Gourd and Squash Artifacts Yield Starch Grains of Feasting Foods From Preceramic Peru

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Gourd and squash artifacts yield starch grains

of feasting foods from preceramic Peru


Neil A. Duncan1, Deborah M. Pearsall, and Robert A. Benfer, Jr.
Department of Anthropology, University of Missouri, Columbia, MO 65211
Edited by Michael E. Moseley, University of Florida, Gainesville, FL, and approved June 8, 2009 (received for review March 26, 2009)

In a study of residues from gourd and squash artifacts, we recovered starch grains from manioc (Manihot esculenta), potato (Solanum sp.), chili pepper (Capsicum spp.), arrowroot (Maranta arundinacea), and algarrobo (Prosopis sp.) from feasting contexts at the
Buena Vista site, a central Peruvian preceramic site dating to 2200
calendar years B.C. This study has implications for the study of
plant food use wherever gourds or squashes are preserved, documents the earliest evidence for the consumption of algarrobo and
arrowroot in Peru, and provides insights into foods consumed at
feasts.

n enhanced understanding of the use of bottle gourds in


aceramic cultures through the examination of starch residues on the gourds interior surfaces provides us with a potential
pathway to understanding behavioral, economic, symbolic, and
ritual contexts in which these artifacts are found. The myriad
uses for the shells of bottle gourd (Lagenaria siceraria) and
squash (Cucurbita sp.) in the archaeological record for both tools
and, for squash, food complicates determination of the precise
use of these materials without direct contextual information,
such as gourd fishing floats attached to nets (1), or encountering
whole or partial gourd artifacts from which one can infer their
use through ethnographic analogy (2). In this study, we extracted
starch from the residues of fragmented squash (Cucurbita) and
gourd (L. siceraria) serving vessels deposited in a ritual context
at the Buena Vista site in central Peru, located 35 km inland
(114351.72S, 76585.45W) from the mouth of the Chillo
n
River (Fig. 1). Overlooking the valley floor, Buena Vista contains monumental late preceramic architecture. Its large, elaborate, isolated space with uniquely patterned astronomical alignments (3) suggest that it served as a small ceremonial center in
the central Chillo
n Valley. The starch residues of edible plants
found on the artifacts and the special archaeological context
from which these artifacts were recovered suggest that the
artifacts were used in a ritual setting for the serving and
production of food. Among archaeologists, there is tremendous
interest in the examination of feasting, defined loosely as the
sharing of food or drink in a special context or for a special
purpose (4), in understanding the social relationships between
emergent elites and commoners. The use of food, particularly
chicha (maize beer), in traditional Andean systems in the
materialization of public labor is well documented (5, 6); however, the prehistoric use of food in religious, social, or economic
rituals in the Andes is less understood (7, 8).
Gourd and squash artifacts for this study were recovered from
a central feature in the Fox Temple at Buena Vista, named after
an incised stylized drawing of a fox in the doorway. At the top
of the temple mound, a large room (15 7 m) contained an
elevated platform within which there was a sunken niche-walled
pit (Fig. 2 A and B) from which the materials in this study were
recovered. The temple was intentionally interred, as is consistent
with the ritual entombment of monumental architecture at other
Andean sites, particularly at Kotosh (911), which shares similar
architectural features with Buena Vista. The room itself was
found clean of refuse, save for a single tree trunk lying adjacent
to the platform in the center. Within the platform, the sunken pit
contained well-stratified and abundant food and other plant
1320213206 PNAS August 11, 2009 vol. 106 no. 32

remains, a few mussel shells, and some small fish remains.


Abundant food remains, including seeds of chili peppers (Capsicum sp.), guava (Psidium guayaba), lucuma (Pouteria lucuma),
and squashes (Cucurbita maxima, C. moschata, and C. ficifolia)
and rind fragments of sweet potato (Ipomoea batatas), manioc
(Manihot esculenta), and potato (Solanum sp.), were recovered
with the remains of nonfood plants such as cotton (Gossypium
barbadense), grass, fragments of agave leaves (Furcraea andina),
and partially burnt twigs and charcoal. Faunal remains are much
less abundant and limited primarily to mussel shells and small
fish vertebrae. Charcoal from separate stratigraphic levels within
the pit provided two radiocarbon dates, both 2200 calendar
years B.C. (Table 1), suggesting that the pit was filled in a short
period. The sequence of burial began with what might be
described as an organic protective layer similar to that described in the temple entombment at Huaca Soledad (12), a 5to 10-cm-deep layer of grass and fine silt that covered the bottom
of the pit, atop which commingled organics, grass, leaves, and
food remains were placed. Small round cobbles and gravel
topped this layer, followed by more mixed organic material, then
capped with small angular gravel. The surrounding room was
then filled with large rocks, some as large as 60 cm in diameter,
to the level of the top of the platform. Last, the entire room was
filled with shicra, cane bags filled with rock.
Archaeological starch grain research has contributed greatly
to the study of plant use and consumption by humans (1315).
However, this is the first study to analyze residue from bottle
gourd or squash artifacts. One cannot underestimate the significance of squash and bottle gourds to humans. Utilitarian bottle
gourds and edible squashes are among the first cultivars in the
Americas (16, 17). On the Peruvian coast, bottle gourds are
common in preceramic contexts and closely associated with
maritime subsistence as net floats (1). However, their use also
enters the realm of the sacred as symbolic containers (18) and
for serving ritual libations. As demonstrated here, residue
analysis can help to determine use.
Results
Starch grains were recovered from squash and gourd artifacts by
employing a method similar to that used to recover microfossils
from stone tools and ceramics (see SI Materials and Methods).
First, the artifact is placed in a sonicating water bath to loosen
and remove adhering residue, which is collected as Sediment 1.
Then, the artifacts interior surface is lightly brushed to remove
any remaining residue, collected as Sediment 2. Starch grains are
then isolated from each of these sediments (Tables 2 and 3).
The artifacts in this study are well-preserved, desiccated
specimens recovered under conditions of exceptional preservaAuthor contributions: N.A.D. and D.M.P. designed research; N.A.D. and R.A.B. performed
research; D.M.P. contributed new reagents/analytic tools; N.A.D. analyzed data; and N.A.D.
wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1To

