Pre-Columbian Earthworks in the Riberalta Region of the Bolivian Amazon

Pre-columbian earthworks in the riberalta region of the bolivian amazon

104 Pre-columbian Pre-columbian earthworks in the EARTHWORKS IN THE riberalta region of the RIBERALTA REGION OF THE bolivian amazon BOLIVIAN AMAZON

s a n n a s a u n a l u o m a

university of helsinki, finlândia

105 Saunaluoma, S.

Abstract PRE-COLUMBIAN EARTHWORKS IN THE RIBERALTA REGION OF THE BOLIVIAN AMAZON Interpretations of the Amazonia prehistory have changed sig- nificantly in the last few decades, as the complexity and diversity of the Amazonian cultures are beginning to be documented and understood. Earthworking, a long-term conscious anthropogenic landscape alteration, was a widespread phenomenon throughout the South American tropical lowlands. A variety of earthworks has been documented in the Southwest Amazon, including ditches and embankments of different shapes and sizes, roads, extensive raised fields, canals, causeways, and artificial wetlands linked to adjacent mounds and forest island settlement sites. A field survey and test excavations were undertaken in the region of Riberalta, in the Bolivian Amazon. The purpose of these investigations was to study the distribution and characteristics of the pre-Columbian occupation in the region. We found different types of sites, some without visible earthworks, indicating fairly dense occupation on river bluffs and terra firme, but lacking long permanence in the same location. The earthwork tradition prevailed in the Riberalta region from at least 100 B.C. until the period of European contact. The function of the less-complex earthworks may have been to enclose the occupation areas, and in some cases, to serve as canals. Compared to the variable layout of the sites, the ceramic assemblages of the region are relatively homogeneous. A central objective for future research will be to determine if the earthwork sites correlate with a single or multiple cultural traditions. Keywords: Southwest Amazonian archaeology, earthworks, ceramic traditions.

Resumo OBRAS DE TERRA PRÉ-COLOMBIANAS NA REGIÃO DE RIB- ERALTA, AMAZÔNIA BOLIVIANA Interpretações da pré-história Amazônica tem mudado significativamente nas últimas décadas, na medida em que a complexidade e diversidade das culturasamazônicas começam a ser documentadas e entendidas. Construções de terra, na forma de alterações consciente de paisagens ao longo do tempo, foi um fenômeno di- fundido através das terras baixas tropicais. Diversas construções

106 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks de terra tem sido documentadas no sudoeste amazônico, incluin- do trincheiras e muretas de diversas formas e tamanhos, estradas, campos de cultivo, canais, caminhos e terras inundáveis artificiais ligadas a assentamentos na forma de montículos e ilhas de flores- tas. Prospecções e escavações-teste foram realizadas na região de Riberalta, na Amazônia Boliviana. O objetivo dessas investiga- ções era estudar a distribuição e características da ocupação pré- Colombiana na região. Encontramos diferentes tipos de sítios, alguns sem obras de terra visíveis, indicando ocupação densa das ribanceiras de rios e terra firme, mas sem longo tempo de per- manência nos mesmos locais. A tradição de construção de obras de terra prevaleceu na região de Riberalta de pelo menos 100 AC até o período do contato europeu. A função dessas obras menos complexas pode ter sido de circundar áreas de ocupação e, em alguns casos, servir como canais. Comparada ao leiaute variável dos sítios, a cerâmica da região é relativamente homogênea. O objetivo principal para as pesquisas futuras será determinar se os sítios com obras de terra correspondem a uma única ou várias tradições culturais. Palavras-chave: Arqueologia do sudoeste da Amazônia, obras de terra, tradições cerâmicas.

Resumen OBRAS DE TIERRA PRE-COLOMBINAS EN LA REGIÓN DE RIBERALTA, AMAZONÍA BOLIVIANA Las interpretaciones de la Amazonia prehistórica han cambiado significativamente en las últimas décadas, como la complejidad y diversidad de culturas amazônicas están empezando a ser docu- mentados y comprendidos. Construcciones de tierra, altaraciones conscientes del paisaje, a largo plazo, fue un fenómeno generaliza- do en las tierras bajas tropicales. Varias construcciones de tierra han sido documentadas en el Sudoeste del Amazonas, incluyendo trincheras y muros de piedra de diferentes formas y tamaños, carreteras, campos agrícolas, canales, carreteras humedales artificales vinculados a los asentamientos en forma de montículos y las islas de bosque. Estudios y excavaciones de prueba se llevaron a cabo en la región de Riberalta en la Amazonia boliviana. El objetivo de estas investigaciones fue estudiar la distribución y características de pre-ocupación precolombina en la región. Nos encontramos con diferentes tipos de sitios, algunos con obras de tierra visible,

Amazônica 2 (1): 104-138, 2010 107 Saunaluoma, S.

lo que indica la ocupación densa de las orillas de ríos y barrancas de tierra firme, pero carecien de larga permanencia en los mismos lugares. La tradición de del movimiento de tierras prevaleció en la región de Riberalta de al menos 100 antes de Cristo hasta el período de contacto con los europeos. La función de estas obras menos complejas puede haber estado ocurriendo alrededor de las áreas de ocupación y, en algunos casos, sirven de canales. En com- paración con la distribucion variable de los sítios, la cerámica de la región es relativamente homogénea. El objetivo principal para la investigación futura será determinar si los sitios con las obras de las tierras corresponden a uno o más de las tradiciones culturales. Palabras clave: Arqueología del sudoeste de la Amazonía, obras de tierra, tradiciones cerámicas.

108 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks

At certain times and places, the results have changed drastically (Heckenberg- of prehistoric environmental manage- er & Neves 2009; Stahl 2002). The vast ment were so profound that they are complexity and diversity of the cultures easily observable even today. Earth- of the tropical lowlands are beginning work engineering, a testimony to long- to be documented and understood. term conscious anthropogenic land- The idea of independent development scape alteration, seems to have been a of Amazonian regional cultural tradi- widespread phenomenon throughout tions, the existence of wide-spread the South American tropical lowlands, and continuous long-term permanent even though the existence of such settlements in the interfluvial areas, as earthworks has only recently been ac- well as on the well documented river- knowledged by scholars. A variety of ine bluffs, and the phenomena of late earthworks have been reported from prehistoric complex chiefdom societ- different parts of the Southwest Ama- ies is increasingly accepted. Resent re- zon1 (Figure 1). These include complex search has focused on flows of cultural earthworks formed by ditches, enclo- influences and traits, networks of ex- sures, and roads of different shapes change and knowledge, and the unique and sizes in the Brazilian State of ability of the pre-Columbian societies Acre (Dias & Teixeira 2008; Pärssinen of the Amazonia to alter and control et al. 2003, 2009; Schaan et al. 2007), their environment (Balée & Erickson semicircular and circular ditches and 2006; Heckenberger et al. 2008; Hill embankments in the north of Bolivia & Santos-Granero 2002; Woods et al. (Arellano 2002; Arnold & Prettol 1988; 2009). Saunaluoma et al. 2002; Saunaluoma & In this article, I present and discuss the Korhonen 2003), ring-ditches on for- results of a survey and test excavations est islands in the savannas in Baures undertaken in the region of Riberalta, and Santa Ana de Yacuma of the Lla- in the departments of Beni and Pando nos de Mojos, in eastern Bolivia (Er- in the Bolivian Amazon. The purpose ickson et al. 1997, 2008; Prümers et of these investigations was to study al. 2006; Walker 2008b), and, perhaps the distribution and characteristics of as the most impressive example of the pre-Columbian occupation in an landscapes of earthworks, extensive area with few recorded archaological raised fields, canals and ditches of sites, and to identify, determine the many types, causeways, and artificial functions of, document the variation wetlands linked to adjacent mound of and date the earthworks in the Rib- and forest island settlement sites in eralta region. This research contributes the Central Llanos de Mojos (Dene- to the knowledge of the pre-Columbi- van 1966; Erickson 1995, 2001, 2006; an cultural sequences not only of the Walker 2008a). northern lowlands of Bolivia but also In the last couple of decades, interpre- of the Southwest Amazon in general. tations of the prehistory of Amazonia

Amazônica 2 (1): 104-138, 2010 109 Saunaluoma, S.

days at a time. The tropical white-water rivers Beni and Madre de Dios, originating in the Andean mountain range, dominate the landscape of the region with their floodplains and numerous tributaries. In the Amazon plain, the meandering of the rivers causes notable processes of erosion and sedimentation in the Figure 1 - Southwest Amazon, main rivers proximity of the riverbanks. The aver- and regions mentioned in the text. age annual migration of the meanders of the Beni River is 30 m, but can ex- RESEARCH AREA tend to as much as 140 m (Gautier et Riberalta was founded in 1894 at the al. 2008). The soils of the region are confluence of the Beni and Madre de mainly weathered clayed latosols typi- Dios rivers, 80 km from the Brazilian cal of the Amazonian plain. border. Established as a rubber boom center, today Riberalta is the primary LOCATION AND DESCRIPTION OF exporter of Brazil nuts in Bolivia, and THE SITES with its population of 95,000 inhabitants, the second-largest town in the Prior to the present study, only few ar- Department of Beni. The region’s chaeological sites in the region of Rib- contemporary ethnic groups, Chaco- eralta were known to scholars: the Tu- bo, Pakawara, Cavineño, and Esse Ejja, michucua earthwork complex (Arnold belong to the Panoan-Tacanan lan- & Prettol 1988; Myers 1988:76), the guage family (Teijeiro et al. 2001). earthworks of the Orthon River Basin (Arellano 2002), and the ruins of a sup- Riberalta is situated 130 m above sea posed Inca fortress2 situated on a bluff level in the Amazonian plains, in a at the ancient confluence of the Beni region covered by evergreen rainfor- and Madre de Dios Rivers, approxi- est on undulating laterite formations mately 4 kilometers to the southwest (Wasson et al. 2002). The region’s of Riberalta. This La Fortaleza de las mean annual temperature is 27 ºC, and Piedras site features a wall constructed mean annual rainfall 1780 mm. The dry of conglomerate blocks that follows season extends from June to August, an old riverbank and a 600-m-long during which time less than 30 mm of curved moat that connects the ends rain falls per month (Myers et al. 2000). of the wall. These defensive features Occasional cold southern fronts, called enclose an area of 10 ha, containing a surazos by the local people, pass over group of small enclosures defined by the region in June and July and can low conglomerate walls (Korpisaari et lower the temperature abruptly by sev- al. 2003:8). The published radiocarbon eral degrees, but only for two or three

