left The dense vegetation in Amazonia still has archaeological secrets to share.

Lost cities of Bolivia: rethinking prehistoric life in the Amazon

It has long been assumed that Amazonia was always sparsely inhabited by humans. Now LiDAR survey in Bolivia has revealed the stunning scale of settlements that lay concealed beneath dense tropical vegetation. Heiko Prümers told Matthew Symonds why it is time to rethink the nature of life in this region.

Start

The Amazon is usually seen as home to pristine nature rather than great human civilisations. Although Amazonia – the area of the great river and its tributaries – is larger than Europe, at first glance it has little to offer human settlers. The tropical rainforest soils are poorly suited to agriculture, while the absence of big game and high-calorie plants seemingly guaranteed a demanding hunter-gatherer existence for those hardy groups scraping a living from the region. Even so, it has long been appreciated that prehistoric peoples were present. Enigmatic earth mounds, rising up to 22m in height, are well known in the region, while platforms attest to modest settlements, and eye-catching pottery can carry colourful decorations that seem rich in symbolism. Extensive networks of earthworks serving as raised, ridged fields have been detected on river floodplains, indicating that some inhabitants found a satisfactory way to till the soil.

BeLOW Recent work in Bolivia has revealed remarkable traces of lost cities. This screenshot from a 3D animation created using LiDAR survey results shows the 147ha extent of the settlement at Cotoca. The mounds at the heart of the site form its civic and ceremonial core, while the defences encircling the city can also be seen. A series of causeways radiate out from the settlement.
Recent work in Bolivia has revealed remarkable traces of lost cities. This screenshot from a 3D animation created using LiDAR survey results shows the 147ha extent of the settlement at Cotoca. The mounds at the heart of the site form its civic and ceremonial core, while the defences encircling the city can also be seen. A series of causeways radiate out from the settlement. IMAGE: H Prümers/DAI (Deutsches Archäologisches Institut)

Such traces of monuments and forgotten fields among the tropical vegetation posed numerous questions. Around the mid-20th century, some archaeologists proposed that the answers could be found in the adjacent uplands: the Andes. This was home to an extraordinarily rich and varied group of cultures (see CWA 110) – most famously the Inca – and the region still dominates discussion of South American archaeology. An argument was advanced that in late prehistory settlers from the Andes had entered the tropical lowlands. Although these people came from sophisticated urban cultures, taming the hostile rainforest environment proved to be beyond their means. As the high heat and low availability of food took a terrible toll, the settlers were no longer able to replicate the range of material manufactured in their Andean homelands. Instead, as life grew harder, so too their artefacts grew simpler, acting as potent illustrations of a losing battle against nature.

Today, it is easy to appreciate that there are many problems with this narrative. One of the most obvious is that pottery from settlement sites in Amazonia can stretch back thousands of years, indicating stable occupation over impressive timespans. Even so, until recently few would have doubted that the tropical lowlands were indeed sparsely occupied – especially when compared to the Andes – and that settlement in the rainforest generally amounted to little more than scattered villages. Now a LiDAR survey in Bolivia has used airborne lasers that can virtually strip away the forest canopy to reveal what lies beneath. The results of this collaboration between the German Institute of Archaeology, University of Bonn, Exeter University, and the Ministry of Planning of the Plurinational State of Bolivia have recently been published. They offer a jaw-dropping glimpse of a previously shrouded world, which exposes a level of human ingenuity and sophistication in Amazonia that was once simply unimaginable (see ‘Further information’ at the end).

RIGHT LiDAR survey has revolutionised knowledge of the archaeology of the Llanos de Mojos area of south-west Amazonia. Here we see the LiDAR scanner being installed in the helicopter (TOP) and a view of the sensor while survey is under way over tropical forest (BOTTOM).
LiDAR survey has revolutionised knowledge of the archaeology of the Llanos de Mojos area of south-west Amazonia. Here we see the LiDAR scanner being installed in the helicopter (above) and a view of the sensor while survey is under way over tropical forest (below). IMAGES: H Prümers/DAI; José Iriarte/DAI

Lost world

‘We have been working in the Llanos de Mojos area of south-west Amazonia, in Bolivia, for a little over 20 years’, says Heiko Prümers, of the German Institute of Archaeology, and lead author of the report. ‘This was home to a group called the Casarabe culture, but when we started it was almost a blank site on the archaeological map. There had been studies by William Denevan and his student Clark Erickson, which built on work by a Swedish ethnographer Erland Nordenskiöld. He did excellent work considering the constraints of his time, with Denevan being the first investigator to use aerial photos, although he focused on the networks of ridged fields. One of the problems facing us when we started was that there was no chronology underpinning the known archaeology. So we built one using the surviving pottery, which now allows us to differentiate between five phases of development and say that the Casarabe culture existed between AD 500 and 1400. This shows that earlier publications had mixed together different phases of activity. They talk about the earth mounds and the ridged fields together, with the focus on the ridged fields as the most important element, because they allowed sustainable agriculture. Now, though, our survey work has revealed the existence of huge settlements associated with some of the mounds, but in areas devoid of ridged fields. There, the landscape was being modified in more subtle ways, using dykes and channels.’

