Look around the Amazon rainforest today and it’s hard to imagine it filled with people. But in recent decades, archaeologists have started to find evidence that before Columbus’s arrival, the region was dotted with towns and perhaps even cities. The extent of human settlement in the Amazon remains hotly debated, partly because huge swaths of the 6-million-square-kilometer rainforest remain unstudied by archaeologists. Now, researchers have built a model predicting where signs of pre-Columbian agriculture are most likely to be found, a tool they hope will help guide future archaeological work in the region.
In many ways, archaeology in the Amazon is still in its infancy. Not only is it difficult to mount large-scale excavations in the middle of a tropical rainforest, but until recently, archaeologists assumed there wasn’t much to find. Amazonian soil is notoriously poor quality—all the nutrients are immediately sucked up by the rainforest’s astounding biodiversity—so for many years, scientists believed that the kind of large-scale farming needed to support cities was impossible in the region. Discoveries of gigantic earthworks  and ancient roads , however, hint that densely populated and long-lasting population hubs once existed in the Amazon. Their agricultural secret? Pre-Columbian Amazonians enriched the soil themselves, creating what archaeologists call terra preta.
Terra preta—literally “black earth”—is soil that humans have enriched to have two to three times the nutrient content of the surrounding, poor-quality soil, explains Crystal McMichael, a paleoecologist at the Florida Institute of Technology in Melbourne. Although there is no standard definition for terra preta, it tends to be darker than other Amazonian soils and to have charcoal and pre-Columbian pottery shards mixed in. Most of it was created 2500 to 500 years ago. Like the earthworks, terra preta is considered a sign that a particular area was occupied by humans in the pre-Columbian past.
By analyzing location and environmental data from nearly 1000 known terra preta sites and comparing it with information from soil surveys that reported no terra preta, McMichael and her team found patterns in the distribution of the enriched soil. The scientists concluded that terra preta is most likely to be found in central and eastern Amazonia on bluffs overlooking rivers nearing the Atlantic Ocean. It’s less common in western Amazonia, where runoff from the Andes tends to add nutrients to the soil naturally, and in highland areas such as Llanos de Moxos in Bolivia, which is home to many impressive pre-Columbian earthworks. By analyzing the environmental conditions most strongly associated with terra preta, the team was able to build a model predicting where undiscovered terra preta sites are most likely to be found. Overall, they suspect that there is probably about 154,063 km2 of terra preta in the Amazon , composing about 3.2% of the basin’s total area, they report online today in the Proceedings of the Royal Society B.
Not only does modeling the likely locations of terra preta reveal possible patterns of human settlement in the Amazon, but it also gives archaeologists “a starting point” for future excavations, McMichael says. “Within a forest of almost 6 million square kilometers, it’s hard for archaeologists to determine site locations for sampling,” she explains. Like the increasingly popular LiDAR —which can find earthworks hidden under the rainforest canopy but can’t sniff out terra preta—“these [statistical] methodologies narrow down the probabilities” of where to find promising archaeological sites.
Other Amazon experts are more skeptical. Michael Heckenberger, an archaeologist at the University of Florida in Gainesville, who was not involved in the research, points out a possible discrepancy in the sampling methods employed by McMichael’s team. The terra preta sites used to make the statistical model, he says, “just happen to be the areas where there’s been intensive archaeological survey.” The areas designated as terra preta-free, on the other hand, were sampled and categorized by ecologists and geologists, often long before anyone was looking for terra preta or other signs of pre-Columbian settlements in the Amazon. Just because a region is labeled terra preta-free now, Heckenberger suspects, doesn’t mean there isn’t any terra preta there. It just means archaeologists haven’t been there to look for it—yet. McMichael’s map “serves as a reminder of what we don’t know” about the Amazon’s past, he says.
McMichael agrees that a terra preta-free label should not be taken as proof that humans never settled a region. The relative lack of terra preta around the Llanos de Moxos earthworks proves that humans didn’t necessarily enrich the soil, or do so in the same way, everywhere they lived, she says. “I would think that cultures adapted differently to the different environmental conditions,” creating terra preta where the natural soil was particularly poor and modifying their environment in other ways in regions where they didn’t necessarily need to enrich the soil to support large populations.
McMichael hopes to use her statistical methods to model all different kinds of ancient human impacts on the Amazon. Her team has a paper in press at the Journal of Biogeography predicting the locations of earthworks, and eventually she hopes to create a map correlating past human settlements with various ecological patterns. If pre-Columbian humans encouraged the spread of particular plants and animals they found helpful in the regions around their settlements, for example, that might affect species distribution in the Amazon today. Soon, scientists might be able to go beyond earthworks and agriculture and read the Amazon’s history in the forest itself.