Perspectives on Early Amazonian Agriculture from Guyana and Venezuela

Abstract

Debates about prehistoric Amazonian subsistence, social organization, and landscape use have ranged from ethnographically based characterizations of relatively mobile foragers to archaeological evidence that suggests sedentary agriculturalists. Recently, great antiquity of settled agriculture and complex social organization has been asserted for portions of the northern Amazon. However, the role of theory and inferences drawn from habitat diversity have receded in these debates. This paper synthesizes the current literature regarding long-term evolutionary changes in Amazon Basin societies with an archaeological case study from mound-building sites of Guyana and ethnoarchaeological data from Venezuelan forager–gardeners to develop a hypothesis regarding the critical role of habitat, aquatic resources, and seasonality in the transition from intensified foraging to cultivation.

1. Introduction: Counterfeit Paradise or Garden Cities?

The ways that researchers define the habitats of the Amazon Basin have had a direct influence on archaeological research into the origins and evolution of agriculture. For decades, scientists described the Amazonian forest habitat as impoverished in soil nutrients and sunlight and therefore scanty in animal and plant resources, arguing that dispersed, mobile human populations who were described in historical and ethnographic accounts represented the pre-colonial lifestyle of Amazonia [1,2,3,4,5,6,7,8,9]. Meggers characterized the lushness of vegetation yet sparsity of obvious food sources in the Amazon Basin as a “counterfeit paradise” [5] that curtailed societal development and limited human groups to slash-and-burn cultivation supplemented by foraging. This Malthusian perspective of ecosystem limitations and opportunities is likely due at least in part to taphonomic bias due to the lack of obvious archaeological indicators of abundant food sources or dense populations (sensu [10]). The purported stressor of scarce resources as a unilineal driver for human settlement and social structure has generally been termed “environmental determinism” [7,11] or the “Standard Model” of Amazon agriculture ([12] (p. 10) after [13,14]).

However, since the 1980s, archaeological discoveries of earthwork features (palisades, geoglyphs, berms, moats, and islands), improved methods for discerning anthropogenic soils termed terra preta or Amazon Dark Earths (hereafter, ADE) [15,16,17], and remote sensing approaches using satellite imagery, LiDAR, and other techniques to discern features [18,19] have formed the basis for alternative arguments that late pre-contact Amazonian populations were dense, sedentary, and supported by intensive cultivation of maize (Zea mays spp.) on river floodplains [20,21,22] and root crops (Manihot esculenta) in the forested uplands [12,23,24,25,26,27]. This “New Model” [24] (p. 12) has been reinforced by ongoing discoveries and the rapid growth of publications that broadly characterize Amazonian populations as complex chiefdom-level societies or Edenic-sounding “garden cities” (sensu [28]; also see [29] for a summary).

However, the “New Model” shows a growing potential for confirmation bias in which the selective pursuit of evidence to re-affirm complex agricultural societies across multiple habitat types and into ever-earlier cultural stages has become a research objective in and of itself (also see Roosevelt’s discussion on the search for maize in the Archaic [11] (pp. 12–13)). Yet inferential arguments that are founded upon archaeological evidence require us to take into account site formation processes that may lead to equifinality [30]. For example, the presence of ADE soils and evidence for earthworks may arise from multiple pasts, of which only one might be a sedentary agricultural chiefdom. We agree with Roosevelt [31] (p. 31) that the results of careful and precise field and lab methodologies may be assessed through the application of independent but related frames of reference. These include evolutionary theory and reference information from ethnographies, which in turn open avenues to derive and re-assess working expectations about adoption of and investment in cultigens.

The aim of this paper is to develop a hypothesis regarding the inception and development of agriculture in the Amazon and implications for later complexity. To do this, we will first summarize the current state of knowledge on this topic and then examine an archaeological case study from early agriculture/earthwork sites in Guyana. This allows us to reconsider inferences about organizational complexity from the archaeological evidence and outline opportunities for investigation into nuanced relationships between diverse habitats, subsistence, and social organization. To address the lack of data on the impacts of seasonal variance upon on the diversity of subsistence and diet breadth, we will develop an ethnographically derived frame of reference using ethnoarchaeological data from Pumé forager–gardeners of central Venezuela.

