Terra preta soils and their archaeological context in the caqueta basin of southeast colombia

Society for American Archaeology

Terra Preta Soils and Their Archaeological Context in the Caqueta Basin of Southeast ColombiaAuthor(s): Michael J. Eden, Warwick Bray, Leonor Herrera, Colin McEwanSource: American Antiquity, Vol. 49, No. 1 (Jan., 1984), pp. 125-140Published by: Society for American ArchaeologyStable URL: http://www.jstor.org/stable/280517 .Accessed: 06/06/2011 22:13

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .http://www.jstor.org/action/showPublisher?publisherCode=sam. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

Society for American Archaeology is collaborating with JSTOR to digitize, preserve and extend access toAmerican Antiquity.

http://www.jstor.org

TERRA4 PRETA SOILS AND THEIR ARCHAEOLOGICAL CONTEXT IN THE CAQUETA BASIN OF SOUTHEAST COLOMBIA

Michael J. Eden, Warwick Bray, Leonor Herrera, and Colin McEwan

Investigation of dark earths from two archaeological sites near Araracuara, on the Rfo Caqueta in Colombian Amazonas, indicates that these soils are anthropic, with characteristics similar to those of the terra preta soils of the Brazilian Amazon. The Araracuara sites belong to the Camani (plainware) phase of the second to ninth centuries A.D. and to the Nofurei phase (ninth to seventeenth centuries) with pottery of the Polychrome Tradition. The Araracuara examples are the first archaeological terra preta soils analyzed from west of Brazil, although similar materials are reportedfrom Ecuador and Peru. The problems of recognizing and dating the first appear- ance of terra preta soils are discussed, and a plea is made for closer collaboration between archaeologists and soil scientists.

Terra preta soils have been reported from many places along the mainstream of the Amazon and its tributaries (Hilbert 1968; Sombroek 1966). The soils are generally free-draining, dark-colored and relatively fertile in character. They normally contain indigenous artifacts and appear to be coincident with areas of relatively dense Precolumbian settlement. In the past, the origin of these soils has been unclear. The possibility of a natural origin has been considered and the assumption made that their favorable fertility served to attract indigenous settlement (Soares 1963); alternatively, the terra preta has been viewed as an "archaeological soil" and its characteristics assumed to derive from the accumulation of domestic organic waste around Precolumbian settlements (Gourou 1949). Detailed analyses of terra preta sites and materials have recently tended to confirm an anthropic origin; on this basis, Sombroek (1966:158) describes a terra preta from the Santar6m region as "a kind of kitchen-midden, developed at the dwelling sites of Pre-Columbian Indians." Similarly, soil investigations by Ranzani et al. (1970), Vieira et al. (1971) and Smith (1980) lead them to accept an anthropic origin for the terra preta.

The distribution of terra preta sites extends widely across the Brazilian Amazon (Brochado and Lathrap 1982; Falesi 1972; Hilbert 1955, 1968; Palmatary 1960; Sim6es 1967, 1974; Smith 1980; Sternberg 1960). They are recorded from Maraj6 island at the mouth of the Amazon, along down- stream tributaries like the Xingui and Tocantins, in the middle Amazon near Santarem and Manaus, and westwards almost to the Colombian frontier at Tapurucuara on the Rio Negro and at Mangueiras on the Rio Japura. Terra preta soils are also reported from Aripuana in northern Mato Grosso, and along the Transamazon highway in the state of Para.

In the past, terra preta soils have been commonly reported along major rivers, but it is now apparent that they also occur on interfluvial sites. The soils have been observed by one of us (Eden) along the northern sector of the Santar6m-Cuiaba highway, while Smith (1980) reports other in- terfluvial occurrences in eastern Amazonia. According to Smith, riverine terra preta soils are deeper and more extensive than those on interfluves. He recognizes a sharply bimodal distribution in site size. The average area of his riverine sites is 21.2 ha, with the largest extending up to 4 km along the riverbank and covering as much as 90 ha. In contrast, the terra preta sites of interfluvial areas consist of circular or lenticular deposits with an average extent of only 1.4 ha, with no site larger than 6 ha. Smith (1980) considers riverine sites to have developed in association with relatively

Michael J. Eden, Department of Geography, Bedford College, London University, London NW] 4NS, England Warwick Bray, Institute of Archaeology, London University, London WCIH OPY, England Leonor Herrera, Instituto Colombiano de Antropologfa, Apartado Nacional 407, Bogotd, Colombia Colin McEwan, Department of Anthropology, University of Illinois, Urbana IL 61801, U.S.A.

American Antiquity, 49(1), 1984, pp. 125-140. Copyright Oc 1984 by the Society for American Archaeology

125

126 AMERICAN ANTIQUITY [Vol. 49, No. 1, 1984]

extensive indigenous settlements, while the smaller interfluvial sites are probably the remains of individual malocas (large multifamily dwellings).

During 1977, scientists of the Colombian Amazonas Expedition undertook fieldwork in the central Caqueta basin of eastern Colombia. Archaeological investigations resulted in the identification of numerous sites of Precolumbian settlement (Bray et al. 1977; Herrera 1981; Herrera et al. 1982). In several places, where detailed excavations were undertaken, terra preta soils were encountered which displayed similarity to those described from the Brazilian Amazon. Such soils, which do not appear to have been previously analyzed from the Colombian Amazon, are also considered to owe their distinctive features to anthropic influences. In the present paper, earlier Brazilian data on terra preta soils and their archaeological context are briefly reviewed. Information from the Colombian sites is then presented and compared, and its broader implications are discussed.

TERRA PRETA SOILS IN THE BRAZILIAN AMAZON

Detailed information on terra preta soils in the Brazilian Amazon is given by Sombroek (1966), Ranzani et al. (1970), Vieira et al. (1971), Falesi (1972) and Smith (1980). Sites are reported to be free-draining in character and often located where higher land comes close to, and has good visibility of, navigable waterways. Smaller sites have also been encountered on interfluvial areas. Terra preta soils are most commonly present on Pleistocene terrace levels, but also occur on the older planalto surface of Plio-Pleistocene age. On the latter surface, at Santar6m, fine-textured terra preta soils are juxtaposed with kaolinitic yellow latosols over a parent material of common age and composition (Sombroek 1966).