LiDAR survey was conducted over six areas, which in total covered 204km² of the roughly 4,500km² territory controlled by the Casarabe culture. This area does not just consist of the jungle that most of us picture in our mind’s eye when we imagine Amazonia. Instead, the landscape resembles a mosaic of rainforest set alongside great swathes of savannah. Neither environment is well suited to traditional survey methods, as in both cases the vegetation can easily obscure archaeological features. As well as allowing this shroud to be digitally stripped away, LiDAR offers another advantage. ‘Most of the land in the Bolivian lowlands belongs to owners that can seem a little like dukes from the Middle Ages’, says Heiko. ‘By that, I mean that they have massive estates, which can cover hundreds of square kilometres. Some are very open minded and support archaeology, but many won’t allow archaeologists on their land. An advantage with LiDAR is that it allows you to document sites from the air that it would be impossible to visit on foot.’

The two largest settlements identified during the LiDAR survey are Landívar (ABOVE), which covers 315ha, and Cotoca (RIGHT), which is 147ha in extent. These two LiDAR maps are not shown to the same scale. In each case, the civic ceremonial core and traces of defensive circuits can be made out.
The two largest settlements identified during the LiDAR survey are Landívar (ABOVE), which covers 315ha, and Cotoca (BELOW), which is 147ha in extent. These two LiDAR maps are not shown to the same scale. In each case, the civic ceremonial core and traces of defensive circuits can be made out. IMAGEs: H Prümers/DAI (generated with ‘Relief Visualisation Toolbox’)

The scale and elaboration of the two largest settlements laid bare by the LiDAR survey are nothing short of breath-taking. One, at Cotoca, covers 147ha, while the other, at Landívar, straddles a river and sprawls over 315ha. Both sites are set within three rings of defences, and are arranged around a monumental core containing civic and ceremonial structures – including great conical mounds and massive U-shaped platforms set on a grand terrace. It does not seem a great stretch to describe such sites as cities, and they certainly lay at the pinnacle of the local settlement hierarchy. Next in the pecking order are sites that sometimes boast a handful of mounds, and were set within polygonal enclosures of 21-41ha. Below them are smaller settlements with circular ditches enclosing up to 2.5ha, while the final tier probably represents temporary campsites, perhaps linked to activities such as logging. Causeways that presumably acted like roads radiate out from Cotoca and Landívar, linking them to the smaller settlements in their hinterland. A 7km-long canal was also constructed to connect Cotoca to the San José lagoon. In total, Cotoca seems to have lain at the heart of a territory spanning roughly 500km², which accounts for a fair chunk of the total territory controlled by the Casarabe culture.

‘Before we started our work, nothing on the scale of Cotoca or Landívar was suspected’, says Heiko. ‘The settlements were always referred to as mounds, because nobody noticed that there was additional architecture. The first step, really, was when we were working at a site called Loma Salvatierra – one of the second-rank settlements – and we saw that there was more to it than just the mounds. When we mapped it, we realised that there were defences surrounding the core. Now that we have the LiDAR, you can look at the plans of all these sites and the scale of them seems obvious. But beforehand it really wasn’t. This is why it was possible for people to say in the past that Amazonia never had complex societies or civilisations. But the important question to ask is “Why should that be the case?” We find complex societies and cities in the tropical forests of Mexico, Cambodia, and Sri Lanka. Africa might seem like a different case, but it has not been extensively studied. I’m sure that more work will reveal large settlements in the rainforest there in the near future. So, when you see it like that, it seems perfectly natural that sizable settlements could exist in Amazonia.’

‘For me, after all these years studying the area, it is nice to be able to say “Hey, these are really important sites, they are not minor settlements”. We cannot yet be certain what was in the area between the site core and the defences at the large settlements of Cotoca and Landívar – the vegetation prevents us from using geophysics – but we should imagine a lot of houses on stilts. You can see on the LiDAR maps that these areas are on lower terrain than the core part of the sites. That means the ground would be inundated with water during the rainy season and muddy for the rest of the year. Assuming, of course, that the climate back then was similar to the climate now – something that is currently under investigation.’