whom correspondence should be addressed. E-mail: neilduncan@mizzou.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/


0903322106/DCSupplemental.

www.pnas.orgcgidoi10.1073pnas.0903322106

Fig. 2. The sunken pit and platform in the Fox Temple. (A) Photograph of
sunken pit facing east. (B) Stratigraphic profile of the pit contents from which
artifacts were studied.
Fig. 1. Location of Buena Vista on the central coast of Peru. (Inset) Location
of Buena Vista in the Chillon Valley.

tion (Fig. S1). Squash and gourd fragments can be distinguished


easily by cross-section (19). One Cucurbita sp. fragment was
studied and yielded several noncucurbit starch grains, indicating
that Cucurbita fruits also were used as food containers or serving
vessels (Table 2). Seven of the 10 artifacts yielded starch. As
expected, we identified starch from the mesocarp of the artifact
itself. There is no significant difference in the number of
Cucurbitaceae starch grains between Sediment 1 (sonicated)
and Sediment 2 (brushed) samples. However, non-Cucurbitaceae starch grains occurred in a higher number in the Sediment
1 samples. Thus, the non-Cucurbitaceae starch grains appear to
occur in greater concentration in the food residue adhering to
the outermost interior surface of the artifacts surviving mesocarp. A strong case can be made that non-Cucurbitaceae economic starch grains in both Sediment 1 and Sediment 2 represent
residues from artifact use and not transfer of starch from soils
adjacent to the artifacts (20). Analysis of soils in the sunken pit

produced only three transient starch grains, all from Level 200,
above the levels containing the starch-bearing gourd and squash
fragments (Table 1). Starch identifications were aided by a
collection of 250 comparative neotropical starch specimens
and regional starch reports and keys (13, 2123). Given the
potential for variation within a taxon, multiple criteria must be
used to identify an individual grain (24). Our identifications are
conservative. We securely identified five taxa: M. esculenta
(manioc), Solanum (potato), Maranta arundinacea (arrowroot),
Prosopis (algarrobo), and Capsicum (chili pepper).
Manioc starch occurs on three artifacts. One grain (Fig. 3B)
is a relatively large (28 m in diameter) sphere with two basal
facets and a large stellate-shaped fissure. Grain size is in the
upper range of domesticated manioc starch and likely too large
to be wild (25). Other hemisphere-shaped grains with two basal
facets are consistent with manioc but not diagnostic on their
own. Facet morphology is distinct from similar grains in Cucurbitaceae and I. batatas (sweet potato); thus, they were identified
as M. esculenta. Another grain identified as M. esculenta is
spherical with a depressed linear fissure and a regular extinction

Artifact nos.

Context

Radiocarbon years B.P.

Date in calendar years B.C.

02, 05, 07
01, 03, 04, 08

Sunken Pit: Unit 10, Feature 1, Level 300


Sunken Pit: Unit 10, Feature 1, Level 400
Sunken Pit: Unit 10, Feature 1, Level 425
Unit 10, Cuad 3, level 100

3,770 80 B.P. (GX-31276)


No date
3,790 80 BP (GX-32177)
No date

24601980 calibrated

09, 10

24702020 calibrated

Standard radiocarbon date from associated charcoal, 2 calibrated result.

Duncan et al.

PNAS August 11, 2009 vol. 106 no. 32 13203

ANTHROPOLOGY

Table 1. Artifacts and radiocarbon dated contexts

Table 2. Starch grain data from squash and gourd artifacts per sediment sample
Artifact no.

Sediment no.

Lagenaria siceraria
Cucurbita sp.
Capsicum sp.
Manihot esculenta
cf. Manihot esculenta
Solanum sp.
Maranta arundinacea
Prosopis sp.
Unidentified
Unidentified root/tuber
Total starch

2*
2

5
2

6
2

8
2

10

1
1
2

2
1
1

1
8

1
1
3

1
1
4

1
3

Sediment 1 was removed from the artifact by sonicating water bath. Sediment 2 was removed by brushing the artifacts interior surface.
*Cucurbita sp. artifact. All others identified as Lagenaria siceraria.

