The Amazon River moves more water and sediment than any other river in the world, and at the mouth of the river, fresh water extends for forty miles out into the Atlantic Ocean. The area drained by this mighty river is about the same size as the continental United States and comparatively little archaeological research has been done there. However, what research has been carried out has shown that the Amazon was home to societies of people who lived in large settlements, made elaborate pottery, and practiced intensive agriculture. As more archaeological data from the Amazon Basin become available, a clearer picture is emerging of what life was like in lowland South America before the arrival of the Spanish.
One question anthropologists often try to answer about a particular group of people is the manner in which they organize themselves. Social organization often governs how resources are divided, how marriage partners are chosen, and where houses are built. For cultural anthropologists, the investigation of social organization is based on participating in the life of the society itself. For archaeologists, this is not an option, and they must focus on the material record left behind by past societies.
Many people in the Amazon were farmers, and of those farmers, many invested their labor in various earthworks that made farming easier or more productive (Fig. 1). These constructions helped move water from rivers to fields, raised crops above floodwaters, and maintained reserves of water during the dry season (Erickson 2001). The study of these prehispanic earthworks has contributed to our knowledge of agriculture in South America and all over the world; it can also help us understand the people who built them. There is a link between the social organization of the prehispanic inhabitants of the Bolivian Amazon and the spatial organization of raised fields that can be seen on the ground today.
A wide variety of these ancient earthworks exists in the Lianos de Moxos, a seasonally flooded tropical savanna in the lowlands of eastern Bolivia (Figs. 2, 3). Although the average temperature does not vary greatly throughout the year, the seasonal changes in the landscape are striking. In the dry season some rivers can be crossed without getting your shirt wet (Fig. 4). In the wet season, the same river is 20 meters deep and several hundred meters wide (Fig. 5). As much as half of the landscape is under water during the months of January, February, and March.
Today, ranchers live in small outposts scattered along the river levees of the Iruyañez River (Fig. 6). Although these ranches are isolated, especially in the wet season (Fig. 7), radios and airstrips help to build a sense of community. Every night in the field we listen to the “messages,” a radio broadcast that contains specific messages for individual ranches and people from their friends and relatives in town. Everything from instructions for immunizing cattle to happy birthday wishes is communicated via these radio messages. Maintaining a herd of cattle is hard work, and for each season of the year there are important tasks that must be attended to. But in July, the ranches are almost deserted as many people head into town to attend the yearly festival (Fig. 8).
Searching for Raised Fields
Over the course of fifteen months in 1996 and 1997, I carried out a program of survey and excavation in a study area of about 340 square kilometers along the Iruyañez River in the north-central part of Moxos, in an area characterized by so-called large raised fields. These fields are broad, long, shallow earthen platforms, averaging 20 meters in width by 200 meters in length and about 20 centimeters in height (Fig. 9), Along the Iruyañez River, raised fields tend to be located in the open savanna, near to the smaller tributary rivers and creeks (Fig. so). In the flood season, which differs quite dramatically from the dry season, raised fields remain dry for the most part, due to their location. They are most easily seen from the air (Fig. ii) and cover several square kilometers in the study area. I measured a sample of 1,711 raised fields using aerial photography and survey, totaling 5.68 square kilometers in area. The total area covered by raised fields in the study area is at least 10 square kilometers, In combination with the neighboring canals, or negative space between platforms, the raised fields cover at least 10 percent of the landscape.
Raised fields are associated with areas of forested high ground along the larger rivers as well as the smaller creeks. Sites in the study area include “forest island sites,” as well as large sites along the river levees. A transect of test excavations yielded ceramics and anthrosol (dark soils accompanied by charcoal, ceramics, and other evidence of human occupation) along more than 330 meters, suggesting a total occupied area greater than 5 hectares (12.3 acres), and perhaps higher than 20 hectares (49.4 acres). Ceramics recovered from the levees include flaring rim vessels, with fine-line painting on the flaring rims. Two sites, both associated with raised fields, were dated using radiocarbon analysis, one to the 14th and I5th centuries(AD, and the other to the 5th and 6th centuries AD. One of these large sites may have represented a village of 1,000 or 2,000 people.
