Pre-Columbian raised-field agriculture and land use in the Bolivian Amazon

Abstract

We present an integrated palaeoecological and archaeobotanical study of pre-Columbian raised-field agriculture in the Llanos de Moxos, a vast seasonally inundated forest–savanna mosaic in the Bolivian Amazon. Phytoliths from excavated raised-field soil units, together with pollen and charcoal in sediment cores from two oxbow lakes, were analysed to provide a history of land use and agriculture at the El Cerro raised-field site. The construction of raised fields involved the removal of savanna trees, and gallery forest was cleared from the area by ad 310. Despite the low fertility of Llanos de Moxos soils, we determined that pre-Columbian raised-field agriculture sufficiently improved soil conditions for maize cultivation. Fire was used as a common management practice until ad 1300, at which point, the land-use strategy shifted towards less frequent burning of savannas and raised fields. Alongside a reduction in the use of fire, sweet potato cultivation and the exploitation of Inga fruits formed part of a mixed resource strategy from ad 1300 to 1450. The pre-Columbian impact on the landscape began to lessen around ad 1450, as shown by an increase in savanna trees and gallery forest. Although agriculture at the site began to decline prior to European arrival, the abandonment of raised fields was protracted, with evidence of sweet potato cultivation occurring as late as ad 1800.

Keywords

Amazonia phytoliths pollen pre-Columbian archaeology raised-field agriculture tropical palaeoenvironments

Introduction

Raised fields are prevalent features of pre-Columbian landscapes across the American tropics (Denevan, 1992, 2001; Gongard, 2006; Sluyter, 1994), and in South America, they are common to both high Andean (Biesboer et al., 1999; Carney et al., 1993; Erickson and Candler, 1989) and lowland regions (Denevan, 2001; Iriarte et al., 2012; Rostain, 2010). Increasing numbers of raised-field sites have been recognized across the Neotropics (Denevan, 2001) because the ancient modifications to soil conditions and micro-relief affect the modern vegetation growing on the raised fields, rendering them clearly identifiable in satellite and aerial imagery. Although pre-Columbian raised fields occur in a variety of forms, such as broad, shallow platforms (Lombardo et al., 2012; Walker, 2004, 2012), small, circular mounds of elevated soils, and platforms in drained wetlands (Denevan, 2001; Sluyter, 1994), they share the common function of improving drainage and soil conditions for the cultivation of food and industrial (e.g. construction materials, dyes, medicines) crops in seasonally inundated landscapes (Carney et al., 1993; Denevan, 2001).

Raised fields comprise a substantial proportion of the pre-Columbian earthworks that characterize the landscape of the Llanos de Moxos, a vast (130,000 km²) seasonally inundated forest–savanna mosaic in the Bolivian Amazon. The wealth of pre-Columbian earthworks in this region also includes habitation mounds, ring ditches, fish weirs, canals and causeways (Denevan, 2001; Erickson, 2000; Lombardo and Prümers, 2010; Lombardo et al., 2013; Mann, 2008; Mayle et al., 2007; Walker, 2004) that point to large, complex societies in pre-Columbian times. The distribution of these earthworks is influenced by the distinctive geo-ecological regions within the Llanos de Moxos (Lombardo et al., 2013). Raised fields dominate the seasonally flooded landscapes west of the Mamoré River, in the south of the region, and in the vicinity of Santa Ana del Yacuma towards the north (Figure 1a). There has been considerable effort in documenting the spatial patterning and geography of the pre-Columbian raised fields (Denevan, 1966; Lombardo et al., 2011, 2013; Walker, 2011), while archaeological enquiries have investigated the period of occupation and cultural artefacts associated with the pre-Columbian raised-field cultures (Walker, 2004, 2012).

Figure 1.

(a) Ecosystem map of northwest Bolivia showing the position of Llanos de Moxos seasonally inundated savannas (shaded white) and the location of the study region (black box); (b) the area surrounding the El Cerro site, indicating the position of the excavated raised-fields units and the two oxbow lakes (LEC: Laguna El Cerrito; LF: Laguna Frontera) cored for pollen and charcoal analyses and (c) detailed map showing the positions of the raised-field and canal units sampled for phytolith analysis (Camellón 1: (1) ridge (C1R), (2) canal (C1C), Camellón 2, (3) ridge (C2R) and (4) canal (C2C)). Also marked are the locations of four soil shovel test pits at the ends of each camellón.

There are large uncertainties, however, regarding raised-field construction, agriculture and land management. There is little conclusive evidence for when the fields were built (Denevan, 2001), but dated occupation layers from associated pre-Columbian settlements show that the areas in which raised fields are situated were inhabited by the 6th century ad (Walker, 2012). Crucially, however, it is unknown whether raised-field abandonment occurred after ad 1492, because of the introduction of new technology and the spread of Old World diseases, as is widely hypothesized (Denevan, 2001; Erickson, 2010; Walker, 2004). Furthermore, whether the management practices of pre-Columbian raised-field farmers involved extensive use of fire (Erickson and Balée, 2006), or its suppression (Iriarte et al., 2012), has direct relevance for both the impact of pre-Columbian land use on global atmospheric carbon emissions (Nevle et al., 2011) and understanding how past fire use has shaped the current savanna-forest mosaic of the Llanos de Moxos.

In addition, the agricultural strategy of pre-Columbian raised-field farmers is largely unknown. The available archaeobotanical evidence suggests that both food and industrial crops were grown in raised fields (Erickson, 1995), and it is also hypothesized that root crops, such as manioc and sweet potato, were cultivated because, compared with maize, they are more tolerant of the relatively poor soil conditions of the Llanos de Moxos (Denevan, 2001; Lombardo et al., 2011). Despite this paucity of archaeobotanical evidence, pre-Columbian raised-field agriculture is nonetheless considered by some as a potential model of sustainable cultivation for poverty alleviation (Renard et al., 2012; Saavedra, 2008), and several experimental sites have been established in an attempt to replicate this ancient form of agriculture in the Llanos de Moxos (Barba et al., 2003; Denevan, 2001; Erickson, 1995; Lombardo et al., 2012; Saavedra, 2008) and elsewhere (Gomez-Pompa, 1978; Muse and Quinteros, 1987; Sanders et al., 1979). Knowledge gained from understanding past agricultural strategies of the pre-Columbian raised-field cultures might provide the basis for improving land productivity in savannas and mitigating flood risk in seasonally inundated environments (Renard et al., 2012; Saavedra, 2008). The environmental and social benefits of raised-field agriculture are particularly pertinent to the Llanos de Moxos today, where unproductive savannas are entirely dominated by cattle ranching, and patches of tropical forest, overlying relatively fertile soils, are cleared for agriculture.