cross. This grain lacks basal facets; however, the type appears
with diagnostic grains and faceted hemispheres in comparative
manioc samples.
Manioc macroremains are widespread in coastal preceramic
contexts; however, manioc is believed to have been domesticated
in the southern Amazon (2627). In Perrys study (22), starch
grains from initial period and late precontact manioc specimens
from coastal Peru lacked the distinguishing stellate fissure (13)
characteristic of lowland Amazonian manioc starch. However,
the large stellate fissure consistent with the lowland morphotype
is present in the Buena Vista assemblage. This grain possibly
represents an earlier imported Amazonian manioc variety that
was subject to selection over time, eventually resulting in the
morphotypes seen in later coastal assemblages. Alternatively,
this starch grain represents an early lowland variety that ceased
to be cultivated or is not related to later coastal forms. Additional manioc starch from early coastal contexts would help to
clarify these issues.
A starch grain from Solanum (potato) tuber was identified
(Fig. 3C). Potato has great significance to modern and ancient
Andean populations. Potato macroremains and microfossils are
reported from a number of contemporaneous sites throughout
Peru (2830). Several potato species and many varieties are and
have been cultivated, and many wild forms exist (31). At this
time, we cannot identify this starch grain to a species.
Capsicum (chili pepper) is represented by a single flattened
lenticular starch grain with an indented base and a linear feature
visible in side view (Fig. 3D). The central linear feature is diagnostic
of starches in this genus (15). Capsicum is reported widely from late
preceramic sites on the coast of Peru (32) and in the highlands (15)
and present in macroremains at Buena Vista.
Two grains identified as M. arundinacea (arrowroot) occur on
two gourd fragments (Fig. 3E). Despite its widespread and long
record of use in the neotropics (26), arrowroot is underrepresented
in Peruvian contexts. The only other evidence for arrowroot in the
central Andes comes from microfossils at Waynuna in the southern
highlands in contexts dating to 20501650 B.C. (28). At several
Table 3. Summary of starch grain data from squash
and gourd artifacts
Number of samples
Sample type

Cucurbitaceae starch

Other starch

Total

Sediment 1
Sediment 2
Total

9 (27.3%)
8 (24.2%)
17 (51.2%)

11 (33.3%)
5 (15.2%)
16 (48.5%)

20
13
33

13204 www.pnas.orgcgidoi10.1073pnas.0903322106

thousand meters above its natural range, the Waynuna arrowroot


represents an imported cultivar. Likewise, wild Maranta is not
known from the coast of Peru, making the starch grains identified
in this study highly unlikely to represent a wild form. With floodplain irrigation, arrowroot could have been cultivated easily. Thus,
the most parsimonious explanation for arrowroot starch on these
gourd artifacts is that it was grown near Buena Vista.
Starch identified as Prosopis (algarrobo) also occurs in the artifact
residue (Fig. 3F). Algarrobo starch has a unique appearance, unlike
anything else in our reference collection. The unusual lobed shape
with protuberances and a deeply depressed area at the hilum, along
with a very irregular extinction cross, correspond closely with starch
produced in Prosopis. This type may be a genus-level indicator of
starch produced in Prosopis pods. Prosopis consists of several closely
related and morphologically similar species with high hybridization
potential; thus, there is considerable confusion in the botanical (33)
and archaeological (34) literature regarding its identification. Currently, the dominant Prosopis on the Peruvian coast is P. pallida,
with P. chilensis in higher elevations of the south and P. juliflora in
the northwest (33). Each species produces similar sweet edible pods.
Species similarities are reflected in the pod starch grains. For
example, a systematic study showed that there are no differences in
starch between P. chilensis and P. flexuosa (35). Algarrobo is
ubiquitous in pre-Colombian Peru, most often as wood for fuel or
construction (30, 36) or more rarely for seeds and edible pods (34).
A long history of algarrobo use occurs in Argentina, beginning
10,000 years ago (35). In Peru, less is understood regarding prehistoric consumption of algarrobo pods, which are today ground
into flour to make a syrup, algarrobina, or for drinks, such as
fermented chicha de algarrobo. To our knowledge, this study
presents the first microfossil evidence and earliest evidence of
algarrobo consumption in Peru.
Discussion
Four of the 10 artifacts sampled yielded starch from multiple taxa
other than Cucurbitaceae and provide clues to preceramic cuisine.
Our sample size is small; however, in future research, patterns of
cooccurrence of taxa on gourd artifacts will help us to better
understand preferences for certain food combinations or certain
combinations limited to ritual use. Alternatively, the multiple taxa
on the artifacts may represent repeated use of the containers in both
ritual and secular contexts. Also, that the containers provide
evidence of manioc and algarrobo, both of which can be fermented
into alcoholic beverages, may be significant (33, 37). The possibility
that these plants could have been used in the production of alcohol
during the preceramic should not be overlooked.
The presence of these artifacts along with plant foods in a special
context is highly consistent with an interpretation of feasting (4).
Duncan et al.

Fig. 3. Archaeological starch granules recovered from gourd and squash artifacts from Buena Vista. (A) Gourd (Lagenaria siceraria) starch hemisphere from
Artifact 3. (B) Manioc (Manihot esculenta) from Artifact 7. Note the central stellate fissure. (C) Potato (Solanum sp.) from Artifact 4. (D) Chili pepper (Capsicum
sp.) from Artifact 6. Note the central linear feature. (E) Arrowroot (Maranta arundinacea) from Artifact 6. (F) Algarrobo (Prosopis sp.) from Artifact 3.