Working Parties and the Organization of Labor
To reconstruct the organization of labor, I hypothesize that communal “work parties” built the raised fields. Four lines of evidence suggest this hypothesis: ethnographic analogy, ethnohistoric data, ceramics, and the raised Fields themselves. First, in farming systems around the world, agricultural yields are used not only to feed farmers and their dependents, but also as payment for communal labor. The coordination of this labor is often accomplished by means of so-called work parties. In cases described by such anthropologists as Erasmus (1965), Netting (1996), and Stone (1996), small communities of farmers pool their labor to perform intensive agricultural tasks such as harvesting crops and raising barns. Since these tasks must be completed in a short period of time, the amount of labor required is often beyond the capacity of any individual household. Therefore, communal work parties combine labor from several households in order to complete the task. As payment, the guests at a work party are usually provided with food and drink (especially alcohol) in exchange for their labor. Sometimes even the performance of difficult tasks can take on a festive, competitive feeling, Since these practices are common among intensive agriculturalists from around the world, they provide the basis for a reasonable hypothesis for how prehispanic farmers might have built their raised fields.
Second, the existence of such work parties in the Llanos de Moxos is implied by early historical sources. In particular, a Jesuit missionary named Agustin Zapata traveled along the Iruya-ñez River in the late 1600s, and discussed in one of his letters the subject of agricultural labor. In this passage, Zapata describes indigenous practice: “Having kept these Indians from their drunken parties, which were payments to those who helped them clear their fields and build their houses…” (Zapata 1906, translation by the author), Such accounts, in this and several other references, imply that work parties like those described in the anthropological literature might have occurred along the Iruyañez River, at least by the time the native peoples encountered the Spanish.
Third, although extensive analysis has yet to be performed, ceramics recovered from surface collections and excavation also suggest the presence of communal work parties. In particular, many sherds are found in dense surface scatters on river levees within several meters of raised fields. The most common vessel form found is a flaring rim bowl or plate, with fine-line painting (Figs. 12. 13). Decorated vessels of this shape are often interpreted as serving vessels, rather than cooking or storage vessels. This painted pottery could have been used at the hypothetical work parties, where workers would have been served food and drink,
Finally, the fourth line of evidence for communal work parties is the agricultural landscape itself, which bears closer examination. The raised field platforms that cover the savanna are both the settings for and the products of agricultural activities. Their location indicates exactly where activities such as soil preparation, planting, and harvesting took place in the past. In comparison with artifacts, raised fields are less affected by post-abandonment transformations. Though platforms erode and canals fill with soil, natural processes do not move raised fields into new patterns. Fields remain where they were abandoned, making it possible to use their spatial patterning to study the spatial patterning of past activities. These patterns can then be compared to theoretical expectations.
Estimating Work Parties from Raised Field Groups
The relationship between labor requirements and the carrying capacity of raised fields suggests that the fields might be organized into groups, For example, while the carrying capacity of a single field is about 3 people, the labor requirement for its construction is about 20 people (Fig. r4), Conversely, the carrying capacity of all of the thousands of raised fields is of course much greater than the labor required to build any one of them. However, there should be a point at which the carrying capacity of a group of raised fields is roughly equal to the labor required to build the largest field in the group. This point can be estimated using the results of experimental construction of raised fields, and also the results of agricultural experiments.
Clark Erickson carried out experimental construction of raised fields in several lowland communities in the early 199Os (Fig. 15). These experiments in Moxos confirmed the results of similar work in the Andean highlands (Erickson 1988), that raised fields can be built, and could have been built, using labor coordinated at the local level. Intensive agriculture of this type does not seem to necessarily depend on outside authority or direction. Studies of time and labor from these experiments are also very useful for archaeological interpretation, as they can be used to estimate the labor that was required to build the prehispanic raised fields.
The amount of food that can be produced per unit area of raised fields is best estimated by using local production figures. Information comes from several agronomic studies, conducted at the Estación Biológica del Beni, a biological research station located in Moxos. The raised fields yielded roughly 25 metric tons of manioc per hectare, a very high figure. The crops on the experimental fields also survived an inundation that wiped out crops in conventional slash-and-burn fields. Recorded yields of agriculture practiced by the Tsimane people are about 18 metric tons of manioc per hectare, confirming that agricultural yields from fields in the region can be high.
Working from these two sets of calculations, we can find the point where the amount of labor that is required to build the largest field in a group is roughly balanced with the number of people that could have been supported by all the fields in the group. Allowing for a great deal of variability, we might expect raised fields to be organized in groups of 7 to 30 fields, corresponding to between 20 and 100 people, Such a social group would have been able to construct each individual raised field in the group of fields and could have been sustained by all of the raised fields in the group of fields. We can now compare this theoretical expectation with the observed raised fields in the study area.
In the Iruyañez study area, raised fields are distributed continuously across the landscape, which makes it difficult to choose an appropriate unit of analysis. We are assuming today that each individual raised field platform can be measured as a discrete unit. Individual fields can be used as units of analysis because the placement of fields and the canals between them suggests that fields were built individually, First, the examination of parallel fields shows that the distance between platforms is quite variable. This suggests that fields were not placed on the landscape in order to maximize the area of land under cultivation. Second, the existence of many empty spaces within groups of fields, as well as the existence of irregularly shaped spaces, suggests that they were not placed on the landscape in a uniform manner. Overall, their placement supports the contention that they were built individually.
