Hornborg_A - Ethnogenesis

7/25/2019 Hornborg_A - Ethnogenesis

1/33

589

C u r r e n t A n t h r o p o l o g y Volume 46, Number 4, AugustOctober2005 2005by The Wenner-Gren Foundation for Anthropological Research. All rights reserved 0011-3204/2005/4604-0005$10.00

Ethnogenesis,Regional Integration,and Ecology inPrehistoric Amazonia

Toward a System Perspective1

by Alf Hornborg

This paper critically reviews reconstructions of cultural develop-ment in prehistoric Amazonia and argues for the primacy of re-gional and interregional exchange in generating the complex dis-tributions of ethno-linguistic identities traced by linguists andarchaeologists in the area. This approach requires an explicitabandonment of notions of migrating peoples in favor of mod-ern anthropological understandings of ethnicity and ethnogene-sis. Further, the paper discusses the significance of such a re-gional system perspective on Amazonian ethnogenesis for theongoing debate on the extent of social stratification and agricul-tural intensification on the floodplains and wet savannas of low-land South America. It concludes that the emergence of Arawa-kan chiefdoms and ethnic identities in such environments afterthe first millennium BC signifies the occupation of a niche de-fined in terms of both ecology and regional exchange but alsothat it transformed both these kinds of conditions. In these pro-cesses, ethnicity, social stratification, economy, and ecology wereall recursively intertwined.

a l f h o r n b o r g is Professor and Chair of the Human EcologyDivision at Lund University (Finngatan 16, 223 62 Lund, Sweden[[email protected]]). Born in 1954, he received hisPh.D. from the University of Uppsala in 1986 and has taught atUppsala and at the University of Gothenburg. He has done fieldresearch in Peru, Nova Scotia, the Kingdom of Tonga, and Brazil.His current research interest is the cultural and political dimen-sions of human-environmental relations in past and present soci-eties, particularly from the perspective of world-system analysis.Among his publications are Dualism and Hierarchy in LowlandSouth America (Uppsala Studies in Cultural Anthropology 9) andThe Power of the Machine: Global Inequalities of Economy,

Technology, and Environment (Walnut Creek: AltaMira Press,2001). The present paper was submitted 7 iii 04 and accepted 26x 04.

[Supplementary material appears in the electronic edition of thisissue on the journals web page (http://www.journals.uchicago.edu/CA/home.html).]

1. This work, as part of the European Science Foundation EURO-CORES Programme OMLL, was supported by funds from the Swed-ish Research Council and the EC Sixth Framework Programmeunder Contract no. ERAS-CT-2003-980409. Many people have en-couraged my interest in indigenous Amazonia over the years, in-cluding Kaj Arhem, Bill Balee, Philippe Descola, Peter Riviere, AnneChristine Taylor, and Eduardo Viveiros de Castro. For valuable ad-

If archeologists and ethnologists will develop anawareness of the kind of assistance they can renderto each other, then the only result can be profit toboth specialties and the advancement of the general

field of anthropology.b et t y meg g er s a n d

c l i f f o r d e v a n s , 1957

To Europeans, the vast, seemingly impenetrable rain for-ests and swamps of Amazonia have long represented theepitome of virgin, chaotic nature. In recent decades, how-ever, anthropological and archaeological research haspresented a completely different picture. For thousandsof years, the tropical lowlands of South America havebeen populated by many different cultures and societiesthat have left their marks on the landscape. I would liketo review some of the evidence that has contributed tothis shift of perspective and to offer an interpretation ofthe processes that, over the millennia, have generatedthe cultural diversity of indigenous Amazonia. This in-

terpretation is based on a synthesis of ecological, eco-nomic, and cultural perspectives on the emergence andinterrelations of the different ethnic groups involved. Afundamental point of departure is the recognition thatAmazonia in the first millennium BC was part of a con-tinentwide trade network and that a variety of phenom-ena, including ethnic identity, linguistic differentiation,cosmology, stylistic diffusion, warfare, kinship, marriagestrategies, stratification, subsistence, and environmentalchange, should to a considerable extent be understoodas products of the dynamics of this larger system.2

The reconstruction of Amazonian prehistory3 raisesissues of general theoretical interest for our understand-ing of the emergence of domestication, sedentism, andsocial stratification in several parts of the prehistoricworld. The focal questions seem almost universally thesame: What kinds of social processes are signified by the

vice and assistance in the context of the present project I am in-debted to Inez de Aguila, Sven Ahlgren, Bill Denevan, Love Eriksen,Bo Ernstsson, Rafael Gasson, Christian Isendahl, Juliana Machado,Betty Meggers, Daniel Morales, Eduardo Neves, Monica Panaifo,Edithe Pereira, Andreas Persson, Jim Petersen, Helena Pinto Lima,Colin Renfrew, Santiago Rivas Panduro, and Per Stenborg. I alsothank five anonymous reviewers for useful comments.2. This view of Amazonia draws inspiration from the applicationof world-system -type perspectives to precapitalist societies (e.g.,Chase-Dunn and Hall1991, Schortman and Urban1992, Sanderson1995, Denemark et al. 2000). Although the need for considerationof regional and interregional relations and linkages hasconsiderable

longevity in archaeology (see Wilmsen1972, MacNeish, Patterson,and Browman 1975, Sabloff andRathje 1975, Earle andEricson 1977,Fry 1980, Francis, Kense, and Duke 1981) and is often recognizedas crucial to an understanding of cultural variation in Amazonia(see Heckenberger1996:426; Viveiros de Castro 1996:188), attemptsto interpret societal development in pre-Columbian South Americain terms of world-system dynamics have been few, tentative, andmostly confined to the Andean area (Gasson2000, Hornborg2000,La Lone2000). Nor have studies of world-systems or interactionspheres in archaeology generally concerned themselves withethnicity.3.My use of the concept of prehistoric does not imply that Am-erindians were ever a people without history (Wolf 1982) butsimply denotes the period of social development for which no writ-ten records are available.


2/33

590F c u r r e n t a n t h r o p o l o g y Volume 46, Number 4, AugustOctober 2005

appearance of domesticates, permanent settlements,long-distance exchange, and evidence of status differ-entiation? How are the geographical distributions of var-ious elements of material culturesuch as ceramicstyles or subsistence technologiesrelated to ethno-lin-guistic or other divisions recognized by prehistoric pop-ulations, and what do these divisions in turn signify? Itis my hope that tracing the ethnoarchaeological conti-nuities in the Amazonian material will provide sounderfoundations for interpretation than are usually availablein the archaeology of Neolithization4 and that the sim-ilarities in the material record will prove significantenough to warrant consideration of possible parallelswith other areas. I believe that generalization mayindeedbe possible at the abstract level of issues such as therelationship between material culture, language, andethnic identity, the relationship between ethnicity, econ-omy, and ecology, and the relationship between trade,social hierarchy, kinship, sedentism, and intensificationof resource use.

We should begin by rejecting the image of Amazoniaas pristine wilderness. The physical evidence aloneforces us to reconceptualize the region as in some re-spects aculturallandscape. Studies in historical ecologysuggest that more than 12% of the supposedly pristineAmazonian rain forests are anthropogenic in origin inthe sense that they would not exist in their present formwithout human intervention (Balee 1993:231). Detailedstudies of landscape change among the Kayapo andKaapor in Brazil (Posey 1985; Posey and Balee 1989;Balee1993, 1994), the Waorani in Ecuador (Rival 1998),and the Nukak in Colombia (Politis 2001) have revealedhow even the unintentional deposition of seeds and nutsfrom wild plants can generate new geographies. In dif-ferent parts of Amazonia, long-abandoned settlementsand gardens constitute important resources for contem-porary indigenous groups (cf. Oliver2001:7273). Inbothdry and wet savannas in Brazil and Bolivia, anthropo-genic islands of forest (such as the apeteof the Kayapo)illustrate how human influence under certain conditionscan enhance rather than reduce biological diversity.Through intentional and unintentional selection of spe-cies, Amazonian hunter-gatherers have for several mil-lennia been shaping their rain forest environment. More-

4. The term Neolithic Revolution or Neolithization is mis-leading in several ways, but the prehistoric processes which it de-notesthat is, the intensification of resource use based on domes-

tication of plants and animals and the emergence of stratifiedsocietiesconstitute real challenges for archaeological research notleast because such processes on different continents seem to havecertain similarities. What I mean by Neolithization is not theoriginof crop cultivation, which by this time had already occurredin Amazonia for several millennia (Oliver 2001:6566; Heckenber-ger 2002:118; Neves et al. 2003:34), but a relatively sudden inten-sification of agriculture in conjunction with the emergence of sed-entary, densely populated, and stratified societies. It is importantto recognize the gradual domestication of food plants such as man-ioc as a processdistinct from and generally muchearlier than Neo-lithization in this sense. Following traditional usage, however,authorities such as Donald Lathrap have characterized the Ama-zonian domestication of manioc as a first step in the NeolithicRevolution of the New World (cf. Neves 1999:225).

over, environmental conditions even in Amazonia canbe locally modified by humans to an extent that makesit difficult to argue that the tropical environment pre-cludes the emergence of complex societies based on in-tensive exploitation of natural resources (Neves 1999:225).