Duncan et al.

research should apply to other areas and time periods in which


gourds and squash rinds are preserved. Gourd and squash artifacts
containing starch residues provide direct evidence of the production and consumption of cultivated or managed plants used for food
or drink. In addition to manioc, potato, and chili pepper, this study
documents the earliest evidence of both algarrobo and arrowroot in
Peru. Also, the unique architectural feature from which the artifacts
were recovered suggests a special ritual use of food in a feasting
context. The social and ritual use of food in emergent complex
society is not well understood, especially in Peru, and this research
enhances our potential for understanding these issues.
Methods
Comparative Squash and Gourd Starch. In studying comparative gourd and
squash specimens, we found abundant starch in the dried mesocarp tissue
but relatively fewer starch granules in the exocarp. In the archaeological
samples, only the outer exocarp layers survive. This may explain the relatively few Cucurbitaceae starch granules in the residues. To differentiate
gourd and squash starch grains from those of other producers, we studied
modern dried gourds collected in Chiclayo, Peru, and domestically purchased or grown squashes (Fig. S2 and Fig. S3). Although there is some
overlap in starch morphotypes in Cucurbitaceae, squashes and gourds can
be distinguished to genus and species based on the diagnostic morphology
of starch grains (38). Starch grains of Cucurbitaceae are also discernable
from nonrelated taxa.
Starch Grain Sampling of Squash and Gourd Artifacts. We adapted procedures
for extracting microfossil residues from stone tools and ceramics (see SI Materials
and Methods), producing two sediment extractions from each tool, Sediments 1
and 2. To obtain Sediment 1, each artifact was submerged in distilled water
within a high-quality, resealable plastic bag, then placed in a sonicating water
bath to release loose residue. To obtain Sediment 2, the interior surface of each
artifact was gently brushed to further dislodge loose residue. The sediments
obtained were then chemically dispersed using a disodium salt solution, then
lightly oxidized with hydrogen peroxide. Finally, the starch was floated out from
sediment using a heavy liquid solution of cesium chloride. Slides were mounted
using distilled water and glycerol and scanned on a Zeiss standard transmitted
light microscope equipped with polarizing light filters at 25 and 40.
ACKNOWLEDGMENTS. Excavation in the Fox Temple in 2005 was carried out
by students in the University of Missouri Field School at Buena Vista and the
University of Frederico Villareal of Lima. We thank the Peruvian National
Institute of Culture for providing the permits to excavate and export archaeoPNAS August 11, 2009 vol. 106 no. 32 13205

ANTHROPOLOGY

Macroremain analysis is ongoing; however, the fragmentary remains of the plant material, including charred and uncharred wood,
grass, and cotton seeds and fiber, particularly the preponderance of
nonedible parts of food plants and broken gourd and squash serving
vessels found in the pit, are counterindicative of offerings but
consistent with the refuse of consumption. Common Andean
offerings of coca leaves are conspicuously absent. Therefore, the
vegetal contents of the pit likely do not represent the original
intended function or symbolism of the pit, for astronomical observation or for ritual offerings of liquid libations such as later ushnu
(39). Rather, the remains are more likely to be the refuse from
feasting associated with the ritual entombment of the temple. Ritual
entombment of monumental architecture was practiced commonly
in the preceramic and later; however, its association with feasting
as a ritual component or to organize and motivate labor, although
ancient, has been investigated only recently. For example, at the
contemporary site of Cerro Lampay (7) on the north central coast,
multiple small scale construction events were preceded by feasting
rituals hosted by informal leaders who lacked the social power to
organize large amounts of labor for more massive building events.
Despite their large size and complexity, many monumental sites
constructed in the preceramic could have been built by a small local
population organized by emergent leaders who used feasts as ritual
and political tools in materializing labor.
At Buena Vista, feasting event(s) before the entombment of the
Fox Temple likely served both political and ritual functions. Politically, leaders could have used feasting as a means to cultivate and
maintain social relationships, whereby labor would be mobilized
toward a communal goal, not unlike in later periods, where feasting
involving maize beer helped to mobilize labor to maintain canals in
the Andes (6). Ritually, feasting before the temples entombment
may have served also as a symbolic act toward canceling the original
function of the temple, and the concentration of refuse in the pit
suggests a deliberate sequestration of these remains, perhaps as a
final symbolic act of cancellation of the pit before the rooms
interment. The concentration of feasting remains in the sunken pit
may reflect the concentration of symbolic power before its burial.
The potential for understanding plant food consumption and
production in aceramic contexts is greatly enhanced through an
analysis of residues on gourd and squash artifacts. This kind of

botanical samples to the United States. We are grateful to Dolores Piperno,


who provided insightful comments and suggestions regarding comparative

cucurbit materials. We also thank Jason Fenton and Eliana Duncan, who
provided helpful criticisms on the manuscript.