Terra preta soils in eastern Amazonia are principally distinctive with respect to their color, which remains relatively dark into lower horizons. Colors range from black in the topsoil to dark brown or strong brown at 70-100 cm in depth. This contrasts with the color of adjacent kaolinitic yellow latosols which are paler at depth (Ranzani et al. 1970; Sombroek 1966). According to Sombroek (1966), the dark coloration of terra preta horizons appears to be associated with only moderately high levels of organic carbon; he suggests that soil darkening may result from a complex formation of organic matter and calcium, forming a coating on the soil particles.

In addition, terra preta soils are commonly less acid than adjacent free-draining soils. A wide range of pH values is encountered, but according to Smith (1980:562), "the pH of sampled black earth sites averaged 5.4 . . ., whereas the pH of oxisols and ultisols in Amazonia is usually under 5." The reduced acidity reflects a relatively high concentration of exchangeable calcium, while exchangeable magnesium may also be slightly increased. In addition, available phosphorus is usually relatively high (Sombroek 1966; Vieira et al. 1971).

Archaeologically, the terra preta phenomenon is a generalized trait, and is not exclusive to any single cultural group or ceramic tradition. Smith (1980) raises the possibility that the deepest levels of some terra preta soils may go back to the preceramic period, though this is frankly speculative. In excavated terra preta sites, the quantity of archaeological material tends to diminish with depth, and the lack of pottery at the base of a few terra preta soils need not therefore indicate great age. Present evidence does not allow us to state unequivocally whether or not the earliest sites of the

ceramic period in Amazonia are of terra preta type. The oldest radiocarbon date for a terra preta is 2400 ? 75 years: 450 B.C. (GrN-4328) from Paredao near Manaus (Hilbert 1968), although on archaeological grounds this figure is generally regarded as too early. Much further west, the Hupa- iya site on the Rio Ucayali in Peru is a terra preta (Lathrap, personal communication) and was occupied between ca. 200 B.C. and A.D. 100 (Lathrap 1970:117-120). By the start of the Christian era the picture becomes clearer, and dated sites are more numerous. The Bom Sucesso site, on Careiro Island, has a date of 2050 ? 120 years: 100 B.C. (L- 198A) on caraipe temper from potsherds found in a terra preta (Carneiro 1974; Sternberg 1960). At the Poc6 site, on a tributary of the Rio Nhamunda, the earliest component (the Poc6 phase) has dates of 2015 ? 95 years: 65 B.C. (SI- 2774), 1840 ? 90 years: A.D. 110 (SI-2776) and 1745 ? 115 years: A.D. 205 (PN-8028) (Hilbert and Hilbert 1979). These figures correspond reasonably well with the ones from Itacoatiara of 1864

REPORTS 127

78o'W 720 w

-4PN

10

20N~~~~~~~~~~~~~~~~~~~2N

IJ Landover m A Z I L

000 > < \a Pedrera l~~~~~~~~~~~~~~~~~~~~~~~~0

EC U A D O R./ t % 9<_ >

{ ,' O~ 100 200 b- 78?,W i ' 9, . 74eW 720W 4 7cera

Figure 1. Southeastern Colombia and adjacent areas.

? 58 years: A.D. 86 (P-372) and from Manacapuru with 1525 ? 58 years: A.D. 425 (P-406) from its upper levels (Hilbert 1968).

At the other end of the chronological range, European trade beads have been found in some terra preta soils of the middle Amazon, and sites with pottery of the Santar&m style can be linked with the historical Tapaj6 Indians (Palmatary 1960).

Taken at face value, this evidence suggests that the first unambiguous terra preta sites (whatever they may mean in terms of land use) do not make their appearance in Amazonia much before the start of the Christian era. However, this may be a false impression, based on inadequate sampling. Few Amazonian sites are reliably dated; archaeologists rarely give competent descriptions of soil profiles and, because of the lack of analytical data, there is as yet no agreement about what constitutes a "minimal" terra preta in terms of its blackness or of its physical and chemical properties.

The sort of problem which arises is exemplified by the controversy over the age of the B6a Vista site, one of the more plausible candidates for early terra preta status (Hilbert and Hilbert 1980). B6a Vista is a settlement of the Poc6 phase and has produced two early radiocarbon dates: 3280 ? 45 years: 1330 B.C. (PN-7542) and 2950 ? 130 years: 1000 B.C. (PN-7453), which are consistent with each other but irreconcilable with the three dates from Poc6 itself. Brochado and Lathrap (1982:25), who see Saladoid affinities in the Poc6 style, are inclined to accept the earlier dates, while the excavators emphasize links with the Taquara and Barrancoid traditions and prefer the later series of dates (Hilbert and Hilbert 1979, 1980). Peter Paul Hilbert informs us (personal commu- nication) that the sherds from B6a Vista were badly eroded and difficult to classify, and that the radiocarbon samples came from the very bottom of the cuts "already out of the Terra Preta in the proper sense, but mixed up with yellowish Kaolinitic latosols."

And there, for the moment, the matter must rest until more sites are excavated and more soil

128 AMERICAN ANTIQUITY [Vol. 49, No. I I 984]

9 ' , - fr4i. ** ' * ' *

.

. . . . , , .

, 1 Eq Vr4 . 1

' d *d - -. -

Figure 2 View eastwards along the Caqueta river from the vicinity of Araracuara.

analyses are made available. All one can say is that the Amazonian terra preta soils have a history

that spans a minimum of 2,000 years, and may eventually be pushed back much further.