110 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks dates from the site suggest that the site together with the smaller semicircular was occupied A.D. 1300-1600 (Siiriäin- Ditch C adjacent to the oxbow lake. en 2003). These two structures are connected by a smaller, ca. 30-m-long ditch. The L- In 2001-2003 and in 2005, mapping shaped, 1350-m-long Ditch B connect- and test excavations were carried out ing the oxbow lake and the floodplain at the following sites in the Province of the River Beni is situated some 100 of Vaca Diez of the Department of m to the north of Ditches C and D. Beni: Tumichucua, Estancia Girasol, The semicircular Ditch E, approxi- Las Palmeras, and Estancia Giese, and mately 300 m to the southwest of the in the Province of Madre de Dios of circular Ditch D, was barely visible in the Department of Pando: El Círculo, the present study. Furthermore, Ar- Chacra Telería, and Candelaria. In ad- nold and Prettol (1988:461-462) refer dition, the following sites were briefly to Ditch A, documented some 230 m inspected during the 2005 field season: to the north of Ditch B, and a partially Estancia Mendez and Estancia Velasco destroyed mound inside the circular in the Department of Beni, and Cha- Ditch D, but these structures had van- cra Carbajal and Dos Palmas in the De- ished by 2002. In addition, we found partment of Pando (Table 1, Figure 2). a solitary ditch roughly 1500 m to the northwest of the Tumichucua commu- TUMICHUCUA nity. This ditch begins from the bank of the oxbow lake and heads towards The earthwork complex of Tumi- the Beni floodplain, disappearing into chucua was discovered in the 1950s the swampy terrain at 70 m distance. during the establishment of the infra- structure of the Summer Institute of Arnold and Prettol excavated a 14-m- Linguistics on the southeast margin long trench, traversing Ditch D in the of an oxbow lake, 18 km to the south- eastern section of the site and reveal- west of Riberalta. Recent land use has ing a few potsherds, unburnt clay balls already begun to affect the site, but and a possible posthole feature on the most of the extensive earthworks are inner edge of the ditch (1988:461- still visible, even in the midst of the 462). During the 2002 field season, we present-day community. placed five 1 m x 1 m test units in the northern section of the site between Dean Arnold and Kenneth Prettol Ditches D and B (Figure 3). Unit 1 in- (1988) mapped and excavated at the site side the circular Ditch D had pottery in the early 1980s. They documented a in abundance in the levels 0.2-0.6 m in complex of earthworks which covers a reddish-brown colored cultural layer, an area of some 125 ha and consists of while Unit 2, situated at the bottom of five distinctive ditches named A to E. the Ditch D, contained only a few pot- The most significant earthwork at the sherds in the levels 0.2-0.8 m. Soil in Tumichucua site is the circular Ditch Units 3 and 4 excavated between the D, approximately 775 m in diameter,

Amazônica 2 (1): 104-138, 2010 111 Saunaluoma, S.

Table 1 Sites investigated during the present study

Site Coordinates Earthwork m² Exca- Number Weight of Type vated of sherds sherds (g) Tumi- 11°08’ S - circular & semicir- 6 374 4813 chucua 66°09’ W cular ditches, canal Estancia 11°10’53” S - double ditch, canal, 3 141 1584 Girasol 66°11’02” W road Estancia 11°13’28” S - straight ditches * 8 527 Mendez 66°14’17” W Chacra 11°03’52” S - U-shaped ditch - - - Carbajal 66°19’13” W Las 10°59’17” S - roughly circular 11 1072 10993 Palmeras 66°00’37” W ditch Estancia 10°58’24” S - semicircular? 7 209 2126 Giese 66°00’52” W embankment El Círculo 11°02’09” S - circular embank- 39 838 21717 66°07’43” W ment, canal Chacra 11°01’25” S - none 1 107 1840 Telería 66°09’22” W Cande- 11°02’87” S - none 2 27 991 laria 66°17’06” W Dos 11°05’13” S - none * 15 654 Palmas 66°18’59” W Estancia 11°09’18” S - none * 7 391 Velasco 66°10’03” W

* superficial collection of sherds

ditches and in Unit 5 at the bottom of 0.1-0.6 m), the major concentration of the Ditch B was culturally sterile. In sherds (43%) occuring in level 0.2-0.3 m. 2005, we excavated an additional Unit The pottery collected at Tumichucua 6 inside the circular earthwork to clar- during fieldwork in 2002 and 2005 ify the dating of the site. This excava- includes 11% diagnostic sherds. The tion was placed some 500 m to the east ware is mainly ground sherd tempered, of the present-day community, in a re- although some sherds have carbon or cently burnt field with many diagnostic sand tempering, and caraipé temper is potsherds on the surface. The Unit 6 rare. Rims are mostly everted, although contained abundant ceramics in a dark direct and inverted rims are also pres- brown colored cultural stratum (levels

112 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks

with short parallel rim-nicking and fingernail incisions. Two small unnotched stone axes and a cylindrical grinding stone were also recovered at the site. An interesting fragment of an anthropomorphic vessel was recovered by a resident of Tumichucua near ditch B connecting the oxbow lake with the Beni floodplain, at some 200 m distance from the lake. It represents a human face with coffee bean-shaped eyes Figure 2 - Location of the investigated (Figure 5). The reddish-yellow ware sites in the region of Riberalta. has ground sherd temper. The fragment in question resembles the modeled anthropomorphic effigies typical of the Barrancoid Tradition (Lathrap 1970:168). Stylistically similar anthropomorphic vessels have also been reported from the Upper Beni River, in the proximity of the Andean piedmon- te (Nordenskiöld 1924, Portugal 1978).

ESTANCIA GIRASOL The Girasol ranch is situated adjacent to the floodplain of the Beni River, 4 km to the southwest of Tumichucua. Figure 3 - Earthworks and location of test units excavated at the Tumichucua site in We documented four earthworks at 2002 (Units 1-5) and 2005 (Unit 6). Map the site (Figure 5). A 240-m-long and partially redrawn from Arnold & Prettol 9-m-wide ditch, oriented in an E-W- 1988:460. direction, connects a smaller stream to the Beni River. Approximately 170 m ent (Figure 4B). The lips are rounded to the south, two partly parallel, badly and thickened or tapered. Some frag- eroded ditches connect to the old riv- ments of flat-based vessels were re- erbank of the Beni. Approximately covered (Figure 4C). Little can be said 500 m to the east of the floodplain, about the surface treatment of the a 15-m-wide road defined by paral- pottery, since the sherds are eroded. lel 1-m-high berms in a N-S-direction The common decorative techniques was recorded, but due to thick second- are incision (Figure 4A) and fine-line ary vegetation we could not follow the incision, in addition to some sherds earthwork’s course or to measure its

Amazônica 2 (1): 104-138, 2010 113 Saunaluoma, S.

Figure 5 - Fragment of an anthropomor- Figure 4 - A-C: Tumichucua assemblage, phic vessel recovered at the Tumichucua incised-design, inverted-rim, and flat-bot- site. Photo S. Saunaluoma. tomed vessel fragments, D: Estancia Gira- sol assemblage, globular (rim diameter 19 sherds are diagnostic. The few recov- cm) and direct (rim diameter 27 cm) body ered rim sherds represent globular or forms with out-turned rims, E-G: Chacra straight body forms with out-turned Telería assemblage, incised sherds repre- tapered or rounded lips (Figure 4D). senting globular and direct body forms. The only basal sherd is flat. The ware The base diameter of the flat bottomed is coarse, mainly cariapé tempered, but vessel (C) is 10 cm, all fragments drawn to the same scale. sand, carbon, and ground sherd tem- pers also occur. total length.

We excavated three 1 m x 1 m test units ESTANCIA MENDEZ during the 2005 field season. Unit 1, 70 m to the north of the eroded double The Estancia Mendez site is located ditch, and Unit 3 placed in between 12 km to the southwest of Tumi- these ditches yielded a few eroded pot- chucua. This earthwork site consists sherds in the 0.1-0.5-m-levels, while of two simple ditches oriented in N- Unit 2, excavated in the bottom of the S-direction, beginning on the bluff outermost ditch, suggested ceramic ac- of the Beni River and disappearing cumulation in the levels 0.2-0.7-m. at some distance from the floodplain. The westernmost ditch is 16 m wide The Estancia Girasol pottery is dete- and, in part, almost 1.8 m deep. A frag- riorated, and only five precent of the ment of caraipé tempered pottery with

114 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks

archaeological remains while briefly exploring the course of the earthwork, but without doubt, a more careful survey would reveal more information on the site.

LAS PALMERAS The community of Las Palmeras is situated 7 km to the northeast of Ri- beralta. Today the distance to the Beni floodplain is about 5 km. The Las Palmeras site was located in 1998 by Ari Siiriäinen and Jorge Arellano. Arel- Figure 6 - Earthworks and location of test lano carried out test excavations at the units excavated in 2001-2005 at the Estan- site the following year, acquiring pot- cia Girasol, Las Palmeras, Estancia Giese, tery which he classified into six distinct and El Círculo sites. types (Arellano 2002:58-65). a pedestal base was found in a cut of a The earthwork of Las Palmeras is a dirt road which crosses the ditch. The roughly circular ditch, approximately other, less prominent ditch is 12 m 12-m-wide, 1-m-deep, and 270 m in wide. The distance between these two diameter (Figure 5). According to the ditches is approximately one hundred landowner, the ditch was more promi- meters and probably they associated in nent when he purchased the property the same earthwork complex, although in the 1970s. The Las Palmeras earth- the full dimensions and shape of these work has suffered considerably from earthworks remain unknown. modern-day land use, making the course of the ditch difficult to distin- guish in certain places. The dirt road CHACRA CARBAJAL crossing the structure has all but de- An earthwork is located approximately 5 km to the southwest of the community of Candelaria and 1.5 km to the east of the Madre de Dios River. This site includes a shallow U-shaped ditch that surrounds a natural hill covered with thick secondary vegetation. The ditch ends in an old river channel, and the open side of the U-shaped earthwork faces in direction of the wa- Figure 7 - Las Palmeras, profile of Trench tercourse. We did not find any other 1 excavated in 2001.