1. Whitaker TW, Bird JB (1949) Identification and significance of the cucurbit materials
form Huaca Prieta, Peru. Am Mus Nov No. 1426, pp 115.
2. Speck FG (1941) Gourds of the Southeastern Indians; a Prolegomenon on the Lagenaria Gourd in the Culture of the Southeastern Indians (New England Gourd Society,
Boston).
3. Benfer RA, Adkins LR The Americas oldest observatory (2007) Astron Mag 35(8):40 43.
4. Dietler M, Hayden B (2001) Feasts: Archaeological and Ethnographic Perspectives on
Food, Politics, and Power (Smithsonian Institution Press, Washington).
5. Isbell BJ (1978) To Defend Ourselves: Ecology and Ritual in an Andean Village (Institute
of Latin American Studies, Univ of Texas, Austin).
6. Morris C (1979) Maize beer in the economics, politics, and religion of the Inca empire.
Fermented Beverages in Nutrition, eds Gastineau CF, Darby WJ, Turner TB (Academic,
New York), pp 2134.
7. Vega-Centeno Sara-Lafosse R (2007) Construction, labor organization, and feasting
during the late archaic period in the central Andes. J Anthropol Archaeol 26:150 171.
8. Jennings J (2005) La chichera y el patron: Chicha and the energetics of feasting in the
prehistoric Andes. Foundations of Power in the Prehispanic Andes, eds Conlee CA,
Ogburn D, Vaughn K (American Anthropological Association, Washington, DC), pp
241259.
9. Izumi S, Terada K (1972) Andes 4: Excavations at Kotosh, Peru, 1963 and 1966 (Univ of
Tokyo Press, Tokyo).
10. Burger RL, Salazar-Burger L (1986) Early organizational diversity in the Peruvian
highlands: Huaricoto and Kotosh. Andean Archaeology, eds Matos R, Turpin S (UCLA
Institute of Archaeology, Los Angeles), pp 65 82.
11. Bonnier E (1997) Preceramic architecture in the Andes: The Mito tradition. Archaeologica Peruana 2Prehispanic Architecture and Civilization in the Andes, eds Bonnier
E, Bischof H (Reiss-Museum, Mannheim, Germany), pp 120 144.
12. Shimada I (1986) Batan Grande and cosmological unity in the prehistoric central Andes.
Andean Archaeology, eds Matos R, Turpin S (UCLA Institute of Archaeology, Los
Angeles), pp 163188.
13. Piperno DR, Holst I (1998) The presence of starch grains on prehistoric stone tools from
the humid Neotropics: Indications of early tuber use and agriculture in Panama. J
Archaeol Sci 25:765776.
14. Chandler-Ezell K, Pearsall DM, Zeidler JA (2006) Root and tuber phytoliths and starch
grains document manioc (Manihot esculenta), arrowroot (Maranta arundinacea), and
lleren (Calathea sp.) at the Real Alto site, Ecuador. Econ Bot 60:103120.
15. Perry L, et al. (2007) Starch fossils and the domestication and dispersal of chili peppers
(Capsicum spp. L.) in the Americas. Science 315:986 988.
16. Dillehay TD, Rossen J, Andres TC, Williams DE (2007) Preceramic adoption of peanut,
squash, and cotton in northern Peru. Science 316:1890 1892.
17. Erickson DL, Smith BD, Clark AC, Sandweiss DH, Tuross N (2005) An Asian origin for a
10,000-year-old domesticated plant in the Americas. Proc Natl Acad Sci USA
102:1831518320.
18. Haas J, Creamer W, Ruiz A (2003) Gourd lord. Archaeology 56(3):7.
19. Cutler HC, Whitaker TW (1961) History and distribution of the cultivated cucurbits in
the Americas. Am Antiq 26:469 585.
20. Haslam M (2004) The decomposition of starch grains in soils: Implications for archaeological residue analyses. J Archaeol Sci 31:17151734.

21. Pearsall DM, Chandler-Ezell K, Zeidler JA (2004) Maize in ancient Ecuador: Results of
residue analysis of stone tools from the Real Alto site. J Archaeol Sci 31:423
442.
22. Perry L (2002) Starch granule size and the domestication of manioc (Manihot esculenta)
and sweet potato (Ipomoea batatas). Econ Bot 56:335349.
23. Reichert ET (1913) The Differentiation and Specificity of Starches in Relation to
Genera, Species, etc. (Chapman, London).
24. Piperno DR (2006) The origins of plant cultivation and domestication in the Neotropics:
A behavioral ecological perspective. Behavioral Ecology and the Transition to Agriculture, eds Kennett DJ, Winterhalder B (Univ of California Press, Berkeley), pp 137
166.
25. Piperno DR, Ranere AJ, Holst I, Hansell P (2000) Starch grains reveal early root crop
horticulture in the Panamanian tropical forest. Nature 407:894 897.
26. Piperno DR, Pearsall DM (1998) The Origins of Agriculture in the Lowland Neotropics
(Academic, San Diego).
27. Olsen KM, Schaal BA (2001) Microsatellite variation in cassava (Manihot esculenta,
Euphorbiaceae) and its wild relatives: Further evidence for a southern Amazonian
origin of domestication. Am J Bot 88:131142.
28. Perry L, Sandweiss DH, Piperno DR, Rademaker K, Malpass MA (2006) Early maize
agriculture and interzonal interaction in southern Peru. Nature 440:76 79.
29. Ugent D, Pozorski S, Pozorski T (1982) Archaeological potato tuber remains from the
Casma valley of Peru. Econ Bot 36:182192.
30. Quilter J, et al. (1991) Subsistence economy of El Paraso, an early Peruvian site. Science
251:277283.
31. Hawkes JG (1990) The Potato: Evolution, Biodiversity and Genetic Resources (Smithsonian Institution Press, Washington, DC).
32. Pearsall DM (2003) Plant food resources of the Ecuadorian formative: An overview and
comparison to the central Andes. Archaeology of Formative Ecuador, eds Raymond JS,
Burger RL (Dumbarton Oaks Research Library and Collection, Washington, DC), pp
213257.
33. Pasiecznik NM, et al. (2001) The Prosopis julifloraProsopis pallida Complex: A Monograph (HDRA,Coventry).
34. Towle MA (1961) The Ethnobotany of Pre-Columbian Peru (Aldine, Chicago).
35. Giovannetti MA, Lema VS, Bartoli CG, Capparelli A (2008) Starch grain characterization
of Prosopis chilensis (Mol.) Stuntz and P. flexuosa DC, and the analysis of their
archaeological remains in Andean South America. J Archaeol Sci 35:29732985.
36. Goldstein DJ (2007) Forests and fires: A paleoethnobotanical assessment of the impact
of middle Sican pyrotechnology on the dry tropical forests of the La Leche River Valley,
Lambayeque, Peru. PhD Thesis (Southern Illinois University, Carbondale).
37. Denevan WM (1974) Campa subistence in the Gran Pajonal, eastern Peru. Native South
Americans (Waveland, Prospect Heights, IL), pp 92110.
38. Piperno DR, Dillehay TD (2008) Starch grains on human teeth reveal early broad crop
diet in northern Peru. Proc Natl Acad Sci USA 105:1962219627.
39. Pino Matos JL (2004) The Inka ushnu and the organization of space in the principal
tampus of the Wamani of the central sierra of Chinchaysuyu (title translated from
Spanish). Chungara Rev Antropol Chil 36:303311.