RIGHT Two images of exactly the same area of the Landívar site, providing a sense of what is visible from above to the naked eye, and what appears when the vegetation is removed virtually: a photo mosaic from drone footage (TOp) and a LiDAR image (BOTTOM).
Two images of exactly the same area of the Landívar site, providing a sense of what is visible from above to the naked eye, and what appears when the vegetation is removed virtually: a photo mosaic from drone footage (ABOVE) and a LiDAR image (BELOW). IMAGES: H Prümers/DAI

Urban jungle

‘We haven’t excavated the defences at any of the sites yet. At Cotoca, though, the dimensions of the defensive structures are very impressive. They also appear to have been modified and amplified over time, so I would suggest that these weren’t just to demarcate the settled area: they had a proper defensive function – at least at certain times. That raises another problem we need to bear in mind when looking at the LiDAR maps. We know there was a long period of activity at Cotoca – almost 900 years – but we can’t yet say what elements of the settlements go with which phase. When you look at the three rings of defences at the site, it is apparent that two of them are now incomplete. That could well be because the site expanded, destroying stretches of earlier defences. If so, it would mirror what happened at Cologne, near where I am. There you have what’s called the first and second ring of defences, with the first ring demolished when the city grew. So I think it’s something like that – the population of the city at Cotoca became larger, so they needed to create a more-extensive defensive circuit.’

Alongside the fortifications, there are traces of plenty of practical features that would have aided the smooth running and success of these settlements. One area of particular interest is the intersection between the defences and the causeways. Platforms were constructed at those points, suggesting that they acted as gateways where access to the city could be monitored, restricted, or denied altogether. In the case of Cotoca, there is also one causeway that leads to a small river. At the end, near the river, another platform was created. In this case, it presumably provided storage for goods coming in by water, before they were transported to the settlement. Reservoirs are important features, too, and have been found at almost every site mapped by the team. At first sight, the need for such facilities seems surprising, as the Amazon is not an area that you would associate with a shortage of rainwater. The area controlled by the Casarabe culture is unusual, though, because of the large stretches of savannah. As a consequence, the driest month of the year in the region is drier than the driest month in Germany. Because of that, an ability to store large quantities of water is essential in order for cities that are home to sizable populations to survive. Alongside meeting the need for drinking water, stores may have been required for the farming of fish or turtle. What is certain is that these reservoirs were impressive features, which can be up to 60m in diameter. Serious labour went into creating them.

The site of Loma Salvatierra, one of the second-rank settlements, seen from above in two views of exactly the same area: a photo mosaic created using a drone (LEFT) and the results of the LiDAR survey (RIGHT).
IMAGES: H Prümers/DAI

This is also true of the civic-ceremonial monuments found at the core of the large settlements. At Cotoca, the earth needed to create the mounds, and other earthworks, was won from a 50-80m wide strip of ground immediately beyond the central terrace. In total, the amount of earth shifted to create this core, which covers 22ha, amounts to an extraordinary 570,690m³. The scale of the undertaking is not the only fascinating feature about these complexes. Another is that the platforms deviate from a simple north–south axis. Some are orientated exactly 19° to the west of this line, with the major structures in the core at both Cotoca and Landívar aligned towards the north–north-west. As the platforms are rectangular, this leaves their eastern side pointing towards the east–north-east. Intriguingly, this arrangement overlaps with findings from excavations at Loma Salvatierra. There, more than 100 graves were investigated. While burial fashions varied over time, there were periods when bodies were placed in extended graves. In such cases, the graves were cut to the same orientation as the monuments in the settlement cores. The east–north-east angle of both the graves and monuments would more or less coincide with the direction of the rising sun at the summer solstice, so perhaps this factor was important to their beliefs and worldview.

If the sheer scale of the two cities seems extraordinary against the backdrop of Amazonia being viewed as a sparsely inhabited region, so too is the quantity of settlements in the wider hinterland. Throughout Casarabe culture territory, you can expect to find approximately 10 sites within a 10km radius of any given settlement. The connection between these sites and the cities is emphasised by the impressive network of causeways, which generally connect smaller settlements to larger ones, rather than creating a web of thoroughfares between sites of a similar size. Once again, Cotoca provides an excellent example, with 18 other monumental sites – including three examples of the second-largest category of settlement – within the 500km² area that seemingly served as its territory. All of this has important implications for how we view the population of the tropical lowlands in comparison to the Andean uplands, which are traditionally believed to have been much more densely inhabited.Population growth

LEFT A circular reservoir, approximately 60m in diameter, next to the settlement at Landívar.
A circular reservoir, approximately 60m in diameter, next to the settlement at Landívar. IMAGE: H Prümers/DAI

‘We can now see that general estimates for the population of Amazonia have been too low’, says Heiko, ‘much too low. We should be thinking of similar population levels in both the Amazon and Andean region. It is increasingly apparent that one of the problems for archaeologists working in Amazonia is that wooden structures are so hard to find. In that regard we’re lucky that large portions of these sites were also built of earth, which shows up clearly on the LiDAR maps. If these cities had only consisted of three-storey wooden houses, we wouldn’t have found anything. But everything about these sites – the scale of the large settlements, the reservoirs, the defences, the sophisticated infrastructure – points to large populations not just living in them, but also building them. This was really a society that knew how to make people work!’

