Shepard jr e ramirez, 2011

“Made in Brazil”: Human Dispersal of the Brazil Nut(Bertholletia excelsa, Lecythidaceae) in Ancient Amazonia1

GLENN H. SHEPARDJR*,2 AND HENRI RAMIREZ3

2Dept. Antropologia, Museu Paraense Emilio Goeldi, Av. Perimetral 1901—Terra Firme, Belém, PA66077-830, Brazil

3Dept. de Letras e Pedagogia, Universidade Federal de Rondônia, BR 425, km 2.5—Jardim dasEsmeraldas, Guajará-Mirim, RO 78957-000, Brazil*Corresponding author; e-mail: [email protected]

“Made in Brazil”: Human Dispersal of the Brazil Nut (Bertholletia excelsa, Lecythidaceae) inAncient Amazonia. The Brazil nut, Bertholletia excelsa, is a colossal tree of terra firmeforest whose seeds represent the most important non-timber forest product in Amazonia.Its peculiarly inefficient dispersal strategy and discontinuous distribution have led some tohypothesize anthropogenic origins, but evidence to date has been inconclusive. Here wepresent results of a multidisciplinary study addressing this question. A review of thegeographic distribution of B. excelsa and comparison with that of similar Lecythis speciessuggest a number of anomalies that are consistent with a recent and wide colonization ofBertholletia. Published studies and field observations indicate that anthropogenic disturb-ance facilitates Brazil nut regeneration. Recent genetic studies showing no sequence dive-rsity and no geographical structuring of within-population variability support a rapid andrecent irradiation from an ancestral population. Historical linguistic analysis of indigenousterms for Brazil nut suggests a northern/eastern Amazonian origin for Bertholletia, with aconcomitant spread of Brazil nut distribution or cultivation to the south and west. Such anexpansion would have been particularly facilitated by the emergence of intensive bittermanioc cultivation and networks of interethnic trade beginning in the first millennium C.E.Together, ecological, phytogeographic, genetic, linguistic, and archeological data reinforcethe hypothesis that ancient Amazonian peoples played a role in establishing this emblem-atic and economically important rainforest landscape.

‘Made in Brasil’: A dispersão antrópica da castanha-do-Pará (Bertholletia excelsa,Lecythidaceae) na antiga Amazônia. A castanha-do-Brasil, Bertholletia excelsa, é umaarvore enorme da terra firme cujas sementes representam o produto florestal não-madeireira mais importante da Amazônia. Alguns pesquisadores, observando sua estratégiaineficiente de dispersão e sua distribuição descontínua, propuseram a hipótese de que suasorigens são antrópicas, mas as evidencias até a data são ambíguas. Aqui se apresentamresultados de um estudo multidisciplinar sobre essa questão. Uma revisão da distribuiçãogeográfica de B. excelsa e uma comparação com as sapucaias (Lecythis spp.) sugeremvárias anomalias compatíveis com uma recente colonização de Bertholletia pela Amazônia.Estudos publicados e observações em campo sugerem que a perturbação antrópica facilita aregeneração de castanhais. Estudos genéticos recentes demonstram nenhuma diversidadegenética de seqüências de cpDNA e nenhuma estruturação geográfica da variabilidadeintra-populacional, o qual sugere uma expansão rápida e recente. Estudos lingüísticossugerem uma origem para Bertholletia no norte/leste da Amazônia, com uma expansãomais recente da distribuição ou cultivação para o sul e o oeste. Tal expansão teria sidofacilitado pela emergência do cultivo intensivo de mandioca amarga e redes de contatointer-étnico especialmente a partir do primeiro milênio dC. Dados ecológicos, fitogeográ-ficos, genéticos, lingüísticos, e arqueológicos reforçam a hipótese de que os povos

1 Received 11 February 2010; accepted 26 January2011; published online ___________.

Economic Botany, XX(X), 2011, pp. 1–22© 2011, by The New York Botanical Garden Press, Bronx, NY 10458-5126 U.S.A.

amazônicos antigos tiveram um papel significante no estabelecimento dessa paisagemamazônica emblemática.

Key Words: Amazonia, non-timber forest products, plant genetics, landscape domestication,historical ecology, historical linguistics, Amazonian archeology.

IntroductionBrazil nut (Bertholletia excelsa Humb. &

Bonpl.) is a signature Amazonian species and animportant resource for local populations. Brazilnut’s nutritious, oil-rich seeds are eaten fresh,roasted, or pressed to produce a milk-like extract.The colossal trees reach up to 60 m in height and16 m in circumference (Pires 1984; Villachica et al.1996). The Brazil nut family, Lecythidaceae,includes the oldest known tree in the Americantropics, a specimen of Cariniana micrantha Duckedated to 1,400 years old (Chambers et al. 1998).Brazil nut trees ~150 cm in diameter have beendated to 270 years (Chambers et al. 1998), whilethe largest individuals (~500 cm in diameter) maybe over 1,000 years old (Pires 1984; cited in Peresand Baider 1997).Carbonized Brazil nuts were identified at Pedra

Pintada, an upper Paleolithic cave site in thecentral Brazilian Amazon that was occupied some11,000 years ago by ancient hunter-gatherers(Roosevelt et al. 1996). Although archeologistshave found a diversity of oily seeds, especiallypalm nuts, in Paleolithic sites throughout lowlandSouth America (e. g., Morcote Ríos et al. 2006),Pedra Pintada is the only one where Brazil nutconsumption is clearly documented.Brazil nuts were introduced to Europe in the late

18th century by Dutch traders, with trade increas-ing greatly in the late 19th century (Mori andPrance 1990a). Today, Brazil nut is Amazonia’smost important non-timber forest product. It isalso the only globally-traded seed crop collectedfrom natural forests (Clay 1997). Historically,Brazil has been the leading producer, but Boliviahas now taken the lead, with 2004 exports valuedat $50 million, compared with $15 million forBrazil and $10 million for Peru (Wander et al.2008). From 2002–2006 annual harvests inBrazil have varied from 24,895 to 30,555 metrictons (IBGE–Instituto Brasileiro de Geografia eEstatística 2004, 2007). Overall, the industryemploys some 200,000 people, mostly forest-based extractivists (Peres et al. 1997).Brazil nut grows in well-drained terra firme

forests throughout the Brazilian Amazon and adja-

cent areas in Bolivia, Peru, Colombia, Venezuela,and the Guianas. The Brazilian state of Pará containsthe largest populations (Müller et al. 1980). Brazilnut trees are found in groves (Fig. 1) of 50–100individuals known as castanhais (Portuguese),manchales or castañales (Spanish), with grovesseparated by considerable distances of compatiblehabitat where the species is completely absent (Peresand Baider 1997). This patchy distribution ledAdolpho Ducke (1946) to suggest that Brazil nutgroves might be plantations left by ancientAmazonian peoples. This “anthropogenic hypoth-esis” has been echoed by numerous authors since(Balée 1989; Müller et al. 1980; Posey 1985;Tupiassú and Oliveira 1967) without empiricaltest or systematic review. Here we review theliterature and present new results from the authors’studies of Brazil nut ecology and genetics, manage-ment practices by local people, and linguisticanalysis of indigenous terms for the species.Based on these findings we suggest that the Brazilnut was spread or facilitated throughout much ofits current distribution by ancient indigenouspopulations.

Botany, Taxonomy, and EcologyBertholletia is a monotypic genus of Lecythida-

ceae, a pantropical family of small to very largetrees. Lecythidaceae in the Americas are foundfrom Mexico to Paraguay and southern Brazil,with diversity and abundance centered on Ama-zonia. The family includes about 200 speciesdivided among ten genera; however, recentgenetic studies demonstrate that major taxonomicrevisions are needed in at least four of them (Moriet al. 2007). Bertholletia excelsa was named in1807 by Alexander von Humboldt and Aimé deBonpland in honor of the chemist L. C.Berthollet; the species epithet refers to its loftystature. A second species, Bertholletia nobilis,described by John Miers in 1874, was laterrejected as synonymous with B. excelsa. Morpho-logical features place Bertholletia closest to thegenus Lecythis, with affinities to L. lurida (Miers)S. A. Mori (Mori and Prance 1990a:135), but

ECONOMIC BOTANY [VOL

genetic studies were unable to elucidate itstaxonomic status within the Lecythidaceae (Moriet al. 2007).

The edible Brazil “nut” is not technically a nut,but rather a seed, encased within a large (11–15 cm.in diameter), heavy (~0. 6 to 1. 2 kg.), woody,exceptionally hard fruit or pyxidium. The fruits ofthe Brazil nut are unique within Lecythidaceaebeing both the hardest in the family and yet alsofunctionally indehiscent. All other Lecythidaceaewith fruits that fall to the ground at maturity eitherhave a fragile fruit wall (pericarp) that soon breaksopen, or else have a “lid” which falls off theoperculum (goblet-like opening), allowing theseeds to disperse. The opercular lid of the Brazilnut falls inward, rather than outward, blocking theopening which in any case is far too small (about1 cm in diameter) to release the seeds, whichaverage 2.0 cm wide by 5.0 cm long (see Peres andBaider 1997). The Brazil nut fruit is so dense andheavy that it sinks in water like a stone, ruling outthe possibility of aquatic dispersal (G. Shepard,

pers. obs.). Fruit development in Bertholletia isextremely slow, taking 15 months, comparedwith half that time for most Lecythidaceae.Bertholletia seeds (i. e., the Brazil “nut”) also takethe longest period of time to germinate of anyLecythidaceae, requiring 12–18 months for nat-ural germination, compared with almost instanta-neous germination for most species (Müller 1981;cited in Mori and Prance 1990a:12).