13206 www.pnas.orgcgidoi10.1073pnas.0903322106

Duncan et al.

Supporting Information
Duncan et al. 10.1073/pnas.0903322106
SI Materials and Methods
Genus-Level Identification of Artifacts. Lagenaria siceraria rind

fragments are distinguished easily from Cucurbita spp. (Fig. S1)


by microscopic examination of their cross-sections (1, 2). This
method has proven much more successful in separating Lagenaria from Cucurbita (3) than using simple measurements of
thickness alone; however, the method is not sufficient for
separating species of Cucurbita. A clean portion of the crosssection was exposed using a razor blade. Specimens were examined using a Zeiss standard reflected light microscope at 160
magnification. To avoid starch contamination, this procedure
was carried out in a laboratory where no comparative starch is
sampled or processed, and care was taken to clean all tools
between samples using a small butane torch to clean tweezers,
probes, and razor blades.
Sampling Artifacts for Starch Residue. We adapted published procedures for extracting microfossil residues from stone tools and
ceramics (46). In developing a technique to sample the gourd
artifacts, one should weigh the costs and benefits of microfossil
recovery to the preservation of the artifact itself. These artifacts
survived this processing with no visible adverse effects; however,
for organic artifacts in more fragile condition, alternative sampling methods, such as in situ sampling (5, 7), should be
considered to avoid damage to the artifact.
Only new, unused plastic bags, test tubes, bottles, and plastic
weighing boats should be used for this procedure. Also, any tools
should be new and unused, such as disposable toothbrushes, or
cleaned, such as metal tweezers and probes, with concentrated
steam or a small butane torch to eliminate starch contamination.
Always use-high quality unpowdered latex, nitrile, or vinyl
gloves, because powdered gloves are often coated in cornstarch.
The procedure used is as follows below.
Sonication. First, an artifact was placed in a new, curation-quality,
resealable polyvinyl bag sized 5 7 inches, with 25 mL of
distilled water or enough to completely submerge the artifact.
Before the bag was sealed, the top of the bag was folded over,
just above the waterline, and the seam was held down with a
binder clip. This removes the air from the bag and helps to keep
the artifact fully submerged in the distilled water. This also helps
keep the bags from floating in the sonicator. The bag containing
the artifact then was placed into an ultrasonic cleaner filled with
water (the amount recommended per the sonicators manufacturer) and sonicated for 5 min to gently dislodge the residue from
the artifact. After 5 min, the bag containing the artifact was
removed from the sonicator. While being careful not to discard
the water in the bag, the artifact was removed with sterile
tweezers and placed in a new plastic container and preserved.
The water and all sediment from the bag then were transferred,
and the bag was rinsed into a new 50-mL test tube. The sediment
dislodged by sonication was labeled Sediment 1. The 50-mL tube
containing the sediment dislodged from the artifact by sonication was topped off with distilled water and placed in a centrifuge
(Damon IEC model HN-SII) for 5 min at 2,500 rpm to concentrate the sample at the bottom of the tube. The supernatant was
then carefully decanted or pipetted off, leaving the sample
undisturbed at the bottom of the tube. This tube containing
Sediment 1 was preserved.
Sediment 2. The interior surface of the artifact then was rinsed
gently with distilled water from a squirt bottle. The rinse water
was collected in the plastic container. To increase recovery of
starch, the interior surface of the artifact was brushed gently with
Duncan et al. www.pnas.org/cgi/content/short/0903322106