‘The culture that is contemporary with the Casarabe in the Andean uplands is the Tiwanaku. It covered an extensive area, running from the south coast of Peru and down into northern Argentina. But – just like the Inca – it doesn’t seem to have extended beyond the foothills into the eastern lowlands. We don’t know why, but it looks like they never ventured into the rainforest. There is also very little sign of traded goods traveling from the uplands to the lowlands. The Tiwanaku pottery is very distinctive – it’s polychrome, you’ll recognise it immediately, but we have never found a single sherd. We do have some sodalite gemstones, though, and there is only one known site where they could have come from in South America: the Cochabamaba area in the Bolivian Andes. These gemstones must have reached the Casarabe culture via some form of trade or exchange, but we do not know if it was direct or indirect. There is also a small amount of copper, which supposedly came from the Andean area, but we don’t know exactly whereabouts. While thinking about the Tiwanaku culture, it is also interesting to compare the quantity of earth moved to build the civic and ceremonial core of Cotoca with that shifted to create the Akapana: the central pyramid in the capital city of the Tiwanaku. Ten times more material had to be moved to build the monuments at Cotoca.’

LEFT Cotoca, as seen on the ground. Here Carla Jaimes Betancourt, one of the co-authors of the Nature paper (see ‘Further information' box on p.22), is descending from the central pyramid in the civic and ceremonial core of the city.
Cotoca, as seen on the ground. Here Carla Jaimes Betancourt, one of the co-authors of the Nature paper (see ‘Further information’ at the end), is descending from the central pyramid in the civic and ceremonial core of the city. IMAGE: H Prümers/DAI

Given the scale of the achievements of the Casarabe culture, and their ability to adapt a seemingly unpromising region to support urban populations, what was it that brought their way of life to an end? ‘We can’t be sure about what caused the collapse,’ says Heiko, ‘but we do know that it wasn’t the Spanish coming in, because 1400 is a century too early for them to be a factor. So we have to look for other reasons. Our own time has plenty of examples of possible explanations. We have wars, pandemics, and shortages of important goods. The answer could lie in any one of those. Given the scale of the defences at the settlements, war could certainly have been a cause. But I think a shortage of water is also a good guess. Today, we can bring essential supplies from one spot to another in a matter of minutes or hours. And there are times when we really need that capability, in order to help avert famine or disease in areas that have been struck by a catastrophe. That option was not available to the Casarabe culture. If the crops had failed for just one or two years, people would need to leave and look elsewhere in order to survive. You don’t have to go too far back in time to find examples of that in Europe. Seen that way, it is not the collapse of these large centres that is surprising, rather it is that they lasted for as long as they did.’

LEFT A virtual section through the major platforms at Cotoca. The brown points show the ground, while the green points show the vegetation that was digitally removed to create the Digital Surface Model (DSM).
A virtual section through the major platforms at Cotoca. The brown points show the ground, while the green points show the vegetation that was digitally removed to create the Digital Surface Model (DSM). IMAGE: H Prümers/DAI

Green cities

The results of this survey work provide a triumphant example of how LiDAR can detect extraordinary new details about the true nature of ancient sites. They show that archaeology still has the potential to deliver real surprises about our understanding of the past. Comparing the current picture of a flourishing urban culture that successfully shaped its environment to its needs with the mid-20th-century perception of upland settlers fighting a losing battle with the environment illustrates how far interpretations can change. Ironically, at the same time that we are beginning to appreciate the traces of these great settlements exist, so too in some places they are in danger of being wiped away.

The dense vegetation in Amazonia still has archaeological secrets to share. IMAGE: H Prümers/DAI

‘Destruction is happening right now in Bolivia’, Heiko notes. ‘Thousands of hectares of land were bought in one area, and it was cleared using machines. Anything archaeological has been destroyed: all that is left is plain. And, of course, it is not just the archaeology that suffers.’ Given that these losses are happening, it raises the question of whether the great settlements at Cotoca and Landívar are exceptions, or if we should be imagining more lost cities scattered through Amazonia. ‘I know there are more out there!’, says Heiko. ‘I hope those data will be published soon.’

FURTHER INFORMATION
H Prümers, C J Betancourt, J Iriarte, M Robinson, and M Schaich (2022) ‘LiDAR reveals pre-Hispanic low-density urbanism in the Bolivian Amazon’, Nature 606: 325-328 (https://doi.org/10.1038/s41586-022-04780-4).