Seed Dispersal and Grove Formation:Natural and Anthropogenic FactorsSome have suggested that certain elements of the

Pleistocene megafauna, including the elephant-likegomphothere, extinct for more than 10,000 years,may have been involved in dispersing certain large-seeded plant species including the Brazil nut(Janzen and Martin 1982). Today, the main andperhaps only significant natural disperser of theBrazil nut is the humble red-rumped agouti(Dasyprocta leporine) and related species. Theagoutis, Dasyprocta spp., are large-bodied, diurnal,terrestrial rodents (infra-order Caviomorpha) withextremely sharp and permanently-growing teeth.After gnawing through the pericarp and removingthe tightly-packed seeds, the agoutis shell andconsume them on the spot or carry individualseeds a short distance to be buried in shallow“scatter-hoards” to be eaten later. Peres and Baider(1997) show that a quarter of Brazil nut seeds areconsumed immediately by agoutis, while 65% arescatter-hoarded an average of 5.0 m (range 0.5–28.4, n=217) from the point of origin. Presum-ably, some scatter-hoards are forgotten, providingshort-range dispersal within existing stands.Arboreal seed predators including the beardedsaki monkey (Chiropotes satanas), the red-neckedwoodpecker (Campephilus rubricollis), and mac-aws (Ara macao and A. ararauna) are able to pryopen green fruits in the forest canopy; however,the seeds at this stage are not viable (Peres et al.1997; Trivedi et al. 2004).

Peres and Baider (1997) assert the fundamentalrole of agoutis in dispersing Brazil nuts and rejectthe anthropogenic hypothesis proposed by Ducke(1946) and others. However, given the lack ofany evidence that agoutis disperse seeds long-distance, Peres and Baider (1997:613) suggestthat macaws or other arboreal seed predators maybe involved in rare, long-distance dispersal events.Once a single individual is established, however,

Fig. 1. Approximately 50 m tall Brazil nut tree in terrafirme forest of the Amanã Sustainable DevelopmentReserve, Amazonas, Brazil (photo © 2001 G.H. Shepard).

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

agouti scatter-hoarding would presumably lead tothe establishment of a new grove.Several published studies as well as our own

field observations indicate that human disturb-ance and intervention greatly facilitate Brazil nutregeneration, and may be crucial for the establish-ment of new groves. Indeed, the weakness ofPeres and Baider’s (1997) argument is that, whilescatter-hoarding agoutis are ubiquitous in Ama-zonia, Brazil nut saplings are exceedingly rare inprimary forest habitats (Pires 1984), requiringsignificant canopy gaps to develop (Myers et al.1996). Peres et al. (2003) have argued thatcommercial over-harvest may be responsible fora “demographic bottleneck,” though this inter-pretation has been criticized in the light ofsignificant human facilitation of Brazil nutrecruitment (see Stokstad 2003). Agoutis appearto disperse Brazil nuts preferentially into gardenfallows, where densities of seedlings and saplings aremuch higher (two and four times, respectively) thanin undisturbed forest (Cotta et al. 2008). Humanpredation of agoutis—an abundant game speciesoften hunted by indigenous peoples in theirgardens (e. g., Ohl-Schacherer et al. 2007)—would only tend to magnify the agouti’s impor-tance as a seed dispersal agent, freeing moreabandoned scatter hoards for germination.In the basin-wide survey of Brazil nut pop-

ulations published by Peres et al. (2003), by farthe highest densities of saplings and trees overallwere registered by Shepard (2002) in a small, fire-impacted grove near Alter do Chão, Pará, with 50individuals/ha, mostly in the size class of 10–60 cm diameter. This was a clear outlier in thedataset, where most groves had 10–100 timeslower densities of Bertholletia, and the majority ofindividuals were larger than 100 cm in diameter.What made the Alter do Chão site unique was itssituation in a drier, central Amazonian climatezone with high susceptibility to fire. Localinformants reported that the region suffered amajor fire in the 1980s, which cleared awaysignificant areas of forest understory but alsocompletely exterminated the local agouti popula-tion. This was the only region surveyed wheremulti-trunked Brazil nut individuals were found:nearly 50% of the trunks surveyed at the site (45of 93 trunks) were fused in groups of 2–5individuals (see Shepard 2002). This was puzzlinguntil multiple seedlings were observed emergingfrom single, unopened seed cases. Withoutagoutis to open and disperse the seeds, the seeds

germinate within the fruit cases. Peres and Baider(1997:599), citing de Souza (1984), claim thatseeds left inside unopened Bertholletia pyxidiararely if ever germinate, succumbing to fungus.Perhaps the drier climate or high levels of humandisturbance in Alter do Chão facilitate germina-tion in these circumstances. Notably, Alter doChão is within the proverbial stone’s throw of thePedra Pintada site where Roosevelt et al. (1996)discovered the earliest evidence of Brazil nutconsumption in ancient Amazonia.Associations between Brazil nut groves and

anthropogenic dark earths have been mentionedin the literature (Balée 1989; Conklin 2001), andwe found similar associations in our field expedi-tions. In the Amanã Sustainable DevelopmentReserve of Brazil, the community of Boa Esper-ança harvests Brazil nuts from a large andproductive grove nearby known as “CastanhalUrumutum.” The community is located within apatch of anthropogenic dark earths where over200 pre-Columbian funerary urns were discov-ered, indicating a significant ancient occupation(Shepard 2001). Guix (2005) found high den-sities of useful, large-seeded plants includingBrazil nut in soils rich with archeological remainsalong the Rio Negro River. Similar observationshave been made in recent archeological andbotanical surveys in the Rio Trombetas (Magalhães2009). Guix (2005) suggests that humans mayhave replaced extinct Pleistocene megafauna spe-cies in dispersing a number of economically useful,large-seeded tree species that might have otherwisegone extinct or suffered range reductions in thesudden climatic and ecological changes thatoccurred approximately 10,000 years ago (Pipernoand Pearsall 1998). Brazil nut trees have also beenfound in association with geoglyphs—square orcircular man-made trenches dated to between1,000 and 2,500 years ago (Pärssinen et al. 2009;Ranzi et al. 2007)—that have become visible innewly deforested areas in Acre. Similar formations,also rich in Brazil nut trees, were observed inRiberalta, Bolivia, near the junction of the Beniand Madre de Dios rivers (H. Ramirez, pers. obs.).Field observations made during a year-long

survey of Bertholletia populations throughout theBrazilian Amazon by Shepard (2002) revealedspecific ways in which local populations havepromoted Brazil nut stands through managementand direct plantation. For example, Ponta daCastanha is a managed Brazil nut grove on Tefélake near the Mamirauá Sustainable Development

ECONOMIC BOTANY [VOL

Reserve. The owner, Vitor Azevedo (Fig. 2), isthe son of a Peruvian immigrant who bought theland in 1944, when Vitor was four years old,from a Turkish merchant who, in turn, hadacquired it as collateral from an indebted cus-tomer. Ponta da Castanha covers about 20 ha ofrich dark earth full of pre-Colombian artifacts. Itoriginally contained 38 productive Brazil nuttrees that were overgrown with lianas. Vitor’sfather cleared the trees of lianas and beganplanting additional trees in small manioc planta-tions he opened through the years to feed hisfamily. Within the first few years of management,production of Brazil nuts had increased from 160to 1,000 liters per year. His father, and later Vitorhimself, eventually planted some 280 additionaltrees throughout the property. Vitor remembersthe exact year many individual trees were planted.Today, the density of Brazil nut there is 20 to 25trees/ha, among the highest densities observedanywhere, and Vitor harvests about 3,000 liters of

Brazil nut per year. Vitor’s father experimentedwith different plantation methods, at first plantingwhole, unopened pixydia—producing multipleindividuals with fused trunks, as noted above—until finally learning to pre-germinate individualseeds. Vitor observed, “You have to take care,protect the trees from vines and remove weeds;otherwise they don’t grow. And this business about15–20 years to fruit is a lie! Maybe in the forest, butin a garden fallow, it’ll give fruit in five or six years.”Pointing at a small fallow area on an adjacentproperty, Vitor remarked, “That fallow there, it waspasture. What, about 10 hectares? You can plant ahundred Brazil nut trees. A hundred trees, that’s acastanhal [Brazil nut grove]! A hundred trees willproduce 100–200 liters per year.”

The approximately 200,000-ha extractivereserve “Rio Preto de Ouro” near Guajará-Mirim,Rondônia, contains a vast and diffuse Brazil nutgrove of over 50,000 ha with an average densityof 1 to 4 Bertholletia individuals per ha (Shepard2002). Through the early 20th century, thisregion had been the economic and culturalheartland of the Wari’ (Fig. 3), a formerly isolatedand bellicose group known for their cannibalisticfunerary customs (Conklin 2001). The Wari’were displaced from Rio Preto de Ouro by rubbertappers during the “Rubber Boom” first in 1895–1917, and then during and after World War II.Traditional Wari’ funeral practices involved thecomplete destruction of all reminders of thedeceased, including consumption of the body.When Wari’ families return to their customaryBrazil nut groves after a death in the family, theyburn the underbrush and discarded fruit capsulesfrom the previous years’ collecting seasons(Fig. 4), altering the appearance of the groveand eliminating physical reminders of pastmoments shared with the deceased (Conklin1989). By generating localized disturbance andfertilizing the soil with ash, this practice certainlyfacilitates Brazil nut sapling recruitment. Conklin(2001) also notes an association between Brazilnut stands and anthropogenic dark earths (terrapreta do índio) in Wari’ territory. Culturallyunrelated Uru-Eu-Wau-Wau Indians in an adja-cent reserve were observed to cut away vines andundergrowth around Brazil nut trees to facilitatetheir healthy growth and longevity. CommercialBrazil nut harvesters throughout Amazonia removeunderbrush in groves to facilitate fruit collectionand to reduce the likelihood of snakebite (ScottMori, pers. comm.). These diverse ethnographic

Fig. 2. Vitor Azevedo, who helped his father plantmost of the Brazil nut trees currently found at Ponta daCastanha, Lago Tefé, Amazonas State, Brazil (photo© 2001 G.H. Shepard).