2. Habitats, Subsistence, and Agriculture in the Amazon

To establish a foundation for descriptive case studies, we first examine the current state of knowledge about the conditioning effects of habitat variability on subsistence options and innovations of foragers, horticulturists, and early farmers (sensu [32]). By 1999 Roosevelt [33] had synthesized the significance of habitat diversity, resource type, and specifically aquatic resources in her chronology of Amazonian subsistence and settlement and continues to incorporate updated information [11,34]. Paleoenvironmental evidence indicates that Amazonian habitats have been mosaics for millennia, predicated on elevation, proximity to river floodplains, local soil characteristics, local climate, and seasonality [33,34,35]. The alluvial floodplains, or várzea, make up the largest area of good-quality, well-watered soils, a significant resource comprising about 25% of the area of Greater Amazonia [33] (p. 372). Upstream, flooded savanna habitats (cerrado in Brazil, llanos in Venezuela) are crossed by smaller drainages that are in turn lined by gallery forests. The timing of rainy and dry seasons varies from one area to the other in the Amazon: in the eastern and southeastern areas, the rainy season runs from December through May, whereas in the northwestern part of the region it is the inverse. Therefore, when the northern tributaries of the Amazon are running low, the southern ones are providing more water during these periods.

Várzea lands are flooded by nutrient-rich whitewater from the volcanic sediment of the Andes Mountains and along the Amazon River itself, in contrast to igapó lands flooded by nutrient-poor blackwater rivers, such as the Xingu, the Tapajos, and the Rio Negro. Várzea forests are defined by their position along the floodplain, where seasonal flooding reaches 4–7 m in depth. These regions are productive habitats for many forest and aquatic bird species, as well as caimans, river dolphins, manatees, large turtles and fish, and anacondas. In upriver areas, herbaceous-dominated lowlands (hereafter called savannas, although there are some differences from typical savanna habitats) are maintained by constant inundation and, recently, anthropogenic fire and grazing (Figure 1).

Sediments from floodwaters enrich soil nutrients, and floodplains have traditionally been some of the most intensively farmed areas of the Amazon Basin. By contrast, at middle elevations the intermittently flooded wet forests are dominated by palms including Mauritia flexuosa and Euterpe oleracea, while in the upland forests (hereafter, terra firme) the well-drained and less nutrient-rich soils are suited for a variety of hardwood tree species. Birds and terrestrial prey predominate and move seasonally among resource patches.

This mosaic character of Amazon habitats has provided a richly varied adaptive theater for human subsistence and settlement from the time of initial Pleistocene colonization forward [11]. Paleoindian (c. 11–8 kyr) foraging emphasized wild tree foods and várzea species such as fish, turtles, and aquatic mammals [11,33,34,35,36] in a largely forested landscape. At about 8000 BP the Middle Holocene/Archaic culture stage saw an overall reduction in tree foods, growing diet breadth, and divergence of subsistence and settlement into upland and riverine lowland adaptations. In the uplands the Middle Holocene dietary spectrum expanded to include an array of small terrestrial prey [33], and the absence of large sites and pottery might indicate high mobility. In the riverine várzea, sites grew larger, and the first pottery appeared. Várzea communities were supported by a strong focus on aquatic resources that is indicated by sizeable shell middens. Significantly, this means that the earliest known Amazonian villages, including the development of the first pottery, were fully supported by wild aquatic resources [11,20,21,37,38,39].

By about 5500–4000 BP, cultivation of manioc, gourds, and other crops augmented by agroforestry was well underway [20,26,34,40,41,42]. Dependence on vegeculture crops co-occurred with a diverse diet of fish, birds, and terrestrial game [33]. Between 4000 and 1500 BP, pottery became adopted in the terra firme uplands. Maize first appears in the archaeological record in flooded lowlands, and anthropogenic changes in forest composition become well-developed. At ca. 1500 BP, evidence for moderate maize cultivation, large-scale earthworks including mounds, and large ceramics appeared near the mouth of the Amazon, the Bolivian Amazon, and the Orinoco Basin [33,34,43]. Roosevelt posits that evidence for complexity at the chiefdom level is not present until late in the first millennium AD [33] (pp. 342–343). The small size of fish remains is cited as evidence of intensive localized fishing that possibly led to over-harvest [33]. Wild foods including arboriculture continued to be important to subsistence [10].

During the late prehistoric period (c. 1500–500 BP), settlements and mounds proliferated in large alluvial floodplains or areas of rich upland soils. Intensive maize cultivation had a “brief reign” [11] (p. 25) as implicated by macrobotanicals, pollens, and C4 stable isotopes in human bone [34,44]. Other crops also appear to have been important, sometimes more than maize in some areas. Growing population densities fueled intensity of cultivation and exploitation of river and forest resources, with deforestation occurring around large settlements [18,36,40,44,45]. Late prehistoric earthworks and anthropogenic enrichment of soil by human refuse became widespread, with significant and lasting influence on landforms and soil quality. Concurrently, warlike political complexes and supra-regional trade expanded in the region. The importance of fish and other aquatic foods continued throughout this period, as evidenced by remains of piranhas, turtles, rays, and shellfish [33], and foraged wild plants may also have reduced the risk of malnutrition from over-reliance on maize [11].