TERRA PRETA SOILS IN THE COLOMBIAN AMAZON

Field investigations were undertaken in the vicinity of Araracuara in the central Caquet basin

(Figure 1). The area has an annual rainfall of approximately 3,000 mm, and generally supports a

REPORTS 129

. . **. t

. **,.

. . . - -

. - : P%*:e.?W. *.; *-. ,' V 4,itp v;r .AYi * SbM -

- .,

2: ' - -K- , -.

- I

. : . - -

. - '.

;1. - -

1;--

. .

,

- - r

Figure 3. Eastern margin of the sandstone massif at Araracuara showing contemporary portage route and

Site 3/4.

climax vegetation of rainforest. To the east of Araracuara, the landscape principally comprises a

dissected alluvial plain at approximately 250-280 m above sea level (Figure 2). The surface is

considered to correspond to the Plio-Pleistocene planalto ofthe Brazilian Amazon (Eden et al. 1982,

1 983). flanking the Caquett river, Pleistocene depositional terraces and a contemporary floodplain

have formed within the older landscape. To the west of Araracuara, an upfaulted sandstone massif

of Cretaceous or earlier age rises abruptly above the general level of the landscape (Gansser 1974).

The massif extends approximately 20 km to the west, and its local summit area is some 380 m

above sea level. Rapids obstruct the incised channel of the Caqueti in this zone, and a portage

route, apparently dating from Precolumbian times, traverses the massif (Figure 3).

At present only 500 to 700 people, mainly Witoto and Andoke Indians and some colonos, live

within a 25 km radius of Araracuara, but a higher Precolumbian population could well have been

supported. Araracuara itself is particularly favorable for settlement. It is situated on a white water

river at the downstream edge of a series of rapids and is simultaneously a good fishing station and

the terminal of a portage route. The existence downstream of patches of floodplain soil, which

receive regular increments of nutrient-bearing sediment (Eden et al. 1 982), reproduces on a smaller

scale the favorable conditions which are considered to have supported dense Precolumbian settle-

ment along the mainstream of the Amazon (Denevan 1 970; Lathrap 1 970; Meggers 1 97 1 ; Myers

I 9 73; Smith I 980).

Evidence ofPrecolumbian settlement is encountered at many places adjacent to the Caquet river

and at low water the beaches around Araracuara are covered with sherds. In all, 22 archaeological

sites were examined. In most cases (excluding middens lying directly on the beaches) the soil materials

were characteristically dark-colored and contrasted with other soils in the area. As would be expected,

all forest soils around Araracuara displayed a darkened mineral topsoil, but only in the archaeological

sites did this darkening persist into the lower horizons. Laboratory analyses also indicate the local

distinctiveness of these soils and their resemblance to the terra preta soils reported from Brazil.

130 AMERICAN ANTIQUITY [Vol. 49, No. 1, 1984]

S S sOm't"' 0~~~~~LaSa

SIte 3/4 aracuara

C A O. LIII~~~~Flood

plain Sandstone massif

1

Kilometres

Figure 4. Sites in the Araracuara region.

Two sites are described, one located near Araracuara on a colluvial parent material derived from the sandstone massif and the other on a low terrace downstream near La Sardina (Figure 4). The site numbers are those of the original field report (Bray et al. 1977).

Analysis of the soils was undertaken using standard procedures (Black et al. 1965). Particle size was obtained using the pipette method. Organic carbon was determined by the Walkley-Black method. Exchangeable metallic cations were extracted with neutral ammonium acetate, and mea- sured by atomic absorption spectrophotometer or flame photometer. Exchangeable aluminum was extracted with KCI and determined by titration with NaOH. Available phosphorus was determined using the Bray method. The pH was measured in a 1:2'/2 suspension in H20.

Site 3/4: Araracuara

Adjacent to Araracuara the relatively flat summit area of the sandstone massif is covered by forest or shrub vegetation. Rock outcrops are common and are interspersed with areas of poorly drained acid soils, consisting of bleached quartz sand overlain by a thick organic horizon. Drainage conditions improve on flanking slopes where sandy colluvial materials give rise to less acidic, brown sandy soils that normally carry rainforest. The colluvial soils have been classified as Inceptisols (J. Morelo, personal communication).

Approximately 1 km northwest of Araracuara, overlooking the river and adjacent to the contem- porary portage route which crosses the sandstone massif, an area of terra preta was encountered (Figure 3). The soil, which is located on the upper flanking slope of the massif, is developed in colluvial materials similar to those of adjacent Inceptisols. The soil is underlain by weathered sandstone at a depth of 100-150 cm.

The investigated area of terra preta is partly covered by mature secondary forest and partly by an experimental manioc plantation. Some 40 small test pits and one larger trench, 2 x 1 m, were excavated down to sterile subsoil. These pits were scattered over more than 2 ha of dark soil, but

REPORTS 131

Table 1. Representative Soil Profile for Terra Preta at Araracuara (Site 3/4).

Location: 1.5 km north-west of Araracuara Altitude: 360 m (estimated) Relief: very gently inclined (10) upper hillslope Parent material: sandstone colluvium Drainage: free draining Vegetation: mature secondary forest

0-7 cm Very dark greyish brown (IOYR3/2) fine sandy loam. Structureless. Friable. Fine charcoal fragments and few sherds. Abundant roots. pH 4.3

7-36 cm Dark reddish brown (5YR2/2) fine loamy sand. Weak fine granular structure. Friable. Common visible pores. Few fine charcoal fragments and sherds. Common roots. pH 4.5

36-49 cm Very dark brown (1 OYR2/2) fine loamy sand. Moderate fine subangular blocky structure. Friable. Common visible pores. Few sherds. Common roots. pH 4.5

49-74 cm Dark brown (7.5YR4/4) sandy clay loam, with abundant strong brown (7.5YR5/8) mottling and incipient iron concretions. Moderate fine subangular blocky structure. Firm. Common visible pores. Few sherds. Common roots. pH 4.8

74-106 cm Strong brown (7.5YR5/6) sandy clay loam, with common to abundant, hardened dark red (1OR3/6) iron concretions. Strong fine subangular blocky structure. Firm. Common visible pores. Few roots. pH 4.8

106 cm + Weathered sandstone

Table 2. Analytical Data for Terra Preta and Adjacent Inceptisols at Araracuara (Site 3/4).