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

observations and other published accounts (Posey1985) attest to how indigenous and other peopleshave facilitated the recruitment of Brazil nut grovesin different parts of the Amazon.

Geographic Distribution and GeneticDiversity

The Brazil nut is found throughout a largegeographic range, from 5° N latitude in the upperOrinoco to 14° S in the upper Madre de Dios(Fig. 5, adapted from RADAM-Brasil 1973–1981; Mori and Prance 1990a:137; Desmoulièren.d. -a, b). Bertholletia is found as far east as theBrazilian Atlantic coast and as far west as theMarañon river in Peru. However, Brazil nut treesfound at the extreme eastern (Atlantic coast) andnorthern (Guianas) limits of distribution wereprobably planted during early colonial times(S. Mori, pers. comm.; see also Mori and Prance1990a). Within this broad range, Brazil nut iscuriously absent from most of the Juruá, Jutaí,and Javarí basins, despite occurring in adjacentareas to the south (Purús), north (Japurá, RioNegro), east (Solimões, lower Purús), and west(Marañon; see Fig. 5).The so-called “sapucaia” or pisonis group

within Lecythis is a morphologically and genet-ically coherent set of species (Mori and Prance1981, 1990b; Mori et al. 2007) including L.ampla Miers, L. lanceolata Poir., L. pisonisCambess. (containing two subspecies), and L.zabucajo Aubl. (= L. tumefacta Miers). The edibleseeds, known in Brazil as sapucaia and appreciatedby local populations, are superficially similar tothe Brazil nut. Unlike the Brazil nut, however,the fruit case is soft and fully dehiscent when ripe,allowing the seeds fall to the ground. Bats mayalso be involved in sapucaia dispersal (Greenhall1965).

Fig. 3. The Wari’ indigenous people of Rondôniagather prodigious amounts of Brazil nut both for do-mestic use and, today, commercial sale (photo © 2000G.H. Shepard).

Fig. 4. The Wari’ first gather and then burn scattered Brazil nut seed cases in the forest after a funeral to helpforget painful memories of dead loved ones (photo © 2000 G.H. Shepard).

ECONOMIC BOTANY [VOL

Comparing the geographical distributions ofspecies within the sapucaia group (Fig. 6) withthat of Brazil nut (Fig. 5), a strikingly differentpattern emerges. The sapucaias are distributedamong well-defined and geographically distinctivepopulations, as might be expected of a lineagewith a long history of dispersal and geographicalisolation between populations. Brazil nut, bycontrast, demonstrates an extensive geographicrange—equal to or exceeding that of the twoAmazonian sapucaias, L. zabucajo and L. pisonisssp. pisonis—and yet shows no internal taxonomicdifferentiation as might be expected of an ancientevolutionary lineage. Given the cumbersomedispersal mechanism of Bertholletia, comparedwith the dehiscent and more easily dispersedseeds of the sapucaias, one would expect Berthol-letia to show more rather than less geographicalisolation between populations, unless of coursethe Brazil nut’s evolutionary history has beenmore recent, and its dispersal process more rapid.

Reinforcing this conclusion, two prior studiesfound exceptionally low levels of genetic diversity

in Bertholletia compared with other tropical trees(Buckley et al. 1988; Kanashiro et al. 1997).Though Buckley et al. (1988) originally attrib-uted this result to special ecological characteristicsof the species, Kanashiro et al. (1997) noted thehypothesized interventions of indigenous peopleas a more likely explanation (see also Mori andPrance 1990a). Both studies, which used nuclearDNA markers, found far greater levels of geneticdiversity within Brazil nut groves than betweenthem, a result that is uncommon for wild woodyplant species (see Buckley et al. 1988) butcommon among cultivated species such as Euca-lyptus globulus Labill. and Camellia sinensis (L.)Kuntze (Kanashiro et al. 1997). Curiously,Kanashiro et al. (1997) found the highest levelsof within-grove phenotype diversity for the centralAmazon Santarém population (adjacent to Alter doChão and Pedra Pintada) and the lowest diversityfor populations from Acre in the western Amazon.Though the authors do not comment on this fact,the data might suggest a central Amazonian centerfor Brazil nut genetic diversity.

Fig. 5. Distribution of the Brazil nut (Bertholletia excelsa). Data synthesized from Mori and Prance (1990a:-137), a spatial analysis of RADAM-Brasil (1973–1981) inventory data conducted by Desmoulière (n.d. -a, b), andauthors’ pers. obs.

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

A Brazilian research group led by RogerioGribel and Maristerra Lemes used contemporarychloroplast gene (cpDNA) sequencing and micro-satellite markers to study genetic diversity of theBrazil nut (Gribel et al. 2007; Shepard 2002).The chloroplast genome, analogous to the mito-chondria genome in animals, is transmitted alongmaternal lines and thus relevant to studying seeddispersal. The results of this study revealed novariation for six non-coding cpDNA markersamplified and sequenced for eight widely sepa-rated (up to 2,800 km apart) Brazil nut popula-tions (Gribel et al. 2007). This result contrastswith cpDNA sequence variability documented atlocal scales (populations separated by as little as30 km) for other Lecythidaceae, including Lecy-this zabucajo (Hamilton et al. 2003). Micro-satellites are highly variable regions of DNAused to study genetic diversity within popula-tions, analogous to paternity testing in humans.Using eight microsatellite markers, Gribel et al.(2007) identified 21 haplotypes for 116 individ-uals from the eight widespread populations. Ananalysis of molecular variance revealed no signifi-

cant geographical structuring, with 93% of thegenetic variation found within populations, rein-forcing the prior findings of Buckley et al. (1988)and Kanashiro et al. (1997) using differenttechniques. If Brazil nut distribution dependedmostly on short-distance seed dispersal by agoutis,with rare, long-distance dispersal events of singleseeds to form new groves, the process would havetaken a very long time, and a geographicallycoherent pattern of genetic variability should haveemerged, as is the case for other Lecythidaceae.Instead, low genetic variability at a large geo-graphical scale suggests a recent and rapidirradiation of the species from a geographicallylimited population origin.Hans Carlos Müller, who has spent decades

gathering Bertholletia throughout Amazonia foragronomic experimentation (see Müller 1981;Müller et al. 1980), suggests the phenotypicvariation he has observed may be the result ofhuman selection (H. C. Müller, pers. comm.).For example, the Brazil nut variety known asabufari produces extremely large seeds (about7 cm in length) arranged like the individual slices

Fig. 6. Distribution of Lecythis spp. in the “sapucaia” or pisonis group. Data adapted from Mori and Prance(1981:72).

ECONOMIC BOTANY [VOL

of a grapefruit. He describes other varieties thatproduce exceptionally large or numerous seeds,that have exceptionally low tree crowns, or thatpresent variable fruiting and maturation dates,characteristics which are reproduced in offspring,ruling out mere ecological variation. Such phe-notypic variables (fruit size, low crown, etc.) aretypical of traits selected for by humans inincipient domestication of managed species(Clement 1990).

Linguistics and Cultural HistoryHistorical linguistics has been used to shed

light on the dispersal of ancient peoples and theircrops, languages, and genes (Bellwood 2001;Bellwood and Renfrew 2002; Brown 2006;Comrie 2002). Though caution is needed ininterpreting such data (Campbell 2002; Mooreand Storto 2002; Roosevelt 1992), proto-languagereconstruction and the study of loan words canprovide evidence about the timing and direc-tion of agricultural, technological, and cultural

innovations (Balée 2000; Balée and Moore1991; Comrie 2002; Urban 1992).

Henri Ramirez has collected a large database ofvocabulary words in numerous South Americanlanguages, including botanical and zoologicalterms (see also Ramirez 2001). Results for theBrazil nut, published and analyzed here for thefirst time (Appendix), suggest an intriguingpattern (Figs. 7 and 8). Of the three majorlanguage families within Brazil nut’s range—Arawak, Carib, and Tupi—only Arawak andCarib have terms for Brazil nut that reconstructto the respective proto-languages. Tupi, on theother hand, shows variable terms for Brazil nutacross different subfamilies that do not appear toreconstruct. The suggested proto-Arawak term forBrazil nut is *maiña or *maina, while thesuggested proto-Carib word is *tutka or *tutuka(the asterisk denotes hypothetical proto-vocabu-lary terms deduced from modern forms). Whenspecific vocabulary items (plants, animals, tools,etc.) reconstruct to the proto-language (barringrecent loan words, which can be detected through

Fig. 7. Indigenous terms for Brazil nut in the Amazon, showing approximate geographical location of eachgroup, color-coded for language family.