Overall, the 12,000-year Amazon regional sequence shows interesting patterning in human–environment interactions that is germane to the onset and intensification of agriculture. Since the terminal Pleistocene, human groups have used different major habitat types—the nutrient-rich várzea, forested uplands, and intermediary savannas and wet forests—in different yet complementary ways that influence the origins and evolutionary pace of agricultural lifeways (

Table 1. Summary of várzea and terra firme prehistoric agricultural sequences (after [11,33]).

Cultural Phenomenon	Várzea	Terra Firme
Predominant prey types	Fish, shellfish, turtles, other aquatic species	Broad spectrum: small-bodied terrestrial prey, some aquatic
Population distribution/density	Rapid increases; densely occupied by c. 1000 BP *	Increases delayed; more dispersed and scantier than várzea
Mobility pattern	Focused on river floodplain; semi-sedentary then mostly sedentary	Frequent mobility followed by long-term semi-sedentism
Emergence of societal complexity	c. 4000 BP to 1000 BP; chiefdom level inferred by c. 1000 BP	Delayed till c. 1000–500 BP
Cultigen types	Arboriculture first and then tubers, followed by maize. Field agriculture before 1000 BP	Primarily tubers. Field agriculture develops late (c. 500 BP)
Ceramics and storage features	Earliest ceramics c. 8000 BP	Ceramics emerge c. 4000 BP
Earthworks	Earliest earthworks c. 1600 BP	ND

).

Important considerations from large-scale temporo-spatial patterning include the following:

• Resource structure and distribution varied strongly according to properties of local habitats such as the várzea and terra firme.
• Sedentism, storage, and ceramics in the várzea pre-date agriculture and were supported by intensive reliance on aquatic foods.
• Maize was introduced early but took centuries to become a staple and mostly in várzea habitats.
• Aquatic prey and foraged terrestrial foods retained importance alongside agriculture including intensive maize farming in the lowlands.
• Wild foods and mobility in the terra firme retained high adaptive value after agriculture.
• Human activities fundamentally altered ecosystems and therefore adaptive options for human populations.

There is a causal and predictable relationship between the availability of aquatic resources and human population density in the Amazon Basin. The changes in mobility in turn facilitate sedentism, the early onset of vegeculture and arboriculture, earthworks, the eventual adoption of seed crops, and growing societal complexity. In upland habitats where aquatic resources were scanty and/or distant and human populations more mobile, these processes were delayed and/or muted.

3. An Archaeological Case Study in Agriculture from Guyana

3.1. Climate, Environment, Settlement, and Subsistence

It is reasonable to expect that habitat types that combine the characteristics of the two main ecosystems would exhibit an intermediate combination of cultural traits and rates of agricultural intensification. In the case of Guyana’s Berbice region, the area of Dubulay and associated areas have moderate-sized rivers, and the coast is a short distance away. As noted above, part-time cultivation, continued reliance on aquatic resources, and semi-sedentism comprised of redundant site occupations over long periods are consistent with habitation mounds and other earthworks that are well documented in the Guianas [46,47].

As observed by Rivière [48], subsistence regimes of the Amerindian populations in Guyana are socially autonomous groups. Forested regions are inhabited by Arawak-Carib populations living in small villages along major river courses [14,20,48,49,50]. Villages consist of several households that garden, hunt, and fish with large manioc fields situated within relatively short distances of structures. In the south savannahs, villages are less aggregated. Seasonal flooding of savannas finds freshwater fish trapped in shallow ponds during the dry season.

Here we review the archaeology of Dubulay where recent investigations argue for the development of agriculture and social complexities in an age range of 6130–3280 BP [51]. Assessing the data in support of these claims, we offer an alternative view using reference to the lifeways of Pumé forager–gardeners to evaluate the premise that the use of aquatic prey may have fostered redundant and long-term semi-sedentary settlement of the region, and mounds and raised fields are associated with an increasing use of domesticates as a fallback to food uncertainty associated with greater agricultural reliance.

