Authors

Joe Hirst¹, Joy Singarayer¹, Umberto Lombardo², and Francis Mayle¹

¹ University of Reading

² University of Bern

Abstract

For decades, scholars have debated the extent to which Amazonian landscapes were modified by pre-Hispanic peoples. Some of the strongest evidence for large-scale pre-Hispanic landscape domestication comes from the South-Eastern sector of the Llanos de Mojos (LM), a seasonally flooded forest-savanna mosaic landscape in lowland Bolivia. Despite the challenges posed by frequent droughts and seasonal inundation, the pre-Hispanic Casarabe culture undertook landscape domestication in this environment to produce over 189 large habitation mounds, interconnected by a dense network of causeways, canals, and lakes (Lombardo and Prümers, 2010). LiDAR scans reveal these mounds to be the remains of large settlement sites, with archaeological evidence suggesting their inhabitants took advantage of the fertile sediments in this sector of the LM to cultivate a wide variety of crops. This evidence has led scholars to propose the Casarabe culture represents the first known case of agrarian-based, low-density urbanism in pre-Hispanic Amazonia, a complex society continuously operating between 400 and 1400 AD (Prümers et al., 2022).

Little is known about either the population size of the Casarabe culture or the extent to which they utilised and modified the surrounding rainforest and savanna ecosystems. Unlike contemporary Amazonian indigenous groups, maize likely formed a key component of their diet (Prümers, 2015), implying that they were able to perform large-scale maize cultivation within the LM. Co-author Lombardo has shown that the Casarabe culture utilised drainage canals to cultivate maize on the open savanna, a previously unseen form of agriculture across Amazonia. However, the extent to which they cultivated maize in forested areas remains unknown. 

In this presentation, I outline the development of, and experiments conducted upon, an agent-based model that has been developed to explore questions about the potential extent and spatial distribution of land transformed by the Casarabe culture for maize based agriculture. Agent-based models are uniquely suited to addressing these types of questions because of their ability to represent macro-scale features such as the extent of human land use (e.g., through deforestation, cultivation) as the emergent products of decisions operating at the individual level. An agent-based approach therefore not only allows us to explore the potential scale of anthropogenic land-use within the South-Eastern LM, but also to gain insight into the processes and reasoning that underlies these human-environment interactions. Our agent-based model functions as a ‘virtual laboratory’, enabling us to explore the extent and spatial distribution of land that is stripped of its vegetation and converted into maize monoculture across a variety of plausible ‘What if?’ scenarios. Encoded within the model is a simplified representation of the South-Eastern LM, parameterised using archaeological, ecological, ethnographic, and topographic data. Inhabiting this virtual landscape are agents representing household units of parents and children, driven by a set of simple rules based on the demand for vital resources. When aggregated, these agents reflect a simplified version of the Casarabe Culture. These pre-Columbian household agents form the primary subjects of our investigation, functioning as the main drivers of landscape modification within the model. 

We conducted experiments with our model to explore the amount of land that household agents appropriate for maize agriculture across a range of different scenarios, as well as to identify the most important model assumptions in determining this output. Our experiments show that households are motivated to transform a far smaller proportion of the landscape than might initially be expected given the Casarabe culture has previously been thought to have driven regional-scale deforestation and landscape domestication (Erickson and Balée, 2006). Over a 1000-year period, Households only converted between 13 and 21% of the terrestrial (non-Water) land surface, more closely aligned with palaeoecological data that suggests deforestation was far more localised (Whitney et al., 2013). This is primarily motivated by the desire of households to preferentially cultivate land that possesses desirable environmental characteristics e.g., higher elevation. While a number of assumptions made by the model influence the extent of modified land, the most important contributing factor is regional population size. 

A second experiment was conducted with the model to determine the spatial distribution of transformed land when different environmental characteristics were valued by household agents, including the preference to cultivate at higher elevation to avoid the impacts of seasonal flooding. The outputs generated by this experiment allowed us to produce spatially explicit maps of land modified by household agents, which demonstrate that households are able to extensively modify localised portions of the landscape for maize agriculture even if large parts of it remain untouched. These outputs can play an important role in guiding future Archaeological and Palaeoecological research on the Casarabe culture, helping to identify sites of interest as viable targets for future fieldwork. They can also be compared to empirical research as it becomes available, improving our understanding of the processes underlying human-environment interactions associated with one of the most complex pre-Hispanic societies yet discovered in lowland South America.

References

Erickson, C., Balée, W., 2006. The historical ecology of a complex landscape in Bolivia., in: Time and Complexity in Historical Ecology: Studies in the Neotropical Lowlands. Columbia University Press, New York, pp. 187–233.

Lombardo, U., Prümers, H., 2010. Pre-Columbian human occupation patterns in the eastern plains of the Llanos de Moxos, Bolivian Amazonia. J Archaeol Sci 37, 1875–1885. https://doi.org/10.1016/j.jas.2010.02.011

Prümers, H., 2015. Loma Mendoza. Las Excavaciones de los años 1999-2002, Kommission für Archäologie Außereuropäischer Kulturen des Deutschen Archäologischen Instituts, 2015. Plural Editores, La Paz.

Prümers, H., Jaimes Betancourt, C., Iriarte, J., Robinson, M., Schaich, M., 2022. Lidar reveals pre-Hispanic low-density urbanism in the Bolivian Amazon. Nature. https://doi.org/10.1038/s41586-022-04780-4

Whitney, B., Dickau, R., Mayle, F., Soto, D., Iriarte, J., 2013. Pre-Columbian landscape impact and agriculture in the Monumental Mound region of the Llanos de Moxos, lowland Bolivia. Quaternary Research (United States) 80, 207–217. https://doi.org/10.1016/j.yqres.2013.06.005