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

careful study), it is presumed that these itemswere present in the cultural and environmentalmilieu at the time the proto-language was spoken(Facundes 2002; Moore and Storto 2002).Contradicting earlier hypotheses, which were

based on fragmentary or flawed evidence (Noble1965; Schmidt 1917), more recent archeologistsand linguists propose that Arawak peoples origi-nated in the northern portion of the Amazonbasin, though opinions are divided as to theprecise center of origin, whether in central(Lathrap 1970; Ramirez 2001:26) or northwest-ern Amazonia (Heckenberger 2002:99; Oliver1989). Arawak speakers began a vigorous expan-sion approximately 3,000 years ago and came tooccupy a vast region from the savannas ofsouthern Brazil, to the Caribbean, to the Andeanfoothills of Peru and Bolivia (Hill and Santos-Granero 2002; Payne 1991). Carib languageswere long thought to have emerged in thesouthern Amazon (Rodrigues 1985; Steinen1894). However, a more recent internal classi-fication by Meira (2006:200) suggests a northernorigin in the Guianas (see also Heckenberger2005; Lathrap 1970; Meira and Franchetto

2005). Regardless, the relatively close linguisticproximity among existing Carib languages sug-gests a relatively recent common ancestor, withperhaps only half the time depth of the Arawak orTupi language families, estimated to have beguninternal diversification more than 3,000 years ago(Payne 1991; Rodrigues 1999). Wherever proto-Carib speakers found themselves some 1,500 to2,000 years ago, Brazil nut appears to have been asalient element of their environment.The Arawak and Carib cases contrast with that

of the Tupi family, for which a proto-word forBrazil nut does not appear to reconstruct.Variable proto-words for Brazil nut reconstructfor some of the Tupi subfamilies (Tupi-Guarani,Tupari) and perhaps other intermediate group-ings (see Appendix). This speculative linguisticevidence suggests that the earliest proto-Tupispeakers might not have known the Brazil nut,but came to know it (either through migration orinterethnic contact) after certain subfamilies haddiverged. Both linguistic and archeological dataprovide strong support that the Tupi languagefamily originated in the southern Amazon, likely inthe upper Tapajos and Madeira rivers in what is

Fig. 8. Preliminary historical/geographical analysis of indigenous terminology for Brazil nut.

ECONOMIC BOTANY [VOL

now the Brazilian state of Rondonia (Heckenbergeret al. 1998; Métraux 1928; Rodrigues 1964). Thisregion currently contains large and importantpopulations of Brazil nut; thus the absence of aclear proto-word is striking and anomalous.Indeed, Eurico Muller, the preeminent arche-ologist of Rondonia’s prehistory (see Miller1992), remarked on the curious absence of Brazilnut remains (though charred palm nuts arecommon) from 4,000-year old sites he hasexcavated in regions where Brazil nut groves area dominant element of the current landscape(E. Muller, pers. comm.).

Diverse language groups near the limits of thecurrent distribution of the Brazil nut refer to itusing loan words from neighboring languages.For example, the Arawak-speaking Lokono in theGuianas call the Brazil nut tutuka, clearly a loanword from neighboring Carib speakers. Likewisethe Tupi-speaking Tembé in Pará (easternAmazon) call it teko-ingwer, the first element ofwhich may be a loan from Carib (an alternativename for Brazil nut in Tembé, zapukaza’i refersto Lecythis pisonis, an example of naming byanalogy; see below). The word for Brazil nut inTikuna, ñoo, appears to be a loan word from thereconstructed Tupi-Guarani ña, associated withthe late western expansion of Tupi speakersalong the main Amazon channel (Rodrigues1999). Other indigenous groups along thefringes of its distribution refer to the Brazil nutwith regional vernacular terms: the Yekuana(Carib) term, wufia, is a loan word from theregional Venezuelan term, jubia, while multiplegroups in southern and eastern Brazil have termsderived from the Portuguese castanha or theregional term tocari, probably of Carib origin(see Appendix, Figs. 7 and 8).

In the Tupi-Guarani subfamily of the Tupifamily, the reconstructed proto-term for Brazilnut, *(i)ña, is close enough to the proto-Arawak*maiña to warrant further scrutiny. The termminata in Kamayurá (within Tupi-Guarani) isespecially similar to the Arawakan form. Maneéhin Maku (northwest Amazon) and méhe inTaruma (Guiana region) are more clear-cut casesof Arawak loan words to unrelated languages.

In some cases, regional loan word patternssuggest longer chains of interethnic contact ormigration. For example, along the Madeira River,the Mura word for Brazil nut, tihii, is similar tothe word among the unrelated but neighboringMatanawi, txipii. In the headwaters of the

Madeira, the Takana-speaking Esse-Eja refer toBrazil nut as xiwiwi, more similar to the Muraand Matanawi words than the cluster of terms(moke, moje, muihe) used by their Takananeighbors. The Mura and Matanawi terms (tihii,txipii) bear at least superficial resemblance tonearby Carib forms such as tetkï, and even to theword for peanut, dihi, among the Leko in theupper Madeira (see discussion below about pea-nut/Brazil nut associations). The Iquito term sahii(Peruvian Amazon), Katawixi sákodia (centralAmazon), and Asurini sa (Tupi-Guarani ofTocantins) also show a superficial similarity.Similarity among vocabulary items can alsoemerge by chance, and thus more systematicinvestigation would be required to test thesespeculations.

A stronger case for chains of linguistic borrow-ing can be made for a cluster of unrelatedlanguage families located in the Purús basin andMadeira headwaters in the southwest Amazon.Terms in the Arawá (not be confused with theArawak) language family such as mowe and moi’diare strikingly similar to the Tupian Arara termmowi, and also resemble nearby Arawak termssuch as Kaixana maihu and Marawá manazi.Further south in the Madeira headwaters betweenBolivia and Perú, the reconstructed proto-Takanaterm *moike is similar both to these modernArawá terms (mowe, moi’di) and to *maïkï asreconstructed for the proto-Piro-Apurinã subgroupof Arawak (see Fig. 8, Appendix). The Harakmbutword morikke is a clear loan word from the proto-Takana *moike.

The Harakmbut presents a particularly inter-esting case, since Brazil nut is virtually absentfrom their current territory on the Manu andupper Madre de Dios rivers in Peru (upperMadeira tributaries). A few isolated individualsof Brazil nut are currently found in the forestinterior several kilometers from the Pakitsa guardpost of Manu National Park (G. Shepard, pers.obs.), far from the commercially viable Brazil nutgroves on the lower Madre de Dios that havebeen considered the southernmost distributionlimit (Mori and Prance 1990a). These isolatedBrazil nut trees in Manu are not likely to havearrived at Pakitsa by natural dispersion, and wereinstead probably brought by indigenous peoplesuch as the Harakmbut-speaking Toyeri whooccupied Manu before being decimated by rubbertappers beginning in the 1890s (Shepard et al.2010). The Harakmbut word for Brazil nut, and

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

the Brazil nut trees themselves, appear to havebeen acquired either through downstream tradingwith the Takana, or as a result of an earlierupstream migration. In the latter regard, Adelaar(2000) suggests tentative linguistic connectionsbetween Harakmbut and the Katukina languagefamily of Brazil. This example suggests a linkbetween processes of linguistic borrowing andactual plant dispersal.A number of unrelated, geographically sepa-

rated languages in the southern and westernAmazon appear to have named the Brazil nutthrough analogy to some other edible nut (seeAppendix), particularly the peanut (Arachis hypo-gea L.). Novel plants or animals are often namedby analogy with more familiar local species(Berlin 1992; Witkowski and Brown 1983).The Portuguese and Spanish words for Brazilnut represent precisely such a case, where theterms castanha and castaña referred originally tothe chestnut (Castanea spp.), and later came torefer to the Brazil nut through analogy. TheMatsigenka (Arawak) live in the Andean foothillsoutside the Brazil nut distribution, and came toknow it only in recent decades through trade; theyrefer to Brazil nut either as inke, literally “peanut,”or else use the Spanish term castaña. In a similarfashion, the linguistically unrelated Sharanahua(Panoan) and Kokama (Tupi), who live near thesouthern and western limits (respectively) ofBrazil nut distribution, refer to it as “largepeanut” (see Appendix). These groups, like theMatsigenka, appear to have encountered theBrazil nut relatively recently. Among the Panoanlanguages there seems to be thorough interchan-geability between terms: while the Sharanahuacall the Brazil nut “large peanut” (tama wan), theChacobo call the peanut a “ground Brazil-nut”(mai tapa).Among multiple subgroups of southern Tupi

languages, the word for Brazil nut is suspiciouslysimilar to the word for peanut: in Makurap, arao(where arawï is “peanut”); in Mondé, mam(where mam kap is “peanut”); Karitiana, mijo(where mĩ’ĩ is “peanut”); and Munduruku, wenïj/wenã (where wenã-bïn ñe is “peanut”). The iso-lated languages Kanoe and Rikbaktsa (Rondônia)show a similar semantic overlap between Brazil nutand peanut (see Appendix). Likewise in the twoNambikwara dialects (southern Amazon), Brazilnut is wana’ and wanakka, respectively, whilepeanut is waiki and waikki. In the relatedSabanê language, kwaiki for Brazil nut is

reminiscent of the word for peanut, wai-se,among the neighboring Arawak-speaking Pareci.The peanut, curiously, was probably first domes-ticated nearby in the dry south Amazon borderregion (see Piperno and Pearsall 1998).Among the more northerly Tupi groups