Amazônica 2 (1): 104-138, 2010 115 Saunaluoma, S.

stroyed the southernmost part of the bottom of a depression on the north- earthwork. In the western part of the ern side of the mound. The majority site, small mounds, approximately 2 m of the potsherds were collected from x 3 m in size, were observed inside the the excavation levels 0.0-0.1 m (34%) circular ditch. Recent agricultural activ- and 0.3-0.4 m (35%). Pottery was ab- ities had exposed numerous potsherds sent from the levels 0.4-0.6 m, but we that were scattered on the surface of obtained the last few sherds from the the mounds. 0.6–0.7-m-level. In 2001, we excavated Trench 1 (1.5 Two 1 m x 1 m test units suggested m x 2.5 m) on the northern side of that the thickness of the cultural layer one of the small mounds, cutting the reaches down to a depth of 0.4 m in- feature into two halves (Figure 7). side the circular ditch in the proxim- While excavating, we noticed that the ity of the above-mentioned mounds. mounds were actually formed by the Unit 4, situated 10 m outside the ditch, backfill coming from oval-shaped de- revealed a cultural layer of a similar pressions measuring approximately thickness, but soil in Unit 5 placed 50 two square meters and situated adja- m to the east of the earthwork was cul- cent to the mounds. The thickness of turally sterile. Unit 3, excavated in the the cultural layer which lay underneath bottom of the ditch on the northern a thin humus layer was only 0.1 m on side of the site, exposed a midden de- top of the mound, but reached down posit: dark-colored culturally altered to a depth of 1.3 m on the side of the soil containing a considerable amount depression. Most of the pottery (64%) of charred organic matter and ceram- was concentrated in the depression, in ics (303 potsherds), which begun at the the levels 0.1-0.3 m. The number of 0.2-m-level and continued to a depth potsherds decreased gradually in the of 1.2 m. next three levels, but increased again in Fifteen percent of the pottery recov- the 0.6-0.7-m-level, which was also the ered at the Las Palmeras site is diag- last level to contain pottery. Immedi- nostic. Most potsherds have caraipé ately underneath, in the 0.75-m-level, temper, but also sand and hematite we observed a 0.15-m-thick feature temper was used. Traces of brown slip consisting of a powdered substance in can be observed, as can incised design the dark brown cultural soil. A sample elements that form straight parallel of this substance was later analyzed, lines, curved lines, and geometric de- revealing it to be calcite3. signs (Figures 8 and 9F). A few sherds Trench 2 (0.5 m x 2 m) was placed feature fingernail impression (Figure on another mound, located some 20 9D), and some of the incised sherds m to the southwest of the Trench 1. have remnants of red paint. Decorated The thickness of the dark brown cul- pottery represents a finer type of ware, tural layer was some 0.2 m on top of while the undecorated sherds, probably the mound and almost 1.3 m on the from large containers, are more robust

116 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks

Figure 9 - A-C: Estancia Giese assemblage, Figure 8 - Las Palmeras assemblage, in- incised rims (diameter 18 cm) and fine-line cised rim (diameter 28 cm) from Unit 3 incised rim (diameter 12 cm), D-F: Las and below incised rim (diameter 20 cm) Palmeras assemblage, fingernail-incised from Trench 1, drawn to the same scale. rim, inverted rim, and globular incised body. The length of the incised body sherd and poorer in quality. Rims are everted, (F) is 10 cm, all fragments drawn to the inverted or direct (Figures 8 and 9E). same scale. Lips are tapered, rounded and thickened or out-turned and squared. Some that we measured and mapped in 2002 fragmentary handles and flat bases are has a 400-m-long perimeter and a di- also present in the assemblage. The ameter of 225 m (Figure 6). Because of main aspects of the ceramics are simi- dense secondary vegetation, we could lar, regardless of the two separated oc- not verify the course of the embank- cupation phases indicated by the 14C ment to the northwest. The difference dates (discussed below). in elevation between the top and the base of the embankment varies between 0.5-1 m, and the embankment’s ESTANCIA GIESE width is 10-15 m. The earthwork prop- The ranch of Roger Giese is situated ably was originally constructed adja- 6 km to the northeast of Riberalta and cent to a minor watercourse. The bed 2 km to the north of the Las Palmeras of this now-abandoned stream runs site. An embankment found on the on the eastern and northeastern sides ranch is eroded due to cane sugar cul- of the embankment at a distance of tivation and earlier cattle farming. The some 100 m. semicircular part of the embankment

Amazônica 2 (1): 104-138, 2010 117 Saunaluoma, S.

While laying the foundations for a hen rated, and only 10% of the shreds are house in the early 1990s, the proprietor diagnostic. The general characteristics found a small polished stone axe with of the pottery are predominant use side notches at a depth of some 0.2 m of cariapé temper, hematite temper below ground level. In 2001, we exca- in lesser quantity, incised and fine-line vated a 1 m x 1 m test unit near the incised decoration, direct rims in finer spot at which the axe had been found, ware (Figure 9A-C), and out-turned 30 m to the east of the embankment. thickened lips in the utility ware. The This unit suggested that the cultural few recovered basal fragments are material was sparse, only a few erod- from flat-bottomed vessels. ed potsherds were observed in the 0.2-0.4-m-level. EL CÍRCULO In 2002, a six additional 1 m x 1 m test units were excavated at the site to study The El Círculo site is situated in the the extension of the occupational area community of Las Piedras, 7 km to the located in 2001. Unit 1 was placed on southwest of Riberalta. The existence top of the embankment and revealed of the El Círculo earthwork has been only one potsherd at a depth of 0.6 m. known to the local people for some The uppermost layer (0.0-0.3 m) exca- time. This curious earthwork was en- vated consisted of soil mixed with he- countered and chosen as an object matite fragments. This deposit of he- of study during the 1997 preliminary matite was observed on the surface of fieldtrip to the region. Jorge Arellano the embankment of the entire north- surveyed the earthwork in 1999, but ern part of the earthwork. The upper did not mention any other indication levels of Unit 2, excavated 20 m to the of cultural remains (Arellano 2002:56). south of the embankment, had been The El Círculo earthwork differs from disturbed by cane sugar cultivation. the other earthworks documented so Three potsherds were recovered, again far in the region as it consists of a cir- in the 0.6-m-level. In Units 3-5, placed cular embankment 350 m in diameter 20 m, 40 m, and 100 m to north of and with an entrance towards the old the Unit 1, evidence of cultural mate- riverbank of the Beni River (Figure rial was scarse, only minor concentra- 6). Eleven mounds measuring 10-18 tions of potsherds in the levels 0.2-0.4 m in diameter were observed attached m. Unit 6, located near the base of the to the inner side of the embankment. embankment on the western side of Contemporary land use at the site the site, contained a cultural layer asso- has caused erosion in some parts of ciated with an accumulation of ceram- the earthwork. Today, the height of ics and organic material, charcoal, and the embankment is approximately 1 carbonized seeds at depths of 0.4-0.8 m. m, and the mean width is 10 m. Ap- The ceramics recovered at the Estancia proximately 150 m to the east of the Giese site are fragmented and deterio- circular embankment, a 15-20-m-wide

118 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks ditch begins on the old bluff of the During the 2003 field season, more Beni floodplain, runs 380 m to the comprehensive excavations were car- north, and then turns and continues ried out at the site to define the lim- some 140 m to the northeast, vanish- its and extent of the cultural strata. ing gradually. The ditch has a depth Nineteen 0.6 m x 0.6 m test units were of 0.5-1 m, and is accompanied by a excavated at 25 m intervals along an low embankment formed of the ditch E-W-line transecting the site beginning backfill on the eastern side. from the western side of the circular embankment and ending at the ditch In 2001, we excavated five 1 m x 1 situated to the east of the site. Three m test units at the site. Unit 1 was additional test units were excavated at placed on one of the westernmost the ends of the ditch, and one test unit mounds attached to the embankment. The cultural layer began at a depth of 0.2 m, where the first few potsherds and dark colored soil were observed. The 0.4-0.6-m-level yielded a similar amount of pottery, but the soil contained more charcoal and burnt clay. The 0.6-0.8-m-level was associated with a hearth feature, consisting of a 0.1-m-thick crust of burnt clay beneath Figure 10 - El Círculo, profile of the trench which a concentration of carbonized excavated in 2003. wood on top of another 0.1-m-thick layer mixed with burned clay and soot was placed 20 m to the north of the was observed. The base of the cul- entrance of the circular embankment. tural deposit (0.8-1.0 m) contained an None of these units contained archae- accumulation of pottery (48% of the ological material. sherds recovered from this unit). Some A well-preserved mound located on fragments of resin were also found in the eastern side of the site was cho- the cultural strata. Unit 2, situated on a sen for excavation to clarify the stra- smaller mound 80 m to the northeast tigraphy of the mounds attached to of the Unit 1, suggested an even dis- the embankment. The mound is 1.5 m tribution of ceramics throughout the high and 18 m in diameter. A 1 m x 12 cultural deposit. Potsherds were found m trench oriented in an E-W-direction between the 0.2 and 0.6 m levels. The was positioned to intersect the highest 0.3-0.4-m-level had an accumulation point and the inner edge of the mound. of fragments of burnt clay and char- The trench was divided into twelve 1 m coal in a sooty soil. Units 3-5 excavated x 1 m units which we excavated in arti- in other parts of the site at some dis- ficial 0.1-m-levels. Culturally sterile soil tance from the embankment turned was reached at a depth of 0.4 m in the out to be culturally sterile. western end of the trench (Unit 1) and

Amazônica 2 (1): 104-138, 2010 119 Saunaluoma, S.