A lack of earthworks connecting raised fields to one another is further evidence for their individual construction. This absence of any larger agricultural infrastructure is a distinctive aspect of this part of Moxos. There are no long canals or causeways, which would have required higher inputs of labor, and would have affected more than one field. Though fields are usually located within 30 or 40 meters of one another, they do not depend on one another for water or for drainage.
In addition, there is no evidence for any mechanism by which farmers living upstream could have controlled the water supply of farmers downstream. There are no structures that impound water, and no structures that improve drainage for more than one field. Though it would have been easy to restrict drainage by building connecting earthworks between fields, thereby closing off canals, no evidence of such constructions exists. The system required little or no water management, and communal decision-making was not necessary for its everyday functioning. These fields are particularly amenable to analysis, because they are all of a single type, and they are of a size that makes them easy to measure.
Returning now to the argument, I suggest that the farmers who built the raised fields organized those individual fields into groups. Now that we have a theoretical estimate for the size of these groups, we can take a map and see if we can systematically sort fields into groups.
Individual fields are placed into groups on the basis of two criteria. First, all of the members must share approximately the same orientation, within a range of 25 degrees. Second, the fields in a group must be contiguous, within 30 or 40 meters of one another. Note that raised fields arranged end-to-end are not considered contiguous. Following these two criteria, I took a sample of 30 groups of raised fields from aerial photographs, and compiled the measurements of their individual member fields (Fig. 16).
Though it was possible to obtain this sample with relative ease using these criteria, the arbitrary nature of this process needs to be emphasized. In contrast to individual fields, which are clearly defined on the ground and in aerial photographs, the definition of raised field groups is less precise.
The 30 groups in the sample represent a total carrying capacity of just under 1,000 people, with about 55 people required to build the largest field. The groups have an average area of just less than 5 hectares. Taking the measured total area of each group and combining it with the estimate of raised field carrying capacity, the groups have an average carrying capacity of 33 people. The standard deviation, however, equals 32 people, with values ranging between 4. and 116 people. The estimates of the number of workers required to build the largest field in the group range from 8 to 55 people. Some groups, such as field group number 6, could have supported enough people to construct any raised field in a short period of time. Other groups could have supported people who probably cooperated with others to build more raised fields. There is great variation in the size of the groups, just as there is great variation in the size of the individual fields.
It is important to remember that this analysis does not claim to describe precisely the social groups that existed in the past. There are too many unknown factors for the results to be interpreted as anything more than estimates of the size of actual groups of people. However, the range of results from the production and labor estimates, combined with the spatial patterns of the raised fields, is illuminating. The evidence suggests that social units of between 20 and 100 people built raised fields in contiguous groups that corresponded to prehispanic social units. Though these hypothetical social groups varied widely in terms of size, they were large enough to build an average raised field in a single event, a single “work party.”
It is difficult to know more about these social groups from the evidence that we now have. However, the ethnographic and historical record from South America provides some interesting analogies. We know that in the highlands to the west, social groups called ayllus were an important part of agricultural, political, and religious life at the time of the Spanish Conquest. To the east, along the Xingu River in Brazil, there are groups of people who have a social organization much more complex than might be expected on the basis of the size of their population. Some anthropologists have suggested that there may be links between the social organization of highland and lowland peoples. The archaeological and ethnohistoric evidence from the Llanos de Moxos could be an important part of this discussion.
This analysis of the agricultural system suggests that it did not require the coordination of large groups of people. In addition to these archaeological implications, the data suggest that the raised fields present in Moxos today could be renovated, or even built in new locations with the labor that could be coordinated by a single community. Raised fields might be part of an effective strategy for agricultural development, and the people of eastern Bolivia will make this determination.
This research would not have been possible without the friendship and support of the people of Santa Ana del Yacuma and Exaltacion. It is part of the Agro-Archaeological Project of the Beni, which has been working in eastern Bolivia since 1990, under the direction of Clark Erickson. Support from the National Science Foundation, the Wenner-Gren Foundation, and TIE Fuibright is gratefully acknowledged. A version of this article was originally presented as a paper at the 2000 Midwest Conference on Amazonian and Andean Archaeology and Ethnohistory in Fort Wayne. I would like to thank Jennifer Quick, Helen Schenck, Jim Mathieu, and two anonymous reviewers whose astute comments, including those not incorporated into the final version, were much appreciated.