Physical Characteristics

Sand Silt Clay (0.05-2.00 mm) (0.002-0.05 mm) (<0.002 mm)

Texture Class (%) (%) (%)

Terra preta a) 0-10 cm Loamy sand 85 8 7

75-80 cm Sandy clay loam 66 10 24

Inceptisols b) 0-10 cm Sandyloam 74 11 15 c) 0-10 cm Sandyloam 78 7 15 d) 0-10 cm Sandy clay loam 54 22 24 e) 0-10 cm Sandy clay loam 46 22 32

Chemical Characteristics

Exchangeable Cations Available Organic (me/l100Og) P Carbon

Ca Mg Na K Al (ppm) (%) pH

Terra preta a) 0-10 cm 0.07 0.04 0.04 0.08 0.71 71 3.8 4.3

75-80 cm 0.07 tr tr tr 1.42 17 1.6 4.8

Inceptisols b) 0-10 cm 0.07 0.04 0.04 0.08 0.93 12 2.0 4.4 c) 0-10 cm 0.07 0.04 0.04 0.08 1.00 13 2.1 4.2 d) 0-10 cm 0.05 0.04 0.17 0.10 1.35 25 4.2 4.5 e) 0-10 cm 0.05 0.04 0.17 0.15 1.57 25 4.4 4.0

132 AMERICAN ANTIQUITY [Vol. 49, No. 1, 1984]

Table 3. Representative Soil Profile for Terra Preta at La Sardina (Site 20).

Location: 400 m inland, right bank of the Caqueta river Altitude: 200 m (estimated) Relief: gently sloping (30) terrace margin Parent material: terrace alluvium, with igneous/metamorphic rock debris Drainage: imperfectly drained Vegetation: mature forest

0-15 cm Very dark grey (10YR3/1) sandy loam. Weak fine granular structure. Very friable. Few fine charcoal fragments and sherds. Abundant roots. pH 4.7

15-90 cm Very dark greyish brown (1 OYR3/2) sandy loam. Weak fine subangular blocky structure. Friable. Few fine charcoal fragments and few sherds. Common to few roots. pH 4.5

90-120 cm Very dark greyish brown (1OYR3/2) sandy clay loam, with common weathered rock fragments. Weak fine subangular blocky structure. Friable. Very fine charcoal fragments and few sherds. Few roots. pH 4.5

120 cm + Weathered rock

subsequent investigations by Andrade and Botero (personal communication) have shown that the site consists of approximately 6 ha of classic black terra preta and 20 ha of brown anthropic soil.

Our test pits outside the area of dark soil produced no archaeological material, but within the black terra preta sherds were present throughout the dark earth, though diminishing in numbers between 40-70 cm in depth. At this site, as in others investigated, the archaeological deposit contained charcoal, chips and spalls of chert, an unusual number of small gravelly pebbles, and many broken fragments of quartz. The chert, and possibly also the quartz fragments, may well be the residue of tool manufacture. Although the stone chips show no clear evidence of human work- manship, such items do not occur to a significant extent in comparable, non-archaeological sites in the area.

A representative soil profile from the site is given in Table 1. In Table 2, analytical data for the profile are provided, together with topsoil analyses from adjacent forest sites where similar colluvial materials are present but terra preta characteristics are lacking.

Site 20. La Sardina

La Sardina is located on the Caqueta river approximately 20 km downstream of Araracuara (Figure 4). In this area, land flanking the river consists mainly of contemporary floodplain or of low depositional terraces of Pleistocene age. Floodplain sediments are mainly silty clay loam to clay in texture, while terrace materials are sandy clay loam to sandy clay, often overlying gravel at depth (Eden et al. 1982). Floodplain and terrace soils in the area have been classified as Inceptisols (J. Morelo, personal communication). Within and adjacent to this stretch of river, there also occur local outcrops of igneous and metamorphic rocks, which are outliers of the Guiana Shield complex.

At La Sardina the terra preta lies in mature forest on the right bank of the Caqueta river. Adjacent to the river is a relatively narrow floodplain zone which gives way approximately 400 m inland to a low terrace level. Weathered igneous/metamorphic rocks outcrop nearby and appear to have contributed to local parent materials. Terra preta soils are present across the floodplain-terrace boundary, and are associated with archaeological materials.

The precise extent of the archaeological site is unknown, although it clearly covers several hectares and is probably within the range of Smith's (1980:563) "riverine" class of sites. Cultural material occurs on the floodplain immediately adjacent to the river, and extends for 500 m inland onto the low terrace. Within the area of the site, intensity of human activity was not necessarily homogeneous. In a pit (Test pit A) on the floodplain, the soil was black and contained sherds throughout, with a marked concentration at 30-50 cm, just above the base of the material. The water table was reached at 55 cm. This zone may perhaps represent a midden or house garden. In Test pit B, on the terrace,

REPORTS 133

Table 4. Analytical Data for Terra Preta and Adjacent Inceptisols at La Sardina (Site 20).

