Phytolith Evidence for Early Holocene Cucurbita Domestication in Southwest Ecuador

See allHide authors and affiliations

Science  14 Feb 2003:
Vol. 299, Issue 5609, pp. 1054-1057
DOI: 10.1126/science.1080365


Cucurbita (squash and gourd) phytoliths recovered from two early Holocene archaeological sites in southwestern Ecuador and directly dated to 10,130 to 9320 carbon-14 years before the present (about 12,000 to 10,000 calendar years ago) are identified as derived from domesticated plants because they are considerably larger than those from modern wild taxa. The beginnings of plant husbandry appear to have been preceded by the exploitation of a wild species ofCucurbita during the terminal Pleistocene. These data provide evidence for an independent emergence of plant food production in lowland South America that was contemporaneous with or slightly before that in highland Mesoamerica.

The transition from hunting and gathering to agriculture was one of the most important economic and social passages of human prehistory, and the topic has long been a focus of archaeological research. Studies in dry highland regions of Mesoamerica show that plant domestication had occurred by 9000 years before the present (yr B.P.) (about 10,000 calendar years ago) (1, 2). However, empirical research in the lowland Neotropics, long hypothesized to have been another independent center of agricultural origins (3), has been impeded by the poor preservation of plant remains. Here we describe early plant domestication in Ecuador using an analysis of phytoliths: microscopic siliceous remains of plants that survive in humid environments over long periods of time (4).

The early Holocene Las Vegas preceramic culture of coastal Ecuador is known from 34 sites on the Santa Elena Peninsula (5,6). Excavations at the type site, OGSE-80 (hereafter site 80) have produced evidence for a broad-spectrum subsistence strategy involving terrestrial, estuarine, and mangrove environments, and fairly sedentary occupations dated from about 10,000 to 660014C yr B.P. (about 11,300 to 7300 calendar years ago) (5–7). Traces of a terminal Pleistocene human presence are found stratified underneath a part of the Las Vegas midden. Cultural materials are sparse in these deep levels, but three14C determinations on shell and charcoal from this zone range from 10,840 ± 410 yr B.P. to 10,300 ± 240 yr B.P. (about 13,800 to 11,000 calendar years ago) (5,6).

M5 A4-67 (site 67) is another Las Vegas site located 15 km further inland than site 80 (6). Midden contents (chipped stone tools, shells, fish bones, and deer bones) support the idea that this is a habitation site. 14C dates range from ∼10,800 to 7250 yr B.P. (Table 1 and table S1) and indicate that occupations at sites 67 and 80 were contemporaneous (8). Although seeds, roots, and nuts were not preserved at sites 80 and 67 and pollen grains are rare, phytoliths were abundant.

Table 1

Phytolith age and size at Las Vegas sites OGSE-80 and M5A4-67. Under “sample number (n),” the first number indicates phytolith length and the second indicates thickness. In the “sample provenience” column, fs is fine silt, cs is coarse silt, and s is sand. All site 80 samples are listed in cm beneath datum (b.d.). Unit 1 column samples from site 67 are listed in cm b.s.; the others are in cm b.d. The date of 3810 ± 40 yr B.P. from site 67 at 3 to 10 cm b.s. is consistent with the presence of a few ceramic sherds dating to the late Valdivia period in superficial contexts at this site. Size ranges for phytolith length and laboratory numbers for radiocarbon dates are available as supporting material on Science Online.

View this table:

At site 80, we found the presence of Cucurbita fruit phytoliths throughout the Las Vegas occupation, and we suggested that wild Cucurbita was exploited during the latest Pleistocene and domesticated by ∼9000 yr B.P. (6,7, 9). However, our interpretations hinged on this single site, and our chronology was based on dating the carbon isolated from tens of thousands of phytoliths of different sizes and taxonomic classes occurring in the same samples withCucurbita (7–9). The dates were closely comparable to those from associated shell and charcoal but were not as precise as 14C analysis of individual macrofragments of domesticated plants might have been.