Aims

We present new palaeoenvironmental data from a raised-field site, El Cerro, located in the northwestern Llanos de Moxos, to investigate land use, agricultural strategy and environmental impact of a pre-Columbian raised-field culture. This study aims to understand: (1) how pre-Columbian farmers modified the savanna landscape for raised-field construction, (2) what agricultural and land management strategies were adopted when the raised fields were cultivated and (3) whether abandonment of raised-field agriculture coincided with European contact, as has been previously hypothesized. We employ an integrated palaeoecological and archaeobotanical approach (Mayle and Iriarte, 2013), combining pollen and charcoal analyses of lake sediment cores with phytolith analysis of raised-field soils, to provide a whole landscape perspective of pre-Columbian land use at a raised-field site.

Materials and methods

Study region and archaeological context

The El Cerro raised-field site is situated in the northwestern Llanos de Moxos, approximately 55 km north of the modern town of Santa Ana del Yacuma and west of the confluence of the Iruyañez River with the Mamoré River (Figure 1a). Natural savannas dominate this region and tropical forest occurs along river margins and in patches where local topography produces better drainage. Typical of the Llanos de Moxos, the climate is characterized by highly seasonal precipitation, the majority of which occurs from December to March in the austral summer (Pouilly and Beck, 2004), and during February, as much as half of the landscape lies under water because of poor drainage (Walker, 2004, 2012) caused by thick, hard-packed Quaternary clays underlying the region (Clapperton, 1993) and very low relief (Lombardo et al., 2013). Annual precipitation in the region of Santa Ana del Yacuma averages 1500 mm (Lombardo et al., 2013) and mean annual temperature is c. 25°C (Hanagarth, 1993).

The raised fields cover an estimated 10% of the 400-km² region in which they are located, including El Cerro (Lombardo, 2010; Walker, 2012). Raised fields in the vicinity of El Cerro are long and broad (c. 200 m × 50 m), and their low elevation platforms (<40 cm) were created with the soil dug from shallow canals (Lombardo et al., 2011). The El Cerro raised fields are closely associated (<500 m) with small (≥1 ha), regular polygonal platforms, which have been tentatively interpreted as pre-Columbian habitation sites, suggesting that pre-Columbian farmers lived close to the fields they cultivated (Lombardo, 2010; Walker, 2004). Unlike many examples of raised-field agriculture in the Neotropics, in which standing water between fields played an important role for raised-field production (Denevan, 2001; Sluyter, 1994), the El Cerro raised fields apparently did not involve water management technologies (Lombardo, 2010; Lombardo et al., 2011).

The abundant raised fields at the El Cerro site surround a highly visible rock formation, the most prominent feature of the landscape. The El Cerro rock formation, measuring c. 40 m in height and covering 10 ha in extent (Figure 1b), is now covered in closed-canopy forest and is surrounded by an additional 110 ha of forested high ground that does not flood during the wet season (Walker, 2004). The gallery forest of the Iruyañez River is the only other large forested area at the site (Figure 1b), although savanna trees have colonized some abandoned raised-field sites. Previous archaeological surveys and excavations (Walker, 2004, 2012) of the terra firme area surrounding the El Cerro rock formation have revealed a substantial anthrosol deposit measuring 500 m across and 120 cm deep, containing abundant charcoal and ceramic sherds, marking the site of a pre-Columbian settlement. It is estimated that 1000–2000 people could have produced the deposit of anthrosol (Walker, 2004). Eight radiocarbon dates on charcoal recovered from four discrete strata of the anthrosol confirm that this site was occupied in the 14th century ad (Walker, 2004, 2012). Currently, a large cattle ranch is located at the base of the El Cerro rock formation, and the savannas around the El Cerro site, including the now-abandoned raised fields, are predominantly used for cattle pasture.

Field sampling

Oxbow lakes

Sediment cores were analysed from two oxbow lakes in the El Cerro area (Figure 1b). This replicate sampling strategy was adopted to confirm that stratigraphic changes in sediment cores reflected landscape change at El Cerro and were not the product of purely local changes in the depositional environment associated with oxbow formation. Sediment cores were extracted from Laguna El Cerrito (LEC) and Laguna Frontera (LF), which are situated within the gallery forest of the Iruyañez River (Figure 1b) and located <500 m from the nearest discernible raised field. Cores were retrieved from a floating platform anchored in the centre of each lake and extracted using a 5-cm diameter drop-hammer Vohnout–Colinvaux modified Livingstone corer (Colinvaux et al., 1999). A Perspex tube and piston were used to collect the uppermost, unconsolidated sediments, which were extruded in the field in consecutive 0.5-cm slices. The Livingstone cores were shipped in their sealed aluminium tubes, extruded in the laboratory and stored in a cold room at 4°C.

Raised-field excavations

Two raised fields, currently located within open savanna (which is largely treeless due to cattle grazing), were selected for sampling from within a cluster that is situated northeast of the El Cerro rock formation (Figure 1b and c). A 25-m baseline was placed midway across each raised field and perpendicular to the orientation of the field (Figure 1c). Surface profiles were measured and drawn for each field. Units measuring 200 cm × 60 cm were excavated to a depth of at least 100 cm from the surface, and soil samples were taken every 10 cm. In addition to the excavation units, 50 cm × 50 cm shovel test pits were dug at both ends of the fields to a depth of 25–30 cm. Any differences in stratigraphy from the main centre excavation units were noted, and a soil sample was taken from 5 to 15 cm below surface.

Vegetation surveys

To aid the identification of pollen and phytolith types, rapid vegetation surveys were conducted on the raised-field complex (July 2010), and around LEC (July 2010) and LF (July 2011) by JDS. Taxa encountered in the survey areas were identified and recorded, and voucher specimens are housed in the herbarium of the ‘Noel Kempff Mercado’ Natural History Museum in Santa Cruz, Bolivia. Survey results are presented in Table 1.

Table 1.

List of species recorded in rapid vegetation surveys around the El Cerro site. Habitat codes refer to location in which a taxon was found: (1) canal; (2) camellón; (3) forest island; (4) gallery forest at LEC; (5) gallery forest at LF; (6) savanna, cerrado; (7) wooded savanna, cerradão; and (8) terra firme forest.