(Tupi-Guarani subgroup), the word for Brazilnut in some languages is closely related to thegeneral word for “seed,” even showing systematicsound correspondence between for example ña(“Brazil nut”) and—a’ïña (“seed”) in Wayampi-Kawahip-Apiaká, and sa (“Brazil nut”) and—a’ïsa (“seed”) in Asurini. Likewise the Chapa-curan (Rondônia/Bolivia) terms tokwe, tokä, tike,teke are very close to the general terms for “seed”(tokwin, toki).Linguistic borrowings depend upon complex

factors involved in sociolinguistic contact (Campbell2002; Comrie 2002; Dixon 1999), and interpre-tation of such data presents numerous challenges.We speculate, based on analysis of loan words andsemantic extension (“nut”/“peanut”), that someof the language groups around the fringes of theBrazil nut distribution, and along certain key riverroutes (e. g., upper Madeira, western Amazon),encountered the Brazil nut relatively recentlythrough migration, trade, or contact. The Tupicase is particularly important, since they arepresumed to have originated in the upperMadeira/Tapajos region, which currently containsvast, commercially productive Brazil nut groves.Our preliminary linguistic analysis suggests thatsome four millennia ago, when the Tupi languagefamily emerged, the Brazil nut may not have beenpresent in their environment.Recent archeological studies (Arroyo-Kalin

2008; Neves et al. 2003) have demonstrated thatthe large patches of anthropogenic dark earths, orterra preta do índio, found mostly in the BrazilianAmazon, resulted from the intensification ofagriculture and the emergence of sedentary life-styles, especially during the first millennium C.E.Arawak peoples have been implicated in thespread of sedentary agriculture in the Amazon(Schmidt 1917), the generation of these darkearth soils (Arroyo-Kalin 2008), and the forma-tion of large-scale interethnic trade networks (Hilland Santos-Granero 2002). Sedentary lifestylesand dark earths are especially associated with thecultivation of bitter manioc, which requires labor-intensive processing that generates large amountsof charcoal during the cooking and toasting ofvarious kinds of manioc flour and other byprod-

ECONOMIC BOTANY [VOL

ucts; bitter manioc cultivation predominates inthe eastern half of Amazonia, the Orinoco basin,and the Guianas (Arroyo-Kalin 2008). Sweetmanioc (“yuca”), by contrast, requires no specialprocessing other than simple cooking, and ispredominant in the western Amazon where semi-nomadic shifting cultivation is the norm, andwhere both the Brazil nut and anthropogenicdark earths are mostly absent. The intensiveagriculture practices required to create largepatches of anthropogenic dark earths provideexactly the combination of anthropic factors thatwould have facilitated the establishment of Brazilnut stands. Brazil nut groves are often associatedwith anthropogenic dark earths, and Arawaklanguages appear to represent an important hubof loan words for Brazil nut to other languagefamilies. Indeed, the distribution of Brazil nutshows striking similarities with the distribution ofknown dark earth sites in the Amazon basin(see Kern et al. 2004:54).

Several authors have noted the conspicuousabsence of the Brazil nut in the Juruá basin (Moriand Prance 1990a; Fig. 5). While soil conditionsmay be a factor, the Juruá is also a region ofcultural disjunction between more geographicallycircumscribed Arawá and Panoans, surrounded tothe north and south by suggested routes of Arawakand multiple Tupi expansions along the Madeiraand Amazon proper (Aikhenvald 1999; Hornborg2005; see also Heckenberger 2002:105). Thus thelimits of the Brazil nut distribution may represent atleast in part the limits of various cultural-linguistic“diasporas” (Heckenberger 2002) associated withthe intensification of agriculture, especially begin-ning in the first millennium C.E. (Arroyo-Kalin2008; Neves and Petersen 2006). Speculativedating of language families (see Appendix)supports a similar time frame (1,500–2,000 yearsago) for the acquisition of loan words or analogyterms (“peanut,” “seed”) for Brazil nut in severalAmazonian language families in the southernand western Amazon (Tupi, Pano, Takana,Nambikwara, etc.).

ConclusionThere is so far no “smoking gun” that proves

Brazil nut groves are the forest plantations ofancient indigenous peoples, as Ducke (1946)once hypothesized. However, a preponderanceof evidence from independent lines of research,

some already published and some presented herefor the first time, lends credence to some degreeof human involvement in the dispersal of theBrazil nut to its current range. A review of thegeographic distribution of B. excelsa, and compar-ison with that of several Lecythis species withsimilar, more easily dispersed seeds, suggest anumber of anomalies that are consistent with arelatively recent colonization of Bertholletiathroughout Amazonia. The dispersal ecology ofthe Brazil nut renders it highly responsive to andperhaps largely dependent on anthropogenicdisturbance for the establishment and expansionof groves, at least given post-Pleistocene ecologicalconditions. Field observations and a review ofethnographic examples suggest how specific cul-tural practices might have facilitated the expan-sion of Brazil nut populations from ancientthrough recent times. Phenotypes observed incertain Brazil nut populations suggest a degree ofselection and incipient domestication. Pastgenetic studies suggesting low degrees of inter-population genetic diversity were confirmed andmade more emphatic by our own more recentstudies of chloroplast DNA, suggesting a recentand rapid dissemination from a restricted pop-ulation of origin.

Historical linguistic analysis of indigenousterms for the Brazil nut reinforces our inter-pretation of previously published genetic(Kanashiro et al. 1997) and archeological(Roosevelt et al. 1996) data, suggesting a northern/central Amazonian origin for Bertholletia, witha more recent spread of Brazil nut distribution(and cultivation?) to the south and west. Such anexpansion would have been particularly facili-tated by the emergence of intensive bittermanioc cultivation and networks of interethnictrade associated with the Arawak diaspora of thefirst millennium C.E. (see Heckenberger 2002).The often-noted association between Brazil nutgroves and anthropogenic dark earths—themselvesa result of intensive pre-Colombian sedentaryagriculture—lends support to such an interpreta-tion. Our arguments contribute to a body ofrelatively recent discoveries challenging the long-standing view of pre-Colombian Amazonianpeoples as small, low-impact nomadic popula-tions, revealing instead the significant legacy ofancient indigenous peoples in shaping modernAmazonian landscapes (Balée and Erickson 2006;Heckenberger et al. 2008; McCann et al. 2001;Roosevelt 1980).

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

AcknowledgmentsThe authors wish to acknowledge Rogerio Gribel

and Maristerra Lemes for support and researchcollaboration in the early phases of this study. Wealso acknowledge Eduardo Góes Neves and theMuseum of Archeology and Ethnology at Universityof São Paulo for support during a later phase of theresearch. We thank Manuel Arroyo-Kalin for manyuseful suggestions on a draft of the paper. We alsothank Sylvain Desmoulière for kindly sharingunpublished geographical analyses of RADAM-Brasil inventory data on the Brazil nut. Thanks also

to Carlos Peres for sharing data, observations, andphotographs during various drafts. We thank ScottMori and two anonymous reviewers for their carefulreading of the manuscript and many helpful com-ments and revisions.We acknowledgeDennyMoorefor urging caution in our linguistic interpretations.Finally, we thank Joshua Birtchall for helpful com-ments on the final draft of some figures. Differentphases of the research were supported by Brazil’sConselho Nacional de Desenvolvimento Científico eTecnológico (CNPq) and Fundação de Amparo aPesquisa do Estado de São Paulo (FAPESP).

AppendixRegional, vernacular, and indigenous terms for Brazil nut (Bertholletia excelsa), researched andorganized by H. Ramirez. All vernacular terms were collected in the field except where bibliographicalsources are noted.

REGIONAL AND VERNACULAR TERMS:

& noix du Brésil (French) < *Latin nuc(e) “fruit of a european tree (Juglans regia),or any similar kind of fruit with an almond” < *Indo-European knu(t).

& Brasil nut (English) < *Germanic knut “hard seed” < *Indo-European knu(t).NOTE: The German and English words for the peanut (Arachis hypogaea) have been formedfrom the same linguistic root: pea + nut in English and erd + nuβ nut of earth in German.

& castanha (Portuguese), castaña (Spanish) < *Latin (nux) castanea “nut of the chestnut tree (Castaneavesca) or any similar kind of fruit < *Greek kástanon < Asiatic language (cf. English chestnut < *OldEnglish chesten nut < *Old French chastaigne “chestnut”).

& almendra (Spanish, cf. Portuguese amêndoa, French amande, English almond) < *Latin amygdala“almond” < *Greek amygdále “ amygdalis”.

& to(ro)cari (Brazil of XVIIIth century) < probable Carib loan, with the non-Carib suffix -ri.& touca (French Guiana) < probable Carib loan: tutuka in Kari’ña.& yuviá, jubia (Venezuelan Spanish) < loanword from an unknown indigenous language.

Symbols, abbreviations, and orthography:

N.P. Brazil nut not present in locality† extinct language[<...] etymological meaning or loan word— / — synonyms(— yrs.) approximative time depth for the proto-language*— hypothetical reconstructed proto-formÏ high central vowelÄ mid central vowely high round front vowel (like u in French)ñ palatal nasalj palatal semivowel (like y in English)x voiceless palatal fricativetx voiceless palatal affricate’ glottal stopã, ẽ,... nasalized vowel

ECONOMIC BOTANY [VOL

INDIGENOUS TERMS:

NOTE: Terms are organized by linguistic families. Roman numbers (I, II, III, etc.) indicate thesubfamilies, while letters (Ia, Ib, Ic, etc.) show further subgroupings. New internal classifications havebeen suggested here by H. Ramirez for Arawá, Arawak, Chapacura, Carib, and Pano families. Whereongoing lexico-statistical study permits, suggestions for language family time depth are provided inparentheses. Supplemental terms for “peanut,” “nut,” etc. are provided where relevant.