Extensive areas of dark, anthropogenic soils (or an-throsols) called terra preta de ndio, generally denselylittered with fragments of pottery, suggest large, seden-tary settlements of considerable longevity along the ma-jor rivers (Neves 1999:22223; Petersen, Neves, andHeckenberger2001; Denevan2001:10410; Lehmann etal.2003; Glaser and Woods 2004). These finds appear tocorroborate the earliest historical reports, for instance,from Orellanas expedition down the Amazon in 1542(Carvajal 1934), which predate what appears to have beena period of massive depopulation following the intro-duction of European-derived epidemics. Sixteenth-cen-tury European explorers were impressed by the populous

chiefdoms lining the banks of the Amazon and farmingits periodically inundated, sediment-rich floodplain orvarzea (see Hemming1978, Roosevelt1993, Porro1994,Carneiro1995).5 Judging from discoveries of earthworkssuch as raised fields (camellones), mounds, and cause-ways in the llanos of Bolivia and Venezuela, significanttracts of wetland savanna were also reclaimed for inten-sive agriculture and habitation (Denevan 2001:21590;1991). Add to this the remains of settlements with de-fensive ditches in various areas (Neves 1999:222; 2001:286; Heckenberger 1996; Heckenberger, Petersen, andNeves 1999; Petersen, Neves, and Heckenberger 2001:97, 99), the many discoveries of prehistoric cemeterieswith large funerary urns (see Guapindaia2001), and the

more than 300 finds of rock art in Brazil alone (Pereira2001) and it will be obvious that Amazonia was thor-oughlyinhabited in 1492. Yet, by the end of the eight-eenth century, following the ravages of epidemics, slav-ery, and European colonization, hardly a trace remainedof the once populous chiefdoms of the floodplains.6

These were among the first indigenous societies to suc-cumb to epidemics and slave raids, and their materialculture, dominated by organic materials such as wood,fibers, skin, and feathers, has left little for the archae-ologists but fragments of pottery. Only in the most iso-lated, upriver areas, such as the Vaupes, eastern Peru,and upper Xingu, were fragments of indigenous societies

able to survive into the age of ethnography. As a con-sequence, European travelers in the nineteenth and

5. As Denevan(2001:1034) observes, the human settlements them-selves would have been located not on the regularly flooded plainsbut on the 1020-m-high bluffs in the vicinity.6. There are scholars who estimate that the pre-Columbian popu-lation of Brazil alone may have reached3 million (Hemming1978:487501). Other scholars regard such estimates as much too high,suggesting instead that all of greater Amazonia (i.e., the lowlandsof Brazil, the Guyanas, Venezuela, Colombia, Ecuador, Peru, andBolivia) originally harbored at most 2 million (Meggers 1992). Re-gardless of the pre-Columbian figures, Hemmings estimate offewer than 100,000indigenous Amazonians in Brazil in 1978 im-plies a reduction by over 90%.


3/33

h o r n b o r g Ethnogenesis, Regional Integration, and EcologyF 591

twentieth centuries have been inclined to perceive thetropical lowlands as the very antithesis of culture.7

Meaningful Places: Fragments of a Prehistoric

Arawakan CartographyIn order to reconstruct the ways in which prehistoricAmazonians may have perceived and engaged with theirregional environment, we might begin by considering therole of place-names among contemporary indigenousgroups. The Ge-speaking Suya of central Brazil, for in-stance, inhabit a familiar landscape rich in memories,meanings, and reference points (Seeger 1977:35355).When he joined a group of Suya for a canoe trip, AnthonySeeger discovered how essential it was for them to knowthe names of significant places along the rivers. Like aSuya child, he was taught a number of anecdotes andother oral history connected with the different places and

was expected to memorize their names. He recounts alist of46 toponyms that were pointed out to him duringa single trip and adds that the list would have been longerif he had not already been acquainted with some of theterritory traversed. Of these46 names,12 refer to placeswhere specific kinds of game are abundant,17 to partic-ular events in the lives of elder Suya, and10to geograph-ical peculiarities such as river bends and rapids. Seegernotes that new names are continuously generated andolder ones may be forgotten but older men generallyknow more names than younger men. He concludes thatthe naming of places serves to socialize and familiarizean extensive geographical area and offers a cultural mapconsisting of both history and practical knowledge of

where to find food and other resources.As it can safely be considered a universal human prac-tice, we must assume that a similar kind of mental car-tography has preoccupied all the populations that haveinhabited Amazonia over the millennia. For our presentpurposes, an interesting aspect of such cartographicalconsciousness is its regional extension in space. This hasrecently been highlighted by the documentation of cer-tain ceremonies among Arawak-speaking groups in thenorthwestern Amazon (Hill 1993, 2002; Vidal 2003).Here, on the margins of the Amazon and Orinoco Basins,Arawak-speakers have for millennia mediated a livelytrade that has connected the central Amazon with theCaribbean in the north and the Andean highlands in the

west. The proclivity to trade, forge alliances, and main-tain far-flung fields of identification is commonly rep-resented as a cultural peculiarity of Arawak-speakinggroups throughout their vast, if fragmented, area of dis-tribution from the Antilles to Bolivia (Hill and Santos-

7. Influential scholars such as Alexander von Humboldt(17691859) were convinced that the rudiments of civilization thatcould be encountered in Amazonia were the results of culturalintrusions from elsewhere, doomed to deteriorate in the tropicalenvironment (see Barreto and Machado 2001:24143). Twentieth-century anthropologists and archaeologists such as Julian Stewardand Betty Meggershave offered similarinterpretations(Meggers andEvans 1957, Meggers 1971).

Granero2002).8 An important element in their mainte-nance of a geographically dispersed Arawakan identity,it seems, may have been the recurrent ceremonial reci-tation of historically significant place-names. Among theWakuenai of the Icana and Guaina tributaries of the RoNegro, Jonathan Hill (2002:23637) has documented alist of ceremonially chanted toponyms that form a chainreaching from the mouth of the Orinoco over the RoNegro all the way to the mouth of the Amazon. Thesenamed places are often referred to as the homes of myth-ical ancestors (in particular Kuwai, the first human) butreflect a living knowledge of riverine geography that isundoubtedly connected to ancient Arawakan traderoutes.

As signs for meaningful places, toponyms are merelywords and in societies without writing risk vanishingwith the people who use them. A more tangible and lessephemeral species of signs is the many petroglyphs androck paintings that have been discovered in various partsof Amazonia (Pereira 2001). The densest concentrationsof rock art are along the upper tributaries of the RoNegro (particularly the Vaupes), some of the northerntributaries of the central Amazon (Urubu, Uatuma, Er-epecuru), and the large tributaries in the southeast(Xingu, Araguaia, Tocantins). The concentrations in thenorthwestern Amazon (Vaupes) are of particular interest,as this is an area where we have access to unusuallydetailed ethnographic information. Local Arawak-speak-ers say that their ancestors carved the petroglyphs whenthe rocks were still soft to commemorate mythicalevents and persons such as Kuwai (Zucchi 2002:2089).Tukano-speakers in the area have also assisted anthro-pologists in deciphering the petroglyphs of the Vaupes.They have for centuries lived in close interaction withArawaks, which has resulted in a cultural affinity thatwas no doubt accentuated by their joint escape upriverto evade European colonialism, slave raiders, and epi-demics (Hill 1996b:15859; Aikhenvald 1999b; Santos-Granero 2002:3536; Heckenberger 2002:111).9 Tuka-noans thus share the Arawakan preoccupation withdistant, named places commemorating the movementsof their ancestors (see Arhem1981:12226), and the Tu-kano-speaking Barasana have identified one of the pet-roglyphs on the Pira-parana River as Nyi, the mythicalplace where the first humans were created (Pereira 2001:220). Such ethnoarchaeological connections are not re-stricted to the significance of archaeological remains in

8.Thus, for instance, it was an Arawak-speaker (Lokono) who pro-vided the Moravians in Surinam with a continentwide digest ofthe location of various ethnic groups and their associated politicalrelationships with each other (Whitehead 2002:73n.10, referringto Staehelin). In southeastern Peru, the historicalmemory regardinga common, Arawakan identity in some cases appears to be moreshallow (Gow 2002), but Arawak-speakers here appear to share asimilar preoccupation with writing history into the landscape(Santos-Granero1998).9. The relative priority of Arawak versus Tukano in the north-western Amazon appears to be a contested issue. Accordingto someaccounts, Arawak-speakers dominated the area long before the Tu-kanoans arrived in the Vaupes (Hill 1996b:159), while other ac-counts suggest the reverse (Aikhenvald1999b:39091).


4/33


contemporary indigenous mythology but include directcontinuities in iconography and material culture. An-thropomorphic petroglyphs by the Vaupes, for instance,are identical to the figure of the mythical yage motherthat the Barasana carve into the floor of the maloca (longhouse) on ceremonial occasions (Hugh-Jones 1979:78).10

Similar continuities have been documented on the upperXingu. Iconographic affinities have also been proposedbetween petroglyphs and prehistoric ceramics, for in-stance, along the upper Madeira and at Monte Alegre andMaraca on the lower Amazon (Pereira 2001:22728).Such comparisons may one day help us generate moredetailed hypotheses about the emergence and diffusionof distinct cultural traditions in prehistoric Amazonia.At this point it will suffice to observe that petroglyphsin Brazil can be divided into two fairly distinct traditionsthat extend throughout vast areas north and south of theAmazon, respectively (Pereira 2001). The northern tra-dition is dominated by anthropomorphic figures, thesouthern by zoomorphic designs. The northern style ex-tends into Colombia, Venezuela, and the Guyanas (Pe-reira 2001) and even into the highlands of Ecuador andnorthern Peru (Williams1985, Polia Meconi1995). Con-sidering the connections with Arawakan cartographyethnographically documented along the upper Ro Negro(see Zucchi 2002:209), it may not be unreasonable tosuggest that much of this rock art may reflect the pro-pensity of Arawaks to mark, name, and memorize sig-nificant places along their extensive trade routes.