Family	Species	Habitat	Family	Species	Habitat
Annonaceae	Duguetia quitarensis	4	Melastomataceae	Desmocellis villosa	2
Annonaceae	Guatteria sp.	5	Melastomataceae	Tibouchina sp.	1, 2
Arecaceae	Bactris sp.	4	Menispermaceae	Abuta grandifolia	8
Asteraceae	Eupatorium sp.	5	Menispermaceae	Cissampelos pareira	3
Asteraceae	Vernonia sp.	1	Myrtaceae	Calyptranthes cf. bipennis	4
Celastraceae	Maytenus cf. erythrocarpa	5		Eugenia sp.	5
Chrysobalanaceae	Hirtella sp.	4		Eugenia cf. egensis	4
	Licania apetala	5		Myrcia sp.	5
	Licania cf. parvifolia	5		Myrcia cf. multiflora	2
Clusiaceae	Calophyllum brasiliense	5	Nyctaginaceae	Neea sp.	4
Clusiaceae	Garcinia brasiliensis	5	Nyctaginaceae	Neea spruceana	5
Combretaceae	Buchenavia cf. oxycarpa	5	Onagraceae	Ludwigia rigida	1
Combretaceae	Combretum laxum	5	Poaceae	Andropogon leucostachys	2
Cyperaceae	Rhynchospora barbata	2		Aristida sp.	2
	Rhynchospora emaciata	2		Aristida capillacea	2
	Rhynchospora sp.	2		Axonopus sp.	1
Dilleniaceae	Curatella americana	2		Axonopus barbigerus	2
	Davilla sp.	4		Axonopus cf. compresus	2
	Davilla cf. elliptica	5		Axonopus fissifolius	1, 2
Eriocaulaceae	Syngonanthus gracilis	2		Chusquea sp.	3
Erythroxylaceae	Erythroxylum anguifugum	4		Guadua cf. paniculata	5
Euphorbiaceae	Chamaesyce sp.	1		Panicum sp.	1
Fabaceae	Acosmium nitens	4, 6		Panicum stenodes	1, 2
	Crotalaria sp.	1, 2		Paspalum plicatulum	2
	Inga poeppigiana	4		Paspalum spp.	1, 2
	Mimosa cf. xanthocentra	5		Paspalum stellatum	1, 2
	Pterocarpus sp.	5		Rhipidocladum sp.	4
	Rhynchospora cf. barbata	1		Sacciolepis angustissima	1, 2
	Senna obtusifolia	5		Setaria gracilis	2
	Zigia divaricata	1, 2		Sporobolus jacquemontii	2
	Zigia spp.	4, 5, 6	Rubiaceae	Alibertia sp.	2
Lamiaceae	Hyptis conferta	1, 2		Duroia sp.	4
Lamiaceae	Hyptis lantanifolia	5		Faramea occidentalis	4, 5
Loranthaceae	Psittacanthus cordatus	7		Psychotria sp.	5
Lyrthaceae	Physocalymma caberrimum	8	Scrophulariaceae	Bacopa sp.	1
Malpighiaceae	Banisteriopsis sp.	3	Scrophulariaceae	Veronica sp.	2
	Banisteriopsis cf. muricata	5	Smilacaceae	Smilax sp.	5
	Byrsonima sp.	2	Smilacaceae	Smilax rufescens	3
Malvaceae	Melochia sp.	2	Vochysiaceae	Vochysia mapirensis	4, 6

Laboratory analyses

Sediment core analysis and chronology

Three radiocarbon dates were obtained from non-calcareous sediments and terrestrial macrofossils from each of the LEC and LF cores. One additional date, obtained from a terrestrial macrofossil from LEC, was rejected on the grounds that it was anomalously young (Table 2). In both records, dates were obtained from overlapping surface and Livingstone cores, which were cross-correlated using high-resolution charcoal data. All dates were calibrated using the IntCal09 calibration curve in OxCal version 4.1 (Bronk Ramsey, 2009), and date ranges are reported with 95% confidence intervals. Age-models were created through linear interpolation between calibrated ¹⁴C dates (Figures 2 and 3) and an assumed modern age for the core top (sediment–water interface). Age estimations were rounded to the nearest 10 years.

Table 2.

Details of AMS ¹⁴C dates from the LEC and LF sediment cores. Where there were multiple calibrated age ranges, the range and intercept marked with * was used for the age–depth model. Intercept ages are presented in parentheses. ** denotes rejected date.

Facility code	Sample ID	Stratigraphic depth	Material	δ¹³C_VPDB (‰)	Conventional ¹⁴C age BP (±1σ)	Calibrated age (ad), ±2σ
SUERC-35219	El Cerro 39.5–40	39.5–40 cm	Sediment	−23.0	471 ± 37	1400 (1440) 1480
**SUERC-34155	El Cerro 39.5–40	39.5–40 cm	Twig	−25.0	208 ± 37	1640 (1666) 1695
						1726 (1796) 1814
						Modern
Beta-345409	EC2-55	54.5–55 cm	Sediment	−18.7	670 ± 30	1280 (1290) 1320*
Beta-345409	EC2-55	54.5–55 cm	Sediment	−18.7	670 ± 30	1350–1390
SUERC-43152	El Cerro 2: 79.5 – 80	79.5–80 cm	Twig	−28.4	1010 ± 38	900 (1060) 1160
Beta-345410	FRONTSURF24	23.5–24 cm	Sediment	−21.7	580 ± 30	1300 (1330, 1340) 1370
Beta-345410	FRONTSURF24	23.5–24 cm	Sediment	−21.7	580 ± 30	1380 (1400) 1420*
Beta-345411	FRONTSURF45	44.5–45 cm	Sediment	−22.1	780 ± 30	1160 (1220) 1260
Beta-345412	FRONT1-61	60.5–61 cm	Sediment	−23.2	1740 ± 30	210 (250) 340

AMS: accelerator mass spectrometry.

Figure 2.

Pollen and charcoal results from Laguna El Cerrito (LEC) plotted with the age-model for the sediment core, which was based on two radiocarbon dates from non-calcareous bulk sediment and one terrestrial macrofossil. Terrestrial pollen types are presented as relative percent abundance. Large, rarely occurring pollen grains, such as Zea mays, which were isolated using the methodology outlined in Whitney et al. (2012), are presented as raw counts which approximately equates to the number of grains per 0.25 cm³. Rare taxa are shown with 5× multiplier curves.

Figure 3.

Pollen and charcoal results from Laguna Frontera (LF) plotted with the age-model for the sediment core, which was based on three radiocarbon dates from non-calcareous bulk sediment. Terrestrial pollen types are presented as relative percent abundance. Large, rarely occurring pollen grains, such as Zea mays, which were isolated using the methodology outlined in Whitney et al. (2012), are presented as raw counts, which approximately equates to the number of grains per 0.25 cm³. Rare taxa are shown with 5× multiplier curves.

Pollen preparations followed standard protocol (Bennett and Willis, 2001; Fægri and Iversen, 1989) and included sieving chemically treated samples to concentrate large crop pollen (Whitney et al., 2012). An exotic marker (Lycopodium spores) was added to each sample to calculate absolute concentrations (Stockmarr, 1971). Counts totalled 300 terrestrial pollen grains. Sediment samples of 1 cm³ were prepared for charcoal analysis by heating for 10 min in 5% (w:v) sodium pyrophosphate to disaggregate clays, and samples were then washed through 250- and 125-µm nested sieves. Charcoal fragments in each size fraction were counted using a stereomicroscope. Pollen and charcoal data were plotted in C2 (Juggins, 2003), and zonal boundaries mark stratigraphic changes in both proxies.

Phytolith analysis

Phytolith extraction followed standard protocols (Piperno, 2006). Raised-field soil samples measuring 100 cm³ were pre-treated to remove clays through deflocculation, agitation and gravity sedimentation. Samples were divided into silt (A/B-fraction, <50 µm) and sand (C-fraction, >50 µm) fractions. Each fraction of 3 mL was digested with 36% HCl and 70% HNO₃, to remove carbonates and organics, respectively. Phytoliths were concentrated by heavy-liquid flotation using ZnBr₂ prepared to a density of 2.3 g/cm³.