3.2. Archaeology and Ethnography of Northeastern Guyana

Two main cultural patterns characterize the prehistoric archaeology of the Berbice area. The Abary Phase [46,48] is found west of the Berbice River and dates to around A.D. 1200 with the occurrence of Mabaruma Phase pottery (Figure 2). Evans and Meggers [48] identified additional ceramic types which include the Tiger Island Plain, Tuarakuli Plain, and Abary Plain. The Tiger Island Plain and Tuarakuli Plain are Cariapé tempered while the Abary Plain is sand tempered, and the Tiger Island Plain is the predominant early type. Abary Phase pottery exhibits some incision and modeling using nubbins with punctuates. Groundstone tools include axes, adzes, hammerstones, manos, metates, and rubbing stones. Excavations at Recht-Door-Zee [52,53] suggest possible Koriabo influences and the use of wattle and daub house construction.

The Hertenrits complex exhibits similarities to the Arauquinoid and is a pattern based on excavations in Suriname [54,55]. It appears to date somewhat earlier than the Abary pattern, dating to A.D. 600 at the Buckleburg Mound in Suriname [54]. Around A.D. 1200, Koriabo culture is seen to have expanded across the Guianas [56]. Recent work in the vicinity of Dubulay in the Berbice country has documented ceramics dating as early as 6270 B.P. [34,51]. In the East Brazilian Amazon, excavations at Stilt Village document an Incised-Punctate/Arauquinoid ceramic tradition appearing to be as old as that in Guianas [57,58].

The age of Abary Phase sites as noted is based on the presence of Mabaruma pottery (steatite-tempered Hotokwai Plain) and the presence of Akawabi and Aruka incised and modeled pottery. Subsistence known from both archaeological and ethnographic sources is based on utilization of resources reflecting both coastal and interior riverine environments (see [39]). Abary and Hertenrits Phase settlements varied seasonally in their use of sand reefs, ridges, and artificial habitation mounds (see [37,46,52,55,59,60,61,62,63,64]).

3.3. Habitation Mounds and Raised Fields

In northeastern Guyana and Suriname, artificial habitation mounds and raised fields are present [46] and may be associated with the Abary and Hertenrits patterns. The Abary sites have been noted by several investigators [64,65]. Habitation mounds of the Hertenrits pattern are well documented by the Buckleburg 1 and Wageningen-1 mounds reported east of the Cortenyne River in Suriname [54,55,64]. Here, many of the mounds rise above two meters and are encircled by moat-like features measuring 20–100 m in width (see [14] (p. 326)).

What has been deemed a habitation mound in Guyana, the Joanna Mound, is located on the Canje River and contains pottery types common in late Hertenrits sites [66]. The presence of wattle and daub construction has been noted as at other sites in Guyana [37].

3.4. The Archaeology of Dubulay

Dubulay is located east of the Berbice River, where it is near historic Fort Nassau. It contains raised fields similar to those described by Parsons and Deneven [67]. Features of this type were first described by Whitehead and Simon [68] who reported on 787 raised fields. The fields have average heights rising 1–2 m and lengths that extend to 8 m. The features are situated in linear arrangements and lack material associations. Raised fields suggest that population aggregates may have existed in the region. This reflects the growing acceptance by regional specialists that anthropogenic forests and large-scale management practices were common in several parts of the Amazon Basin in later prehistoric times [16,18,46,49,61,69,70,71,72].

In 2009 the University of Florida conducted a pilot assessment of the Fort Nassau area [62]. This was followed by additional fieldwork in 2011 and 2014 [73]. Large-scale excavations were conducted in the vicinity of Dubulay Ranch near the Berbice River. Two major loci were identified. Locus 1 includes an area of c. 200 × 60 m that includes what is referred to as the large Dubulay mound (75 × 200 m), a highly stratified mid-Holocene context that contained substantial ADE. This configuration extends upward from a relatively flat area toward the adjacent ridge line above the Berbice River. Dates range from 6130 to 3280 B.P. (two age ranges, midpoint). A large ceramic assemblage (>10,000 sherds) was analyzed and contains some decorated wares. Excavations provided for the identification of a Dubali Complex that consisted in two ceramic phases. Dubali 1 is associated with single-coil appliqué, contrasting with Dubali II characterized by double-coil appliqués having complex crosshatch and fretwork designs [51,74]. The latter designs called fretware by Williams [39] have been found in Kaurikreek in Suriname [46,64].

Locus 2 is a ceramic cache of eight vessels. A third locus is a presumed public area extending between Locus 1 and Locus 2. Test excavations were also conducted at the site of Hitia, which is located to the northeast of Dubulay. Additional test excavations north of the auger locations within Locus 1 were conducted by the University of Guyana in 2019. These confirmed the presence of ADE to a varied depth of 50–60 cm below the surface. The sediment contained extensive undecorated ceramic sherds intermixed with some colonial period material. The record of Locus 1 has been interpreted by Shearn, Heckenberger, and Simon as follows: “population increase led to an elaboration of routine agricultural practices, which took on additional ritual and social meaning as simultaneously, the growing population became more dependent upon the success of a harvest”, as a way to explain ceramic innovation and design [51] (p. 16). In Locus 2, the ceramic cache is interpreted as a “ritual cache” relating to agricultural fertility that represents an analogous form of agricultural innovation to the mounds [75].