Physical Characteristics

Sand Silt Clay (0.05-2.00 mm) (0.002-0.05 mm) (<0.002 mm)

Texture Class (%) (%) (%)

Terra preta a) 0-10 cm Sandy loam 73 16 11

75-100 cm Sandy loam/ 61 19 20 sandy clay loam

Inceptisols b) 0-10 cm Sandy clay loam 57 14 29 c) 0-10 cm Sandy clay loam 53 19 28 d) 0-10 cm Loam 39 36 25 e) 0-10 cm Loam 44 32 24

Chemical Characteristics

Exchangeable Cations Available Organic (me/ 100 g) P Carbon

Ca Mg Na K Al (ppm) (%) pH

Terra preta a) 0-10 cm 0.20 0.08 tr 0.10 0.71 36 2.2 4.7

75-100 cm 0.10 tr 0.09 0.10 0.57 15 1.2 4.5 Inceptisols

b) 0-10 cm 0.12 0.04 0.04 0.10 1.00 11 2.3 4.0 c) 0-10 cm 0.10 0.04 0.04 0.13 1.07 14 2.9 4.4 d) 0-10 cm 0.12 0.04 0.04 0.10 1.64 24 3.0 4.1 e) 0-10 cm 0.07 0.04 0.09 0.10 1.99 16 3.5 4.0

the soil was less dark in color. Sherds were found from surface to bedrock at 108 cm, with a dense concentration of unusually large fragments, including one complete and several almost intact jars, at 50-70 cm. Locality B may have been the site of a structure. Compared with Site 3/4 at Araracuara, La Sardina is not merely larger but is also much more densely packed with large sherds and archaeological debris. Disturbance is minimal, and the site was evidently a large settlement with fairly continuous occupation.

A terra preta profile from the terrace margin at La Sardina is given in Table 3. In Table 4, analytical data for the profile are provided, together with topsoil analyses from adjacent forest sites where similar terrace materials are present but terra preta characteristics are lacking.

DISCUSSION

The terra preta soils described from the Caqueta basin are developed over parent materials of different origins, namely, sandy colluvium at Araracuara and terrace alluvium at La Sardina. In each site, the terra preta is juxtaposed with Inceptisols which have developed over the same parent material as the terra preta. No consistent drainage variations are evident between the terra preta soils and the Inceptisols examined. From these conditions and from the frequent occurrence of artifacts in the terra preta soils, it is concluded that the latter are anthropic in origin and are the products of previous Indian settlement. In this respect, the soils are similar to those reported from the Brazilian Amazon.

Apart from the presence of artifacts, the most prominent feature of the terra preta in the Caqueta

134 AMERICAN ANTIQUITY [Vol. 49, No. 1, 1984]

Table 5. Soil Colors in the Araracuara Region.

Mineral Topsoil Subsoil (0-10 cm) (70-100 cm)

a) Terra preta, Site 3/4 Very dark greyish brown Dark brown (7.5YR4/4) to Araracuara (1OYR3/2) strong brown (7.5YR5/6)

b) Adjacent Inceptisols Dark brown (10YR3/3 and Brownish yellow (1OYR6/6) to lOYR4/3) yellowish red (5YR5/8)

a) Terra preta, Site 20 Very dark grey (1 OYR3/ 1) Very dark greyish brown La Sardina (10YR3/2)

b) Adjacent Inceptisols Very dark greyish brown Yellowish red (10YR5/6) to light (10YR3/2) to dark yellowish yellowish brown (10YR6/4) brown (1 OYR4/4)

basin is its coloration, which is usually dark grey to dark brown. Other soil types in the vicinity of Araracuara commonly display a dark mineral topsoil, but the terra preta is distinctive in that its dark coloration persists into the subsoil. Adjacent Inceptisols commonly display yellowish to reddish subsoils (Table 5).

No general contrast exists between terra preta and other soils around Araracuara with respect to the organic carbon content of the mineral topsoil. All values cited fall within the normal topsoil range for the area of 2.0-4.5%. Terra preta subsoils also display relatively high organic carbon content, and in this respect they differ from adjacent soil types. The terra preta sites give subsoil values of 1.6% and 1.2% organic carbon respectively (Tables 2 and 4), while other subsoils analyzed in the area average 0.4%, with a range of 0.2-0.6% (n = 8).

Terra preta soils around Araracuara are also distinctive with respect to available phosphorus. This element, which is present in small quantities in most Amazonian soils (PRORADAM 1979; Sanchez et al. 1982), displays higher values in the terra preta sites. In the latter, topsoil values of 71 ppm and 36 ppm are recorded, which compare with an average of 17.5 ppm (range 11-25 ppm) for adjacent Inceptisols (Tables 2 and 4). Improved phosphorus status accords with the conditions described in the Brazilian Amazon (Smith 1980; Sombroek 1966) and is in general to be expected on old settlement sites (Eidt 1977; Griffith 1980).

In most other respects, the terra preta soils were not found to be distinctive. Present data suggest that the relatively high levels of exchangeable aluminum, which are characteristic of most Amazonian soils (PRORADAM 1979; Sanchez et al. 1982), are somewhat reduced in terra preta soils, but in general the latter still display the high acidity (< pH 5.0) and very low base content which characterize most soils in the region.

The Archaeological Context

Changes in pottery style have allowed two successive periods of occupation to be defined for the Araracuara region.

The older phase (named Camani) is characterized by small and medium-sized vessels, consisting mainly of simple bowls and jars, predominantly with caraipe (Carneiro 1974) and charcoal temper, and with no decoration apart from brownish or reddish slip (Figure 5). Fire dogs and pottery griddles indicate the processing of bitter manioc, and carbonized fruits of the milpeso palm (Jessenia polv- carpa Karst) were also found in Camani deposits. Both these foods are still important today, and the subsistence pattern in the Camani period was probably not very different from that recorded in more recent times.

The time span of the Camani phase is fixed by six radiocarbon dates from three different sites: 1815 ? 105 years: A.D. 135 (Beta- 1503); 1800 ? 85 years: A.D. 150 (IAN- 113); 1690 ? 55 years: A.D. 260 (Beta-1504); 1480 + 95 years: A.D. 470 (Beta-1509); 1420 ? 70 years: A.D. 530 (Beta- 1505) and 1120 ? 65 years: A.D. 830 (Beta- 1508). Camani pottery has no exact parallel elsewhere,

REPORTS 135

( ) )S

0 5 10 _ I I cms

Figure 5. Camani pottery from Araracuara.

but its closest links seem to be with the plainware component of the Japura style from Mangueiras and other sites some 370 km downstream, on the Brazilian side of the frontier (cf. Hilbert 1968: Abb.83, 4-6).