Approximately 15 mg of extracted phytolith residue from each sample was mounted in Entellan^® on a microscope slide. If processed samples yielded less than 15 mg, the entire residue was mounted and scanned. For the A-fraction, phytoliths were examined, described and photographed at 500× magnification. A minimum of 200 phytoliths were counted, and the rest of the microscope slide was scanned to identify any other diagnostic types. For the C-fraction, the entire slide was scanned at 200× magnification and all diagnostic phytoliths counted. Data were plotted using C2 (Juggins, 2003).

Results and interpretation

Pollen and charcoal

Pollen and charcoal results from the sediment cores of the two lakes are presented alongside their age-models to demonstrate the error range for the estimated age of each zonal boundary (Figures 2 and 3). Pollen is poorly preserved in a lower stratum of the LEC core, possibly due to oxidizing conditions related to hydrological change associated with the formation of the oxbow. Thus, only data from above 80 cm (ad 1060) are presented, where we are confident of continuous sedimentation. The topmost sediments (0–60 cm) were analysed from the second site (LF) to replicate the changes observed in LEC. The lowest dated horizon (60 cm) from LF is much older (ad 310) than the base of LEC, and the two records together provide an environmental history of the El Cerro site for the past c. 1700 years.

The modern (topmost) pollen samples of LF and LEC show that both lake sites reflect vegetation from the surrounding gallery forest and savanna ecosystems. Fringing aquatic vegetation, including grasses and sedges, are absent from the shoreline of both lakes, and the surrounding gallery forests lack terrestrial herbs in the understorey. Therefore, we are confident that the high abundance of Poaceae pollen in both records reflects the wider savanna landscape. Further evidence that both records capture a savanna pollen signal is shown by the presence of Curatella americana pollen, a key indicator of savanna ecosystems (Jones et al., 2011). Although the oxbow lakes are hydrologically connected to the Iruyañez River, the relative contribution of upstream river-borne pollen is negligible because floristic differences in the local gallery forests at the two sites are clearly reflected by compositionally distinct surface pollen assemblages of each lake.

Both records contain abundant charcoal and little arboreal pollen in the lowest zone, dating from >ad 1060 to 1290 (LEC-1) and >ad 310 to 1220 (LF-1). Regional pollen and charcoal records from southwestern Amazonia (Burbridge et al., 2004; Mayle et al., 2000) show a trend of rainforest expansion and concomitant decline in burning, consistent with increasing precipitation through the late Holocene. However, at LEC and LF, the pattern of high charcoal values, followed by a rapid decline as recently as ad 1220–1290, is contrary to the expected trend if climate was the key control on fire at this time. Furthermore, peak charcoal correlates with maize pollen at both lakes (Figures 2 and 3), strengthening our interpretation that the charcoal records at both sites largely reflect anthropogenic burning. The low abundance of maize pollen at both sites is unsurprising, given the limited dispersal of maize pollen (Lane et al., 2010) and the distance of several hundred metres from the lake sites to the nearest fields. Nonetheless, the presence of maize pollen is a key indicator of pre-Columbian agriculture in pollen records. Therefore, we are confident that greater levels of burning, combined with evidence of maize cultivation, reflects human activity from as early as ad 310.

The low levels of arboreal pollen in zones LEC-1 and LF-1, compared with the modern (core top) assemblage, indicate that the gallery forest which currently surrounds both lake sites was not present when anthropogenic fires were more prevalent, and also that the gallery forest was cleared before ad 310. Absence of savanna tree pollen, such as C. americana, also indicates that trees were cleared from nearby savannas by this time. This low abundance of arboreal pollen suggests that the use of fire suppressed regrowth of cleared trees in the El Cerro area, in both savanna and forested ecosystems. High levels of anthropogenic burning declined at broadly contemporaneous periods at LEC (ad 1280–1320) and LF (ad 1160–1260). Arboreal pollen, however, does not show any substantial rise associated with the decline in charcoal in either record, which suggests that, in addition to the suppression of tree regrowth, the elevated charcoal concentrations in LEC-1 and LF-1, compared with subsequent zones, may have originated from burning in savannas to build and clear the fields.

The El Cerro raised fields were continuously cultivated despite the reduction in anthropogenic fire activity, as shown by the presence of maize pollen at this time. Ipomoea batatas-type (sweet potato) pollen is present in LEC-2 after the decline in charcoal. The genus Ipomoea contains c. 500 Neotropical species, the pollen of which have overlapping morphologies, and several wild species have the characteristic bottle-shaped echinae of I. batatas (Colinvaux et al., 1999; Herrera and Urrego, 1996). Thus, in the absence of extensive taxonomic research on this genus, we cannot distinguish I. batatas from all of its wild relatives. Although other Ipomoea pollen morphotypes are present throughout the entire core, I. batatas-type grains are mostly constrained to pre-Columbian horizons. We infer, therefore, that the presence of I. batatas-type pollen reflects sweet potato cultivation.

Approximately 150 years after the decline in charcoal, arboreal pollen rises in both records. Absolute concentration data (not shown) confirm that the rise in arboreal pollen in both records is not driven by the decline in Poaceae pollen within a closed sum. At LEC, this horizon dates from ad 1450 to 1530. This re-establishment of gallery forest is characterized by a sharp increase in Myrtaceae to values reaching 10–26%. Myrtaceae (Eugenia spp.) was identified in our gallery forest survey at LEC (Table 1). Combretaceae/Melastomataceae appears in the record (1–4%), and Ouratea, previously less abundant, achieves values exceeding 10%. Although not identified in our surveys, both of these pollen types represent species that occur in gallery forest. C. americana, a diagnostic savanna tree (Jones et al., 2011), and Vochysia, common to the savannas around El Cerro, also become established at the same time. The methodology to concentrate large crop pollen (Whitney et al., 2012) also isolated Inga pollen at LEC from c. ad 1400, just prior to the period in which trees become re-established until c. ad 1700. Although Inga poeppigiana was identified in the modern gallery forest at LEC (Table 1), its pollen was not recovered in the uppermost horizons of the lake core. We conclude, then, that Inga must have been more common at LEC from c. ad 1400 to 1700.

The pattern of rising arboreal pollen 150 years after the decline in charcoal is mirrored at LF. The horizon in which gallery forest became re-established at LF dates from ad 1410 to 1450. At both sites, the age range of the forest rise overlaps at ad 1450, which suggests that gallery forest expansion occurred contemporaneously at both sites. As shown at LEC, Myrtaceae (4–7%), which currently forms part of the gallery forest at LF (Table 1), and Ouratea (1–3%) both increase from previously negligible values. Cecropia also rises in abundance to above 5%. As shown in LEC, pollen from the savanna tree, C. americana, appears in the record. Once established, the gallery forest at LEC and LF demonstrates little variability until the present day. Charcoal values remained low at both sites.