The broad distribution of the Arawakan linguistic fam-ily contrasts with the more consolidated distribution ofother large language families in South America, such asthe Tukano, the Pano, the Carib, and the Ge.11 In contrastwith the predatory cosmology attributed to other Am-azonian groups (see Viveiros de Castro 1992,1996; Des-cola 1993, 1994; Arhem 1996), the Arawaks ethos em-phasizes peaceful relations with other Arawak-speakerseven over great geographical distances (Santos-Granero2002). The prohibition of endo-warfare has been codifiedin ritualized greetings serving as reminders that, what-ever their genealogical or geographical distance, Arawaksdo not kill each other (Renard-Casevitz 2002:130).12 Ar-awaks have also shown a characteristic willingness toincorporate other ethnic groups into their communities,as is exemplified by the history of the upper Ro Negro

10.Yage is the local name for the hallucinogenic vine Banisteriop-sis,which is used, for example, in male initiation ceremonies (Yu-

rupari) focused on a set of sacred flutes or trumpets that may onlybe handled by men. Similar sacred trumpets also occur among Ar-awaks (Hill 2002) and were encountered by the Portuguese on thelower Amazon, giving Rio Trombetas its name (Gomes 2001:148).Yurupari is the Nheengatu (Tup-based trade-language) name forKuwai (Vidal 2002:268).11. The Tup language family was also very widely dispersed butmade up of two contiguous blocks on either side of the Ge, onedominating the area south of the main Amazon River and the otheroccupying the entire east coast of Brazil south of the mouth of theAmazon.12. According to Renard-Casevitz, Campa chiefs meeting for thefirst time traditionally say, We are Ashaninka [a-shaninka: a pinclusive we;shaninkap origin of humans] and the Ashaninka donot kill each other.

(Hill1996b, Aikhenvald1999b, Neves2001) and the up-per Xingu (Heckenberger 1996:255, 259, 26263, 420;Seki1999) since preconquest times. Such cultural insti-tutions have undoubtedly been significant for their ca-pacity to integrate remote areas of the South Americanlowlands into a common, continentwide trade network.From their point of origin somewhere in the northwest-ern Amazon, Arawak-speakers during the second mil-lennium BC expanded northward along the Orinoco tothe Caribbean and south along the Ro Negro to the cen-tral Amazon (Heckenberger 2002:1047). From the Am-azon area, Arawakan languages continued to spreadsouthward along the Purus and Madeira Rivers to thelowlands of Peru and Bolivia, where Arawak-speakinggroups established themselves as middlemen in the tradebetween the lowlands and the Andean highlands.

The pervasive presence of Arawak-speakers, with theircharacteristic cultural emphasis on river navigation,trade, intensive agriculture, hierarchy, and geographi-cally extended identities, undoubtedly played a crucialrole in the emergence of a regional exchange system inprehistoric Amazonia. The Arawak phenomenon, how-ever, poses a number of theoretical and methodologicalproblems yet to be resolved. To what extent can archae-ological remains of material culture such as ceramicstyles be identified with specific ethno-linguistic groups?To what extent can linguistic groups be assumed to rep-resent ethnic entities? To what extent can ethnic groupsbe assumed to represent some kind of demic, biological-genetic continuity over time, as implied by conventionalcategories that tend to reify peoples or cultures? Anew synthesis of Amazonian prehistory will require notonly a careful consideration of existing data but also atheoretical perspective that adequately accounts for therelations between material culture, language, andidentity.

Economy, Ecology, and Ethnicity in aRegional Perspective: The Significance ofTrade

Until recently, anthropological attempts to account forcultural variation in Amazonia in processual terms haveresorted to notions of environmental determinism, dif-fusion, migration, or combinations of these factors(Schmidt1917; Nordenskiold 1930; Meggers and Evans1957; Steward and Faron 1959; Lathrap 1970; Meggers1971; Brochado 1984; Oliver 1989; Zucchi 1991, 2002;Wilson1999; Myers2004). For several decades, however,most Amazonianist anthropologists have been less con-cerned with comparative or regional understandings ofindigenous cultures over time than with what Christo-pher Crocker (1977:256) has called the tautological her-meneutics of individual local groups in the ethno-graphic present. Against this background, the recentappearance of a volume on the comparative history ofArawakan societies (Hill and Santos-Granero 2002) isvery gratifying. The challenge for a renewed concern


5/33


Fig. 1. The recursive relation between socio-ecological niche and ethnic identity construction, indicating themain categories of traces left by such processes in prehistory and the different academic fields required to re-cover them.

with diachronic cultural processes in a regional perspec-tive is to take due account of earlier concerns with ecol-ogy, diffusion, and migration while acknowledging moresubtle and intangible factors such as politics, exchange,identity, and the autonomous logic of symbolic systemsemphasized by most modern anthropologists.13

The best way to go about this, I believe, is to begin byassuming that cultural and linguistic variation in Ama-zonia has been generated through a continuous, dynamicinteraction between ecology, economy, and ethnicity(fig.1). Various natural environments have afforded particularpopulations different options regarding subsistence, eco-nomic specialization, and exchange with other groups,and these economic activities in turn have modified theenvironment as well as provided foundations for ethnicidentity construction. Viewed as a regional system ofexchange, Amazonia had by the first millennium BC de-veloped a differentiated political-economic structure inwhich the geographical positions of specific populationscontributed to shaping the roles that they came to playwithin this system. It must nevertheless be emphasizedthat the indigenous groups engaged in this exchangeshould be viewed not as passive recipients of impulsesfrom either their economic or their ecological environ-

13.With obvious reference to Roosevelt (1994), Viveiros de Castro(1996:195) writes: As to the hopes of a theoretical new synthesis,I believe that any unification still lies somewhat ahead. Althoughresearchers from opposite traditions, united by the unanimous de-sideratum of transcending the classical antinomies between natureand culture, history and structure, political economy of change andanalysis of monads in cosmological equilibrium, mentalism andmaterialism, and so on, are certainlyand auspiciouslyedgingcloser, it is difficult not to see the persistence of attitudes that werecharacteristic of earlier phases of the discipline.It seems a paradox,

I would add, that the anthropological preoccupation with cosmo-logical monads to such an extent has eclipsed the far-rangingperspectives of earlier generations (e.g., Schmidt, Nordenskiold,Tello, Steward, Meggers, Lathrap, scholars who allowed themselvesto sense connections between Cuba and the Beni and betweenChavn and Marajo) in the very decades that have seen the birthand growth of world-systems analysis.

ment but as agents creatively developing their own cul-tural responses to the economic and ecological nichesthat were available to them. It is precisely their statusas agents or subjects that obliges us to include ethnicidentity construction and historical self-consciousnessas central factors in our account. Ethnic identity is aproduct of the dialectic between externally attributedand internally experienced qualities, often closely inter-woven with traditional modes of subsistence and thespecific kinds of landscapes within which they are con-ducted (Barth 1969). Ethnogenetic processes (cf. Ren-frew 1987, Hill 1996a) therefore involve ecology andeconomy as well as culture, language, politics, andhistory.

The point of departure of any account of the ecologyof Amazonia must be the fact that it harbors the largestcontiguous rain forest in the world (approximately 5 mil-lion square kilometers) as well as the worlds most vo-luminous river (with a flow of water 5times greater thanthat of the Congo and 12 times that of the Mississippi).The area is customarily described as composed of 98%terra firme (older, poorer, and slightly higher land) and2% varzea (fertile, periodically inundated floodplainsalong the shores of major rivers). The ecology of the Am-azon Basin is actually much more diverse and complexthan is suggested by this simple distinction (cf. Moran1993), but the fertile sedimentary soils of the varzea inany event seem to have been a scarce and coveted re-source for prehistoric populations. Intensive cultivationof maize and peanuts may have been conducted on thefloodplains as early as the first millennium BC (Roose-velt 1993; Oliver 2001:6566). In addition, these sameriverbanks offer the greatest abundance of fish, turtles,manatees, and other aquatic resources. Historicalsources indicate that the floodplains were densely pop-ulated and intensively exploited when the first Europe-ans traveled down the Amazon River in 1542, mention-ing series of huge settlements obeying a paramountchief,extensive cultivation, and numerous turtle corrals (Car-


6/33


vajal 1934; Porro 1994). Archaeological excavations inthe vicinity of Santarem and Manaus have confirmedsignificant prehistoric population densities, but the ac-tual size of individual settlements is a contested issue.Roosevelt (1993:274) believes that some sites may havehad tens of thousands of inhabitants, while Meggers(1992:3536) considers such concentrations ecologicallyinfeasible.

The occurrence of complex, stratified societies alongthe main rivers of prehistoric Amazonia shouldnot, how-ever, be reduced to a direct reflection of the fertility oftheir floodplains. These rivers were simultaneously themain arteries in a continentwide trade network the ex-tent of which we have only begun to appreciate. Accessto rare prestige goods from remote areas was, here aselsewhere, an important foundation for ritual and polit-ical-economic authority. Such authority, in turn, waswhat enabled chiefs to maintain densely populated set-tlements and persuade their followers to intensify theexploitation of natural resources (Heckenberger 2002:118).14 Although most of this trade undoubtedly con-sisted of organic plant and animal products that had novalue for Europeans and quickly decomposed in the trop-ical climate, leaving no traces in either the historical orthe archaeological record, there is plenty of early his-torical as well as archaeological evidence for such long-distance exchange throughout Amazonia (Loven 1928;Gade 1972; Lathrap 1973; Oberem 1974[1967]; Roth1974[192429]; Camino 1977; Lyon 1981; Myers 1983;Renard-Casevitz et al.1986; Boomert1987; Burger1992;Santos-Granero 1992; Whitehead 1993, 1994; Dreyfus1993; Arvelo-Jimenez and Biord1994; Hill 1996b; Gas-son1996, 2000, 2002; Kurella1998; Taylor1999; Shady1999; Renard-Casevitz2002).