The more recent phase at Araracuara (named Nofurei) has a much wider variety of forms, ranging from small cups and bowls to large jars that may have been used for fermented drinks. The full range of shapes, including the plainwares, is illustrated in Herrera et al. (1982) and is not repeated here. A good deal of Nofurei pottery has elaborate decoration carried out in several different techniques: incision, excision, grooving or channeling, impression, zoned punctate, bands of red paint, and complex designs in which red-painted zones are outlined by shallow grooves (Figure 6). Some of these red-zoned wares also have a fugitive white paint on the background areas, and are therefore true polychromes. All Nofurei sites produced abundant fragments of manioc griddles indistinguishable from those of the previous phase, and generally with leaf-imprints on the underside.

In excavations on the bank of the river at Araracuara (Site 15) Camani plainwares were strati- graphically overlain by Nofurei sherds, and a radiocarbon date of 1145 ? 80 years: A.D. 805 (Beta- 1507) marked the replacement of one style by the other. Two further dates carry the Nofurei style into the historical period: 705 + 60 years: A.D. 1245 (Beta-1506) and 350 + 50 years: A.D. 1610 (Beta- 1510). Nofurei pottery is clearly a regional variant within the Polychrome Tradition already known from the Rio Napo in Ecuador (Evans and Meggers 1968), around Leticia in the Colombian Trapecio (Bolian 1972, 1975), and throughout the Amazon basin (Hilbert 1968; Lathrap 1970).

136 AMERICAN ANTIQUITY [Vol. 49, No. I I 984]

.4

e

. ..

p

[5Cm

Figure 6. Nofurei pottery from sites in the Araracuara region.

REPORTS 137

At Araracuara and at La Sardina, the bulk of the pottery in the terra preta soils belongs to the Nofurei style, but at both sites a little Camani pottery was present in the lowest level. These terra preta soils, therefore, must have begun to form no later than A.D. 800, and probably go back centuries before that date. This figure is consistent with the single date of 1318 ? 59 years: A.D. 632 (P-588) for one of the nearest recorded terra preta sites to Araracuara, at Mangueiras on the Japura (the Brazilian portion of the Rio Caqueta), where the pottery seems transitional between the modelled-incised Barrancoid tradition and the true Polychrome Tradition to which Nofurei belongs (Hilbert 1968; Lathrap 1970:121).

Araracuara and La Sardina are at present the only terra preta sites outside Brazil for which soil analyses are available. Similar deposits may be expected elsewhere in the Amazon region, and this impression is confirmed from a search of the archaeological literature and by the comments of those with field experience in the area.

At La Pedrera, on the Caqueta close to the Brazilian frontier, the upper 40 cm of the archaeological stratum was a "tierra negrusca" which the excavator also identified as a terra preta (von Hildebrand 1976:151). The pottery from this locality has clear Barrancoid affinities, as has the Hupa-iya style from a reported terra preta site on the Rio Ucayali (Lathrap 1970:117-120, and personal com- munication). In the Casiquiare-Rio Negro region of Venezuelan Amazonas, Sternberg reports large terra preta sites with as yet undescribed pottery (in Klinge et al. 1977).

Elsewhere, outside Brazil, possible terra preta soils are reported mainly from sites of the Poly- chrome Tradition. At Nuevo Rocafuerte, on the Rio Napo in Ecuador, the deposit is described as a "black sandy loam" (Evans and Meggers 1968:34). In Peru, Lathrap (personal communication) informs us that all the Caimito occupation that covers the shore of Imariacocha consists of terra preta, though these soils are neither as extensive nor as deep as those of the Central Amazon. Caimito pottery has two radiocarbon dates in the fourteenth century A.D., and, as Weber (1975) notes, is closely related to the Ecuadorian Napo wares and to Bolian's Zebu Complex from the Finca Rivera in the Colombian Trapecio. The radiocarbon dates for Finca Rivera range from 920 ? 90 years: A.D. 1030 (1-5776) to 435 ? 90 years: A.D. 1515 (1-5773), showing a good temporal overlap with Nofurei (Bolian 1975:47). The final link is provided by a trade sherd of Zebu negative painted ware (or something very close to it) from our excavations at La Sardina.

Although unsupported by soil analyses, these reports suggest that the terra preta phenomenon may be more widespread than is generally recognized. Indeed the scope of enquiry will eventually have to extend to sites outside the Amazon basin in order to encompass, for example, the midden of "tierra negra" at the Saladero site on the Orinoco (Cruxent and Rouse 1958:214-215), or the "sites of ancient villages marked by a deep black soil mixed with broken pottery" from the forests of the River Corentyne in Guyana (Brown 1876:339). In a rather different environment, what appear to be terra preta soils have been observed in the Llanos Orientales of Colombia (Pedro Botero, personal communication). For these areas there are no analytical data whatsoever and comparisons are for the moment impossible.

CONCLUSIONS

Soils displaying distinctive terra preta characteristics have been identified on contrasting riparian sites in the middle Caqueta basin of Colombia, and are considered to represent the location of previous Indian settlement. The darker color and higher organic and phosphorus status of the soils are significant characteristics that appear to have resulted from the accumulation of domestic waste products within settlement areas. This accords with the presence of pottery and other artifacts in the soil. While terra preta soils, once formed, represent favorable areas for cultivation, there is no reason to assume that the soils themselves were originally the direct result of agricultural activity.