Phytoliths and stratigraphy in raised-field ridges and canals

The stratigraphy of the profiles of the excavated ridges and canals contain three generalized units; the lowest stratum (c. 20–110 cm depth) is characterized by mottled red-orange clays, which are similar to unmodified soils in the region (Walker, 2004). Thus, these strata likely represent the subsoils underlying the raised fields. This lowest stratum is overlain with an eluviated, white-grey clay silt, approximately 10 cm in depth, visible in the canal and ridge sequences of Camellón 1. In Camellón 2, the second stratum is composed of compacted grey silt of similar thickness, reaching a depth of c. 20 cm in the ridge unit. Both the eluviated layer and the mottling of red-orange clays with grey (gleyed) soil in the lowest stratum are characteristic of soils that have been repeatedly inundated. The upper stratum in all profiles is characterized by grey silt, with roots visible in the upper few centimetres. Generally, the upper two strata are similar among the canals and ridges, although these layers are slightly thicker in the ridges. No artefacts were recovered from any of the units. These stratigraphic changes were not radiocarbon dated because of the lack of sufficiently large macroscopic charcoal fragments or organic material for dating. The lack of ceramic sherds and low charcoal concentrations corroborate previous raised-field excavations by Walker (2004, 2012).

Phytoliths were most abundant in the upper two strata and the topmost section of the underlying subsoil, which corresponds to the upper 30–50 cm and 20–30 cm of the ridge and canal units, respectively. Abundance declines below these levels, where <50 diagnostic phytoliths were found per sample, so percent abundances were not calculated for these samples (Figure 4). The exception is C1C where, similar to the other units, phytolith concentrations are low (<50) in depths from 30 to 50 cm, but in levels examined below 70 cm, phytolith abundance is high enough (>50) to allow taxonomic frequencies to be reliably calculated (Figure 4). In the phytolith preparations, diatoms were encountered in all samples, possibly reflecting regular annual inundation of the area and the deposition of fluvial silts.

Figure 4.

Phytolith results of the ridge (top) and canal (middle) units from the two excavated raised fields (or ‘camellón’), and the raised-field profile (bottom). Relative percent abundance of each taxon is presented with black bars in samples with phytolith counts of 200 or greater, and with white bars in samples with phytolith counts > 50 and < 200. For both population sizes, taxa with abundances <2% are marked with (+). Occurrences of taxa in samples with <50 phytoliths are marked with (x). Horizontal grey dashed lines indicated stratigraphic divisions. The presence of Zea mays, determined by discriminant analysis of a minimum of 20 Poaceae cross-body phytoliths, is marked by a solid black circle.

The upper two strata in all units are characterized by very high abundance of Poaceae phytoliths (>80%), which are dominated by Panicoideae bilobates. In C1C, Oryzoideae bilobates achieve 5–14% abundance, possibly reflecting recent wetter conditions in the canal. Phytolith diagnostics of other herb types are rare or absent in the upper strata, with the exception of C2C where Scirpus-type (3–5%) and Cyperus/Kyllinga (2–8%) achene phytoliths are present in low abundance. The few arboreal phytoliths present in the upper strata are mostly dominated by globular granulates.

Generally, arboreal phytoliths are most abundant in the lower strata of the units, specifically globular granulates. Arecaceae (palm) phytoliths also comprise a substantial part of the arboreal sum of the lower strata in C1C and C2R, where globular echinate palm phytoliths reach upwards of 20%. We interpret the arboreal signal to reflect past presence of savanna trees that grew here prior to the creation of the raised fields and were removed during construction (Figure 5). The upper two layers contain fewer arboreal phytoliths compared with the lowest stratum, which is contrary to the expected trend if savanna trees became re-established after field abandonment. However, unlike some raised-field sites in the region, the sampled raised-field area is currently used for cattle pasture, and occasionally burned, both of which would have suppressed any recent tree growth.

Figure 5.

Schematic of raised-field construction and use. (a) Savanna trees, reflected in the phytolith assemblages of the underlying subsoils, were cleared to build the fields; (b) soils dug from the canals, containing arboreal phytoliths, were heaped up to build the raised-field ridges; (c) after construction, maize cultivation and weedy grasses contributed to the majority of the phytolith signal on the raised-field ridges and canals, the latter of which filled with eroding soils from the raised-field ridges. The relative depth and width of canal and ridge features are exaggerated for illustrative purposes.

Discriminant function analysis of cross-body phytoliths produced in Poaceae leaves (Pearsall and Piperno, 1990; Piperno, 2006) shows the presence of Zea mays within the canal unit at Camellón 2, and at two of the shovel test pits (data not shown). The low number of contexts from which maize was identified is not surprising, given the conservative design of the discriminant function analysis (Iriarte, 2003; Pearsall and Piperno, 1990) and the high amount of other Poaceae cross-bodies in savanna agricultural fields. No other diagnostic domesticate phytoliths were identified in the excavated unit soils. Although we have no evidence of other food or industrial crops, the phytolith assemblages vary among the excavated units. This suggests there was inter-field variation and that the fields were not exclusively mono-cropped.

Discussion

Raised-field construction and agriculture

Although the raised-field excavations could not be radiocarbon dated due to the insufficient quantities of organic material and large enough charcoal fragments in the units, pollen analysis of our two lake records indicates that people were growing maize in the El Cerro site since at least ad 310. The first maize at LF also pre-dates archaeological excavations at San Juan, located 12 km southwest of El Cerro, where occupation layers associated with raised fields dated to the 6th century ad (Walker, 2004). Our phytolith data suggest that savanna trees were first cleared for the construction of raised fields, after which soils removed from the canals were heaped into a platform to create the raised-field ridges (Figure 5). The soil profiles, however, are undated, and the mottled clays of the lower strata could have formed thousands of years before the arrival of people in the region. The arboreal phytoliths in these lower strata are present in similar concentrations to modern phytolith assemblages recovered from terra firme savannas (Dickau et al., 2012), thus these phytoliths likely originated from savanna trees. Further evidence supporting our interpretation that savanna trees were more abundant prior to raised-field construction is shown by the arrival of C. americana after the decline of human activity in the study region, which is suggestive of savanna tree recovery.

The raised fields are currently elevated c. 30 cm higher than the canals, but this height difference does not likely reflect the original morphology of the raised fields. Instead, the canal profiles exhibited upper strata of less compacted, light grey silt, similar to that of the ridges, which suggests that the upper strata of the canals are soils from the raised-field ridges that have eroded into the canals (Walker, 2004). Phytolith evidence of Zea mays in a canal unit (C1C) confirms that these strata represent erosion of the raised-field platforms (Figures 4 and 5). The current raised-field morphology is the product of centuries of erosion following abandonment and does not likely reflect their original form. Based upon their current morphology, Lombardo et al. (2011) argue that the raised fields in the NW Llanos de Moxos were built for drainage rather than the improvement of soil fertility. Without soil improvements, however, it is unlikely that maize would have been grown on the savanna soils, given its requirement for higher soil nitrogen compared with root crops (Denevan, 2001). Instead, similar to other raised- and drained-field agricultural systems in the Americas (Iriarte and Dickau, 2012), we show that maize was one of the staple crops at the pre-Columbian El Cerro site, which implies that soil improvements achieved in the construction of raised fields are not fully reflected in their current remnant state. As well as centuries of physical erosion into the canals, nitrogenous fertilizer added to the raised fields, such as household waste and compost, also would have long since leached from the system, as shown by analysis of modern raised-field soils (Lombardo et al., 2011, 2013).