Although inferences from historical evidence aboutprecolonial conditions are always risky, the complexways in which many of these trade relations reflect long-established and altogetherindigenousdemands and con-sumption patterns often suggest time depths antedating1492. For instance, green stone amulets (called muira-quitaon the lower Amazon and takourave in the Guy-anas) were traded for gold objects and other productsfrom the Vaupes, the Orinoco, and the Roraima (Boomert1987; Whitehead 1994:38). These amulets were oftenshaped like stylized frogs and may have been connectedwith a frog cult in eastern Amazonia (Whitehead 1993:29596).15 From three main manufacturing areas

coastal Surinam, the lower Amazon (Nhamunda-Trom-

14. In indigenous Amazonia, in particular, many anthropologistshave confirmed Clastress (1987) observation that the standard re-sponse to any pretense of authority is to walk away and set up anew village elsewhere. To understand the emergence of stratifiedsocieties in Amazonian prehistory, we therefore need to reconstructthe outlines of integrative ideologies capable of counteracting suchcentrifugal forces. The ceremonial life of contemporary Arawakangroups in the northwestern Amazon (cf. Hill 2002) suggests suchan ideological system, embedding social hierarchy in compellingconstructions of inclusive ethnic identities ultimately founded onfamiliar notions of consanguinity.15. The color green, frogs, and water in much of Amazonia sym-bolize femaleness and fertility (Boomert 1987:36).

betas-Tapajos),16 and the Virgin Islandsthey weretraded widely and used, for instance, as prestigious or-naments in womens necklaces and as a medium of eliteceremonial exchange, including bride-price and deathcompensation (Boomert1987:3641). Representations ofwhat may be muiraquita on ceramics from Santarem,where many such amulets have been found, suggest thatthey may also have been used as ornaments on womensheadbands (Gomes2001:141).17

Historically, three extensive trade networks convergedon the central Orinoco llanos (Hill 1996b:14950; Spen-cer 1998:109). From the north and east, Caribs came totrade blowguns, arrows, baskets, arrow poison (curare),dyes, and pearls for shell beads (quirpa), turtle oil,smoked fish, gold, and salt. Arawaks traded forest andsavanna products for gold, salt, and cotton textiles fromthe Chibchan chiefdoms in the northern Andes, and inthe south they traded, among other things, gold and cu-rare for shell beads, turtle oil, and smoked fish. Shellbeads were used both as prestigious ornaments and as amedium of exchange. They were particularly in demandamong Arawak-speakers such as the Achagua in the Ori-noco area and were still being produced at the beginningof the nineteenth century by the Otomac in Uruana, onthe central Orinoco, where people from all over the lla-nos regularly came to trade (Gasson 2000:595). Notablein all this trade is the occurrence of foodstuffs. The highpopulation densities and intensive food production onthe floodplains should in part be understood against thisbackground. Significant portions of the fish, turtles, andeven cassava were evidently produced for export (White-head1994:36).

The significance of regional trade networks for theformation and reproduction of economically and eco-logically specialized ethnic groups can hardly be over-estimated. In eastern Peru, the salt cakes traditionallyproduced by the Campa of the sub-Andean montanahave served as a medium of exchange along the traderoutes reaching deep into Amazonia (Renard-Casevitz1993; 2002:13136). Arawak-speaking groups such asthe Campa, the Mojo, and the Piro regularly visited theInca capital, Cuzco, to trade forest products such asmedicinal herbs, birds, and tropical hardwoods for An-dean metalwork and other highland products. Chiefsamong, for example, the Conibo and the Cocama basedtheir power in part on their access to objects of silverand gold obtained from Cuzco through Piro traders

(Taylor1999:199).

18

Different lowland groups have been

16.Boomert (1987:4041) suggests that the manufacturing centersin the Nhamunda-Trombetas-Tapajos area can be associated withthe archaeological complexes known as Kondur and Santarem.17. Another kind of lithic artifact suggesting long-distance ex-change networks is the characteristic stone figurines attributed tothe Kondur culture on the lower Amazon (see Nimuendaju 2004:13638, figs. 4850), which show stylistic similarities with thestone sculptures of San Agustn in southern Colombia and of Pucaraand Tiwanaku in highland Bolivia (see Lathrap n.d.:17; Aires Atadeda Fonseca 2004:30, referring to Preuss).18. Judging from archaeological discoveriesalong the upperUcayali,bronze axes from the Andes may also have served as prestige goodsin the lowlands (Lathrap1970:17778).


7/33


specialized in specific kinds of products. For example,Arawaks in the area were known for their cotton tex-tiles with decorative motifs or feathers woven into thefabric, canoes, and pearls, while Panoans were knownfor their painted pottery, mats, hammocks, and gourds(Renard-Casevitz 2002:133). Such trade reinforces eth-nic boundaries while institutionalizing their transgres-sion and cementing interethnic alliances to this day.The Arawak-speaking Yanesha (also known asAmuesha) recognize as real human beings (achen) notonly other Arawaks such as Campa and Piro but alsoriverine Panoans such as Shipibo and Conibo. Amongthe qualities that qualify them for inclusion in this cat-egory are their drinking manioc beer and wearing thecushma (p. 133). Piro count both Campa and Yaneshaas people like us, while marginalized Panoans suchas Amahuaca and Yaminahua are classified as wildIndians who walk about naked and eat raw, unsaltedfood (Gow 2002:155).19 Such moral barriers within thelowlands were generally more difficult to overcomethan the ethnic boundaries distinguishing highland andlowland groups with a mutual interest in trade. Alongthe eastern slopes of the Andes (fig. 2), there has forseveral centuries been lively interaction between peo-ple of the mountains and the lowlands, for instance,between the Quechua and Aymara and the Campa, Piro,and Mojo in southern Peru and Bolivia, between theQuijos and the Omagua in Ecuador and northern Peru,and between the Chibcha (or Muisca) and the Achaguain Colombia (Oberem 1974[1967]; Renard-Casevitz etal. 1986; Taylor 1999:199201; Loven 1928; Kurella1998; Gasson2002). From Colombia to Bolivia, the pri-mary agents of this trade have often been Arawak-speakers.20

Indigenous trade in eastern Peru, the northwesternAmazon, and the upper Xingu area has encouraged al-liances and a certain degree of cultural homogenizationwithout dissolving the ethno-linguistic identities of in-dividual groups. On the contrary, trade and speciali-zation should be regarded as central factors in the veryemergence of such ethnic identities. Material cultureis an important medium for expressing ethnic specific-ity but can, as we have seen, be shared by several eth-nically distinct groups speaking different languages. El-ements of material culture such as ceramic style arefrequently adopted from other ethno-linguistic groups(see DeBoer and Raymond 1987, DeBoer 1990). There

are many indications that language generally consti-tutes a more profound core of ethnic identity than ma-

19. Similar indigenous categories glossed in English as wild In-dians recur among many Arawak-speaking groups throughoutAmazonia, for instance, the Mehinaku and other Arawaks on theupper Xingu (Gregor 1990:10910; Heckenberger 1996:249, 259,269, 272). Such moral, hierarchical distinctions between Arawaksand non-Arawaks in some areas seem paradoxical in view of actualpower relations, as in the case of the historically marginalized sub-Andean Arawaks of Peru (see DeBoer 1990, Santos-Granero1998).20. There are even indications that Arawak- and Tukano-speakersin the northwestern Amazon at one time maintainedtrade relationswith Quechua-speakers in the Andes (Hill 1993:221 n. 8; Dixon andAikhenvald1999b:10; Aikhenvald2002:15).

terial artifacts (see Dixon and Aikhenvald1991b:8; Ne-ves 2001:26768). In the northwestern Amazon, forinstance, Tukanoans use their different dialects to de-fine the boundaries for exogamous marriages (Sorensen1974[1967], Jackson 1983, Aikhenvald 1999b), while

speakers of various Arawakan dialects are included inthe common ethno-linguistic category of wakuenai,people of our language (Hill1996b:144). In both casesa common language is a fundamental criterion for draw-ing societal boundaries. To shift language is no doubtgenerally a more profound transformation of identitythan to adopt new elements of material culture.

By implication, historical linguistics should be ableto tell us important things about social processes in thepast. However, substantial inferences from linguisticsregarding prehistoric cultural processes in Amazoniahave been few, contested, and inconclusive. Althoughmuch detailed linguistic research has been carried outin the area (cf. Klein and Stark 1985, Campbell 1997),

Dixon and Aikhenvald 1999a), there seems to be littlegeneral agreement on how to interpret the results. Lin-guists have generally been more hesitant than archae-ologists to draw processual conclusions from linguisticdata. The latter (e.g., Lathrap 1970; Meggers1987; Bro-chado 1984; Rouse 1986; Oliver 1989; Zucchi 1991,2002) have been inclined to view contemporary lan-guage distributions in Amazonia as fairly straightfor-ward reflections of past migrations and the associateddiffusion of material culture such as pottery. Linguistshave rightly remained skeptical of such essentializingequations of language groups, ceramic styles, and ge-netically coherent (demic) populations but often seemto have lacked the social theory with which to replacethem. While emphasizing the ubiquity of areal diffu-sion through borrowing and language shifts (Campbell1997:34652; Aikhenvald and Dixon 1998; Dixon andAikhenvald 1999b; Facundes 2002), their accounts ofthe social processes historically resulting in such lin-guistic change rarely make use of anthropological per-spectives on ethnic identity formation in relation toregional exchange (e.g., Barth 1969; Jones 1997).21 Ren-frews (1987) modification of conventional migrationtheory shares Lathraps (1970) conviction that languagedistribution may reflect the expansion of agriculture22

but also opens European archaeology to wider theoret-ical questions regarding ethnogenesis that have hith-

erto been lacking in Amazonian archaeology. Recent

21. Dixons punctuated-equilibrium model of language develop-ment (Dixon and Aikhenvald 1999b:1619) seems anthropologi-cally naive in assuming long periods of equilibrium during whicheach geographical zonein the tropical forests, on the grasslands,on the mountains, and so on, would have been inhabited by anumber of small political groups of similar size, each with its owntraditions, religion, laws and language, where no one group orlanguage would have substantially greater prestige than any other.Such an image of prehistoric social conditions is difficult to rec-oncile with modern anthropological theory.22.Dixon and Aikhenvald (1999b:17) and Bellwood (2001) also pro-pose a connection between the adoption of agriculture and lin-guistic expansion but without necessarily involving migration.