a soft-bristled toothbrush. Then, the artifact and the brush were


rinsed again into the same container. To avoid crosscontamination, only a new, unused disposable toothbrush was
used for each artifact. The artifact was then removed and allowed
to dry thoroughly before returning to curation. The water was
transferred, and the container was rinsed into a new 50-mL
centrifuge tube or a new 250-mL bottle if the quantity of water
was 50 mL. Then, the tube was centrifuged for 5 min at 2,500
rpm to concentrate the sample at the bottom of the tube. The
supernatant then was decanted carefully, leaving the sample
undisturbed at the bottom of the tube. This tube containing
Sediment 2 was preserved.
Dispersion. To the Sediment 1 and Sediment 2 tubes, 10 mL of
0.1% EDTA disodium salt (Na2H2EDTA) solution was added,
and the capped tubes were placed in a reciprocating shaker for
2 h to disperse the sediments. After being removed from the
shaker, the tubes were topped off with distilled water and
centrifuged for 3 min at 2,500 rpm, and the supernatant was
decanted carefully. The water rinse was repeated two more times
to remove any remaining chemical.
Oxidation. If the sediment at the bottom of the tubes and the
remaining water appear clear, then oxidation can be skipped.
Otherwise, gentle oxidation will help to remove superfluous
organic matter. To each tube was added 10 mL of a 5.75%
solution of hydrogen peroxide, and the reactions were monitored
for a maximum of 10 min. The tubes were then topped off with
distilled water and centrifuged for 3 min at 2,500 rpm. The
supernatant then was decanted carefully, leaving the sample
undisturbed at the bottom of the tube. The rinse was repeated
two more times to completely remove any remaining hydrogen
peroxide.
Flotation. To recover starch from the remaining sediments, flotation is recommended. Cesium chloride (CsCl) was dissolved in
distilled water to a specific gravity of 1.6. Then, 5 mL of the CsCl
solution was added to the samples and gently mixed. The tubes
were then centrifuged for 5 min at 2,000 rpm. The supernatant
containing the suspended starch sample then was decanted into
a fresh 50-mL tube and preserved. The flotation was repeated,
decanting the suspended starch solution to the starch sample. To
rinse, the starch samples were topped off with distilled water to
reduce the specific gravity and centrifuged for 5 min at 2,000 rpm
to concentrate the starch at the bottom of the tube. The
supernatant was decanted very carefully, leaving 15 mL of
water and the starch sample undisturbed at the bottom of the
tube. The rinse was repeated two more times to completely
remove remaining CsCl, and the starch sample was preserved.
The remainder of the sediment sample was topped off with
distilled water and centrifuged for 5 min at 2,000 rpm, and the
rinse was repeated two more times. The sediment was reserved
for phytolith extraction to be completed later.
Starch slide mounts and microscope scanning. A 22.5-L aliquot of
concentrated starch sample was placed on a clean glass slide with
two drops of 100% glycerol. A coverslip was placed on top, and
the edges were sealed with nail polish. Slides were scanned on a
Zeiss standard transmitted light microscope equipped with
polarizing light filters at 25 and 40. Polarized light is necessary to observe birefringent properties of starch.
Comparative Gourd (Lagenaria siceraria) Starch. In L. siceraria,
starch grains vary in shape from spherical to hemispherical to
oblong hemispheres (bell-shaped). The oblong hemispheres
(Fig. S2 A and B) may have pronounced, thick lamellae in distinct
1 of 5

Comparative Squash (Cucurbita spp.) Starch. Cucurbita ficifolia. Starch


grains in C. ficifolia, grown in Missouri by D.M.P., include oblong
hemispherical grains from 10 to 16 m in length having faint
pleat-like lamellae with slightly crenate sides and a deeply
indented basal facet, often with a lipped edge (Fig. S3A). The
pleated lamellae are much less pronounced and narrower than
those in oblong hemispherical grains produced in L. siceraria.
These grains also differ from bell-shaped grains with pleated
lamellae produced in the modern butternut variety of C. moschata, which tend to be smaller and have more pronounced
lamellae and an irregular base. The oblong hemispheres pro-

duced by C. ficifolia differ from traditional varieties of C.


moschata as well in that the latter produces oblong hemispheres
with pleat-like lamellae lacking crenate sides (8).
Other bell-shaped hemispheres with smooth sides are present,
and these typically also have a deeply indented basal facet (Fig.
S3B). These grains range in size from 10 to 22 m in length.
Spherical grains with smooth surfaces, lamellae, and an open
hilum are also present, from 10 to 20 m in diameter. There is
considerable variation in starch produced in C. ficifolia and C.
maxima (see below), and both produce significant numbers of
potentially diagnostic types as observed in this study and by D.R.
Piperno (personal communication). For example, flattened
hemispheres (Fig. S3C) having a deeply indented base and a
bump along the bottom edge of the long axis overlap types in L.
siceraria; however, certain flattened hemispherical grains in C.
ficifolia are distinctively round when viewed from the top of the
dome (Fig. S3C) and diagnostic to this species. These grains are
typically 15 to 20 m in diameter.
Cucurbita maxima. Starches in C. maxima, obtained from a comparative specimen from Chiclayo, Peru, and locally purchased
buttercup squash, contained hemispherical starch grains with a
deeply indented basal facet and compound hemispheres with
double basal facets (Fig. S3D), types frequently found in other
Cucurbitaceae. Irregularly shaped spheres are common (ranging
from 5 to 20 m in diameter) with centric, open hila and weak
lamellae and often with one or more pressure facets (Fig. S3E).
This species also produces distinctive double or triple compound
starch (Fig. S3F). The compound granules may exhibit only a
small fissure under transmitted light, but under polarized light,
one can see individual extinction crosses in each grain. These
compound granules range in size from 10 to 20 m in diameter.
Cucurbita moschata. Cucurbita moschata, locally purchased butternut squash, contains oblong hemispherical grains with crenate
sides and a deeply indented basal facet (Fig. S3 G and H). The
lamellae are more pronounced than those in bell-shaped grains
of C. ficifolia but also differ from those of L. siceraria in that the
lamellae on the C. moschata grains occur over two-thirds of the
length of the grain (Fig. S3H). Other bell-shaped grains with
pleated lamellae in C. moschata are distinct in having a smaller
base than dome. In our comparative samples, these grains range
in size from 8 to 12 m in length. Other starch granules in C.
moschata include spheres and hemispheres ranging in size from
5 to 15 m with shallow lamellae and open hila (Fig. S3I). The
hemispheres typically have an indented base characteristic of
other Cucurbitaceae.
Although the oblong hemispherical grains produced in our
comparative butternut squash can be distinguished from those of
other Cucurbitaceae, these grains with thickly pleated lamellae
and crenate sides do not occur in traditional varieties of C.
moschata (D. R. Piperno, personal communication). Starch
grains from modern butternut squash do not represent traditional South American varieties of C. moschata, which produce
diagnostic bell-shaped hemispheres with thin pleat-like lamellae
and lacking crenate sides (8); therefore, the modern types are not
expected archaeologically.