Aikana jiry (wikere / wita peanut)Arawá (2,000 yrs.)

I) PAUMARI moi’di (mowa flower)II) ZOROWAHA namï-wasazu (namï high?, wasazu inajá [Attalea sp.])III) YARAWARA-DENI

IIIa) YARAWARA mowe (mowe flower) (cf. Arara-Aripuanã [Tupi] mowi)IIIb) DENI wato

Arawak (Aruak, Maipure) (4,500 yrs.) *maiña/*maina

I) KAIXANA maíhu / maikï (sïmi / sumi seed)II) †BAHUANA miñi’iIII) WAPISHANA minaïIV) MAWAYANA mijaV) †MARAWÁ manazi (usi seed) (data from Tastevin 1920)VI) PIRO-APURINÃ (2,000 yrs.) *maï(-)kï (cf. proto-Takana *moike, Harakmbut morikke,

Paumari [Arawá] moi’di)

VIa) PIRO mïxi / janajsi (-xi seed)VIb) INAMPARI mïhï (-hï seed)VIc) APURINÃ makï / make / mitjatakuru (-kï seed)

VII) LOKONO tútuka (< Carib)VIII) PARECI tokali-se / tokware-se (< Carib) (wai-se peanut)IX) CAMPA/MATSIGENKA inke (= peanut), kastaña (<Sp.)

Cahuapana xiwako’, monopi, tanpa’pi nuts spp.

Carib (2,500 yrs.) *tut(u)-ka (cf. Mura-Piraha tíihí)

I) GUIANA

Ia) † PALMELLAS tutuko (data from Fonseca 1880–1881)Ib) WAIWAI tïtko

HIXKARYANA tutkoKAXUYANA tutko

Ic) TRIÓ tuhka / tuuka

† OYARICOULÉ tura-tura

Id) APALAI tutuko

WAYANA tutukä / tutukoWAIMIRI-ATROARI tetkïKARI’ÑA tutuka

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

II) YEKUANA wufia (< Venezuelan Spanish ‘jubia’)III) ARARA toromoIV) KUIKURO N.P. (tigite peanut)

Chapacura (1,300 yrs.) *toke/*tike (< seed / nut?; cf. Carib *tutka?)

I) CHAPAKYRA

Ia) WARI’ tokwe (tokwi-n seed), kaji-tokwe peanut (< kaji introduced,tokwe nut)

Ib) MIGUELENHO tike (toki seed)Ic) ORO-WIN teke (toki seed)

II) CHAPAKURA (MORÉ) tokä (toki-n seed)

Harakmbut morikke (cf. Takana *moike, Piro-Apurinã *maïkï)(bogpi[h] peanut)

Iquito sahii (Lev Michael, pers. comm.) (cf. Katawixisákodia, Asurini sa)

Jabuti (1,500 yrs.) *ore

I) JABUTI oreII) ARIKAPU orä

Kanoe epy (epy-kwã peanut, -kwã seed)†Katawixi sákodia / sakudga (cf. Iquito sahii, Asurini sa) (data

from Tastevin 1909/1920)Kayapo pi’ï (pi tree)Kwaza kũc (c rĩ peanut)Leko N.P. (dihi-wo peanut, -wo round, cf. Mura-Piraha

tíihí Brazil nut)Maku (nadëb) maneéh (< Arawak)†Matanawi txipií (cf. Mura tíihí) (data from Nimuendajú 1925)Moseten N.P. (dabah peanut, cf. Pano tapa)Munku tjuka-i (< Pareci)Mura-Piraha tíihí (cf. Leko dihi-wo peanut)Nambikwara (2,500 yrs.)

I) SABANÊ kwaiki (cf. Pareci wai-se peanut)/ tokali’ (< Pareci)

II) NAMBIKWARA

IIa) NORTHERN wana’ (wai-ki peanut, -ki seed < Pareci wai-se)IIb) SOUTHERN wanakka (waik-ki peanut, -ki seed < Pareci wai-se)

Pano (1,500 yrs.) *tama/*tapa peanut, Brazil nutNote: Pano is now considered genetically related to Takana

I) KASHARARI tama Brazil nutII) CENTRAL PANO

IIa) CHACOBO tapa Brazil nut (mai-tapa peanut < mai earth, tapa nut)IIb) SHARANAHUA tama-wan Brazil nut (< tama peanut, wan big)IIb) OTHERS (Brazil-Peru) N.P. (tama peanut: Amahuaca, Yora, Kashinawa,

Katukina-Pano, Shanenawa, Yawanawa, Shipibo-Conibo, Capanawa, Cashibo, Wariapano)

ECONOMIC BOTANY [VOL

Rikbaktsa pitsi (pitsi-pïrïk peanut, pïrïk trunk, stem)Takana (1,500 yrs.) *moike (cf. Harakmbut morikke, Piro-Apurinã *maïkï)Note: Takana is now considered genetically related to Pano

I) CAVINEÑA mokeII) TAKANA CENTRAL

IIa) TAKANA PRÓPRIO muiheIIb) REYESANO muiheIIc) ARAONA moje / ewi

III) ESSE-EJJA xiwiwi (cf. Mura-Piraha tíihí)

Taruma méhe (<Arawak?)Tikuna ñoó (< Tupi-Guarani) / torenïTupi (4,000 yrs.)

I) MUNDURUKU-KURUAYÁ wenï(j) / wenã / wa(i)nai / waeraña / erai(wenã-bïn n peanut)

II) JURUNA (i)jãIII) TUPI-GUARANI

IIIa) TUPI-GUARANI *jã / *(i)ña (mandu[w]i / mundu[w]i peanut)(i)jã / ñã (Wayampi, Kawahip, Apiaká)(-a’ïñ[a] seed)

ñï / dja (Asurini Xingu)sa /so (Asurini Tocantins) (-a’ïs[a] seed;cf. Iquito sahii, Katawixi sákodia)txu (Parakanã)kãtãi (Urubu) (< Portuguese ‘castanha’)mina-ta (Kamayurá) (cf. proto-Arawak

*maiña/*maina)zapuka-z-a’i (Tembé) (< zapukaj sapucaia

[Lecythis sp.], a’i seed)teko-ingwer (Tembé) (< Carib?)muni-watsu (Kokama) (< muni peanut,

watsu big)ï’wat / ïwa-ete (Kayabi) (< ïwa tree , ete true)

IIIb) MAUÉ wẽ’ẽñã (-ã’ïñ[a] seed)

IV) TUPARI-MAKURAP

MAKURAP arao (arawï peanut)TUPARI

TUPARI aráo-’a / kãnã (hiráp peanut, kit seed)WAYORO karã (aragwi peanut, kiit seed)SAKIRABIAT karã (araakwi peanut, kiit seed)

V) MONDÉ mam / mom (mam-kap peanut, kap seed)VI) ARARA / RONDÔNIA ijãVII) KARITIANA mijo (mĩ’ĩ peanut)VIII) ARARA / ARIPUANÃ mowi (cf. Yarawara [Arawá] mowe)IX) PURUBORÁ mam-ka / ham-ka ([h]e’ -kap / i-kap peanut, kap

seed)

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

Literature CitedAdelaar, W. F. H. 2000. Propuesta de un nuevo

vínculo genético entre dos grupos lingüísticosindígenas de la Amazonía occidental: Harakmbuty Katukina. Pages 219–236 in L. M. Esquerre,ed., Actas del I Congreso de Lenguas Indígenasde Sudamérica, Vol. 2. Universidad RicardoPalma, Lima.

Aikhenvald, A. Y. 1999. The Arawak languagefamily. Pages 65–106 in R. M. W. Dixon andA. Y. Aikhenvald, eds., The Amazonianlanguages. Cambridge University Press, Cam-bridge.

Arroyo-Kalin, M. 2008. Steps towards an ecologyof landscape: A geoarchaeological approach tothe study of anthropogenic dark earths in thecentral Amazon region, Brazil. Ph.D. thesis,Dept. of Archeology, University of Cam-bridge, Cambridge, U.K.

Balée, W. 1989. The culture of Amazonianforests. Pages 1–21 in D. A. Posey and W.Ballee, eds., Resource management in Ama-zonia: Indigenous and folk strategies. NewYork Botanical Gardens, New York.

———. 2000. Antiquity of traditional ethno-biological knowledge in Amazonia: The Tupi-Guarani family and time. Ethnohistory 47(2):399.

——— and D. Moore. 1991. Similarity andvariation in plant names in five Tupi-Guaranilanguages (Eastern Amazonia). Bulletin of theFlorida Museum of Natural History: Biolog-ical Sciences 35(4):209–262.

——— and C. Erickson. 2006. Time andcomplexity in the Neotropical lowlands:Explorations in historical ecology. New York:Columbia University Press.

Bellwood, P. 2001. Early agriculturalist popula-tion diasporas? Farming, languages and genes.Annual Review of Anthropology 30:181–207.

——— and C. Renfrew. 2002. Examining thefarming/language dispersal hypothesis. Cam-bridge, U.K.: McDonald Institute forArchaeological Research.

Berlin, B. 1992. Ethnobiological classification:Principles of categorization of plants and animalsin traditional societies. Princeton UniversityPress, Princeton.

Brown, C. H. 2006. Glottchronology and thechronology of maize in the Americas. Pages647–663 in J. Staller, R. Tykot, and B. Benz,eds., Histories of maize: Multidisciplinaryapproaches to the prehistory, linguistics, bio-geography, domestication, and evolution ofmaize. Academic Press, Burlington.