8/33


Fig. 2. Indigenous groups (142) and archaeological sites (4362) mentioned in the text.1, Achagua (Arawak);2, Amahuaca (Pano); 3, Arua (Arawak); 4, Baniwa (Arawak); 5, Barasana (Tukano); 6, Bare (Arawak); 7,Campa/Ashaninka (Arawak); 8, Chane (Arawak); 9, Chibcha (Muisca); 10, Chipaya (Arawak?); 11, Cocama(Tup); 12, Cocamilla (Tup); 13, Conibo (Pano); 14, Curucirari (Tup?); 15, Guahibo; 16, Guaran (Tup); 17,Kaapor (Tup); 18, Kallawaya (Arawak?); 19, Kaua (Arawak); 20, Kayapo (Ge); 21, Lokono (Arawak); 22, Manao(Arawak);23, Maya; 24, Mehinaku (Awarak); 25, Mojo (Arawak); 26, Nukak (Maku); 27, Omagua (Tup); 28,Otomac (Arawak); 29, Palikur (Arawak); 30, Piro (Arawak); 31, Quijos; 32, Shipibo (Pano); 33, Suya (Ge); 34,Taino (Arawak); 35, Tariana (Arawak);36, Taruma (Arawak); 37, Tupinamba (Tup); 38, Uru (Arawak?); 39,Wakuenai (Arawak); 40, Waorani; 41, Yaminahua (Pano); 42, Yanesha/Amuesha (Arawak); 43, Acutuba; 44, Al-

tamira;45, Belterra; 46, Chavn de Huantar; 47, Chiripa; 48, Cuzco; 49, Hatahara; 50, Hupa-iya; 51, Juriti; 52,Kondur; 53, Kotosh; 54, Los Barrancos; 55, Manacapuru; 56, Pukara; 57, Saladero;58, San Agustn; 59, Santa-

rem;60, Tiwanaku;61, Valdivia; 62, Wari.

applications of this concept to Amazonia (Hill 1996b,Schwartz and Salomon 1999, Hill and Santos-Granero2002) have usefully reconceptualized historical pro-cesses in the area, but its ramifications for archaeologyhave yet to be consistently worked out. Although thereare recurrent concessions to the fluidity and provision-

ality of ethnic identities in the history of indigenouspopulations in the region, it has proven difficult, inpractice, to abandon notions of essentialized, boundedpeoples as coherent, persistent entities to be identi-fied in the archaeological record (see Heckenberger1996).


9/33


Linguistic Distribution and SocietalStructures

The archaeologist Donald Lathrap (1970:7479) sug-gested that the adoption of manioc horticulturepropelled

the initial expansion of proto-Arawak and proto-Tupfrom the central Amazon around 3000 BC, the formerprimarily toward the north and west and the latter pri-marily toward the south and east. He thought that thedistribution of language families in Amazonia had beengenerated by an outward movement of horticulturalistsfrom the central Amazon along the major tributaries andlater also along the northern and eastern coasts of SouthAmerica. This pattern of expansion, he argued, differedfrom, for example, that of the Pano-speakers, who appearto have moved overland across the drainage basins of theMadre de Dos, Purus, Jurua, Ucayali, and other riversin western Amazonia (p. 81). In recent years, as Heck-enberger (2002:103) observes, linguists have been able to

identify probable areas of origin for Arawak (the north-western Amazon), Tup (Rondonia), Carib (the Guyanauplands), Ge (the central Brazilian uplands), and Pano(the borderlands between Peru and Brazil).23 Tup, Carib,and Ge all appear to have founded their linguistic iden-tity in upland areas and originally traveled primarilyon foot, as did the Panoans. This suggests a conspicuouscontrast to the Arawak, with their ancient connectionto rivers, canoes, and water transport (Schmidt 1917;Lathrap 1970:7374; Neves 2001:273; Heckenberger2002:1046; Hill 2002:228).

The distribution of Arawakan languages suggests a pat-tern of expansion along the very barriers that surroundand separate other linguistic families: the Orinoco, RoNegro, Amazon, Ucayali, Purus, and Madeira Rivers, thellanos of Venezuela and Bolivia, and the coastal areas ofGuyana (cf. Heckenberger 2002:1056). Considering thediscovery of pottery near Santarem dating from the sixthmillennium BC (Roosevelt et al. 1991; Neves1999:219),however, it is unlikely that the highly productive flood-plains and aquatic resources in these areas were unoc-cupied prior to the Arawakan expansion beginning in thesecond millennium BC. Whether earlier populationswere displaced by or incorporated into the Arawakannetwork, the linguistic distribution maps suggest thatthe Arawakan expansion created ethnic wedges that con-tributed to the geographical demarcation of other, spa-

tially more consolidated linguistic families such as theCarib, the Tukano, and the Pano. In some cases, it iseven possible to detect how a wedge of Arawakan lan-guages has split a language family, as in the case of thePanoan groups on either side of the Arawaks along thePurus and the Madeira (cf. Erikson 1993:55). The dis-persed pockets of Arawakan dialects that have been doc-umented along the river systems from the lower Orinocoto the upper Madeira appear to be the remains of a net-work of Arawak-speaking societies that in prehistoric

23.For details of these reconstructions, see Dixon and Aikhenvald(1999a).

times spanned the whole of western Amazonia. In viewof their role in integrating regional exchange, these Ar-awak-speakers should be viewed less as ethnic wedgesthan as the social glue of ancient Amazonia.

Although Arawakan groups have been characterizedas comparatively peaceful and as the undeserving vic-tims of raids by Caribs and others, their societies wereundoubtedly the most powerful and expansive politiesof pre-Columbian Amazonia.24 Their identity was basedon having appropriated most of the fertile floodplains,dominated trade along the major rivers, and establisheddensely populated and allied chiefdoms along these riv-erine trade routes. Several of the other Amazonian lan-guage families such as Carib, Pano, and Ge appear at leastoriginally to have been denied access to these resource-rich areas. The difference between the river-based Ara-wak and the more marginalized upland groups can tothis day be detected in elements of their kinship systems(Hornborg 1998:17980). Kinship terminologies and mar-riage rules among most Carib, Pano, and Ge express lo-cally endogamous, atomized, and introverted exchangerelations, while many Arawak groups emphasize exog-amy (cf. Gow1991; Hill1993:910;1996b:144). In somecases, an inclination to extend the category of classifi-catory siblings codifies the encouragement of distantalliances beyond an expansive field of consanguineouskin (Gregor1977:277).25 Although Arawaks in the north-western Amazon are generally patrilineal, the apparentexpansiveness of Arawakan identity may in some areasin part have derived from an inclination among offspringof interethnic marriages to identify with their more pres-tigious, Arawakan parent, irrespective of whether thisimplied patri- or matrifiliation. The Arawakan preoc-cupation with genealogy and extended kinship categoriesis rare among Amazonian societies except the Tuka-noans, with whom, as we have seen, Arawak-speakershave maintained close cultural interchange for centuries.It is no doubt connected with their equally rare emphasison social stratification and with their ambition to in-corporate rather than confront neighboring groups. Ar-chaeologically, the concern with descent and ancestorsis obvious in the many finds of elaborately ornamentedburial urns of probable Arawakan affiliation, stylisticallydistinguished as Guarita, Maraca, Ariste, Arua, etc. (seeGuapindaia2001).

Neither Carib-, Pano-, nor Ge-speakers appear to haveachieved the same degree of political centralization or

social stratification as the Arawak (Heckenberger 2002:

24.The capacity of Arawak groups to expand through military con-quest is exemplified by the Tariana invasion of the Vaupes (cf. Aikh-envald1999b:390, referring to Bruzzi).25.Caribs and Pano exemplify the typical Amazonian, Dravidianor two-line system of kin classification initially defined by Ri-viere (1973) as geared to bilateral cross-cousin or sister-exchangemarriage and subsequently confirmed for a number of indigenousgroups in the region, while the Ge-speakers approximate a similarlyintroverted kin-affine dualism in the spatial layout of their villages(see Hornborg1988[1986], 1993). The equation of siblings andcross-cousins exemplified by the Mehinaku and neighboring groups onthe upper Xingu is referred to as a Hawaiian or generationprinciple or kin classification.


10/33


115). The Tup-speakers developed extensive chiefdomsto the south and east of the Arawakan sphere of influ-ence, but their expansion seems to have been based moreon military conquest than on trade and diplomacy (Bro-chado 1984).26 From their point of origin in Rondonia,between the upper Madeira and the upper Tapajos, theTup-speakers expanded eastward and conquered vastterritories south of the Amazon and in southeastern Bra-zil, where the Tupinamba shortly prior to the Europeanarrival had displaced Ge-speakers along the coast (cf.Morey and Marwitt 1978:254).27 A westward expansionof Tup languages was also under way when the Euro-peans arrived, as Tup-speakers had recently establishedthemselves in a narrow zone along the upper Amazonas far west as the Ucayali (the Cocama) and the Huallaga(the Cocamilla) in Peru. The most powerful of theseTup-speaking groups was the Omagua, which at thetime of Orellanas expedition in 1542 controlled thefloodplain between the Napo and Jurua Rivers. The Oma-

gua are reported to have launched annual war expedi-tions up the tributaries (Lathrap 1970:152; Morey andMarwitt 1978:251). Which ethnic groups or languages theTup-speakers had displaced or incorporated along thePeruvian floodplains is not known, but Heckenberger(2002:122 n. 6) suggestsunfortunately without muchsupporting evidencethat the Omagua, Cocamilla, andCocama were former Arawak-speakers who had recentlyadopted a Tup lexicon.28 In the sixteenth century, at anyrate, Tup-speaking societies controlled the southernbank of the central and lower Amazon. Although theevidence is inconclusive, Arawak-speakers may at onepoint have dominated the opposite, northern bank, sug-