The desire to see Dubulay as evidence of another complex development is represented by the presumed existence of a “public area” situated between Locus 1 and Locus 2. These claims are unsubstantiated and reflect the all-too-common simplicity of thinking regarding the nature of Amazonian adaptations. It remains easy to broadly interpret the presence of ongoing occupations associated with ADE deposits along with decorated ceramics as meaningful to the understanding of the origins of agricultural activity, but this approach continues the archaeological dispositions of 1950s culture historians. As has been and continues to be common within the region, many archaeologists view the appearance of new, particularly decorated wares as indicative of the migrations of ethnic populations [46,55,75,76,77].

3.5. Costs and Productivity of Agricultural Field Construction

An important issue regarding the emergence of agriculture in the Berbice area is the use of ridged fields and ADE as proxies for agricultural development, particularly as it is tied to a specific period of use. In the case of Dubulay, a pattern dating as early as c. 6000 BP establishes the date for agricultural prominence earlier than elsewhere in Guianas. Roosevelt [11] (p. 13) notes that many Formative/Horticultural period sites have inadequate sample collections that hamper solid conclusions regarding taxa, a problem made worse by data recovery from disturbed contexts and the use of arbitrary levels of excavation. At present, there is an absence of reporting on macrobotanical data, much less pollen and starch data. Assuming that ridged fields do signify a reliance on domesticated plants, no data exist that would allow us to determine when most of the fields were constructed and over what intervals.

Further, there have been no attempts to examine the productivity of the fields that vary by construction type or environmental setting (lowland vs. upland settings). Erickson’s [78] experimental work in raised field constructions in Bolivia estimated 900 person-days were required to construct fields and canals within a single hectare, which requires careful thinking about the circumstances that would facilitate time investments in the construction and maintenance of ridged fields, though to some extent determined by environmental context. The tendency has been to associate increased population with numbers of fields raised and ADE. While the known extent of such fields may well be associated with periods of greater sedentism, the question remains whether time investments in field construction/maintenance and crop production have been the most optimal exploitation of resources. Assessing the length of time required to meet the dietary requirements of a large population needs to be determined and must take into account the maturation rates of cassava and other domesticated crops. If population size remained relatively stable over lengthy periods of time, this is based in large part upon productive fisheries, and the fields that were no doubt constructed over a period would be supplemental to fishing productivity. In this way, fields were constructed over extended periods and would have fallen short of the dietary requirements of increasing and more sedentary populations.

The issue here is not with the presence of ADEs, raised fields, or even mounds but how the site area is interpreted. Though recognition of Amazonian people having innovatively altered and built upon natural landscapes is important, it must be understood in terms of the multiplicity of factors that have undoubtedly changed over time and varied by region [29,49]. At Dubulay the use of the term “mound” is itself deceptive, as it implies intentional construction. The area as noted is an open and relatively flat area that inclines toward an upper ridge line above the Berbice River (Figure 3), where occupation appears to have been most intense, a pattern common in the Amazon and in this area [12,24,48]. Importantly, Dubulay provides no real archaeological evidence of population increase—statements regarding population increase are little more than assertions. In fact, the short intervals of occupation documented within Locus 1 might be interpreted as comings and goings at this location. Shearn [74] and Shearn et al. [51] presume that the early residents of Dubulay were most probably semi-sedentary groups but then argue for a more permanent occupation associated with intensive agriculture—an assertion relating to the transition that they make but cannot (yet) demonstrate.

At present there is no botanical evidence suggesting the use of domesticated crops at Dubulay, although this has been the presumption [25,75]. Equally problematic is the absence of faunal data. Though local conditions in the region may influence levels of preservation, as with botanical remains, excluding consideration of the importance of animals in the diet breadth remains all too common even if terrestrial mammals do not predominate. This is particularly true regarding the importance of fish and other aquatic prey. Gragson [79,80] long ago demonstrated the importance of fish to the diet of Amazonian peoples in his work with the Pumé, where he demonstrated that 17% of total kcals and 64% of all protein annually was provided by fish (whereas terrestrial game provided 19% of the total). Gragson [79] suggests that fish appear to be more dependable than game, as fish are significantly more abundant in time and productive over time than game (see also [81]).