1. Whitaker TW (1948) Lagenaria: A pre-Columbian cultivated plant in the Americas.


Southwest J Anthropol 4:49 68.
2. Cutler HC, Whitaker TW (1961) History and distribution of the cultivated cucurbits in
the Americas. Am Antiq 26:469 585.
3. Ford RI (1986) Reanalysis of cucurbits in the ethnobotanical laboratory, University of
Michigan. Mo Archaeol 47:1331.
4. Loy TH (1994) Methods in the analysis of starch residues on prehistoric stone tools.
Tropical Archaeobotany: Applications and New Developments, ed. Hather JG (Routledge, London), pp 86 114.

5. Chandler-Ezell K, Pearsall DM (2003) Piggy-back microfossil processing: Joint starch


and phytolith sampling from stone tools. Phytolitharian Newsletter 15(3):2 8.
6. Perry L (2004) Starch analyses reveal the relationship between tool type and function:
An example from the Orinoco valley of Venezuela. J Archaeol Sci 31:1069 1081.
7. Loy TH, Spriggs M, Wickler S (1992) Direct evidence of human use of plants 28,000 years
ago: Starch residues on stone artefacts from the northern Solomon Islands. Antiquity
66:898 912.
8. Piperno DR, Dillehay TD (2008) Starch grains on human teeth reveal early broad crop
diet in northern Peru. Proc Natl Acad Sci USA 105:1962219627.

concentric rings perpendicular to the long axis to the grain. In


profile, the sides of the grain appear crenate. Hila are usually
open and eccentric but always in the dome of the hemisphere.
Grain size in our comparative specimens ranges from 8 to 15 m
in length. These grains are diagnostic to L. siceraria. Bell-shaped
grains with pleated lamellae are present in C. moschata (8), C.
ficifolia, and Cyclanthera pedata; however, these grains can be
distinguished based on characteristics individual to each taxon as
discussed below. For example, bell-shaped grains with pleated
lamellae produced in Cyclanthera pedata are smaller than those
in squashes, averaging 6 m more narrow and lightly pleated
with three or four lamellae and nearly smooth sides that are not
as distinctly crenate as those of L. siceraria and C. ficifolia.
Spherical grains in L. siceraria are generally round in outline
with smooth surfaces (Fig. S2 I and J). The centric to slightly
eccentric hilum is usually open and may be subject to tearing
(Fig. S2 J). Note that unlike spheres in manioc, for example,
lamellae are present and pronounced in larger grains. Spherical
grains range in size from 8 to 25 m in our samples. The unique
lamellae occurring on these spheres likely make the grain an
additional diagnostic morphotype produced only by L. siceraria.
Hemispherical starch grains have smooth surfaces and usually
closed centric to slightly eccentric hila (Fig. S2 CE). Small
hemispheres, 10 m, are common (Fig. S2C), but larger grains,
up to 20 m, are more distinctive, usually having visible lamellae
(Fig. S2E). The base of the hemisphere is concave or deeply
indented. This type, a hemisphere with a deeply indented base,
also is found in the Cucurbita studied here and in Sicana odorifera
and Cyclanthera pedata. This type may be a family-level diagnostic. Less abundant starch morphotypes include smallcompound and double-faceted hemispheres with smooth margins (Fig. S2D) and compound granules consisting of two to
three individual grains, each of which will exhibit individual
extinction crosses under cross-polarized light.
Flattened hemispherical grains with deeply indented bases
(Fig. S2 FH) also are present. When viewed from the top of the
dome, these grains are longer than wide. From the side, the
grains have a distinctive bump on the bottom edge of the long
axis. The hilum is located on the side of the grain along a surface
feature that resembles a small ridge running parallel or in a v
perpendicular to the long axis of the grain. Lamellae are
generally absent. This grain type overlaps types in C. ficifolia but
may be a useful indicator of L. siceraria.

Duncan et al. www.pnas.org/cgi/content/short/0903322106

2 of 5

Fig. S1. Three archaeological specimens of squash and gourd artifacts from Buena Vista, Peru. The interior surface of each is displayed on the bottom row.
See Table 2 for taxa of starch identified in the residues of (A) Artifact 2, identified by cross-section as Cucurbita sp., and (B, C) Artifact 4 and Artifact 6, both
identified as Lagenaria siceraria by cross-section. (Scale bar, 2 cm.)

Duncan et al. www.pnas.org/cgi/content/short/0903322106

3 of 5

Fig. S2. Comparative starch grains of Lagenaria siceraria from specimens collected near Chiclayo, Peru. Description in SI Materials and Methods. Each grain
is photographed under transmitted light (Left) and cross-polarized light (Right). (Scale bar, 10 m.)

Duncan et al. www.pnas.org/cgi/content/short/0903322106

4 of 5

Fig. S3. Comparative squash starch from Cucurbita ficifolia (AC), C. maxima (DF), and C. moschata (GI). See SI Materials and Methods for descriptions. Each
grain is photographed under transmitted light (Left) and cross-polarized light (Right). (Scale bar, 10 m.)

Duncan et al. www.pnas.org/cgi/content/short/0903322106

5 of 5

You might also like