26.An example of the kind of nonmilitaristic expansion that seemsto have beengenerally characteristic of Arawaks is the involvementof the Arawak-speaking Piro in the trade networks of eastern Peru(Gow2002:164). The Arawakan groups in the area presumably en-tered into trade alliances with other indigenous groups in pre-Co-lumbian times in much the same way as they have more recentlyallied themselves with European trade partners.27.Such non-Tup groups in the area were often referred to as Ta-puya, a category deriving from the Tup word for enemies.28.The language most closely related to Cocama/Omagua is in factthe Tupinamba-derived trade language Nheengatu (Jensen 1999:12931), which was widely adopted by former Arawak-speakers inwestern Amazonia. Jensen (p.129, referring to Cabral) is convincedthat the Cocama language is a result of a language shift from anunknown non-Tup language to Tupinamba. Interesting little cluesregarding the ethnic identity of the sixteenth-century Omagua areCarvajals (1934) references to their exquisite polychrome pottery

and, not least, Cristobal de Acunas observation in 1639 that theOmagua obtained this pottery from the Curucirari in exchange forcotton cloth (Hemming 1978:233). In Peru, cotton cloth has tra-ditionally been produced by Arawaks and painted pottery by Pan-oans (see Renard-Casevitz 2002). The polychrome decoration of re-cent Panoan pottery should probably be traced to their closecoexistence with the Cocama during the seventeenth to nineteenthcenturies (Lathrap 1970:184; Myers 1976; Brochado 1984:304;DeBoer and Raymond1987:12829; DeBoer1990:87,103). Lathrap(1970) and Brochado (1984) have suggested a general connectionbetween the Amazonian Polychrome Tradition and Tup-speak-ers. It should be noted, however, that Marajo Islandthe source ofthe earliest known polychrome ceramics in Amazoniawas in-habited by Arawak-speakers (Arua) at the time of European contact(cf. Meggers and Evans 1957; Aikhenvald1999a:6669).

gesting that the main Amazon here served as an ethnicboundary.29 Demographically and politically, the Ara-waks in this area seem to have suffered greater lossesfrom European diseases and slave raids than the Tup orCarib, two ethno-linguistic groups that at this time ap-pear to have expanded at the expense of the enfeebledArawak. It is significant, not least for our reconstructionsof linguistic diffusion along prehistoric trade routes, thatthe Nheengatu trade language that was adopted by Ar-awakan groups along the Ro Negro in the colonial periodhad Tup roots (Sorensen 1974[1967]:15253; Aikhenvald1999b:38788; Santos-Granero 2002:35; Hill 2002:240).Although beyond the scope of this paper, it appears thatthe argument against simple migrationism could alsobe applied to the spread of Tup languages (cf. Brochado1984:402), which in many areas appear to have sup-planted Arawak in the late precontact and contactperiods.30

A regional systemic view of prehistoric Amazonia alsoprompts us to rethink the occurrence of warfare and theideology of predation that has been posited as genericto indigenous cultures in the area (cf. Viveiros de Castro1996). It would not seem correct to attribute a preda-tory cosmology to Arawak-speaking groups in general(cf. Santos-Granero2002). To the extent that it has beenapplicable to, for example, Carib- or Tup-speakers, thismust to some extent be understood historically in thecontext of the disorienting demographic, economic, po-litical, and cultural convolutions of the colonial period,involving indigenous groups in slave raids on their neigh-bors, competition for European trade goods, and the ri-valry between Dutch, Spanish, and Portuguese in SouthAmerica (see Whitehead1990,1992; Arvelo-Jimenez andBiord 1994). The European arrival rapidly reorganized theentire regional system and in many areas induced com-plete reversals in the balance of power between ethnicgroups. Seventeenth-century sources from the Orinocomention extensive fortifications among the Achagua andother Arawaks on the floodplains, who had been deci-mated by their encounters with Europeans and were now

29. According to J. Alden Masons linguistic map in volume 6 ofthe Handbook of South American Indians (Steward1950), almostthe entire northern bank of the central Amazon was originally in-habited by Arawak-speakers. In Terrence Kaufmans (1994) maps,however, this bank appears to be more or less empty at time ofcontact. Considering that the worst ravages of European-derivedepidemics probably occurred long before the actual arrival of Eu-ropean colonizers, both maps may be valid, reflecting the situation

around the years 1500 and 1650, respectively.30. Brochado (1984:4023) writes that the importance of the ac-culturation and incorporation of alien groups by the Guarani andTupinamba . . . cannot be overstated. These processes constituteone of the most important mechanisms of population increasethrough colonization. . . . In many cases, acculturation and com-plete physical and cultural absorption of the vanquished was theend result in the long run. These processes of guaranization ortupianization included the adoption of either the Guarani or theTupinamba pottery, which makes them identifiable archaeologi-cally. The widespread transition from Barrancoid to AmazonianPolychrome ceramics in many areas from the lower Amazon to theUcayali (Lathrap 1970, Myers 2004) suggests notso much migrationand demic replacement as an ethno-linguistic shift from Arawakto Tup in the centuries before and after the European arrival.


11/33


being subjected to systematic predation by previouslymarginal groups such as the Carib and the Guahibo(Morey and Marwitt 1978:25153). Although archaeo-logical discoveries of defensive ditches in various areasof Amazonia confirm that warfare was by no means apostconquest invention (Heckenberger 1996:423; Neves1999:222; 2001:286; Heckenberger, Petersen, and Neves1999; Petersen, Neves, and Heckenberger 2001:97, 99),in many areas it was undoubtedly exacerbated by socialupheavals following the European arrival. To some ex-tent, competition over scarce resources focused on thesame kinds of things as before (e.g., prestige goods, nowof European origin [cf. Gasson 2000:590]), but the con-ditions and rules of the game had been decisively trans-formed.

Although a recent and authoritative summary of Am-azonian linguistics (Dixon and Aikhenvald 1999a) ad-vocates extreme skepticism with regard to higher-levelgenetic groupings, several studies have suggested variousdegrees of affinity between the four most important lan-guage families in Amazonia: Arawak, Tup, Carib, andGe.31 To the extent that there is a foundation for any ofthese studies, it would lend support to the hypothesisthat at least these families are to be seen as products ofregional ethnogenetic processes rather than traces ofmigrations from other parts of the continent.32 The pos-sible significance of ecological factors in such processesalso deserves to be considered. Meggers (1982,1987) hassuggested that drought-related fluctuations in the extentof forest vegetation have contributed to the geographi-cal distribution of different language families, some ofwhich (e.g., horticulturalists such as the Arawak, theTup, the Pano, and the Carib) were originally confinedto distinct forest refugia but subsequently expanded, atthe expense of savanna-dwelling hunter-gatherers, with

31. For references to some of these studies, see Klein and Stark(1985), Klein (1994), Kaufman (1994), Campbell(1997:170205),andDixon and Aikhenvald (1999b:1116).32. Geneticists have expected to find a correlation between lan-guages and genes in Amazonia, but Cavalli-Sforza, Menozzi, andPiazza (1994:341) report that it is very difficult to make inferencesabout the order of entry of the people who today speak Carib, Equa-torial [Greenbergs (1987:83) macro-category including Arawak andTup], Ge, and Panoan, on the basis of genetic data. They add thatit seems natural to suggest that they entered in the order in whichthey are found in South America, those located farther south beingfirst, implying that Panoan, Equatorial, and Ge are of roughlythe same age. This is highly improbable for several reasons. A gen-eral objection would be that the major language families in Ama-

zonia have more discontinuous distributions . . . than . . . is foundin any other part of the world (Dixon and Aikhenvald 1999b:1),which makes it impossible to order them in terms of spatial suc-cession. Both Carib and Arawak, for instance, range from the An-tilles to the upper Xingu. Among more specific objections wouldbe the conclusion of historical linguists that the Panoan languages,rather than equaling Arawak or Tup in age, indicate a fairly shal-low time-depth and recent expansion and split (Loos 1999:227).To their credit, Cavalli-Sforza and colleagues admit that their con-siderations could have more weight if there was a good correlationbetween linguistics and genetics in South America but that, un-fortunately, there is not, or it has not yet been found. Earlier andsimilarly futile attempts to find such correlations include Steg-gerdas (1950) observations on the virtually identical anthropom-etry of Arawak, Carib, Tup, and Pano.

the recovery of the rain forest as the climate grew morehumid. Dixons punctuated-equilibrium model similarlyimplies that linguistic families such as Arawak, Carib,and Tup, prior to the punctuation represented by theadoption of agriculture, originated as the result of rela-tive confinement within specific geographical zones(Dixon and Aikhenvald 1999b:17). However, ecologicalfactors can be assumed to be significant for the distri-bution of, for example, Arawak (wetland agriculturalists)and Ge (hunter-gatherers of the dry savanna) without anyreference to paleoclimatic fluctuations or ecologicallyinduced isolation. On the contrary, an anthropologicalperspective on the formation of ethno-linguistic identi-ties would emphasize ecologically induced interactionrather than isolation (Barth 1969). The geologically re-cent expansion of Arawak-speakers on floodplains andwet savannas from the llanos of Venezuela to the llanosof Bolivia over the course of a millennium and a half (seeHeckenberger 2002:1067) suggests the systematic ex-

ploitation of an existing socio-ecological niche in inten-sive interaction with the populations of other such zonesrather than reflecting post-Pleistocene changes in bio-geography or long periods of homogeneous, egalitarianequilibrium. Furthermore, as studies in historicalecology show, the relation between ecology and culturalidentity cannot be a matter of one-way causality whenthe biophysical environment is continuously trans-formed by human activity.

A convincing account of the genesis of ethno-linguisticdivisions in prehistoric Amazonia needs to recognize therecursive relation between ecological and economic spe-cialization within regional exchange systems, on onehand, and ethnic and cultural creativity, on the other.

