism not previously documented in Amazonia. The lidar covered around 200 square kilometers of a culture area estimated at some 4,500, and within that window it mapped twenty-six sites, eleven of them never before recorded — and two of a size that changed the question.

Cotoca and Landívar

The two principal centers are Cotoca and Landívar. Cotoca covers roughly 315 hectares, Landívar about 147 — figures that put them well beyond anything earlier work in Mojos had reckoned with, and that read less like overgrown villages than like planned ceremonial cities. Each sits at the heart of a built core whose architecture is monumental and deliberate. Broad stepped platforms raise the ground; on top of them stand U-shaped structures and rectangular platform mounds; and rising above the whole, conical earthen pyramids climb to as much as twenty-two meters — the height of a six- or seven-story building, built entirely of moved and shaped earth.

These cores are not open. Each is wrapped in ranked concentric polygonal banks — earthen ramparts and moats laid out in nested rings, defensive in form and emphatic in effect, marking the center off from the surrounding land and from the lesser sites that depend on it. The orientation is consistent across the network; the platforms and U-shaped structures align, suggesting a shared cosmological grammar imposed on the plain rather than an accretion of unrelated villages. The settlement system is tiered — the great centers like Cotoca at the top, then secondary sites, then smaller nodes — a four-level hierarchy of the kind archaeologists associate with centralized political organization. One large center appears to have presided over an area on the order of 500 square kilometers.

The earthen pyramids invite the obvious comparison with the masonry monumentality of the Andes that loomed to the west — with the sunken temples and stelae of Chavín, the carved gateways and platforms of Tiwanaku on the high Titicaca plateau, the administrative cities of the Wari. The comparison is one of scale and ambition only, not of descent: the Casarabe builders worked in earth, not dressed stone, in a flooded lowland rather than a mountain basin, and there is no line of continuity running up the eastern slopes to join them to the highland states. What the comparison registers is that monumental, planned, hierarchical settlement was not the property of the Andes alone. The lowland answered the same human impulse in its own medium — packed earth raised against the floodwater rather than dressed stone set in a mountain basin.

The engineered plain

The pyramids are the part that rises; the more consequential engineering is the part that lies flat. Radiating from the centers, straight raised causeways run for kilometers across the savanna, lifting paths above the seasonal flood and tying the settlements into a single connected fabric — roads that doubled as dikes. Their very straightness is the signature of design: a causeway holds its line across terrain that gives the eye no reason to, joining a great center to its dependent sites by the shortest visible path, so that the network reads on the bare-earth model as a wheel of spokes anchored on each hub. Between and around them runs a system of canals and reservoirs that reads as a deliberate management of the plain’s hydrology: channels to carry water off the high ground in the rains, and impoundments to hold it through the dry months when the rivers withdraw. The same earthworks that drained also stored; the same lines that moved people also moved water.

The integration is the point. A causeway is a road, a dike, and a canal bank at once; a reservoir is a fishery, a dry-season water store, and a flood buffer at once; the high ground of a mound is house plot, refuge, and ceremonial platform at once. None of these elements is unique to Casarabe — raised fields, ditches, fish weirs, and forest islands had long been recognized across the Llanos de Mojos, and the geometric ditched enclosures dug into the forest earthworks of Acre to the north testify that the southwestern Amazon was being shaped by human hands well before and well beyond this culture — but here they cohere around the monumental centers into a single articulated system, a countryside laid out as one design rather than a patchwork of unrelated improvements. The plain was not merely lived on; it was instrumented, every contour bent toward holding the year’s water on the near side of disaster at both ends of the season.