Buckley, D. P., D. M. O’Malley, V. Apsit, G. T.Prance, and K. S. Bawa. 1988. Genetics ofBrazil nut (Bertholletia excelsa Humb. &Bonpl.: Lecythidaceae). Theoretical andApplied Genetics 76:923–928.

Campbell, L. 2002. What drives linguistic diver-sity and language spread? Pages 49–63 in P.Bellwood and C. Renfrew, eds., Examiningthe farming/language dispersal hypothesis.McDonald Institute for ArchaeologicalResearch, Cambridge.

Chambers, J. Q., N. Higuchi, and J. P. Schimel.1998. Ancient trees in Amazonia. Nature391:135–136.

Clay, J. W. 1997. Brazil nuts: The use of akeystone species for conservation and develop-ment. Pages 246–282 in C. H. Freese, ed.,Harvesting wild species: Implications for bio-diversity and conservation. John HopkinsUniversity Press, Baltimore.

Clement, C. R. 1990. Origin, domestication andgenetic conservation of Amazonian fruit treespecies. Pages 249–263 in D. A. Posey and W.Overall, eds., Ethnobiology: Implications andapplications. Proceedings of the First Interna-tional Congress of Ethnobiology. Museu Para-ense Goeldi, Idi/CNPq, Belém, Para.

Comrie, B. 2002. Farming dispersal in Europeand the spread of the Indo-European languagefamily. Pages 409–419 in P. Bellwood and C.Renfrew, eds., Examining the farming/languagedispersal hypothesis. McDonald Institute forArchaeological Research, Cambridge.

Conklin, B. A. 1989. Images of health, illness anddeath among the Wari’ (Pakaas Novos) ofRondonia, Brazil. Ph.D. thesis, MedicalAnthropology Program, University of California,San Francisco.

———. 2001. Consuming grief: Compassionatecannibalism in an Amazonian society. Austin:University of Texas Press.

Yanomami hawariYurakare N.P. (sebbe peanut, cf. Pano tapa, Moseten dabah)

ECONOMIC BOTANY [VOL

Cotta, J. N., K. A. Kainer, L. H. O. Wadt, andC. L. Staudhammer. 2008. Shifting cultiva-tion effects on Brazil nut (Bertholletia excelsa)regeneration. Forest Ecology and Management256(1–2):28–35.

de Souza, M. L. 1984. Estudos de processostecnológicos para obtenção de produtos daCastanha-do-Brasil (Bertholletia excelsa, H. B.K.). Ph.D. thesis, Dept. Tecnologia deAlimentos, Universidade Federal do Ceará,Fortaleza.

Desmoulière, S. J. M. n.d. -a. Castanha daAmazônia: Inventario padronizado. Unpub-lished GIS product (map) summarizing specialdata on Bertholletia from RADAM-Brasil(1973–1981) botanical inventories. GIS Labo-ratory, Instituto Nacional de Pesquisa daAmazônia (INPA), Manaus.

———. n.d. -b. Castanha da Amazônia: Notexto descritivo da parcela. Unpublished GISproduct (map) summarizing special data onBertholletia from RADAM-Brasil (1973–1981)botanical inventories. GIS Laboratory, Insti-tuto Nacional de Pesquisa da Amazônia(INPA), Manaus.

Dixon, R. M. W. 1999. Introduction. Pages 1–22 in R. M. W. Dixon and A. Y. Aikhenvald,eds., The Amazonian languages. CambridgeUniversity Press, Cambridge.

Ducke, A. 1946. Plantas de cultura précolombi-ana na Amazônia brasileira. Boletin Técnicodo Instituto Agronomico do Norte 8:2–24.

Facundes, S. D. S. 2002. Historical linguisticsand its contribution to improving the knowl-edge of Arawak. Pages 74–96 in J. D. Hill andF. Santos-Granero, eds., Comparative Arawa-kan histories: Rethinking language family andculture area in Amazonia. University of IllinoisPress, Urbana.

Fonseca, J. S. da. 1880–81. Viagem ao redor doBrasil, 1875–1878 (2 vols. ). Oficina tipográf-ica de Pinheiro, Faria & Companhia, Rio deJaneiro.

Greenhall, A. M. 1965. Sapucaia nut dispersal bygreater spear-nosed bats in Trinidad. Carib-bean Journal of Science 6:167–171.

Gribel, R., M. R. Lemes, L. G. Bernardes, A. E.Pinto, and G. H., Jr. Shepard. 2007. Page 281Phylogeography of Brazil-nut tree (Bertholletiaexcelsa, Lecythidaceae): Evidence of humaninfluence on the species distribution. Associa-tion for Tropical Biology and Conservation,Morelia.

Guix, J. C. 2005. Evidence of old anthropiceffects in forests at the confluence of theCaurés and Negro Rivers–NW Amazonia: Therole of Indians and Caboclos. Grupo deEstudos Ecologicos, São Paulo.

Hamilton, M. B., J. M. Braverman, and D. F.Soria-Hernanz. 2003. Patterns and relativerates of nucleotide and insertion/deletionevolution at six chloroplast intergenic regionsin new world species of the Lecythidaceae.Molecular Biology and Evolution 20(10):1710–1721.

Heckenberger, M. J. 2002. Rethinking theArawakan diaspora: Hierarchy, regionalityand the Amazonian Formative. Pages 99–122in J. D. Hill and F. Santos-Granero, eds.,Comparative Arawakan histories: Rethinkinglanguage family and culture area in Amazonia.University of Illinois Press, Urbana.

———. 2005. The ecology of power: Person,place, and culture in Southern Amazon, AD1250–2000. New York: Routeledge.

———, E. G. Neves, and J. B. Petersen. 1998.De onde surgem os modelos? As origens eexpansões Tupi na Amazônia Central. Revistade Antropologia 41(1):69–96.

———, J. C. Russell, C. Fausto, J. R. Toney, M.J. Schmidt, E. Pereira, B. Franchetto, and A.Kuikuro. 2008. Pre-Columbian urbanism,anthropogenic landscapes, and the future ofthe Amazon. Science 29(321):1214–1217.

Hill, J. D. and F. Santos-Granero. 2002. Com-parative Arawakan histories: Rethinking lan-guage family and culture area in Amazonia.University of Illinois Press, Urbana.

Hornborg, A. 2005. Ethnogenesis, regional inte-gration and ecology in prehistoric Amazonia:Toward a system perspective. Current Anthro-pology 46(4):589–620.

IBGE—Instituto Brasileiro de Geografia e Esta-tística. 2004. Produção da extração vegetal eda silvicultura–2003. Instituto Brasileiro deGeografia e Estatística, Rio de Janeiro.

———. 2007. Produção da extração vegetal e dasilvicultura–2006. Rio de Janeiro: InstitutoBrasileiro de Geografia e Estatística.

Janzen, D. H. and P. S. Martin. 1982. Neo-tropical anachronisms: The fruits the gom-photheres ate. Science 215:19–27.

Kanashiro, M., S. A. Harris, and A. Simons.1997. RAPD diversity in Brazil nut (Bertholle-tia excelsa Humb. & Bonpl.: Lecythidaceae).Silvae Genetica 46(4):219–223.

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

Kern, D., G. D’aquino, T. Rodrigues, F. Frazao,W. Sombroek, T. Myers, and E. Neves. 2004.Distribution of Amazonian dark earths in theBrazilian Amazon. Pages 51–75 in J. Leh-mann, D. Kern, B. Glaser, and W. Wodos,eds., Amazonian dark earths. Kluwer Aca-demic Publishers, Dordrecht.

Lathrap, D. W. 1970. The Upper Amazon.Praeger, New York.

Magalhães, M. P. 2009. Evolução antropomorfada Amazônia. Revista de História da Arte eArqueologia 12:4–38.

McCann, J. M., W. I. Woods, and D. W. Meyer.2001. Organic matter and anthrosols inAmazonia: Interpreting the Amerindian leg-acy. in R. M. Rees, B. C. Ball, C. D.Campbell, and C. A. Watson, eds., Sustain-able management of soil organic matter, pp180–189. CAB International, Wallingford.

Meira, S. 2006. Cariban languages. Pages 199–204 in K. Brown, ed., Encyclopedia oflanguage and linguistics, 2nd edition. Elsevier,Oxford.

——— and B. Franchetto. 2005. The SouthernCariban languages and the Cariban family.International Journal of American Linguistics71(2):127–192.

Métraux, A. 1928. La civilisation matérielle destribus Tupi-Guarani. Librairie Orientaliste, Paris.

Miller, E. (ed.). 1992. Arqueologia dos empreen-dimentos da Eletronorte: Resultados prelimi-nares. Brasilia: Eletronorte.

Moore, D. and L. Storto. 2002. As línguasindígenas e a pré-história. Pages 73–92 in S.D. J. Pena, ed., Homo Brasilis. FUNPECEditora, São Paulo.

Morcote Ríos, G., S. Mora Camargo, and C. E.Franky Calvo. 2006. Pueblos y paisajes anti-guos de la Selva Amazónica. UniversidadNacional deColombia/Instituto de CienciasNaturales, São Paulo.

Mori, S. A. (lead author) et al.; D. Swarthout(topic editor). 2007. Brazil nut family (Lecy-thidaceae) in the New World. In Encyclopediaof Earth, ed. C. J. Cleveland and D. Swarth-out. Environmental Information Coalition,National Council for Science and the Environ-ment, Washington, D.C.

——— and G. T. Prance. 1981. The “sapucaia”group of Lecythis (Lecythidaceae). Brittonia 33(1):70–80.