at 1.5 m in the eastern end (Unit 12), incised decoration is sparse, although on top of the mound (Figure 10). several fingernail-incised sherds (Fig- ure 11B) were recovered. Traces of a Underneath a thin humus layer, we dark brown as well as red all-over slip encountered a culturally sterile light and red painting on white slip are still reddish-brown soil. The thickness observable on several sherds (Figure of this layer was 0.15 m in the west- 11F). Rims are direct or everted with ern end of the trench and 0.40 m on squared or rounded and thickened lips top of the mound. Below this sterile (Figure 11A). Flat basal fragments as layer, we observed a 0.15-0.45-m-thick well as pedestal bases (Figure 11C) are yellowish-red (5 YR 4/6) stratum. In present, as are spindle whorls (Figure the 0.6-1.3-m-levels, from Unit 5 to 11D-E). Reconstructed vessels bod- the eastern end of the trench, a red- ies are globular in shape, and open- dish-brown (5 YR 4/4) layer was en- mouthed bowls are also present in the countered. This layer mixed with the assemblage. Some fragments of grind- yellowish-red one in Units 7, 11, and ing stones made of laterite slabs and 12. Both layers contained accumula- small fragments of resin4 were also re- tions of potsherds and fragments of covered. Surface collection at the site burnt clay as well as organic material, yielded a re-utilized polished stone axe. such as carbonized seeds. In Unit 12, at a depth of 1.3 m, we encountered a thin dark reddish-brown (5 YR 2.5/2) CHACRA TELERÍA layer mixed with charcoal reaching to a depth of 1.45 m. The Chacra Telería site is located about 3 km to the northwest of the El Círcu- In Units 4 and 5, the 0.4-0.5-m-level lo site and the present-day Beni flood- consisted of a hearth feature with a plain. The proprietor of the land had concentration of charcoal and burnt noticed abundant potsherds scattered clay. Another hearth was found in Unit on the surface of his field, but he was 8, level 0.7-0.8 m. Unit 6 yielded one unaware of any earthworks related to third of an undecorated ceramic vessel these finds. To clarify the chronology which had remains of organic material and variation of the cultural material and carbonized seeds inside. A total of of this site in relation to the earthwork 658 sherds (19 kg) were obtained from sites of the region, a surface collection the trench excavation, most of which of the diagnostic pottery was made, (53%) were found in Units 6-8. The and a 1 m x 1 m test unit was excavated major accumulation of pottery was in during the 2005 field season. The test the levels 0.5-0.8 m. unit exposed a cultural layer at a depth The ceramics of the El Círculo site of 0.2-0.5 m associated with abundant differ from the other ceramic assem- ceramics. blages of the Riberalta region. Twelve Thirty-three percent of the sherds re- percent of the sherds are diagnostic. covered from the Chacra Telería site The ware is caraipé tempered, and

120 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks

Figure 11 - El Círculo assemblage, A: rims (diameters 18-21 cm), B: fingernail-incised rim, C: pedestal base, D-E: spindle whorls, F: painted rim (diameter 14 cm). The diameter of the discoidal spindle-whorl (D) is 4,5 cm, all fragments drawn to the same scale. are diagnostic. The ware is caraipé, surface. This site does not contain any sherd, and sand tempered. Rims are visible earthworks. In 2005, we exca- everted, but also some direct rims are vated a 1 m x 1 m test unit which represent. Lips are rounded and thick- vealed a dark reddish-brown culturally ened, and the base sherds are flat. Rim altered soil continuing to a depth of diameters vary between 10 cm and 20 0.4 m, underlain by a culturally sterile cm in decorated pottery, and larger un- latosol. The composition of the cul- decorated rimsherds have a diameter turally altered soil, perceived as darker up to 40 cm. The few defined vessel in color and containing sand, differs forms represent globular and round from the soil of the other sites we ex- bodies. Frequently used decoration cavated. Pottery was found only in the techniques are incision and fine-line 0.0 m-0.2-m-levels. The depth of the incision (Figure 4E-G). Dark brown cultural layer is more superficial than all-over slip occurs on some sherds. at the other sites we investigated in the region.We found two kinds of pottery at the Candelaria site: a coarse ware and CANDELARIA a fine-grained ware, both of which are Adjacent to the bluff of the Madre de caraipé tempered. The small sample of Dios River, near the small community ceramics (27 potsherds) from the site of Candelaria, local inhabitans had lo- does not include diagnostic sherds. cated abundant potsherds, including a small pedestal base, scattered on the

Amazônica 2 (1): 104-138, 2010 121 Saunaluoma, S.

DOS PALMAS new tradition of robust, almost non- decorated ceramics, interpreted as hav- Dos Palmas, a small community of ca. ing had its roots in the Central Ama- 100 inhabitants, is situated 5 km to the zonian Polychrome Tradition, arrived southwest of the community of Can- in the region (Lathrap et al. 1985:47). delaria. A brief surface collection car- The period A.D. 300-800 corresponds ried out during the 2005 field season to the arrival and dominance of the in a manioc field yielded few undiag- Panoans along the Ucayali River. They nostic potsherds and a sandstone flake. brought with them the Pacacocha Tra- The Dos Palmas site does not have any dition of technologically simple ce- earthworks. ramics, featuring all-over red slip, zoo- morphic adornos, corrugation, everted ESTANCIA VELASCO rims, and globular forms (Lathrap et al. 1985:47-48). The hallmarks of Halfway between Tumichucua and Es- the subsequent tradition, Cumancaya tancia Girasol, we noticed archaeologi- (A.D. 800-1250), include corrugated, cal remains on a stream bank. A sur- fine-line incised, and brushed decora- face collection at the site yielded some tion, as well as use of caraipé temper fragmented grinding stones made of (Myers 2002:76). The appearance of laterite slabs, a fragment of an abraded the region’s last recognized prehistoric stone artifact, a fragment of a pottery ceramic tradition dates to around A.D. vessel base, as well as some fragments 1200, when Tupi-speaking groups, of fine-grained ware with caraipé tem- producers of highly elaborated poly- per. This site has no visible earthworks. chrome ceramics, invaded the floodplains of the Ucayali River (Lathrap et al. 1985:91). CERAMIC TRADITIONS OF THE SOUTHWEST AMAZON Ondemar Dias, who carried out the first archaeological investigations in Information about the ceramic se- the State of Acre in the 1970s, named quences of the Southwest Amazon the ceramic tradition related to the is scarce. So far, the most completely sites of the Juruá River Acuriá and documented are those of the River the ceramic tradition of the eastern Ucayali Basin (De Boer 1972, Lathrap part of the State of Acre in the River 1970, Myers 2002, Roe 1973, Weber Purus Basin Quinari. He subdivided 1975). Tutishcainyo, the earliest ceram- the Quinari tradition into four phases: ic style recognized in the region, was Iquiri, Xapuri, Iaco, and Quinari (Dias replaced around 300 B.C. by Hupa-Iya 2006). The characteristics of the Qui- modeled-incised style pottery, whose nari tradition include a great variabil- introduction is interpreted as result- ity of vessel forms, cylindrical forms ing from Arawakan migration into the dominating, caraipé used as the most area (Lathrap et al. 1985:46-47). Ap- common tempering material, and red proximately A.D. 90, Yarinacocha, a

122 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks slip and red lines on white slip used as 2006:280). the predominant decorative technique. The ceramic traditions of the northern Incised potsherds are also found, but Bolivian lowlands are poorly known. in lesser quantities. Typical of the Qui- The fragmentary pottery collected at nari tradition is an anthropomorphic sites with earthworks along the Or- vessel in the form of a cylinder on a thon River is generally tempered with globe, representing a human face in caraipé, hematite, ground sherds, and which the design is executed using the shell. The few defined vessel forms are appliqué technique. globular in shape with everted rims. The common traits of the ceramic se- Additionally, the use of all-over slip, in- quences of the Central Llanos de Mo- cision, and excision has been reported jos are: the use of ground sherd, shell, (Arellano 2002). The Finnish-Bolivian and cauxi temper, incised-modeled, excavations at the La Fortaleza de Las incised, and polychrome painted deco- Piedras site yielded a handful of Inca- ration, globular tripode and tetrapode style ceramics, a much larger quantity vessels, and pedestal bases as well as of typical regional caraipé and sand large discoidal plates and grater plates tempered utility ware, a few sherds (Calandra & Salceda 2004:159, Dough- of painted and incised pottery, as well erty & Calandra 1981, Jaimes 2009). In as larger fragments of more sophisti- addition, pottery related to the Barran- cated pottery, most likely ceremonial coid Tradition and the Central Amazo- bowls, decorated with incised geomet- nian Polychrome Tradition has been ric designs and bearing some common recovered at some mound settlements features with the Sivia-style pottery of (Lathrap 1970:124-125,159, Norden- the Upper Amazon of Peru (Siiriäinen skiöld 1913). The mound settlements et al. 2002; Korpisaari et al. 2003). have been dated to around the time span A.D. 350-1300 (Dougherty & Calandra 1982) and A.D. 600-1400 CERAMIC STYLES OF THE RIBERAL- (Jaimes 2009). The late pre-Columbi- TA REGION an ceramics from the Baures region, Unfortunately, the quite fragmented in northeastern Llanos de Mojos, are and eroded state of the ceramics col- mainly cauxi, kaolin, or ground sherd lected in the Riberalta region did not tempered and fine-line incised. Figu- allow us to accurately define the ves- rines, appliqué technique, supports, sel body forms and surface treatment and painting are also present, but to a of many sherds. The few recon- lesser extent. (Dougherty & Calandra structed body forms are globular and 1985:50, Prümers et al. 2006:274-279). rounded. Thick undecorated sherds, Some fragments of flat bottomed probably from large containers, with vessels with impressions from palm rim diameters up to 40 cm are com- fiber basketry were recovered in the mon in all Riberalta assemblages. The Bella Vista complex (Prümers et al. documented ceramic assemblages dis-

Amazônica 2 (1): 104-138, 2010 123 Saunaluoma, S.

play some common traits typical of upstream from the Central Amazon Amazonian ceramics, such as caraipé occurred only after A.D. 900, and pos- and ground sherd temper, incision, sibly influenced the ceramic complexes fine-line incision, all-over slip and, of the Bolivian tropical savanna low- to a lesser extent, fingernail incision, lands around A.D. 1000 (Brochado while the most distinctive attributes 1984:329-330). Curralinho pottery is of the Incised-Punctated and the also characterized as predominantly Amazonian Polychrome Tradition are having caraipé temper (Simoes & lacking. The El Círculo ceramics dif- Lopes 1987), which is rare in Tumi- fer from other assemblages in having chucua pottery. painted ware, mainly red designs on white slip, open-mouthed-bowls, and spindle whorls. Otherwise, the stylistic INTERPRETATIONS AND CHRONOL- differences among the ceramics of the OGY OF AMAZONIAN EARTH- investigated sites are slight, regardless WORKS of the chronological dispersion. The Scholars have interpreted the function Riberalta ceramic assemblages can be and significance of the Amazonian considered as regional styles which earthworks in different ways, taking have some features typical of late pre- into consideration the location, tech- historic ceramic traditions of the Up- nical attributes, dating, and cultural per Purus River Basin and the Ucayali affliation. One of the most frequent River Basin, although the corrugation explanations for the different shapes occurring in the Pacacocha and Cu- of ditches is that they protected the mancaya traditions is not present. occupation areas (Heckenberger et al. Arnold and Prettol (1988:462) men- 1999, 2008; Pärssinen et al. 2003; Pe- tion that Tumichucua pottery is com- tersen et al. 2001:97). Arnold and Pret- parable to the Curralinho ceramic tol (1988:463) also explain the function complex of the lower Madeira River, of the earthworks at the Tumichucua while Thomas Myers (1988:76) finds site as defensive moats. This inference it related to Miracanguera, a subtradi- is supported by the fact that their exca- tion of the late prehistoric Amazonian vations revealed a possible posthole on Polychrome Tradition. However, our the inner edge of the circular ditch D, dates of the Tumichucua site (72 ± leading them to suggest that this sector 81 cal. B.C. and cal. A.D. 98 ± 43) do of the site was encircled by palisades not support these inferences. The Cur- (Arnold & Prettol 1988:461-462). ralinho complex, which pertains to the Arellano (2002) argues that in the less- Incised-Punctated Tradition, has three complex earthwork sites located along- dates: A.D. 840 ± 60, A.D. 885 ± 90, side the Orthon River, the occupation- and A.D. 1451 ± 55 (Simoes & Lopes al area is surrounded and protected by 1987:122). Furthermore, the spread of semicircular or square-shaped ditches the Amazonian Polychrome Tradition adjacent to the bluff of the river. Con- sequently, this kind of protection has