The various cultural traits and institutions that enabledArawak-speakers to integrate long-distance trade net-works in ancient Amazonia should be understood notonly as prerequisites for but also as products of theseexchange systems. Cultural patterns do exhibit a certaindegree of autonomy and inertia, acknowledged in notionssuch as ethos or tradition, but rather than replaceenvironmental determinism with cultural essentialismwe should ask how the cultural creativity of the proto-Arawak may have constituted a response to the eco-nomic niche afforded them by the opportunities of riv-erine trade. Heckenberger (2002:121) asserts that he doesnot wish to replace ecological with cultural determin-ism, but his recurrent references to the underlying sym-bolic structure and other cultural features (pp.11013) of the Arawaks at times seem to echo the es-sentialist approaches of early diffusionists such asSchmidt (1917) and Nordenskiold (1930). Heckenbergersuggests (p.121) that if Julian Steward and his followershad recognized the historical relationship between thevarious Arawakan chiefdoms in South America, as didSchmidt, they might have come to the startling con-clusion that culture, as much as ecology or demog-raphy, plays a key role in differential cultural develop-ment in Amazonia. The distributional patternthecorrelation between language and cultureis unmistak-


12/33


able: Where we find Arawak speakers we typically alsofind social hierarchy, sedentism, and regionality.

Although Heckenbergers understanding of the Ara-wakan expansion is probably the best account to date,he does not offer an analytically distinct alternative to

the conclusion of earlier diffusionists that Arawak highcultureunique in the lowlandswas already pre-sent in the ancestral Arawakanpeoplesof ancient Ama-zonia (p.110, emphasis added). Thus he ultimately failsto bridge the theoretical gap between traditional diffu-sionism and modern anthropological theories of eth-nicity.33

Instead of treating either ecology or culture as an in-dependent variable, a more dynamic ethnogenetic per-spective can illuminate the emergence of cultural traitsin a regional and historical perspective by focusing onthe historical structures of regional exchange systemsrather than ecology or culture in themselves. This wouldseem to be fertile ground for a synthesis of processual

and constructionist perspectives in the study of prehis-toric ethnogenesis. Viewed from a regional perspective,as Hill (2002:229) observes, the proto-Arawakan terri-tories around the Icana and Guaina Rivers in the north-western Amazon were actually centrally located inrelation to the riverine connections between the Orinocoand Amazon Basins. If this was indeed the area in whichArawakan traits and institutions originally developed, itis not difficult to imagine a close connection betweentheir cosmopolitan ethos (cf. Santos-Granero 2002) andtheir role as long-distance traders. When this contagiousethos became entrenched along the trade routes, how-ever, it would be misleading to represent its diffusion interms of the movement of peoples.

Language, Ceramics, and PrehistoricIdentities: Toward a Nonessentialist Approach

A central challenge for Amazonian archaeology has longbeen to trace connections between ethno-linguisticgroups and ceramic styles. Pioneers such as Norden-skiold (1930) observed similarities between pottery fromthe Caribbean and the lower Amazon and suggested acommon, Arawakan origin. Meggers and Evans (1957)discovered similarities between ceramics from the Napo

River in Ecuador and from Marajo Island at the mouthof the Amazon and interpreted them as evidence of a

33. In fact Schmidts approach occasionally seems more in agree-ment with modern understandings of ethnic identity constructionand cultural hybridization than some recent accounts of Arawakanmigrations. For instance, he made several observations on indig-enous language shifts (e.g., among the Kaua and Chane) and ex-plicitly suggested that Arawak served as a trade language in thenorthwestern Amazon (Schmidt 1917:1921, referring to Koch-Grunberg and Nordenskiold). He also emphasized (pp. 3661) therole of elite gift exchange and (male) exogamy in the expansion ofArawakan culture and observed that the incorporation of non-Arawak women would have resulted in the assimilation of ceramicstyles and other features from neighboring groups.

downstream migration of groups with Andean roots.34

Lathrap (1970:15051) reinterpreted these same similar-ities as the result of an upstream migration of Tup-speakers. He went on to suggest a number of other cor-respondences between the distribution of ceramic stylesand presumed migrations of linguistic groups, such asbetween various fine-line incised traditions (Santarem,Paredao, Milagro, etc.) and Carib-speakers (pp. 16470)and between Cumancaya pottery and proto-Panoans (p.140). More central to his argument, however, is the con-nection between Saladoid-Barrancoid pottery and the mi-grations of Arawak-speakers (pp. 112, 123).35 An impor-tant link in this argument is the Guarita style from thecentral Amazon, which appears to be related to the Bar-rancoid ceramics from the Orinoco (pp. 15659). Recentarchaeological excavations at sites such as Acutuba andHatahara, near Manaus, confirm this ceramic continuity(Petersen, Heckenberger, and Neves 2003).

Most experts today seem to agree that there is indeeda connection between the distribution of Saladoid-Bar-rancoid ceramics and Arawakan languages but do notshare Lathraps (1970:112, 120) conclusion, followingfrom his model of prehistoric migration routes, thatthese pottery styles originated in the central Amazon.36

Oliver (1989) has attempted to find support for Lathrapshypothesis that ancestral Arawak-speakers brought earlyceramic styles from the central Amazon northward alongthe Ro Negro to the Orinoco, but the absence of chro-nologically relevant finds along the Ro Negro and theAmazon contradicts such an interpretation (Neves 1999:22829; Zucchi 2002; Gasson 2002:27071). The oldestpolychrome ceramics in Brazil may well be the Mara-joara style from Marajo Island, dated at around AD400(Schaan 2001; Oliver 2001:62; Petersen, Neves, andHeckenberger 2001:101). Such polychrome pottery (redand black on white) also occurs along the lower reachesof larger tributaries such as the Tocantins, the Xingu,and the Madeira but generally only up to the first rapids,suggesting a connection with navigable waterways andvarzea environments (see Meggers et al. 1988:288).

The occurrence of Guarita-related pottery on the mainrivers from the lower Amazon to the upper Ro Negrotoward the end of the first millennium AD probably re-flects intense trade relations along these waterways.Guarita-type pottery was manufactured along the entireextent of the Ro Negro up until European colonization.As the whole area along the Ro Negro was inhabited by

Arawak-speakers (Taruma, Manao, Bare, and others) at

34. This interpretation was shared by Hilbert (1968). Rather thana culturally marginal region passively receiving impulses from up-stream, however, the lower Amazon was clearly a major center ofinnovation. Pottery finds near Santarem may be the oldest on thecontinent (Roosevelt et al.1991; Neves 1999:219).35.These ceramic styles are named after the sites Saladero and LosBarrancos on the lower Orinoco.36. In fairness to Lathrap, however, we should add that he doesconsider the possibility that the earliest Barrancoid ceramics orig-inated on the network of waterways connecting the Ro Negroand the Ro Orinoco, spreading along with the Maipuran branchof Arawakan languages mainly during the first millennium B.C.(p. 127).


13/33


Fig. 3. Approximate chronology of some relevant ceramic styles and archaeological sites in ten regions ofSouth America, 1800BCAD 1500. Italics, sites or styles proposed as representing Saladoid-Barrancoid ceramic

influences or for other reasons suggesting involvement in an interregional, Arawak-mediated exchange network.#, sites associated with raised fields; , sites associated with terra preta; /, stylistic discontinuity; M.M., Masi-cito Mound; P., Paredao. Data based on Lathrap (1970), Lathrap, Collier, and Chandra (1975), Bruhns (1994),Heckenberger (1996, 2002), Bauer (2001), Denevan (2001), Petersen, Heckenberger, and Neves (2003). The main

period of consolidation of the Arawakan trade network appears largely to coincide with the so-called Early Ho-rizon (ca. 1000200BC) and the most significant period of reorganization with the close of the Middle Horizon(ca. AD 1000). The Early Horizon is defined by the diffusion of the lowland-oriented Chavn style across the

Andean crest and the close of the Middle Horizon by the collapse of the highland polities Tiwanaku and Wari.Rather than Amazonian developments being viewed as repercussions of political developments in the Andes orvice versa, the two processes are better approached as recursively connected within a single system of mutual

interdependence.

this time, it seems legitimate here to assume a connec-tion between the Guarita style and Arawakan languages(Neves 2001:275). Even today, in the upper Ro Negroarea, there is a sharp distinction between the ceramicsmanufactured by the Arawak-speaking Baniwa on theIcana River and the dark, monochrome pottery made byTukanoans on the Vaupes (p.274; cf. Aikhenvald1999b:389, referring to Galvao).

Several additional indications of correspondences be-tween ceramic styles and Arawakan languages have beenpresented in recent years.37 For instance, Zucchi (2002)has identified a Saladoid-related ceramic complex (par-allel line incised tradition) that she suggests was man-ufactured by Arawak-speakers on the Icana and otherupper Ro Negro tributaries around1500BC. On the Bra-zilian coast just north of the mouth of the Amazon,

37.Less visible aspects of prehistoric ceramics, such as temperingtechniques, may also reflect ethnic boundaries. Although cauxi(sponge-spicule) tempering is generally predominant along the ma-jor rivers, cariape (tree-bark-ash) appears to have been in use si-multaneouslyduring the periodAD 8001400 in geographically sep-arate but apparently Arawak-related contexts such as the Ipavuphase on the upper Xingu (Heckenberger1996:13637), the Guaritaphase on the lower Ro Negro (Petersen, Heckenberger, and Neves2003:252), and the Mazagao phase in Maraca (Meggers and Evans1957:596).

Guapindaia (2001:16971) has confirmed the connection,posited much earlier by Nimuendaju (1926:23), betweenAriste-style pottery from about AD400and the Arawak-speaking Palikur documented both historically andethnographically in the area. Finally, Heckenberger(2002:109) has postulated connections between Saladoid-Barrancoid ceramics and Arawak-speakers in four widelyseparated areas: the West Indies, the Orinoco, the centralAmazon, and the upper Xingu.38

There thus seems to be little doubt that pioneers likeSchmidt and Nordenskiold are being vindicated in hav-ing suggested that the distribution of pottery styles can

tell us something about the distribution of Arawak-speakers in different parts of South America (fig. 3), butthe implications of such correlations are unquestionablymore complex than the patterns of prehistoric migra-

38.Moreover, Heckenberger posits a connection between Saladoid-Barrancoid pottery and circular village layouts wherever it has beenpossible to reconstruct settlement patterns in these areas. Suchring villages, in which a circle of extended households surroundsa ceremonial plaza have long been associated with Ge-speakers (seeHornborg1990), but Heckenberger (2002:109) argues that they wereequally characteristic of Arawaks from the Antilles to eastern Peruand the upper Xingu, and elsewhere (Heckenberger 1996:219) sug-gests a connection with defensive works.