In general, the Colombian terra preta soils appear similar in origin to those reported elsewhere in Amazonia. Consultations with our Brazilian colleagues (vide Acknowledgments) produced a consensus; in their experience, as in ours, all terra preta localities are archaeological sites, although not all archaeological sites have terra preta soils. Some spatial variability occurs in the soils. Intrasite

138 AMERICAN ANTIQUITY [Vol. 49, No. 1, 1984]

variation, for example, is evident in the middle Caqueta basin, as demonstrated by our own research at La Sardina and the work of Andrade and Botero at Araracuara. Variations also occur between sites. Thus, some Brazilian terra preta soils contain significantly higher levels of available phosphorus than those reported in the present study area, and also display enhanced levels of exchangeable calcium and magnesium (Falesi 1972; Sombroek 1966). Such variability is to be expected, and presumably reflects differential rates of accumulation of domestic waste, as well as variations in the length of site occupance. Because weathering and leaching processes have continued since the formation of individual terra preta soils, the age and subsequent history of the soils, including the possibility of later cultivation, will also have influenced their chemical and other characteristics.

While terra preta soils can readily be discussed in cultural and pedological terms, the question arises as to their precise status from a soil taxonomic point of view. As yet, rather few sites have been investigated and little comparative information is available on the soil material. Using the American soil classification, Ranzani et al. (1970) have described the Brazilian terra preta as a "plaggen epipedon," which develops as a surface agricultural horizon (U.S.D.A. 1960, 1975). While a terra preta is better described in that classification as an "anthropic epipedon," developed at a place of settlement (U.S.D.A. 1975), the objective of Ranzani et al. (1970) to place the material within a general soil classification is appropriate, since it provides an independent framework for the terra preta and also draws attention to the fact that comparable soils exist in other parts of the world.

Although few generalizations can currently be offered with respect to terra preta soils, considerable scope exists for further investigation of their nature, origin and distribution, as an aid to the reconstruction of Precolumbian settlement and population patterns (Beckerman 1979:555; Denevan 1970; Lathrap 1970; Meggers 1971; Myers 1973; Smith 1980). The situation emphasizes once again that archaeologists should not work in isolation, classifying their potsherds while ignoring the soil matrix in which these occur. The urgent need now is for more laboratory analyses and for a much closer collaboration between archaeologists and soil scientists.

Acknowledgments. Acknowledgment of assistance in the field is made to the Colombian military leader of the Colombian Amazonas Expedition 1977, Major Alberto Rubio, and to his British colleagues, Captains John Saunders and Adrian Goldsack; also to Ana Maria Falchetti, Elizabeth Reichel de von Hildebrand, Tomas Lombo, Julio Morelo, William Nieto, Valentin Oyola, Dr. Jaime Restrepo, Geoff Tancred, Jose Luis Varela, and to our Yucuna friends at La Sardina.

For comment, literature and information we are indebted to Ulpiano Bezerra de Meneses, Igor Chmiz, Ondemar Dias, Robert Eidt, Donald Lathrap, Euricio Th. Miller, Pedro Ignacio Schmitz, Mario Sim6es, Marlene Suano and Peter Paul Hilbert. Angela Andrade and Pedro Botero made available the results of their own investigations in the Araracuara region. None of these scholars should be held responsible for any conclusions we have drawn from their data. The palm fruits were identified by Jesus Idrobo. For laboratory and cartographic work, we thank Margaret Onwu and Claire Wastie. For permission to carry out the research we are grateful to the Instituto Colombiano de Antropologla and its then Director, Alvaro Soto Holguin, and, for other assistance, to Carlos Caceres, Director of the Proyecto Radargrametrico del Ainazonas.

Financial assistance is acknowledged by Eden and Bray from the Central Research Fund, University of London, and bv Brav frnm the British Academv

REFERENCES CITED

Beckerman, Stephen 1979 The Abundance of Protein in Amazonia: A Reply to Gross. American Anthropologist 81:533-560.

Black, C. A., D. D. Evans, J. L. White, L. E. Ensminger, and F. E. Clark 1965 Methods of Soil Analysis (Parts 1 and 2). American Society of Agronomy, Madison.

Bolian, Charles 1972 An Archaeological Survey of the Trapecio de Amazonas, Colombia. Paper presented at the Northeastern

Anthropological Association meetings, April, 1972, Buffalo, N.Y. 1975 Archaeological Excavations in the Trapecio of Amazonas. The Polvchrome Tradition. Ph.D. disserta-

tion, University of Illinois at Urbana-Champaign. University Microfilms, Ann Arbor. Bray, Warwick, Colin McEwan, and Leonor Herrera

1977 The Archaeology of the Region of Araracuara (Colombian Amazonas): A Preliminary Report. Ms. on file, Instituto Colombiano de Antropologia, Bogota.

REPORTS 139

Brochado, Jose P., and Donald W. Lathrap 1982 Amazonia. In Chronologies in New World Archaeology, edited by C. Meighan. Academic Press, New

York, in press. Brown, C. Barrington

1876 Canoe and Camp Life in British Guiana. Edward Stanford, London. Carneiro, Robert L.

1974 "Caraip6": An Instance of the Standardization of Error in Archaeology. Journal of the Steward An- thropological Society 6(1):71-75.

Cruxent, J. M., and Irving Rouse 1958 An Archaeological Chronology of Venezuela, vol. 1. Pan American Union, Social Science Monographs

6. Washington, D.C. Denevan, William M.

1970 The Aboriginal Population of Western Amazonia in Relation to Habitat and Subsistence. Revista Geograjica, Rfo de Janeiro 72:61-86.

Eden, Michael J., Duncan F. M. McGregor, and Julio A. Morelo 1982 Geomorphology of the Middle Caqueta Basin of Eastern Colombia. Zeitschrift fur Geomorphologie

26:343-364. 1983 Semi-quantitative Classification of Rainforest Terrain in Colombian Amazonia using Radar Imagery.

International Journal of Remote Sensing, in press. Eidt, Robert C.

1977 Detection and Examination of Anthrosols by Phosphate Analysis. Science 197:1327-1333. Evans, Clifford, and Betty J. Meggers

1968 Archaeological Investigations on the Rfo Napo, Eastern Ecuador. Smithsonion Contributions to An- thropology No. 6. Washington, D.C.