Our charcoal results show that savanna and raised-field burning was a common practice and that levels of burning in pre-Columbian times were much higher than present, despite the regular annual burning instigated by modern cattle ranchers across the Llanos de Moxos (Erickson, 2010). The co-occurrence of maize and charcoal in both lake sediment records suggests that the use of fire in the savannas played a role in raised-field agriculture. It is interesting to find that the use of fire at El Cerro contrasts with a similar study from French Guiana (Iriarte et al., 2012), where pre-Columbian farmers suppressed fire in savannas, pointing to a diversity of agricultural strategies among different cultures across Amazonia. The practice of high levels of burning associated with maize cultivation at El Cerro declines around ad 1300.

Land-use change in the pre-Columbian era

The pre-Columbian agricultural strategy at the El Cerro site changed c. ad 1260–1280, after which savannas were not burnt as frequently. Because of the poorly dispersed nature of many large crop pollen grains, it is difficult to assess whether there was a substantial alteration in the food crops produced, but the initial presence of I. batatas-type (sweet potato) pollen, beginning c. ad 1320 in LEC, suggests that there was a shift in agricultural strategy associated with the decline in charcoal. However, this second phase of raised-field agriculture also involved maize cultivation, as evidenced by maize pollen in both records, and points to the exploitation of a greater mix of resources at this time. Intriguingly, these two distinctive phases of land use identified in our palaeoecological data correlate with the two different periods of occupation shown in the archaeological excavations (Walker, 2004, 2012).

Further evidence of a change in resource strategy after c. ad 1280 comes from the abundance of Inga in the pollen record, which was concentrated through the sieving methodology of Whitney et al. (2012), intended to isolate large domesticate pollen grains. Most species of Inga produce seeds with edible flesh, in particular ‘ice-cream bean’ (Inga edulis), a common food resource of Native Amazonians (Boom, 1987; Clement, 1999; Métraux, 1942). It is argued that indigenous groups domesticated their landscape through the selective management of economically important trees (Erickson, 2010; Levis et al., 2012). Erickson and Balée (2006) reported the presence of several useful Inga species on, and adjacent to, the forested pre-Columbian habitation mounds in the Monumental Mound Region (Lombardo and Prümers, 2010) of southeastern Llanos de Moxos. Given that Inga pollen cannot be resolved to species level, we cannot verify whether the presence of Inga pollen reflects a cultivated food resource. Inga poeppigiana was identified in our vegetation survey (Table 1), and it is possible that pre-Columbian people actively selected this tree species in the gallery forest of the Iruyañez River. Nitrogen-fixing legumes, particularly species of Inga, are important successional trees in the recovery of degraded Neotropical forests (Montagnini, 2008). Thus, post-disturbance, succession may be a possible alternative explanation for the peak in Inga pollen. Nonetheless, whether or not the high abundance of Inga was caused by human agency, it is likely that the pre-Columbian raised-field farmers would have exploited this resource all the same.

We infer that gallery forest was re-established around the lakes at c. ad 1450. Thus, although the final abandonment of the raised fields at El Cerro may have been the result of population decline due to Old World diseases, or the introduction of new technologies, such as steel axes, that influenced agricultural strategy (Denevan, 2001; Erickson and Balée, 2006; Walker, 2012), our study suggests that the environmental impact of pre-Columbian activities at El Cerro had begun to lessen prior to ad 1492. Despite the forest recovery at ad 1450, the presence of Ipomoea and Inga in the pollen assemblages after ad 1450 shows that El Cerro was continually occupied and that agriculture continued into the colonial period. The last recovered I. batatas-type pollen grain dates to ad 1800, which suggests that the abandonment of raised fields at El Cerro occurred over a protracted period.

Pre-Columbian impact and legacy

The absence of gallery forest at both LEC and LF prior to c. ad 1450 points to earlier clearances by the raised-field culture, possibly related to settlement and raised-field construction. However, their impact extended beyond gallery forest clearances. The appearance of C. americana and Vochysia pollen at ad 1450 suggests that savanna trees were removed for the construction of raised fields, which then subsequently recovered. This pattern is borne out in our raised-field phytolith assemblages, where arboreal phytoliths are more abundant in the underlying subsoils (Figure 5), and corroborates geographic information system (GIS)-based analysis of the position of raised fields relative to local topography (Lombardo, 2010), which showed that they are predominantly situated in areas of higher elevation that also might have naturally supported trees. The dominance of Cyperaceae and Poaceae in the pollen assemblages during the period of high anthropogenic fires (prior to ad 1300) indicates that both savanna and gallery forest vegetation communities were heavily impacted by human activities. Our results corroborate the findings of Erickson and Balée (2006), who observed that the modern landscape of the Llanos de Moxos reflects less intense management of the landscape, in particular, savanna burning, compared with pre-Columbian times.

After the establishment of gallery forest, c. ad 1450, there was little alteration of the floristic composition of the forest, with the exception of Inga pollen, which diminishes at c. ad 1800. Although the important fruit resources provided by Inga were likely exploited by pre-Columbian people, the prevalence of Inga pollen is not maintained through the post-colonial period and declines c. 200 yr BP. This trend is contrary to expectation if the legacy of past human management is imprinted on the modern floristic composition (Erickson and Balée, 2006). Inga, however, is one of the few arboreal resources present in our records from pre-Columbian times. During the period when the fields were intensively farmed for maize and burned frequently, gallery forest resources were not maintained but almost cleared entirely. Given the shortage of timber in this open savanna-dominated landscape, restricted to gallery forests lining the rivers, it is perhaps surprising that this presumably valuable resource was completely cleared at El Cerro, rather than intensively managed and maintained. These data suggest that, unlike the Monument Mound Region, where forests comprise a greater proportion of the landscape (Erickson and Balée, 2006; Lombardo and Prümers, 2010; Prümers, 2009; Whitney et al., 2013), forest resource management was not a key strategy employed by the raised-field culture of the northwest Llanos de Moxos. Instead, our data suggest that crop cultivation was the dominant land-use strategy in pre-Columbian times.

Conclusion

Our integrated landscape reconstruction of a raised-field site in the Llanos de Moxos shows that, despite highly infertile soils that are hypothesized to have restricted maize cultivation (Denevan, 2001; Lombardo et al., 2011), maize crops were grown at El Cerro in pre-Columbian times, corroborating previous studies of raised- and drained-field agricultural systems across the American tropics (Iriarte and Dickau, 2012). Given the higher nitrogen requirements of maize compared with root crops, the construction of raised fields increased soil fertility as well as drainage, but evidence of these ancient modifications is no longer reflected in the current soil properties (Lombardo et al., 2011), due to leaching over the centuries following abandonment. Evidence of both maize and sweet potato cultivation at El Cerro indicates that raised-field agriculture was successful in producing staple food crops and supports the assertion that raised-field agriculture could serve as a model for sustainable economic development in the region today (Renard et al., 2012).