124 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks been considered advantageous dur- are situated in the landscape, suggest ing the Amazonian late prehistory ritual functions that were part of a tra- which is often described as a period dition of shared collective perception of intensifying waves of migration and and ideology. ethnic expansion, accomplished by While some ditches reserve water dur- Tupi-Guarani groups (Brochado 1984; ing the rainy season, Martti Pärssinen Heckenberger 1996; Lathrap 1970:78- and colleagues (2003:101,130) sug- 79; Lathrap et al. 1985:91-94; Pärssin- gested that the ditches of the State of en et al. 2003; Wüst & Barreto 1999:6). Acre could have also been used for Moreover, the Pano-speaking groups aquaculture, as ethnographic literature have a lengthy history of conflicts with mentions the raising of small fishes, the neighboring ethnic groups, as well molluscs, and turtles in artificial aquat- as of constant endo-warfare (Santos- ic depressions near the indigenous vil- Granero 2002:29). Villages surrounded lages. by moats and palisades in the eastern lowlands of Bolivia are reported in the Only a small number of the earth- ethnographic record (Denevan 1966; work sites have been dated. In the Erickson et al. 2008; Metraux 1948:82; Llanos de Mojos, indications of hu- Nordenskiöld 1918), so the building man settlements and the alteration of of protective palisades may have been the savanna began by 900 B.C., but the practiced also in pre-Columbian times, active use of the massive earthworks at least during the late prehistoric pe- related to agricultural activities took riod. place between 400 B.C. and A.D. 1500 (Erickson 2006: 253). In the commu- The ring-ditch sites of the savannas of nity of Bella Vista, situated at the con- the Baures region may have enclosed fluence of the Blanco and San Martin settlements, or been constructed for Rivers, in the northeastern Llanos de defense, elite residences, cemeteries, Mojos, a site containing burials and ritual spaces, and/or special gardens surrounded by ditches was dated to (Erickson 2006:259, Erickson et al. A.D. 1300-1400 (Prümers et al. 2006). 2008, Prümers et al. 2006). Some of The Upper Xingu region’s earthwork the more complex earthwork sites sites also date to the late prehistoric found in the State of Acre possibly period (Heckenberger et al. 1999), as served as ceremonial centers or meet- does one of the occupation sequences ing places (Pärssinen et al. 2009) – of the Fazenda Colorada, one of the or both – and would thus be related most complex earthwork sites in the to the cosmology and socio-political State of Acre, having been dated to concerns of the peoples that occupied the end of the thirteenth century A.D. them. The repetitive pattern of precise (Pärssinen et al. 2003:122-123). Ce- geometric forms of the earthworks, ramics recovered at the earthwork sites carefully planned and constructed, and of Xipamanu and Alto Alegre, locat- the way in which these constructions ed in the eastern part of the State of

Amazônica 2 (1): 104-138, 2010 125 Saunaluoma, S.

Acre, in the Purus Basin, have yielded precisely because of the presence of thermoluminescence dates of approxi- the existing earthwork. mately 1200-0 B.C. and 200 B.C.-A.D. The radiocarbon dates5 recovered at 200, respectively (Nícoli 2000:131). the Estancia Giese earthwork site (cal These published dates reveal that the A.D. 194 ± 54 and cal A.D. 330 ± 57) Amazonian earthwork engineering is a indicate that the initial stage of the use cultural tradition covering an extended of this site was roughly coincident with period. Las Palmeras’ first occupation phase. The Candelaria site, situated adjacent to the floodplain of the Madre de Dios DATING OF THE INVESTIGATED River, dates to cal A.D. 327 ± 56, and SITES is contemporaneous with the later date One of the principal goals of our re- of occupation of the Estancia Giese search in the Riberalta region was to site. Based on a single date, the Estan- collect radiocarbon samples from se- cia Girasol site seems to be much more cure archaeological contexts. So far, recent (cal A.D. 1428 ± 13). 18 samples from seven sites have been In total, we have eight radiocarbon dated (Table 2). dates from the El Círculo site. The Tumichucua and Chacra Telería sites 14C dates indicate that the occupation represent the earliest occupation of the of the site lasted for slightly over one sites discussed here, dating from the hundred years (from cal A.D. 1272 ± first century B.C. to the first century 11 to cal A.D. 1349 ± 44)6. It is remark- A.D. At the Las Palmeras site, we dated able that the site was inhabited for only two different components of the site: such a relatively short time, just for a one of the small mound/depression few generations. The active river chan- features situated inside the ditch and nel of the Beni is nowadays situated the cultural layer in the circular ditch at some hundreds of meters from the a depth of 0.6 m. The sample collected site. The migration of the river could from cultural soil located directly be- have caused the site to lose its principal low the calcite feature in Tench 1 is late attraction, the proximity to a navigable (cal A.D. 1582 ± 57). In contrast, the water route, gradually leading to the date obtained from the ditch, is much abandonment of the site. Similar cir- earlier (cal A.D. 159 ± 51). Thus, the cumstances may have been at play at mounds/depressions correspond to the Estancia Girasol and Candelaria sites. a later phase of human activity at the Based on the radiocarbon dates, two site. Based on the moderate depth (0.4 tentative separate periods of hu- m) of the cultural layer inside the area man occupation can be proposed for demarcated by the ditch, the occupa- the archaeological sites of the Riber- tion of the site was not continuous. alta region: an initial period dating to Perhaps the abandoned site was cho- 100 B.C.-A.D. 400 and a later period sen to be re-occupied in a later phase continuing from A.D. 1200 until the

126 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks

Table 2 Radiocarbon dates of the sites.

Site Unit, Context Lab id C14-age bp Calibrated Level date Tumichucua 2002 Unit 1, Charcoal in Hela-702 1905 ± 40 Cal 98 ± 43 - 40 cm cultural layer AD Tumichucua 2005 Unit 6, Charcoal in Ua-24932 2045 ± 65 Cal 72 ± 81 - 60 cm cultural layer BC

Estancia Girasol Unit 1, Charcoal in Ua-24929 475 ± 35 Cal 1428 ± 2005 - 40 cm cultural layer 13 AD

Las Palmeras Trench 1, Charcoal in Ua-24076 285 ± 35 Cal 1582 ± 2001 -96 cm cultural layer 57 AD Las Palmeras Unit 3, Charcoal in Ua-24930 1850 ± 40 Cal 159 ± 51 2005 - 60 cm cultural layer AD

Estancia Giese Unit 4, Charcoal in Hela-708 1815 ± 45 Cal 194 ± 54 2002 - 40 cm cultural layer AD

Estancia Giese Unit 6, Charcoal in Hela-709 1695 ± 40 Cal 330 ± 57 2002 - 60 cm cultural layer AD Estancia Giese Unt 1, Charcoal Hela-707 10355 ± 80 Cal 10318 ± 2002 - 80 cm under cultural 211 BC layer El Círculo Unit 2, 50 Ceramic tem- Hela-570 1790 ± 75 Cal 233 ± 93 2001 - 60 cm per AD El Círculo Unit 1, Hearth Hel-4585 600 ± 60 Cal 1349 ± 2001 - 85 cm 44 AD El Círculo Unit 7, Charcoal in Poz-9523 680 ± 30 Cal 1326 ± 2003 - 20 cm cultural layer 45 AD

El Círculo Unit 7, 40 Soot on sherd Poz-9426 715 ± 30 Cal 1272 ± 2003 - 50 cm surface 11 AD El Círculo Unit 6, Charcoal under a Poz-9524 650 ± 30 Cal 1334 ± 2003 - 53 cm ceramic vessel 41 AD El Círculo Unit 8, Hearth Poz-9427 660 ± 30 Cal 1331 ± 2003 - 80 cm 42 AD El Círculo Unit 10, Carbonized Poz-9428 685 ± 30 Cal 1324 2003 100 seed in cultural ± 46 AD - 110 cm layer

Amazônica 2 (1): 104-138, 2010 127 Saunaluoma, S.

El Círculo Unit 10, Soot on sherd Poz-9429 645 ± 30 Cal 1335 2003 120 - 130 surface ± 40 AD cm Chacra Teleria Unit 1, Charcoal in Ua-24931 1940 ± 40 Cal 57 2005 - 42 cm cultural layer ± 44 AD Candelaria Unit 1, Charcoal in Ua-24928 1700 ± 40 Cal 327 2005 - 25 cm cultural layer ± 56 AD

period of European contact7. What, mainly inside the area delineated by the then, happened before and between earthworks in the Tumichucua, Estan- these two periods? Assuming that the cia Girasol, Las Palmeras and Estan- settlement pattern was preferably riv- cia Giese sites. Moreover, the ditches erine and bearing in mind the active contained quantities of cultural debris. meandering on the floodplains of the The moderate depth of the archaeo- Beni and Madre de Dios Rivers, many logical deposits and lack of marked archaeological sites today may today be stratigraphy did not indicate centuries- located far from the active river chan- long uninterrupted occupation at the nels, and some have been destroyed by investigated earthwork sites. The occu- the relocations of the watercourses. pation persisted for a few generations, The present survey found a few of the or the use of the sites was more spo- possibly hundreds of archaeological radic than continuous. sites in the region. Therefore, the ini- By building enclosures, people may tial stage of human occupation could have wanted to protect themselves have begun centuries, or even millen- from the natural and supernatural nia earlier, and the region may have worlds, too. The vast rainforest encir- been inhabited more or less continu- cling the villages was not completely osly with settlement locations along tamed, although the prehistoric in- navigable water routes. habitants of Amazonia were certainly acquainted with their environment. EARTHWORK FUNCTIONS AND Many activities occurred in the village SETTLEMENT LAYOUT at night when darkness transformed the surrounding rainforest into a place I argue that the primary purpose of more mysterious and dangerous than simple, individual earthworks, such as in daylight. Even today, the forest is those of the Riberalta region, was to home to many inexplicable entities and visibly delimit and mark the area of supernatural beings or spirits such as occupation. In some cases the earth- caboclinho da mata, the powerful guard- work enclosure could have been more ian of the forest, recognized through- symbolic than practical. Based on out present-day Amazonia (Virtanen the test excavations, the archaeologi- 2008:126). Any understanding of the cal deposits seem to be concentrated spiritual life of the pre-Columbian