Falesi, Italo Claudio 1972 0 estado atual dos conhecimentos s6bre os solos da Amaz6nia brasileira. In Zoneamento agricola da

Amaz6nia (la aproxima,co). Boletim Tecnico do Instituto de Pesquisa Agropecuaria do Norte 54:1-67. Gansser, Augusto

1974 The Roraima Problem (South America). Verhandlungen en der Naturforschenden Gesellschaft in Basel 84:80-97.

Gourou, Pierre 1949 Observa,6es geograficas na Amaz6nia. Revista Brasileira de GeograJia 11:354-408.

Griffith, M. A. 1980 A Pedological Investigation of an Archaeological Site in Ontario, Canada, 1. An Examination of the

Soils in and Adjacent to a Former Village. Geoderma 24:327-336. Herrera, Leonor

1981 Relaciones entre ocupaciones prehispanicas y suelos negros en la cuenca del Rio Caqueta en Colombia. Revista CIAF 6:225-242.

Herrera, Leonor, Warwick Bray, and Colin McEwan 1982 Datos sobre la arqueologia de Araracuara (Comisaria del Amazonas, Colombia). Revista Colombiana

de Antropologfa 23(1980-81): 183-251. Hilbert, Peter Paul

1955 A ceramica arqueol6gica da regiao de Oriximina. Instituto de Antropologia e Etnologia do Para, Publication 9. Belem.

1968 Archaologische Untersuchungen am Mittleren Amazonas. Marburger Studien zur Vo'lkerkunde 1. Die- trich Reimer, Berlin.

Hilbert, Peter Paul, and Klaus Hilbert 1979 Archaologische Untersuchungen am Rio Nhamunda, unterer Amazonas. In Beitrdge zur Allgemeinen

und Vergleichenden Archdologie, edited by Hermann Miiller-Karpe, Band 1, pp. 439-450. Deutsche Ar- chaologisches Institut: Verlag C. H. Beck. Munich.

1980 Resultados preliminares da pesquisa arqueol6gica nos Rios Nhamunda y Trombetas, Baixo Amazonas, Boletim do Museu Paraense 'Emilio Goeldi," n.s. 75: 1-11.

Klinge, H., E. Medina, and R. Herrera 1977 Studies on the Ecology of Amazon Caatinga Forest in Southern Venezuela. 1. General Features. Acta

Cientufica Venezolana 28:270-276. Lathrap, Donald W.

1970 The Upper Amazon. Thames and Hudson, London. Meggers, Betty J.

1971 Amazonia: Man and Culture in a Counterfeit Paradise. Aldine-Atherton, Chicago. Myers, Thomas P.

1973 Towards the Reconstruction of Prehistoric Community Patterns in the Amazon Basin. In Variation in Anthropology edited by D. W. Lathrap and J. Douglas, pp. 233-252. Illinois Archaeological Survey.

140 AMERICAN ANTIQUITY [Vol. 49, No. 1, 1984]

Palmatary, Helen C. 1960 The Archaeology of the Lower Tapaj6s Valley', Brazil. Transactions of the American Philosophical

Society, n.s., vol. 50, Part 3. Philadelphia. PRORADAM

1979 La Amazonia Colombiana y sus recursos (Tomo 1). Proyecto Radargrametrico del Amazonas, Bogota. Ranzani, G., T. Kinjo, and 0. Freire

1970 Ocorrencias de "plaggen epipedon" no Brasil. Noticia Geomorfol6gica, Camnpinas 10:55-62. Sanchez, Pedro A., Dale E. Bandy, J. Hugo Villachica, and John J. Nicholaides

1982 Amazon Basin Soils: Management for Continuous Crop Production. Science 216:821-827. Sim6es, Mario F.

1967 Considera~oes preliminares s6bre a arqueologia do Alto Xingu (Mato Grosso). Prograina Nacional de Pesquisas Arqueologicas; Resultados Prelininares do Pritneiro Ano 1965-6. PublicaQ5es A vulsas 6:129-15 1. Museu Paraense "Emilio Goeldi," Belem.

1974 Contribui,co a arqueologia dos arredores do baixo Rio Negro, Amazonas. Programa Nacional de Pesquisas Arqueologicas 5:165-188. Museu Paraense "Emilio Goeldi," Belem.

Smith, Nigel J. H. 1980 Anthrosols and Human Carrying Capacity in Amazonia. Annals of the Association ofAinerican Geog-

raphers 70:553-566. Soares, Lucio de Castro

1963 Amaz6nia. Conselho Nacional de Geografia, Rio de Janeiro. Sombroek, W. G.

1966 Amazon Soils. Centre for Agricultural Publications and Documentation, Wageningen. Sternberg, Hilgard O'Reilly

1960 Radiocarbon Dating as Applied to a Problem of Amazonian Morphology. Comptes Rendus du XVIII Congres International de Geographie, vol. 11:399-424. Centro de Pesquisas de Geografia do Brasil, Uni- versidade do Brasil, Rio de Janeiro.

U.S.D.A. 1960 Soil Classification. A Comprehensive System. Soil Conservation Service, United States Department of

Agriculture, Washington, D.C. 1975 Soil Taxonomy. Soil Conservation Service, UJnited States Department of Agriculture, Handbook 436.

Washington, D.C. Vieira, Lucio Salgado, Nioma Viegas de Carvalho y Oliveira, and Terezinha Xavier Bastos

1971 Os solos do Estado do Para. Cadernos Paraenses 8. Belem. von Hildebrand, Elizabeth de

1976 Resultados preliminares del reconocimiento del sitio arqueol6gico de La Pedrera (Comisaria del Amazonas). Revista Colombiana deAntropologia 20:145-176.

Weber, Ronald Le Roy 1975 Caimito: An Analysis qf the Late Prehistoric Culture of the Central Ucayali, Eastern Peru. Ph.D.

dissertation, University of Illinois at Urbana-Champaign. University Microfilms, Ann Arbor.