Furthermore, we demonstrate that frequent burning was not integral to the cultivation of raised fields in the Llanos de Moxos, but instead that pre-Columbian agriculture relied on different fire-use strategies in two distinct agricultural phases. Although the earlier phase of agriculture (ad 310–1300) relied on high fire use, the second phase shows a land-use pattern similar to that revealed by Iriarte et al. (2012) in French Guiana, whereby fires were actively suppressed. The two very different land-use strategies employed by the raised-field cultures at El Cerro demonstrate that caution needs to be exercised when estimating the impact of pre-Columbian land use on atmospheric carbon emissions (Nevle et al., 2011). We found that a reduction in burning occurred at El Cerro almost two centuries before ad 1492. Therefore, assumptions regarding the scale of pre-Columbian atmospheric carbon emissions must not solely consider changes associated with the ‘European Encounter’ but also consider cultural change in the pre-Columbian era which may have had equally profound effects on land-use strategies and their associated environmental impacts.

Most forest patches were cleared at El Cerro prior to ad 310, both in savannas and along river margins. The re-establishment of forest at ad 1450 shows that the scale of human impact on the landscape had begun to decline prior to the arrival of Europeans, for reasons as yet unknown. Thus, we show that, contrary to the hypothesis that raised-field abandonment occurred after ad 1492 due to repercussions of the European Encounter (Denevan, 2001; Erickson, 2010; Walker, 2004), a shift away from raised-field agriculture may have already begun prior to ad 1492. However, we also show that the full abandonment of raised-field agriculture at El Cerro was possibly drawn out over a long period, with the cultivation of sweet potato occurring as late as the 19th century. Rather than a simple narrative of timeless indigenous settlement destroyed by the ‘European Encounter’, the history of the raised-field site at El Cerro is one of nearly two millennia of shifting land-use strategies and anthropogenic landscapes.

Footnotes

Acknowledgements

Fieldwork logistical support was provided by the ‘Noel Kempff Mercado’ Natural History Museum, Santa Cruz, Bolivia, and ‘Programa de Conservación de la Paraba Barba Azul’, Trinidad, Beni Department, Bolivia. We also thank Oscar Saavedra for his assistance in the field, and Charles Clement and André Junqueira for their helpful advice on the use of Inga by the indigenous people of Amazonia. We extend special thanks to Beatriz Keller de Iriarte and family, and residents of the El Cerro ranch, for their kind hospitality. We also thank the two anonymous reviewers whose comments improved this manuscript.

Funding

This research was funded by a Leverhulme Trust research project grant (F/00158/Ch) awarded to FEM and JI. Radiocarbon dates were granted by the NERC radiocarbon facility to FEM (allocation numbers 1527.1010 and 1623.0312).

References

Barba

Canal-Beeby

Garcia

. (2003) Moxos: Una Limnocultura. Cultura y Medio Natural en la Amazonia Boliviana. Barcelona: CEAM.

Bennett

Willis

(2001) Pollen. In: Smol

Birks

HJB

Last

(eds) Tracking Environmental Change Using Lake Sediments: Terrestrial, Algal, and Siliceous Indicators. Dordrecht: Kluwer Academic Publishers, pp. 5–32.

Biesboer

Binford

Kolata

(1999) Nitrogen fixation in soils and canals of rehabilitated raised-fields of the Bolivian Altiplano. Biotropica 31: 255–267.

Boom

(1987) Ethnobotany of the Chácobo Indians, Beni, Bolivia. New York: New York Botanical Garden.

Bronk Ramsey

(2009) Bayesian analysis of radiocarbon dates. Radiocarbon 51: 337–360.

Burbridge

Mayle

Killeen

(2004) Fifty-thousand-year vegetation and climate history of Noel Kempff Mercado National Park, Bolivian Amazon. Quaternary Research 61: 215–230.

Carney

Binford

Kolata

. (1993) Nutrient and sediment retention in Andean raised-field agriculture. Nature 364: 131–133.

Clapperton

(1993) Quaternary Geology and Geomorphology of South America. Amsterdam: Elsevier.

Clement

(1999) 1942 and the loss of Amazonian crop genetic resources. I. The relation between domestication and human population decline. Economic Botany 53: 188–202.

10.

Colinvaux

De Oliveira

Moreno

PJE

(1999) Amazon Pollen Manual and Atlas. Amsterdam: Harwood Academic Publishers.

11.

Denevan

(1966) The Aboriginal Cultural Geography of the Llanos de Mojos of Bolivia. Berkeley, CA: University of California Press.

12.

Denevan

(1992) The pristine myth: The landscape of the Americas in 1492. Annals of the Association of American Geographers 82: 369–385.

13.

Denevan

(2001) Cultivated Landscapes of Native Amazonia and the Andes. Oxford: Oxford University Press.

14.

Dickau

Whitney

Iriarte

. (2013) Differentiation of Neotropical ecosystems by modern soil phytolith assemblages and its implications for palaeoenvironmental and archaeological reconstruction. Review of Palynology and Palaeobotany 193: 15–37.

15.

Erickson

(1995) Archaeological methods for the study of ancient landscapes of the Llanos de Mojos in the Bolivian Amazon. In: Stahl

(ed.) Archaeology in the Lowland American Tropics. Cambridge: Cambridge University Press, pp. 66–95.

16.

Erickson

(2000) An artificial landscape-scale fishery in the Bolivian Amazon. Nature 408: 190–193.

17.

Erickson

(2010) The transformation of environment into landscape: The historical ecology of monumental earthwork construction in the Bolivian Amazon. Diversity 2: 618–652.

18.

Erickson

Balée

(2006) The historical ecology of a complex landscape in Bolivia. In: Balée

Erickson

(eds) Time and Complexity in Historical Ecology: Studies in the Neotropical Lowlands. New York: Columbia University Press, pp. 187–233.

19.

Erickson

Candler

(1989) Raised fields and sustainable agriculture in the Lake Titicaca basin of Peru. In: Browder

(ed.) Fragile Lands of Latin America: Strategies for Sustainable Development. Boulder, CO: Westview Press, pp. 230–250.

20.

Fægri

Iversen

(1989) Textbook of Pollen Analysis. Chichester and New York: John Wiley & Sons.

21.

Gomez-Pompa

(1978) An old answer to the future. Mazingira 5: 50–55.

22.

Gongard

(2006) Campos elevados en llanuras humedas del modelado al paisaje camellones, waru warus o pijales. In: Valdéz

(ed.) Agricultura ancestral. Camellones y albarradas: Contexto social, usos y retos del pasado y del presente. Quito: Abya Yala, pp. 25–53.

23.

Hanagarth

(1993) Acerca de la Geoecología de las Sabanas del Beni en el Noreste de Bolivia. La Paz: Instituto de Ecologia.

24.