128 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks communities needs to be grounded Las Palmeras site, the earthworks may in archaeological data, but one can have had completely different function imagine the multitude of mythical or meaning for the new inhabitants. creatures, some benevolent, some evil, The small mound/depression fea- which formed part of the universe of tures inside the perimeter of the Las the past Amazonian peoples. Palmeras ditch are elements that we Some of the ditches can be interpret- did not find at the other investigated ed as evidence of water management. sites. Arellano (2002:67) interprets Ditch B at the Tumichucua site may them as habitation mounds, but our have served as a canal facilitating ac- excavations revealed that these forma- cess between the oxbow lake, the river, tions are more likely depressions filled and the floodplain (Arnold & Prettol with abundant sherds and midden de- 1988:463). The Estancia Girasol and bris. The small mounds were probably El Círculo sites also have simple ditch- formed by backfill removed from the es which pass within several hundred depressions. In addition to ceramics we meters of the occupation sectors and observed a calcite feature in one of the could have been used as canals to link depressions, in Trench 1 (Figure 7), but smaller streams with more primary wa- the implication of this feature remains tercourses. uncertain. Today, calcite has more uses than almost any other mineral as con- In addition, the ditch at the Las Palm- struction material, a pigment for paint, eras site apparently served as a dump a field dressing, and in medication. for domestic waste, judging by the in- tensely dark color of the cultural layer At the Estancia Girasol site, we docu- found on its bottom. This charcoal mented a road, a structure that has so rich cultural layer was different from far not been observed at any other that found inside the earthwork. At earthwork site in the Riberalta region. Tumichucua we also found fragments Since we only mapped the Estancia of pottery and charcoal in the circular Girasol road partially, and did not ex- Ditch D, which was also observed by cavate or date the feature, it may either Arnold and Prettol (1988:461), as in belong to the same complex as the the outer double-ditch of the Estancia ditches or be a remnant of an earlier Girasol site. Still, the possible utiliza- or later occupation. Roads are a basic tion of these earthworks in the dispos- element of the earthwork sites of the al of refuse should not be considered State of Acre (Pärssinen et al. 2003, as one of their central functions, but 2009; Schaan et al. 2007) and Llanos de rather as one of the secondary ways in Mojos (Erickson 2001, 2006) probably which the ditches situated in the prox- due to the complexity of these two re- imity of the settlements could have gions’ ancient settlement networks and been used. Furthermore, if the earth- to environmental factors, at least in the work sites were later re-occupied, as case of seasonally inundated savanna. seems to have been the case with the The distinctive earthworks are linked

Amazônica 2 (1): 104-138, 2010 129 Saunaluoma, S.

by roads, forming a complex entity of central plaza as the political and ritual clusters of occupational sites, which is center of the village can still be seen not the case of the earthwork sites of in the Upper Xingu area among the the Riberalta region. present-day Kayapó (Turner 2002) and Kuikuro, of which the last-mentioned In regard to the settlement plan, the group may be preserving the cultural El Círculo site was apparently differ- characteristics of their ancestors, al- ent from the other sites studied in though the size of the villages today the region. I interpret the El Círculo is much smaller (Heckenberger et al. mounds as the remains of residences 2008:1215). The circular village plan because our excavations revealed in- linked with socioceremonial organiza- tense concentrations of ceramic and tion is also common among the con- domestic debris (Figure 10), including temporary Gê and Bororo groups of hearth features, in the mounds form- Central Brazil (Nimuendajú 1946:37, ing part of the embankment. The Wüst & Barreto 1999:3-4). circular occupation zone for domestic activities encloses an area which All the sites surveyed in the Riberalta was presumably kept clean as a plaza region lack the dark anthropogenic soil, since the test units excavated inside or Amazonian Dark Earth (ADE), that this zone contained no archaeologi- is often found in the pre-Columbian cal remains. The earthwork type and riverine sites of the Central and Low- the settlement plan, together with the er Amazonia (Glaser & Woods 2004, presence of a distinct pottery style Lehmann et al. 2003, Petersen et al. contrast with the other pre-Columbian 2001, Woods et al. 2009). Despite of sites found so far in the region of Ri- the absence of ADE, and the circular beralta, suggesting that the occupants rather than linear organization of the of the El Círculo site may have had a sites, Tumichucua, Estancia Girasol, different ethno-cultural background8. Estancia Mendez, El Círculo, Cande- laria, as well as the River Orthon sites The settlement form and size (350 can be regarded as bluff sites (Dene- m in diameter) of the El Círculo site van 1996) situated on upland terraces resemble the ring villages of Central adjacent to the floodplains of the main Brazil in which a circular, elliptical, or river channels. However, the locations semicircular arrangement of houses of the sites of the Riberalta region are enclose the central plaza (Wüst & Bar- not exclusively riverine. For example, reto 1999). The ring villages prevailed the sites of Las Palmeras, Estancia around A.D. 800-1700 and are associ- Giese, Chacra Telería, and Dos Palmas, ated with the Aratu and Uru ceramic are situated a distance from the naviga- traditions (op. cit.). The circular plaza ble main rivers. However, because of was the public, communal sector of the extensive meandering and migra- the village, reserved for certain gath- tion of the rivers of the Riberalta re- erings, festivities, rites, and ceremonial gion, these sites may have been located displays. The habit of considering the

130 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks closer to a triburitary or main channel are easily transferred from one ethnic when occupied in the distant past. group to another, which can be seen, for example, among the modern-day Piro who belong to the Maipuran- DISCUSSION Arawakan linguistic group but whose Construction and use of earthworks ceramic decorative/artistic tradition is was not restricted to certain prehistoric Panoan, typical of the Shipibo-Conibo periods or areas of the South Ameri- (Lathrap et al. 1985:33). However, the can tropical lowlands. The earthwork connection between certain ceramic tradition manifests itself in slightly dif- traditions and ethno-linguistic families ferent ways at different times in a num- is appealing. Donald Lathrap and col- ber of regions, determined by natural leagues (1985) and Myers (2002:105) and cultural factors. Clark Erickson propose that analogues of the Pacaco- (2006) argues that the monumental cha (Panoan) ceramics of the Ucayali earthworks of the seasonally inun- are to be found in the tropical lowlands dated savanna plains of the Bolivian of Bolivia, and that a continuous long- Amazon are domesticated landscapes term stream of people moved from shaped by a long, complex history of the Eastern Bolivian Panoan heartland past human activities, a profound pro- towards the floodplains of the Peru- cess that was driven by social demands. vian Amazon. Lathrap (1970:81) also Thus, the reasons for constructing the describes the interfluvial ecological earthworks were not only practical, but dispersal typical of Panoan groups, ex- also aesthetic, ideological, social and tending across the upper watersheds of political. The earthworks were planned the Madre de Dios, Purus, Juruá, and and constructed intentionally, with the Ucayali Rivers. Taking into account the purpose of altering the environment incised, fine-line incised and fingernail permanently and visually. Simultane- impressed ceramic styles found at the ously, the domination and manipula- Riberalta sites, the hypothesis of the tion of the landscape were symbols of Panoan connection requires careful territoriality, demonstrating to others consideration. any given group’s ability and right to On the other hand, the circular plaza occupy a given territory. These activi- village has been considered as a socio- ties left important signatures embed- cultural pattern characteristic of the ded in the landscape – traces which Arawak (Heckenberger 2002:109), al- remain long after the sites were aban- though today the circular village plan is doned and the groups that carried out also found among the Central Brazil- the landscape engineering moved away ian Gê and Bororo groups (Wüst and or ceased to exist. Barreto 1999:3). The settlement layout Cultural traditions and ethnicity are of the El Círculo site does bear a re- not necessarily related (Lucy 2005). semblance to Central Brazilian ring vil- The characteristics of material culture lages, but more sites of this type must

Amazônica 2 (1): 104-138, 2010 131 Saunaluoma, S.

be studied before any conclusions of We also recorded sites without earth- the ethnicity and/or socioceremonial works that are contemporaneous with aspects of the El Círculo inhabitants the earthwork sites. One objective for can be drawn. future research would be to determine The earthworks in the Riberalta re- if these two types of sites are associ- gion have a variety of forms and are ated with each other, being used by the found in slightly different locations, same people but for different purpos- indicating separate time periods, pos- es, if the sites were hierarchically relat- sibly distinct cultural affiliations, and ed, and if the earthwork sites correlate obviously diverse functions. Although with a single or multiple cultural and/ only a small number of the possible or ethnic traditions. pre-Columbian sites of the Riberalta region were recorded during the pres- ACKNOWLEDGEMENTS ent study, we found different types of sites indicating fairly dense occupa- I wish to thank Clark Erickson, Zbig- tion, but lacking long permanence in niew Fiema, Antti Korpisaari, Mar- the same location. The earthwork tra- tti Pärssinen, and one anonymous re- dition prevailed in the Riberalta region viewer for reading in detail the draft from at least 100 B.C. until the period of this article and providing valuable of European contact. Compared to comments, Wesa Perttola for assis- the variable layout of the sites, the ce- tance with the maps, Martti Lehtonen ramic assemblages of the region are for the X-ray diffraction analysis, and relatively homogeneous. Except for Margot Whiting for the language re- the El Círculo assemblage, which has vision. I am grateful to my colleagues a small number of incised sherds and Juan Faldín, Risto Kesseli, Jussi Ko- painted sherds in abundance, a uni- rhonen, and Antti Korpisaari for their form ceramic style prevailed in the Ri- company and assistance during the beralta region througout the late pre- fieldwork seasons in Bolivia. I would Columbian period. The environmental also like to thank the following per- conditions alone apparently did not in- sons in the Riberalta region for their fluence the position of the earthworks, contribution to the fieldwork: Onorio as some sites are situated on the river Amutari, the proprietor of the Estan- bluffs and others at some distance cia Girasol site, Roger Giese, the pro- from the navegable river channels. The prietor of the Estancia Giese site, Juan earthworks of the Riberalta region are Quenebo, the proprietor of the Las less complex and diverse than the ones Palmeras site, Juan Navi, Grover Bani, registered in the State of Acre and José Barba, Manuel Canamari, Juan Central Llanos de Mojos. The func- Cartagena, Yolanda Chau, Carlos Chi- tions of the structures may have been punawi, Floria Cuajera, Amparo Cuel- simply to enclose the occupation areas lar, Luis Medina, and Walter Telería of and, in some cases, to serve as canals. the Las Piedras community, Johnny Farfan, Carlos Imanareco, Humberto