This is the heart of why the Casarabe landscape matters. The plain that flooded and parched by turns was not endured but rebuilt — its water rerouted, its ground raised, its surplus banked against the lean season. The economy that all of this fed was, against long expectation, an agricultural one centered on maize. Stable-isotope analysis of human remains has since shown that maize was the dietary staple, not a supplement, and that the people also managed animals, including Muscovy ducks; pollen and land-use evidence indicates maize was cultivated locally for well over a millennium, with the savanna kept deliberately open against the encroaching forest. The population that the system supported has been estimated, on current evidence, somewhere in the broad range of ten thousand to a hundred thousand. A maize-fed, water-engineering, mound-building society of that size, sustained for the better part of a thousand years on a seasonally drowned savanna, is precisely the kind of thing the Amazon was long held to be incapable of producing.

Rethinking the green desert

The earthworks of Mojos had long sat awkwardly against the mid-twentieth-century assumption that the Amazon’s thin, acid soils set a hard ceiling on population and complexity: Denevan’s raised fields and Erickson’s causeways were already difficult to square with it. The Casarabe lidar made the discrepancy impossible to manage. Here was urban-scale planning, monumental architecture, a settlement hierarchy, and continental-grade water engineering, on the very kind of ground the ceiling theory said could bear none of it.

The correction is not that the forest was secretly easy. It is that the people remade the ground rather than submitting to it — that the Casarabe landscape is an artifact, the savanna’s hydrology rewritten in earth so thoroughly that the plain a visitor crosses today is in large part their construction, decommissioned and overgrown. To call such a place wilderness is to mistake a ruin for an absence. The other side of that correction is a caution the makers’ descendants insist on: this is the ancestral country of living peoples — the Mojeño and Baure among them, alongside the Movima, Canichana, Cayuvava, and Itonama whom the first Spaniards met on the plain — and its monuments belong to a continuous Indigenous history, not to any vanished lost-world fantasy or to some invented colony of Andean exiles dropped onto the plain. The mounds were dug, the causeways walked, the savanna burned and planted by ancestors of people still on the land.

Scholarship and sources

The defining publication is Heiko Prümers, Carla Jaimes Betancourt, José Iriarte, Mark Robinson, and Martin Schaich, “Lidar reveals pre-Hispanic low-density urbanism in the Bolivian Amazon,” Nature 606 (2022): 325–328, which presents the bare-earth terrain models of Cotoca, Landívar, and the surrounding network and frames the find as tropical low-density urbanism. The institutional accounts from the German Archaeological Institute and the University of Exeter, distributed via EurekAlert, summarize the survey method, the team, and the site dimensions for a general readership; Science News places the result within the wider reassessment of Amazonian settlement.

The subsistence question was settled separately. Carbon-isotope work on Casarabe human remains, and the synthesizing paper “Maize monoculture supported pre-Columbian urbanism in southwestern Amazonia,” Nature (2024), established maize as the dietary staple rather than a marginal crop and tied the agricultural regime to the engineered water system. Land-use and population modeling — the localized maize chronology near Loma Suárez and the working population estimate of roughly ten thousand to one hundred thousand — extend that picture. The deeper background lies in a century of Mojos research: Erland Nordenskiöld’s excavations of 1908–9 near Trinidad and Loreto; William Denevan’s recognition in the 1960s of the raised fields, causeways, and canals; Clark Erickson’s landscape-archaeology argument for an engineered savanna; and Carla Jaimes Betancourt’s definition of the Casarabe ceramic sequence across its phases.

A managed countryside

Why the great centers were given up around 1400, before Europeans reached the plain, is not yet resolved on the evidence in hand; shifts in climate and in the rhythm of flood and drought are among the factors under study, and the abandonment of mounds like Loma Suárez left the forest to close back over the once-open ground. What the laser recovered is not a riddle of disappearance but a plan — the visible blueprint of a city laid out across a flooded savanna, its pyramids and ramparts at the center and its causeways and reservoirs reaching out to bind a managed countryside into one. The plain near the lower Mamoré is not, and was not, an empty quarter. It is a drawing in earth, large enough that it took an aircraft and a laser to read the whole of it at once.

→ Related: Marajoara Culture · Acre Geoglyphs · Santarem Tapajos Culture · Tiwanaku · Wari · Chavin