Herrera

Urrego

(1996) Atlas de polen de plantas útiles y cultivadas de la Amazonia Colombiana. Bogotá: Tropenbos.

25.

Iriarte

(2003) Assessing the feasibility of identifying maize through the analysis of cross-shaped size and three-dimensional morphology of phytoliths in the grasslands of southeastern South America. Journal of Archaeological Science 30: 1085–1094.

26.

Iriarte

Dickau

(2012) As culturas do milho? Arqueobotânica de las sociedades hidráulicas das terras baixas sul-americanas. Amazônica 4: 30–58.

27.

Iriarte

Power

Rostain

. (2012) Fire-free land use in pre-1492 Amazonian savannas. Proceedings of the National Academy of Sciences of the United States of America 109: 6473–6478.

28.

Jones

Mayle

Pennington

. (2011) Characterisation of Bolivian savanna ecosystems by their modern pollen rain and implications for fossil pollen records. Review of Palaeobotany and Palynology 164: 223–237.

29.

Juggins

(2003) User Guide C2: Software for Ecological and Palaeoecological Data Analysis and Visualization, User Guide Version 1.3. Newcastle upon Tyne: University of Newcastle.

30.

Lane

Cummings

Clark

(2010) Maize pollen deposition in modern lake sediments: A case study from Northeastern Wisconsin. Review of Palaeobotany and Palynology 159: 177–187.

31.

Levis

de Souza

Schietti

. (2012) Historical human footprint on modern tree species composition in the Purus-Madeira Interfluve, Central Amazonia. PLoS ONE 7: e48559.

32.

Lombardo

(2010) Raised fields of northwestern Bolivia: A GIS based analysis. Zeitschrift für Archäologie Außereuropäischer Kulturen 3: 127–149.

33.

Lombardo

Prümers

(2010) Pre-Columbian human occupation patterns in the eastern plains of the Llanos de Moxos, Bolivian Amazonia. Journal of Archaeological Science 37: 1875–1885.

34.

Lombardo

May

Veit

(2012) Mid- to late-Holocene fluvial activity behind pre-Columbian social complexity in the southwestern Amazon basin. The Holocene 22:1035–1045.

35.

Lombardo

Canal-Beeby

Fehr

. (2011) Raised fields in the Bolivian Amazonia: A prehistoric green revolution or a flood risk mitigation strategy? Journal of Archaeological Science 38: 502–512.

36.

Lombardo

Denier

May

J-H

. (2013) Human-environment interactions in pre-Columbian Amazonia: The case of the Llanos de Moxos, Bolivia. Quaternary International 312: 109–119. Available at: https://dx-doi-org.web.bisu.edu.cn/10.1016/j.quaint.2013.01.007.

37.

Mann

(2008) Ancient earthmovers of the Amazon. Science 321: 1148–1152.

38.

Mayle

Iriarte

(2013) Integrated palaeoecology and archaeology – A powerful approach for understanding pre-Columbian Amazonia. Journal of Archaeological Science. Available at: https://dx-doi-org.web.bisu.edu.cn/10.1016/j.jas.2012.08.038.

39.

Mayle

Burbridge

Killeen

(2000) Millennial-scale dynamics of southern Amazonian rain forests. Science 290: 2291–2294.

40.

Mayle

Langstroth

Fisher

. (2007) Long-term forest–savannah dynamics in the Bolivian Amazon: Implications for conservation. Philosophical Transactions of the Royal Society B: Biological Sciences 362: 291–307.

41.

Métraux

(1942) The Native Tribes of the Eastern Bolivia and Western Matto Grosso. Washington, DC: Smithsonian Institution.

42.

Montagnini

(2008) Management for sustainability and restoration of degraded pastures in the Neotropics. In: Myster

(ed.) Post-Agricultural Succession in the Neotropics. New York: Springer, pp. 265–295.

43.

Muse

Quinteros

(1987) Experimentos de reactivación de campos elevados, Peñón del Río, Guayas, Ecuador. In: Denevan

Mathewson

Knapp

(eds) Pre-Hispanic Agricultural Fields in the Andean Region. Oxford: British Archaeological Reports International Series 359, pp. 249–266.

44.

Nevle

Bird

Ruddiman

. (2011) Neotropical human-landscape interactions, fire, and atmospheric CO₂ during European conquest. The Holocene 21: 853–864.

45.

Pearsall

Piperno

(1990) Antiquity of maize cultivation in Ecuador: Summary and reevaluation of the evidence. American Antiquity 55: 324–337.

46.

Piperno

(2006) Phytoliths: A Comprehensive Guide for Archaeologists and Paleoecologists. San Diego, CA: AltaMira Press.

47.

Pouilly

Beck

(2004) Geografía general. In: Pouilly

Beck

Moraes

al.

(eds) Diversidad Biológica en la Llanura de Inundación del Río Mamoré: Importancia Ecológica de la Dinámica fluvial. Santa Cruz: Centro de Ecológia Simón I. Patiño, pp. 15–26.

48.

Prümers

(2009) ¿‘Charlatanocracia’ en Mojos? Investigaciones arqueológicas en la Loma Salvatierra, Beni, Bolivia. In: Kaulicke

Dillehay

(eds) Procesos y expresiones de poder, identidad y orden tempranos en Sudamérica. Lima: Segunda Parte, pp. 103–116.

49.

Renard

Iriarte

Birk

. (2012) Ecological engineers ahead of their time: The functioning of pre-Columbian raised-field agriculture and its potential contributions to sustainability today. Ecological Engineering 45: 30–44.

50.

Rostain

(2010) Pre-Columbian earthworks in coastal Amazonia. Diversity 2: 331–352.

51.

Saavedra

(2008) Culturas Hidráulicas de la Amazonia Boliviana. La Paz: Oxfam.

52.

Sanders

Parsons

Santley

(1979) The Basin of Mexico: Ecological Processes in the Evolution of Civilization. New York: Academic Press.

53.

Sluyter

(1994) Intensive wetland agriculture in Mesoamerica: Space, time, and form. Annals of the Association of American Geographers 84: 557–584.

54.

Stockmarr

(1971) Calcium tablets with spores used in absolute pollen analysis. Pollen et Spores 13: 615–621.

55.

Walker

(2004) Agricultural Change in the Bolivian Amazon. Pittsburgh, PA: University of Pittsburgh Latin American Archaeology Publications.

56.

Walker

(2011) Social implications from agricultural taskscapes in the southwestern Amazon. Latin American Antiquity 22: 275–295.

57.

Walker

(2012) Regional associations of a ceramic assemblage from the fourteenth century Llanos de Mojos. Andean Past 10: 241–261.

58.

Whitney

Dickau

Mayle

. (2013) Pre-Columbian landscape impact and agriculture in the Monumental Mound region of the Llanos de Moxos, lowland Bolivia. Quaternary Research 80: 207–217.

59.

Whitney

Rushton

EAC

Carson

. (2012) An improved methodology for the recovery of Zea mays and other large crop pollen, with implications for environmental archaeology in the Neotropics. The Holocene 22: 1087–1096.