The Guyana case study indicates that interpretations of archaeological evidence can become circular: ADEs, ridged fields, ceramics, and storage evidence, mounds, and macrobotanical remains are cited as irrefutable evidence for early agricultural-based complexity, and then the search is on for additional “proof”. Yet the Guyanese evidence from the early Formative period is also consistent with a long-term pattern of seasonal semi-sedentism, re-occupation, and gradual build-up of significant locations, based on a mixed economy of fish, crops cultivated in a shifting agricultural pattern, and wild resources. Further, the naïve presumption that agriculture is always and necessarily predicated solely upon the use of domesticated plants continues to constrain our thinking about the varied strategies that may have supported population growth and increased sedentism, while falling short of fostering the emergence of complex polities. Recent comparative global studies suggest the need to assess the use of raised fields in a broader context [82].

4. Ethnoarchaeological Frames of Reference for Foraging, Fishing, and Cultivation: The Pumé of Venezuela

In addition to spatial diversity in habitats, seasonal fluctuations in rainfall and river flows of the Amazon and environs were an important factor in the onset and development of cultivation. Reference information from ethnographically known societies can enrich working expectations about seasonal conditions that influenced diverse pathways of early agriculture in the Amazon. Warnings about using ethnographic information as direct analogies for the past [35,83] are well taken, but the use of relevant but independent areas of knowledge allows us to develop alternative explanations for patterning in archaeological data [30,84,85,86]. If relationships between linked variables can anticipate patterning in the archaeological record, certain characteristics of hunter–gatherers can serve as frames of reference for investigating the material remains of the foraging–cultivation interface [30].

4.1. Climate, Environment, Settlement, and Subsistence

The Pumé (formerly, Yaruro) are an indigenous ethnic group (c. 9400 persons as of 2011 [87]) with traditional territory in the neotropical savannas of south–central Venezuela (Figure 4). Historically, Pumé communities occupied two major ecosystems: riverine, which was focused on aquatic resources and focused camp locations on or near riverbanks, and savanna or llanos, in which terrestrial game fostered higher mobility in the grasslands and stabilized dunes of the interior [79,80,88,89,90,91,92]. Savanna Pumé people occupy a flooded savanna and gallery forest ecotone, subsist mainly on wild resources, practice manioc–tobacco–hallucinogen horticulture, and are currently experiencing societal pressure to adopt sedentary cultivation of maize. Among these mobile communities, large wet season camps of c. 60 persons disperse to smaller mobile dry season camps of c. 10–15 persons. Savanna Pumé people are somewhat less acculturated than river Pumé people and still rely primarily on wild resources taken using traditional methods and technologies. In the Venezuelan llanos as with the Amazon Basin, aquatic and terrestrial productivity varies according to the wet versus dry season. Seasonal variance in Pumé subsistence tactics is discussed below.

4.2. Lessons from Pumé Subsistence: The Wet Season

Seasonal transitions are times of scarce and unpredictable resources, and the wet season onset marks a time of acute nutritional stress among the Pumé [86,88]. December marks the beginning of heavy rains, and people aggregate in large wet season camps positioned on small dunes. In rivers and tributaries, an influx of oxygen and nutrients brings a drop in water temperature. In the larger drainages, many fish migrate inland from the ocean to access increased foraging opportunities. In the smaller drainages, dry season ponds fill and then connect, leading to annual synchronous mass fish spawning events. As the wet season proceeds from January to May, aquatic prey disperses throughout the flood landscape and becomes less predictable and harder to encounter. Terrestrial game becomes constricted on reduced dry land, improving human ability to intercept them [79,80,88,93,94]; thus, fishing is reduced and hunting increases.

However, the Pumé’s major wet season dietary staple is wild plants, primarily tubers. Wild roots grow large, nutritious, and tasty, and wild fruit comes into season. In the 1992–1993 wet seasons, Doro Aná Pumé women typically provided over half the community’s food by weight. Daily returns of up to 35 kg of wild roots and other plant foods were common [85,88,93,95]. Post-reproductive age Pumé women were particularly busy, visiting multiple patches per trip and making several trips per week. One Pumé tactic of plant intensification is vegetatively propagating wild tubers and corms by re-planting stem sections after harvest (see Figure 5). While gathering, women often direct family members to weed around wild food plants to reduce competition for sunlight and soil nutrients (Figure 5). Other measures include opening up the canopy with low-burning fires or propagating wild species near natural openings or human-made ones like manioc gardens [85,88,93,95]. Easy access to cultivated and wild plant species in one spot is clearly desirable, as this reduces opportunity costs to the foraging lifestyle. This is another argument in favor of the ecotonal interface between várzea and terra firme as a likely zone of incipient agriculture [86].