132 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks

Marupa, Ciro Rossell, and Ricardo Ya- context, and description of the La For- mara of the Tumichucua community, taleza de las Piedras site, see Siiriäinen and and Hugo Vargas of the community Pärssinen 2001:64-70. of Puerto Rico. 3 The X-ray diffraction analysis of this The author’s first acquaintance with sample was carried out on March 10, 2005 the archaeological sites of the Bolivian in the University of Helsinki. Amazon was in 1997 during a prelimi- 4 The contemporary Cavineño use resin nary excursion to Riberalta that was re- to glaze their pottery (Plaza & Carvajal alized under the auspices of the proj- 1985:70). ect “Prehistory of the Chiriguano and Tacano”. The fieldwork seasons 2001- 5 The third date (10355 ± 80 B.P.) from the 2003 formed part of the “The Ama- Estancia Giese site is a clear outlier. The zonian Interests of the Incas” project. 14C sample was collected below the cultur- Both projects were funded by the Uni- al deposit with the objective of dating the versity of Helsinki and coordinated initial establishment of the embankment. by Ari Siiriäinen. The investigations Unfortunately, this charcoal sample turned carried out in 2005 formed part of out to actually date some ancient natural the multidisciplinary research project burn event. “Man and Nature in Western Amazo- 6 One date (cal A.D. 233 ± 93) is an excep- nian History” of the University of Hel- tion. The material dated was organic tem- sinki, Universidade Federal do Acre, per from a non-diagnostic potsherd. The and Museo Paraense Emilío Goeldi, sample was collected in 2001 from test unit financed by the Academy of Finland 2 at a depth of 0.6 m, which was the last and coordinated by Martti Pärssinen level containing ceramics. I consider that and Denise Schaan. While putting the this potsherd does not belong to the actual finishing touches on this article, I was earthwork but rather is an evidence of an supported by a Kone Foundation re- earlier occupation in the same area, taking search grant. The Ella and Georg Eh- into account that the Chacra Telería site, rnrooth Foundation provided funds situated 3 km from the El Círculo site, for processing the radiocarbon samples dates to more or less the same period (cal of the El Círculo site. A.D. 57 ± 44). 7 The La Fortaleza de Las Piedras site was NOTES possibly occupied until A.D. 1600 (Siiriäin- en 2003). 1 By Southwest Amazon, I refer to the region defined by the triburitary areas of the 8 Corresponding earthworks have been rivers Ucayali, Juruá, Purus and the white- discovered in eastern Acre. The Coqueiral water triburitaries of the Madeira River, site consists of seventeen mounds 1.5 m flowing in the alluvial extension zone of high and 15-20 m in diameter in the form the Amazon Basin. of a circle. Furthermore, the Sol de Maio site features numerous mounds forming a 2 For the detailed bibliography, historical

Amazônica 2 (1): 104-138, 2010 133 Saunaluoma, S.

circle, as well as a simple ditch in the prox- Díaz-Andren, M., S. Lucy, S. Babic & D. imity of the mounds. Both sites yielded Edwards (eds.). 2005. The archaeology of iden- surface ceramics which bear some resem- tity. Approaches to gender, age, status, ethnicity blance to El Círculo pottery (Schaan & and religion. London & New York: Rout- Bueno 2008:37, 78). ledge. Dias, O. 2006. As estruturas arqueológicas REFERENCES de terra no estado do Acre – Amazônia Ocidental, Brasil. Um caso de resiliência? Arellano, J. 2002. Reconocimiento arqueológico in Estudos Contemporaneos de Arqueologia, Ed- en la cuenca del rio Orthon, Amazonia Boliviana. ited by O. Dias, E. Carvalho & M. Zim- Quito: Museo Jacinto Jijon y Caamaño/ mermann, pp. 59-168. Palmas: Unitins, Taraxacum. IAB. Arnold, D. & K..Pret¬tol. 1988. Aborigi- Dias, O. & E. Teixeira de Carvalho. 2008. nal earthworks near the mouth of the As estruturas de terra na arqueologia do Beni, Bolivia. Journal of Field Archaeology Acre, in Arqueologia da Amazônia Ocidental: 15(4):457-465. Os Geoglifos do Acre. Edited by D. Schaan, A. Balée, W. & C. Erickson (eds). 2006. Time Ranzi & M. Pärssinen, pp. 45-56. Belém: and complexity in historical ecology. Studies in Editora Universitária Universidade Federal neotropical lowlands. New York: Colombia do Pará - UFPA; Rio Branco: Biblioteca da University Press. Floresta Ministra Marina Silva. Brochado, J. 1984. An ecological model of the Dougherty, B. & H. Calandra. 1981. Nota spread of pottery and agriculture into Eastern preliminar sobre investigaciones arque- South America. Ph.D. Diss., University of Il- ológicas en Llanos de Moxos, Departa- linois at Urbana-Champaign. mento del Beni, República de Bolivia. Re- vista del Museo de La Plata VIII 53: 87-106. Calandra, H. & S. Salceda. 2004. Amazo- nia boliviana: arqueología de los Llanos de ____. 1981-82. Excavaciones arqueológi- Mojos. Acta Amazonica 34 (2):155-163. cas en la Loma Alta de Casarabe, Llanos de Moxos, Departamento del Beni, Bolivia. DeBoer, W. 1972. Archaeological explo- Relaciones de la Sociedad Argentina de Antropología, rations on the Upper Ucayali River, Peru. N.S., XIV(2): 9-48. Ph.D. Diss., Department of Anthropol- ogy, University of California at Berkeley. ____. 1984-85. Ambiente y arqueología en el Oriente Boliviano. La provincia de ____. 1966. The aboriginal cultural geography Itenez del departamento del Beni. Rela- of the Llanos de Mojos of Bolivia. Berkeley: ciones de la Sociedad Argentina de Antropología University of California Press. XVI (N.S.):37-61. ____. 1996. A bluff model of riverine Erickson, C. 1995. Archaeological per- settlement in prehistoric Amazonia. An- spectives on ancient landscapes of the Lla- nals of the Association of American Geographers nos de Mojos in the Bolivian Amazonia, 86(4):654-681. in Archaeology in the American Tropics: Current

134 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks

Analytical Methods and Applications. Edited in the shadow of empire: Sociopolitical by P. Stahl, pp. 66-95. Cambridge: Cambridge change in the upper Xingu of Southeast- University Press. ern Amazonia, A.D. 1400-2000. Ph.D. Diss., Department of Anthropology, Uni- ____. 2001. Precolombian roads of the versity of Pittsburgh. Amazon. Expedition 43(2):21-30. Heckenberger, M. & E. Neves. 2009. Ama- ____. 2006. Domesticated landscapes of zonian Archaeology. Annual Review of An- the Bolivian Amazon, in Time and complexity thropology 38: 251-266. in historical ecology. Studies in Neotropical Low- lands. Edited by W. Balée & C. Erickson, Heckenberger, M., J. Petersen & E. Neves. pp. 237-278. New York: Colombia Univer- 1999. Village size and permanence in sity Press. Amazonia: Two archaeological examples from Brazil. Latin American Antiquity Erickson, C., W. Winkler & K. Candler. 10(4):353-376. 1997. Las investigaciones arqueológicas en la region de Baures en 1996. La Paz: University Heckenberger, M., J. Russell, C. Fausto, J. of Pennsylvania and Dirección Nacional Toney, M. Schimdt, E. Pereira, B. Franchet- de Antropología and Arqueología. Unpub- to & A. Kuikuro. 2008. Pre-Columbian ur- lished report. banism, anthropogenic landscapes, and the future of the Amazon. Science 321:1214-1217. Erickson, C., P. Álvarez & S. Calla. 2008. Zanjas circundantes: Obras de tierra monumen- Hill, J. & F. Santos-Granero (eds.). 2002. tales de Baures en la Amazonia Boliviana. La Comparative Arawakan Histories: Rethinking Paz: Unidad Nacional de Arqueología, La Paz. Language Families and Culture Area in Ama- Unpublished report. zonia. Urbana: University of Illinois Press. Gautier, E., D. Brunstein, C. Garcia, P. Jaimes, C. 2009. La cerámica de la Loma Salvati- Vauchel, O. Fuertes & J. Guyot. 2008. erra, Beni, Bolivia. Ph.D. Diss. University of Bonn. Meander dynamics of a tropical river: the Korpisaari, A., J. Faldín, R. Kesseli, J. Ko- Rio Beni (Bolivian Amazonia). Geophysical rhonen, S. Saunaluoma, A. Siiriäinen & M. Research Abstracts. 10, EGU2008-A-04299. Pärssinen. 2003. Informe preliminar de las Glaser, B. & W. Woods. 2004 (eds). Explo- investigaciones arqueológicas de la tem- ration in Amazonian Dark Earths. Berlin: porada 2002 en el sitio de La Fortaleza de Springer-Verlag. Las Piedras, in Reports of the Finnish-Bolivian Archaeological Project in the Bolivian Amazon. Heckenberger, M. 2002. Rethinking the Edited by A. Siiriäinen & A. Korpisaari, Arawakan diaspora: Hierarchy, regionality pp. 7-34. Helsinki: Department of Archae- and the Amazonian formative, in Compara- ology, University of Helsinki. tive Arawakan Histories: Rethinking Language Families and Culture Area in Amazonia. Ed- Lathrap, Donald. 1970. The Upper Amazon. ited by J. Hill & F. Santos-Granero, pp. 99- London:Thames and Hudson. 122. Urbana: University of Illinois Press. Lathrap, D., A. Gebhart-Sayer & A. Mes- Heckenberger, M. 1996. War and peace ter. 1985. The roots of the Shipibo art

Amazônica 2 (1): 104-138, 2010 135 Saunaluoma, S.