——— and ———. 1990a. Taxonomy, ecologyand economic botany of the Brazil nut

(Bertholletia excelsa Humb. & Bonpl.: Lecy-thidaceae). Advances in Economic Botany8:130–150.

———. 1990b. Lecythidaceae—Part II. Thezygomorphic-flowered New World genera(Couroupita, Corythophora, Bertholletia, Coura-tari, Eschweilera, & Lecythis). Flora Neotropica21(2):1–376.

Müller, C. H. 1981. Castanha-do-Brasil: Estudosagronômicos. EMBRAPA/Centro de PesquisaAgropecuária do Trópico Úmido, Belém, Pará.

———, I. A. Rodrigues, A. A. Müller, and N. R.M. Müller. 1980. Castanha-do-Brasil: Resul-tados de pesquisas. Belém, Pará: EMBRAPA/Centro de Pesquisa Agropecuária do TrópicoÚmido.

Myers, G. P., A. C. Newton, and M. Melgarejo.1996. The influence of canopy gap size onnatural regeneration of Brazil nut (Bertholletiaexcelsa) in Bolivia. Forest Ecology and Man-agement 127(1–3):119–128.

Neves, E. G. and J. B. Petersen. 2006. The politicaleconomy of pre-Columbian Amerindians: Land-scape transformations in central Amazonia. Pages279–310 inW. Balée and C. Erickson, eds., Timeand complexity in the Neotropical lowlands:Explorations in historical Eecology. ColumbiaUniversity Press, New York.

———, ———, R. N. Bartone, and C. A. daSilva. 2003. Historical and socio-culturalorigins of Amazonian dark earths. In Amazo-nian dark earths: Origins, properties, manage-ment, ed. J. Lehmann, D. Kern, G. Glaser,and W. I. Woods, 29–50. Dordrecht: KluwerAcademic Publishers.

Nimuendajú, C. 1925. As tribus do alto Madeira.Journal de la Société des Américanistes deParis 17:166–172.

Noble, G. K. 1965. Proto-Arawakan and itsdescendants. International Journal of Ameri-can Linguistics 31(3):1–129.

Ohl-Schacherer, J., G. H., Jr. Shepard, H.Kaplan, C. A. Peres, T. Levi, and D. W. Yu.2007. The sustainability of subsistence hunt-ing by Matsigenka native communities inManu National Park, Peru. ConservationBiology 21(5):1174–1185.

Oliver, J. R. 1989. The archaeological, linguisticand ethnohistorical evidence for the expansionof Arawakan into northwest Venezuela andnortheastern Colombia. Ph.D. thesis, Dept. ofAnthropology, University of Illinois, Urbana-Champaign.

ECONOMIC BOTANY [VOL

Pärssinen, M., D. Schaan, and A. Ranzi. 2009.Pre-Columbian geometric earthworks in theupper Purús: A complex society in westernAmazonia. Antiquity 83(322):1084–1095.

Payne, D. L. 1991. A classification of Maipuran(Arawakan) languages based on shared lexicalretentions. Pages 355–499 in D. C. Derbyshireand G. K. Pullum, eds., Handbook of Amazo-nian languages. Mouton de Gruyter, The Hague.

Peres, C. A. and C. Baider. 1997. Seed dispersal,spatial distribution and population structure ofBrazilnut trees (Bertholettia excelsa) in south-eastern Amazonia. Journal of Tropical Ecology13:595–616.

———, L. C. Schiesari, and C. L. Dias-Leme.1997. Vertebrate predation of Brazil-nuts, anagouti-dispersed Amazonian seed crop (Ber-tholletia excelsa, Lecythidaceae): A test of theescape hypothesis. Journal of Tropical Ecology13:69–79.

———, C. Baider, P. A. Zuidema, L. H. O.Wadt, K. A. Kainer, D. A. P. Gomes-Silva, R.P. Salomão, L. L. Simões, E. R. N. Franciosi,F. Cornejo Valverde, R. Gribel, G. H.Shepard Jr., M. Kanashiro, P. Coventry, D.W. Yu, A. R. Watkinson, and R. P. Freckle-ton. 2003. Demographic threats to the sus-tainability of Brazil nut exploitation. Science302:2112–2114.

Piperno, D. and D. M. Pearsall. 1998. Theorigins of agriculture in the lowland Neo-tropics. Academic Press, San Diego.

Pires, J. M. 1984. The Amazonian forest. Pages581–602 in H. Sioli, ed., The Amazon:Limnology and landscape ecology of a mightytropical river and its basin. Dr. W. JunkPublishers, Boston.

Posey, D. A. 1985. Indigenous management oftropical forest ecosystems: The case of theKayapo Indians of the Brazilian Amazon.Agroforestry Systems 3:139–158.

RADAM-Brasil. (1973–1981). Levantamento derecursos naturais: Geologia, geomorfologia,solos, vegetação, uso potencial da terra (22volumes). Rio de Janeiro: Ministério de Minase Energia, Departamento Nacional de ProduçãoMineral.

Ramirez, H. 2001. Línguas Arawak da Amazôniasetentrional: Comparação e descrição. Univer-sidade do Amazonas, Manaus.

Ranzi, A., R. Feres, and F. Brown. 2007. Internetsoftware programs aid in search for Amazoniangeoglyphs. Eos 88(21):226–229.

Rodrigues, A. D. 1964. A classificação do troncolingüístico tupí. Revista de Antropologia 12(1–2):99–104.

———. 1985. Evidence for Tupi-Cariban rela-tionship. In South American languages: Retro-spect and prospect, ed. H. Klein and L. Stark,371–404. Austin: University of Texas Press.

———. 1999. Tupí. In The Amazonian lan-guages, ed. H. Klein and L. Stark, 107–124.Cambridge, U.K.: Cambridge University Press.

Roosevelt, A. C. 1980. Parmana: Prehistoric maizeand manioc subsistence along the Amazon andOrinoco. Academic Press, New York.

———. 1992. Arqueologia Amazônica. In His-tória dos Índios no Brasil, ed. M. M. Carneiroda Cunha, 53–86. São Paulo: Companhia dasLetras.

———, M. Lima da Costa, C. Lopes Machado,M. Michab, N. Mercier, H. Valladas, J.Feathers, W. Barnett, M. Imazio da Silveira,A. Henderson, J. Sliva, B. Chernoff, D. S.Reese, J. A. Homan, N. Toth, and K. Schick.1996. Paleoindian cave dwellers in the Ama-zon: The peopling of the Americas. Science272:373–384.

Schmidt, M. 1917. Die Aruaken: Ein beitrag zumproblem der kulturverbreitung. UniversitätLeipzig, Leipzig.

Shepard, G. H. Jr. 2001. Relatório preliminarsobre sítio arqueológico e cemitério indígenana Reserva Amanã. Report submitted to 1ªSuperintendência Regional do IPHAN (Memonº 0308/2001). Instituto de Patrimônio His-tórico e Artístico Nacional (IPHAN), Brasilia,DF, Brazil.

———. 2002. Estudos sobre a biologia reprodu-tiva de plantas Amazônicas e suas aplicações nouso, manejo e conservação dos recursos natur-ais. Manaus, AM, Brazil: Conselho Nacionalde Pesquisa (CNPq).

———, K. Rummenhoeller, J. Ohl, and D. W.Yu. 2010. Trouble in paradise: Indigenouspopulations, anthropological policies, and bio-diversity conservation in Manu National Park,Peru. Journal of Sustainable Forestry 29:252–301.

Steinen, K. von den. 1894. Unter den natur-völkem Zentral-Brasiliens. Berlin.

Stokstad, E. 2003. Too much crunching onrainforest nuts? Science 302:2049.

Tastevin, C. 1909/1920. Vocabulaires Katawishi.Unpublished vocabulary. Archives of ProfessorPaul Rivet, Musée de l’Homme, Paris.

SHEPARD & RAMIREZ: HUMAN DISPERSAL OF THE BRAZIL NUT2011]

———. 1920. Dialecte Marawa. Unpublishedvocabulary. Archives of Professor Paul Rivet,Musée de l’Homme, Paris.

Trivedi,M. R., F. H. Cornejo, and A. R.Watkinson.2004. Seed predation on Brazil nuts (Bertholletiaexcelsa) by macaws (Psittacidae) in Madre deDios, Peru. Biotropica 36(1):118–122.

Tupiassú, A. and N. V. C. Oliveira. 1967. Acastanha do Pará. Instituto de DesenvolvimentoEconômico-Social do Pará, Belém, Pará.

Urban, G. 1992. A história da cultura Brasileirasegundo as linguas nativas. Pages 87–102 inM. M. Carneiro da Cunha, ed., Índios noBrasil. Companhia das Letras, São Paulo.

Villachica, H., J. E. Urano de Carvalho, C. H.Müller, S. C. Diaz, and M. Almanza. 1996.Frutales y hortalizas promisórios de la Amazonía.Tratado de Cooperación Amazónica/Food andAgriculture Organization (FAO), Lima, Peru.

Wander, A. E., R. Gazzola, C. H. M. Coelho, J.Gazzola, and G. da Silva e Souza. 2008. Brazilnut almonds: Nutritional and market aspects.Tropentag 2008. Stuttgart-Hohenheim, Ger-many: Competition for Resources in a ChangingWorld: New Drive for Rural Development.

Witkowski, S. L. and C. H. Brown. 1983.Marking-reversals and cultural importance.Language 59(3):569–582.

ECONOMIC BOTANY [VOL