The Pumé people also cultivate small amounts of “bitter” manioc, which has larger roots during the wet season but can be collected during the dry season as well. This strain of manioc contains high levels of cyanic acid and must be extensively processed (peeled, grated, squeezed of excess liquids, and heated) to make it fit for human consumption. Pumé gardens are left alone for lengthy periods seasonally, indicating that the opportunity cost of cultivars has been reduced by the increasing productivity of early domesticates and preserving useful adaptive characteristics (such as toxicity/resistance to disease and predation, drought tolerance, ability to flourish in local soils, etc.; see also [86]). Bitter manioc fits the bill in that toxicity renders it safe from predation, and the plant is well-adapted to local micro-organisms, soils, and water availability. In this way, indigenous toxic cultivars are managed in a balanced state by humans, to mutual advantage. Toxicity that can be eliminated or neutralized by cultural practices is compatible with the mobility needed to access other resources (for heavy reliance upon non-native, non-toxic cultivars, sedentism may be a pre-requisite: foraging peoples of Australia and Pacific Northwestern North America fully adopted wheat flour and potatoes only after enforced settlement [96,97]).

The Pumé subsistence strategy of combining wild and cultivated tubers has been in place since at least the time of initial European contact [87]. Greaves and Kramer [95] explore the persistence of wild root foraging among savanna Pumé as a strategic decision based on considerations of energy expenditure and return rates [95] (p. 264). They assessed the expectation that the incorporation of manioc into South American hunter–gatherer diets may indicate one of two phenomena: a transition to horticulture or the incorporation of a complementary resource to support the foraging economy. In energetic terms, if cultivated foods are more nutritious or offer labor savings in search or handling costs (or have a trade value), they are expected to replace wild foods. Alternatively, if cultivated foods do not offer these advantages, hunters and gatherers may either ignore them or incorporate them into the diet as complementary or fallback foods [88] (p. 265).

When the return rates in kg/hour were compared between wild tubers and domesticated manioc, Greaves and Kramer found little statistical difference [95] (p. 266). However, when garden preparation and maintenance and de-toxification procedures were factored in, cultivated manioc dropped below wild tubers in return rates. In the Pumé case, there is selective pressure for a mixed strategy: Greaves and Kramer determined that simply expanding foraging ranges to increase wild root returns may not be as economical as combining foraging with minimal cultivation of local patches of manioc. In this way the savanna Pumé promote plant productivity of nearby locations where roots are not otherwise available and increase dietary diversity during the wet season [88,89].

These interesting data from the Pumé are referable to early Amazonian agriculture. Cultivated tubers would have supplemented wild roots and terrestrial game in the terra firme uplands where high-ranked aquatic resources were distant, and tuber crops would be a useful fallback option in the várzea during wet season when aquatic prey were harder to procure. Thus, small-scale cultivation to enhance local plant-based productivity is feasible in both major habitat types, with year-round utility in the uplands and higher adaptive value during the middle to late wet season. It is worth noting that the savanna Pumé people have been strongly encouraged to become sedentary and cultivate maize but have thus far strongly resisted it due to stated opportunity costs to mobility and desired wild resources [93].

4.3. Lessons from Pumé Subsistence: The Dry Season

With the arrival of the dry season in May, wild plant foods shrivel and lose their flavor, and Pumé women turn their work effort toward technological pursuits like raw material gathering, weaving, and carving [86,88,93]. The early dry season is the time for manioc field preparation and maintenance tasks. Large wet season camps located on local dunes break up dispersed small dry season settlements located along drainages. Pumé mobility increases, and group sizes decrease.

As water levels drop, many larger fish emigrate to the ocean to breed, and resident species become hyper-concentrated when smaller tributaries shrink into chains of ponds. Fish, turtles, eels, and the occasional mammal become predictably concentrated in time and space, which reduces search and transport costs [88,89,98]. The Pumé people use bow and arrow (Figure 6), hook and line, traps, and piscicide. The latter is highly effective in pond settings or after small dams have been built; the Pumé people pound and swish the roots of Tephrosia sinapou (bi in Pumé), which blocks oxygen intake. Fish are scooped by the basketful in a festive mood by the whole community [79,88,93]. Gragson [79,80] and Greaves [88] documented that aquatic prey constitutes a major nutritional source annually, providing the bulk of protein especially during the dry season.