14/33


tions that they and their followers (e.g., Meggers, Lath-rap, Brochado, Oliver, Zucchi) have imagined. Whetheror not connected to ecological arguments like those ofMeggers or Lathrap, earlier models of ceramic and lin-guistic diffusion in Amazonia have assumed a too sim-ple relation between diffusion and migration. It mayseem reasonable to believe that actual (demic) Arawak-speaking groups brought Saladoid-Barrancoid potterywith them during their initial southeastward expansionalong the Ro Negro, as seems to be the consensus evenamong modern Arawakan scholars (cf. Hill and Santos-Granero 2002), but at the moment we have no way ofknowingwhatactually expanded, whether a populationor merely particular ways of talking and making pottery.

Considering the rapidity and apparent ease with whichindigenous populations along the Ro Negro were ableto adopt the Nheengatu trade language and Europeanutensils in the colonial era and a great number of otherAmazonian examples of language shifts and interethniccultural adoptions (cf. Aikhenvald 1999b, 2002),itispuz-zling that we should continue to think about pre-Co-lumbian cultural processes in terms of reified peoplesmigrating across the Amazon Basin. Nheengatu is spo-ken as a first language by some populations on the upperRio Negro to this day (Jensen1999:127).39 In view of itsconspicuously riverine distribution pattern, there is adistinct possibility that the Arawakan language family(i.e., proto-Arawak) similarly originated as a trade lan-guage of prehistoric Amazonia (see Schmidt 1917).40 Atany rate, it should be obvious that more recent ceramicstyles were able to spread quite rapidly in different di-rections along the network of trade routes controlled byArawak-speakers up until the arrival of Europeans. Fromthis perspective, at least, there is no contradiction be-tween concluding that Arawakan languages and Sala-doid-Barrancoid pottery styles originally spread eastwardalong the Ro Negro and the lower Amazon, on one hand,and suggesting that other ceramic styles employed byArawak-speakers at times may have spread in the op-posite direction, on the other. An influential and con-tagious Arawakan identity may indeed have beenfounded on a fairly coherent, if provisional, constellationof traits, including language and ceramic style, but thisdoes not allow us to draw any conclusions about popu-lation movements. Rather than continuing to reproduce

39.Nheengatu spread throughout the whole of the Amazon basin,reaching the border of Peru in the west, penetrating Colombia on

the Rio Vaupes in the northeast, and reaching Venezuela alongthe Rio Negro (where it is called Yeral). It is still spoken fairlyextensively along the Rio Negro and elsewhere in pockets in theAmazon region (Campbell1997:23). Similarly, even Quechua orig-inally spread through the central Andean area as a trading language(Schwartz and Salomon1999:457; Torero2002:91105; Stark1985:181).40.Arawakan languagesthe geographically most extended fam-ily in South Americatypically show more structural similari-ties to their neighbours than to their genetic relatives (Aikhenvaldand Dixon 1998:241). This is exemplified by the diffusion of evi-dentiality (a grammatical system whereby for every statement ofa certain type made in the language, the evidence on which it isbased must be stated [p. 244]) from Tukanoan and Panoan intonorthern and southern Arawak, respectively (p. 253).

the billiard-ball model of migrating, essentialized peo-ples pushing each other across the Amazon Basin andthus generating our linguistic and archaeological distri-bution maps, we should be asking ourselves what pre-historic linguistic and stylistic diffusion could tell usabout communicative processes within a pan-Amazo-nian system of exchange relations.41

Traces of Intensification

To dissociate the Arawakan ethos and its various stylis-tic markers from the notion of a biologically delineatedpopulation is not to stop asking questions about materialprocesses in Amazonian prehistory. Whatever genes theymay have carried, the Arawak-speaking potters of thefloodplains in the first millennium BC were engaged ina process of social transformation that also had majorecological repercussions. The tropical landscape still car-

ries imprints of these transformations. As in other areasof the New World (see Cronon 1983:4951), the pristinenature or wilderness discovered by the Europeans hasto a large extent proven to be a cultural landscape (seeHornborg2001). The black (terra preta de ndio) or darkbrown (terra mulata) anthropogenic42 soils that occuralong most larger rivers are less acidic and contain morehumus, nitrogen, and phosphorus than surrounding soilsand are appreciated by both indigenous and nonindige-nous farmers for their high fertility. Terra preta occursin patches normally varying between1and100hectaresin extent, with an average around 21 hectares. Somesites, however, are considerably larger, measuring 500hectares at Santarem,350 hectares at Juriti, west of San-

tarem, and 200 hectares plus1,000hectares ofterra mu-lataat Belterra, by the Tapajos (Denevan2001:105). Sitesare typically elongated and run parallel to riverbanks, forinstance, at Altamira, on the Xingu (1.8 km # 500 m,90hectares), and at Manacapuru, near Manaus (4km #

41. The ceramic style known as Guarita provides an interestingexample in that it suggests a continuity in material culture that insome areas survived a major linguistic shift. Although the (Barran-coid) roots of this style appear to have been among Arawak-speakersof the northwestern Amazon (Lathrap 1970:15657; Neves 2001:275), in southern and eastern Amazonia it seems itself to be an-cestral to the so-called Amazonian Polychrome Tradition, subse-quently associated with Tup-speakers (Brochado 1984:31321).Rather than debating whether all Guarita pottery was made byArawak- or Tup-speakers, we might therefore conclude that, at

some point in space and time, a population manufacturing Guarita-related ceramics appears to have shifted from an Arawak to a Tuplanguage.42. Experts disagree on whether Amazonian dark earths were de-liberately created as a technology of agricultural intensification(Myers et al.2003, Myers2004) or merely an unintentional artifactof long-term human occupation (Balee 1992:42; Neves et al. 2003:44). Considering recent research on Amazonian dark earths andfollowing Denevans (1992) observation on the excessive labor re-quirements of shifting cultivation using stone axes, many nowagree that shifting cultivation became prevalent in Amazonia onlyafter the European introduction of metal axes, implying that pre-Columbian horticulture in Amazonia was generally of a more per-manent character (Myers et al. 2003; Mora2003; Neves et al.2004:44; Myers 2004:74; Denevan 2004).


15/33


200 m, 80 hectares). A recent survey maps almost 400sites in Brazil alone (Kern et al. 2003). The deposits maybe up to2 m in depth. Most researchers agree that terra

preta was formed in connection with dense, sedentary,and extended human habitation, indicating more per-manent and often larger communities than those thathave been documented ethnographically in Amazonia.43

It has been proposed that the black soils are the resultof human habitation, while the dark brown soils are for-mer agricultural land (Herrera et al. 1992:102, referringto Andrade; Petersen, Neves, and Heckenberger 2001:100). Terras pretas are usually associated with artifactssuch as pottery, whereasterras mulatasare not (Kern etal. 2003:73). The most fertile (black) anthrosols appearto be the result of continuous deposition of householdgarbage, ashes, feces, urine, bones, shells, and other or-ganic material.

Terra preta begins to form at roughly the same timealong the larger rivers in Amazonia a few centuries BC.

At Hupa-iya, near Yarinacocha, by the Ucayali River inPeru, the oldest deposits of terra preta have been datedto 200 BC.44 At the large site of Acutuba on the lowerRo Negro, terra preta may have begun to form around360 BC. (Petersen, Neves, and Heckenberger 2001:97,100). In view of the correlations between ceramics andlinguistics mentioned above, the abundant finds of Bar-rancoid ceramics at both Hupa-iya and Acutuba, as atother sites withterra pretaalong the Amazon (see Myers2004:91), could be mustered in support of the hypothesisof an Arawak-speaking population on the intensivelycultivated floodplains from at least 500 BC up to its iden-tification as such by Europeans. These anthrosols indi-cate a population density that could not have been sus-

tained with the current (shifting) agricultural practicesof indigenous people in the area (Oliver2001:73; cf. Roo-sevelt1993). The evidence suggests that the floodplainsof Amazonia in the first millennium BC experienced anunprecedented concentration of human population inconjunction with significant economic intensificationand that these material processes were part and parcelof complex new social formations associated with whatwe have referred to as an Arawakan culture and identity.

Another lasting imprint of complex social systems andeconomic intensification in Amazonia is the earthworksthat have been identified in association with pre-Colum-bian settlements and cultivation systems. The most con-spicuous of these are the extensive drainage systems that

have been discovered in the seasonally inundated wetsavannas of Bolivia, Colombia, and Venezuela and inother waterlogged areas (floodplains, deltas, coastalzones, lake shores, marshes) in the Guyanas, highlandColombia, Ecuador, Peru, and highland Bolivia (Denevan2001:21590; Parsons 1985). In these widely separatedareas, a similar method of drainage was employed, the

43.A notable exception is Meggers, who attributes extensive areasof terra preta to discontinuous occupation by small groups (e.g.,1992).44. Whether Hupa-iya should be considered a terra preta site hasrecently been questioned, however (Myers 2004:79).

basic idea of which was to construct artificially raisedfields to protect the crops from periodic inundations.Such raised or ridged fields have been given variousnames in different parts of South America, but a com-monly used name is camellones, which was already inuse in the sixteenth century by Spaniards in the Orinocollanos and in highland basins of Colombia and Ecuador.It was also being used at least by 1674 for the famouschinampas of the Aztecs in Mexico, which appear tobelong to the same agricultural tradition. The term ca-mellones, which refers to camel humps, was obviouslyinvented by the Spaniards after1492but is used to thisday in the Beni area in Bolivia (Denevan 2001:217,220,23738, 252). Raised fields had several functions in ad-dition to drainage, including soil aeration, reduction ofroot rot, increased nitrification, pest reduction,reductionof acidity, moisture retention (

Documents

Hornborg_A - Ethnogenesis