style: Three waves on Imiríacocha of there Federal Fluminense, Rio de Janeiro. were “Incas” before The Incas. Journal of Nimuendajú, Curt. 1946. The Eastern Latin American Lore 11(1):31-119. Timbira, in University of California Publica- Lehmann, J., D. Kern, B. Glaser & W. tions in American Archaeology and Ethnology. Woods (eds). 2003. Amazonian Dark Earths. 41. University of California Press. Origin, Properties, Management. Dordrecht, Nordenskiöld, E. 1913. Urnengräber und Boston, London: Kluwer Academic Publishers. Mounds im bolivianischen Flachland. Lucy, S. 2005. Ethnic and cultural identi- Baessler-Archiv 3(6):205-255. ties, in The archaeology of identity. approaches to Nordenskiöld, E. 1918. Palisades and gender, age, status, ethnicity and religion. Edited “Noxious Gases“among the South-Ameri- by M. Díaz-Andren, S. Lucy, S. Babic & can Indians. Ymer 3: 220-243. D. Edwards, pp. 86-109. London & New York: Routledge. ____. 1924. Finds of graves and old dwell- ing-places on the Rio, Beni, Bolivia. Ymer McEwan, C., C. Barreto & E. Neves (eds). 43: 229-237. 2001. Unknown Amazon: Culture in Nature in Ancient Brazil. London: British Museum Press. Pärssinen, M., A. Ranzi, S. Saunaluoma & A. Siiriäinen. 2003. Geometrically pat- Métraux, A. 1948. The Guarani, in The terned ancient earthworks in the Rio Bran- Tropical Forest Tribes. Handbook of co region of Acre, Brazil: New evidence South American Indians. Edited by J. of ancient chiefdom formations in Ama- Steward. Bureau of American Ethnology Bulle- zonian interfluvial terra firme environment, tin 143(3):69-94. Washington, D.C.: Smith- in Western Amazonia – Amazônia Ocidental. sonian Institution. Multidisciplinary Studies on Ancient Expan- Myers, G., A. Newton & O. Melgarejo. sionistic Movements, Fortifications and Sedentary 2000. The influence of canopy gap size Life. Edited by M. Pärssinen & A. Kor- on natural regeneration of Brazil nut (Ber- pisaari (eds.), pp. 97-133. Helsink: Renvall tholletia excelsa) in Bolivia. Forest Ecology and Institute Publications 14. Management 127(1-3):119-128. Pärssinen, M., D. Schaan & A. Ranzi. Myers, T. 1988. Vision de la prehistoria 2009. Pre-Columbian geometric earth- de la Amazonia Superior, in I Seminario de works in the upper Purus: a complex Investigaciones Sociales en la Amazonia, pp. 37- society in Western Amazonia. Antiquity 87. Iquitos, Peru: Editorial Ceta. 83(322):1084-1095. ____. 2002. The Pacacocha and Cumancaya Petersen J., E. Neves & M. Heckenberger. traditions at Yarinacocha, Peru. University of 2001. Gift from the past: terra preta and Nebraska State Museum. Unpublished report. prehistoric Amerindian occupation in Amazonia, in Unknown Amazon: Culture in Nícoli, I. 2000. Estudo de cerâmicas de sítios Nature in Ancient Brazil. Edited by C. McE- com estrutura de terra circulares do alto curso do wan, C. Barreto & E. Neves, pp. 86-105. rio Purus, por meio de métodos geoquímicos: Data- London: British Museum Press. ção e caracterização. Ph.D. diss., Universidade

136 Amazônica 2 (1): 104-138, 2010 Pre-columbian earthworks

Plaza, P. & J. Carvajal. 1985. Etnias y lenguas Piccoli. 2007. Geoglifos da Amazônia oci- de Bolivia. Instituto Boliviano de Cultura. La Paz. dental: Evidência de complexidade social entre povos da terra firme. Revista de Ar- Portugal, Max. 1978. La Arqueologia de la Re- queologia 20: 67-82. gion del Rio Beni. La Paz: Editora Universo. Schaan, D., A. Ranzi & M. Pärssinen (eds.). Prümers, H., C. Jaimes & R. Plaza. 2006. 2008. Arqueologia da Amazônia Ocidental: os Algunas tumbas prehispánicas de Bella geoglifos do Acre. Belém: Editora Universi- Vista, Prov. Iténez, Bolivia. Zeitschrift für tária UFPA; Rio Branco: Biblioteca da Flo- Archäologie Außereuropäischer Kulturen 1:251-284. resta Ministra Marina Silva. Roe, P. 1973. Cumancaya: Archaeological Schaan, D. & M. Bueno. 2008. Geogli- excavations and ethnographic analogy in fos do Acre. Primeiro relatório parcial de the Peruvian Montaña. Ph.D. Diss. De- campo Janeiro-Julho. Rio Branco: Uni- partment of Anthropology, University of versidade Federal do Pará & Universidade Illinois at Urbana-Champaign. Federal do Acre. Unpublished report. Santos-Granero, F. 2002. The Arawakan Siiriäinen, A. 2003. Towards the chronol- matrix: Ethos, language, and history in ogy of the Las Piedras fortress in the Bo- native South America, in Comparative Ar- livian Amazon: Four radiocarbon datings, awakan Histories: Rethinking Language Fami- in Reports of the Finnish-Bolivian archaeological lies and Culture Area in Amazonia. Edited by project in the Bolivian Amazon. Edited by A. J. Hill & F. Santos-Granero, pp. 25-50. Ur- Siiriäinen & A. Korpisaari (eds.). Vol. II, bana: University of Illinois Press. pp. 1-6. Helsinki: Department of Archae- Saunaluoma, S., J. Faldín, A. Korpisaari ology, University of Helsinki. & A. Siiriäinen. 2002. Informe preliminar Siiriäinen, A. & M. Pärssinen. 2001. The Am- de las investigaciones arqueológicas en la azonian interests of the Inca State (Tawantin- región de Riberalta, Bolivia, in Reports of suyu). Baessler-Archiv, Band 49(45-78). the Finnish-Bolivian archaeological project in the Bolivian Amazon. Edited by A. Siiriäinen & Siiriäinen, A. & A. Korpisaari (eds.). 2002. A. Korpisaari, vol. I, pp. 31-52. Helsinki: Reports of the Finnish-Bolivian archaeological Department of Archaeology, University project in the Bolivian Amazon, vol. I. Helsin- of Helsinki. ki: Department of Archaeology, University of Helsinki. Saunaluoma, S. & J. Korhonen. 2003. In- forme preliminar de las investigaciones Siiriäinen, A., J. Faldín, M. Jalkanen-Mäkelä, arqueológicas realizadas en la región de A. Korpisaari & S. Saunaluoma. 2002. The Riberalta, Bolivia, en 2002, in Reports of fortress of Las Piedras in the Bolivian Ama- the Finnish-Bolivian archaeological project in the zon, in Reports of the Finnish-Bolivian Archaeo- Bolivian Amazon. Edited y A. Siiriäinen & logical Project in the Bolivian Amazon. Edited by A. Korpisaari, vol. II, pp. 55-71. Helsinki: A. Siiriäinen & A. Korpisaari (eds.), vol I, pp. Department of Archaeology, University 1-30. Helsinki: Department of Archaeology, of Helsinki. University of Helsinki. Schaan, D., M. Pärssinen, A. Ranzi & J. Siiriäinen, A. & A. Korpisaari (eds.). 2003. Re-

Amazônica 2 (1): 104-138, 2010 137 Saunaluoma, S.

ports of the Finnish-Bolivian archaeological project in Walker, J. 2008a. The Llanos de Mojos, the Bolivian Amazon, vol. II. Helsinki: Depart- in Handbook of South American Archaeology. ment of Archaeology, University of Helsinki. Edited by H. Silverman & W. Isbell, pp. 927-940. New York: Springer. Simões, M. & D. Lopes. 1987. Pesquisas ar- queológicas no baixo/médio Rio Madeira Walker, J. 2008b. Pre-Columbian ring (AM). Revista de Arqueologia 4(1):117-134. ditches along the Yacuma and Rapulo Riv- ers, Beni, Bolivia: A preliminary review. Silverman, H. & W. Isbell (eds.). 2008. Journal of Field Archaeology 33(4):1-15. Handbook of South American Archaeology. New York: Springer. Wasson, J-G., S. Barrera, B. Barrere, D. Bi- net, D. Collomb, I. Gonzales, F. Gourdin, Stahl, P. 2002. Paradigms in paradise: Re- J-L. Guyot & G. Rocabado. 2002. Hydro- vising standard Amazonian prehistory. Re- ecoregions of the Bolivian Amazon: A view of Archaeology 23:39-50. geographical framework for the function- ____. (ed.). 1985. Archaeology in the American ing of river ecosystems, in The ecohydrology Tropics: Current analytical methods and applications. of South American rivers and wetlands. Edited Cambridge: Cambridge University Press. by M. McClain, pp. 69-91. IAHS Special o Steward, J. (ed.). 1948. Handbook of Publications n . 6. South American Indians. Bureau of Ameri- Weber, R. 1975. Caimito: An analysis of the can Ethnology Bulletin 143, Vol. 3. Washing- late prehistoric culture of the Central Ucayali, ton, D.C.: Smithsonian Institution. eastern Peru. Ph.D. Diss. Department of Sullivan, L. (ed.). 2002. The native religion and Anthropology, University of Illinois at cultures of Central and South America, Anthro- Urbana-Champaign. pology of the Sacred. New York: Continuum. Woods, W., W. Teixeira, J. Lehmann, C. Teijeiro, J., F.Santalla, S. Ajacopa & T. Laime. Steiner, A. WinklerPrins & L. Rebellato 2001. Atlas étnico de investigaciones antropológicas (eds.). 2009. Amazonian Dark Earths: Win "Amazonia Boliviana”. La Paz, Bolivia. Sombroek’s view. Berlin: Springer. Turner, T. 2002. The sacred as alienated Wüst, I. & C. Barreto. 1999. The ring vil- social consciousness: Ritual and cosmol- lages of Central Brazil: A challenge for ogy among the Kayapó, in The native religion Amazonian Archaeology. Latin American and cultures of Central and South America, an- Antiquity 10(1):3-23. thropology of the sacred. Edited by L. Sullivan, pp. 278-298. New York: Continuum. Recebido em 16/12/2009. Virtanen, P. 2008. Observacôes sobre as Aprovado em 26/02/2010. possíveis relacôes entre os sítios arque- ológicos de Acre e um povo Aruak con- temporâneo, in Arqueologia da Amazônia Ocidental: Os Geoglifos do Acre. Edited by D. Schaan, A. Ranzi & M. Pärssinen, pp. 120- 133. Belém: Editora Universitária UFPA.

138 Amazônica 2 (1): 104-138, 2010