The Pumé dry season subsistence pattern also has interesting implications for the early cultivation in the Amazonian várzea, suggesting that forager intensification of aquatic prey and de-emphasis on tubers (wild or domesticated) likely occurred in the várzea and peripheral drainages during the mid-to-late dry season. Aquatic prey was favored primarily during the late dry season in the tributaries and the early wet season on the major rivers, where people aggregated to take advantage of seasonal fishing opportunities. Thus, during the early phases of agriculture, it is expected that small-scale cultivation to enhance local plant-based productivity would not be indicated in the várzea during the mid-to-late dry seasons, due to high productivity of aquatic species.

In sum, the Pumé data offer reference information that suggests intensified foraging and initial experimentation with domesticants, particularly tubers, would have been more favored during seasonal transitions—especially the early wet season—when water levels rose, aquatic resources were dispersed and difficult to procure, and plant productivity was low but increasing. Resource shifts during the inter-seasons reduced predictability, increasing the desirability of a consistently reliable fallback like cultivated tubers (Figure 7).

5. Summary and Implications

The archaeological and paleoenvironmental record indicate that the mosaic of Amazon habitats and seasonal variance creates diverse foraging opportunities that have a deep past. This was foundational to subsistence decision making and the origins of agriculture in the Amazon Basin. The story is far from a straightforward march from “simple” foraging to complex agricultural chiefdoms. The archaeological case study from Guyana indicates that caution is warranted when making inferences about complexity from cultigens, earthworks, ceramics, and ADEs. A focus on aquatic prey could well have fostered redundant, long-term semi-sedentary settlement of the Guyanese savanna/rainforest ecotone, with resultant build-up of mounds and other settlement indicators. This settlement type is compatible with the adoption of domesticated crops, under conditions where a complementary and reliable dietary fallback is needed. The Pumé ethnoarchaeological data provide intriguing reference information about the important role of seasonal variation and conditions that would favor the adoption of crops: the earliest use of cultigens likely occurred during seasonal transitions in the ecotones between the várzea and terra firme uplands, just above the flood zone. In these locations the soil was productive but not flooded, and terrestrial and aquatic prey would be roughly equidistant.

Our working hypothesis is that the development of intensified agriculture in the Amazon Basin was predicated on periodic concentrations of people accessing seasonally productive aquatic resources in the várzea. Human groups were incentivized to position themselves in particular locations, increasing local population densities and constructing mounds and other earthworks to extend the seasons of utility in the várzea and the flooded savanna (also see [10] (p. 132)). Once populations became focused on this subsistence mode and grew denser and more sedentary, tuber crops would shift from fallback options to staples, and maize would become a useful complement.

As the sequence of sedentarization proceeded, farmers began to cultivate maize intensively as a laborious but productive complement to tubers and arboriculture. In the terra firme uplands, growing population density shifted tubers from fallback foods to reliable staples augmented by other terrestrial resources. We note that any reduction in the labor force needed for maize cultivation would necessitate a “downshift” to less intensive forms of subsistence, including tuber vegeculture and wild resources [11,48,86,99]. In a recent paper, Boone and Alsgaard [100] evaluated issues relating to intensification and surplus production in the development of social complexity with assessment of variance across coastal and terrestrial habitats. They posit that the coordinated labor and economics of scale define social complexity; thus, the evolutionary dividing line between foraging and the food production dividing line is over-rated. This aligns with our hypothesis.

To conclude, the archaeological case study from Guyana and ethnoarchaeological data on seasonal variance in Venezuelan forager–gardeners elucidate three main characteristics of ancient Amazon agriculture: (1) the importance of aquatic resources, (2) the need for fallback foods to expand diet breadth during inter-seasonal food stress, and (3) resulting habitat- and density-dependent variability in site re-occupation and sedentism. As pointed out by Roosevelt [34], the inference of these phenomena from archaeological remains is not straightforward; therefore, assertions about societal complexity and agriculture in the Amazon Basin should stem from improved sampling strategies, better methodologies for detection, diagnostics, and identification, and testable hypotheses that draw strategically from ethnographic and other actualistic data.

These findings have larger relevance. Roosevelt [34] asserts that Amazonian ecosystems are variable and dynamic and if allowed sufficient recovery can be resilient to disturbances, human and otherwise. The lack of irrefutable evidence for “garden cities” in all parts of the Amazon or very early in the chronological sequence is not a disappointing vacuum but rather an opportunity for scientifically rigorous understanding of the diverse, dynamic, and patterned interplay between traditional human societies, climates, and habitats [10]. The predictable variability in Amazonian subsistence from the várzea to terra firme and habitats in between reveals innovative adaptive strategies developed by tropical human societies over thousands of years to sustain organizational levels appropriate to local neotropical forested habitats, resources, and social conditions.