In the present study, we isolated and studied Cucurbitaphytoliths in three additional sediment samples from the terminal Pleistocene occupation at site 80, as well as 12 samples representing the earliest to latest cultural activity at site 67 (4,8). Ten samples from site 67 were from a column sample removed from a deep stratigraphic cut, Unit 1, which was located well away from burials; the remaining two were recovered from undisturbed contexts in other portions of the site as the excavations proceeded. To assess the status of the Cucurbita remains as wild or domesticated, we used a large modern reference collection comprising 157 mature fruits from 115 different populations, land races, and varieties from most known wild and all domesticated species ofCucurbita. We examined 18 fruits from 18 different populations of wild and semi-domesticated varieties from the only knownCucurbita native to Ecuador, Cucurbita ecuadorensis (10–12). We dated archaeological Cucurbita remains using refined methods of phytolith 14C study (8) (figs. S1, A and B).

All archaeological sediment samples yielded spherical scalloped phytoliths specific to the fruit rinds of the genusCucurbita (7, 9, 13,14). An increase in mean and maximum size above those recorded in modern wild plants is a standard criterion for separating wild from domesticated macrobotanical remains (such as seeds and peduncles) of Cucurbita archeologically (2). Our data show that size criteria are useful in phytolith studies, because mean and maximum phytolith length and thickness in modern wild species are substantially smaller than in domesticated varieties (Fig. 1, A and B). There is also a clear relation between phytolith size and fruit size (Fig. 1C). The sensitivity of phytolith size to domestication can also be seen in individual fruits from wild taxa (C. argyrosperma ssp.sororia and C. pepo ssp. fraterna) that exhibit signs of introgression with domesticated species, such as nonbitter flesh. These fruits have somewhat larger phytoliths than do wild examples, with no such evidence of domesticated germ plasm (Fig. 1, A and B, arrows). Moreover, fruits from different wild species and populations sampled over wide geographic areas have similar sizes.

Figure 1

(A) Length and (B) thickness of scalloped phytoliths in nine wild, six domesticated, and one semidomesticated species of Cucurbita. Names of domesticated species are shown in bold print. Each triangle represents the mean phytolith size from an individual fruit, usually from a separate population, for which at least 50 phytoliths were measured. Numbers above and below the triangles are size ranges recorded for each species. For detailed information on the localities where these plants were collected or grow, see (14). (C) Relationship between phytolith length and fruit breadth for three wild, two domesticated, and one semidomesticated species ofCucurbita. Pearson's correlation coefficient = 0.88;P < 0.001.

Phytoliths with a mean length of about 82 to 86 μm or longer and a mean and maximum thickness of at least 68 and 90 μm, respectively, are contributed only by semi-domesticated C. ecuadorensisfruits at least 14 cm in height or 16 cm in diameter. A maximum phytolith thickness of 100 μm or more was seen only in a C. ecuadorensis fruit that was 14 cm in height and 18 cm in diameter, and similarly in the largest fruits of C. moschata andC. ficifolia studied, with diameters of 17 to 25 cm and heights of 21 to 36 cm (15). Phytoliths of such sizes also distinguish all other domesticated from wild taxa and provide a secure baseline for identifying domesticated varieties of archaeologicalCucurbita in Ecuador.

In three different terminal Pleistocene samples from site 80, where14C ages on charcoal and shell range between 10,840 ± 410 yr B.P. and 10,300 ± 240 yr B.P., phytolith size is indicative of a wild species of Cucurbita (Table 1and table S1). In the three deepest samples from Unit 1 at site 67 [between 70 and 100 cm beneath the surface(b.s.)], where phytoliths are dated at or somewhat before 10,820 ± 250 yr B.P., phytolith size likewise falls within or close to the upper ends of the means and ranges of wild species. Because no other wild species ofCucurbita has been located in Ecuador, the evidence suggests that a wild form of C. ecuadorensis was present on the landscape and used by the pre–Las Vegas people.

Previous analyses of site 80 indicated that Cucurbitaphytoliths with sizes substantially exceeding those found in modern wild specimens made their first appearance near the base of the Las Vegas–phase midden in the 110- to 120-cm level of Unit E8-9 (6, 7, 9). Our initial determination of phytolith age in this sample from a phytolith extract overwhelmingly represented by grass phytoliths isolated from the fine (5 to 20 μm) and coarse (20 to 50 μm) silt was 9080 ± 60 yr B.P. (7, 9) (Table 1, box). Our new results indicate that this date is somewhat too young. WhereCucurbita phytoliths are large (with a mean length of 80 μm or more) and numerous, they can be consolidated in large numbers nearly entirely in the sand fraction of a sample, in such a way that few other types of phytoliths are present. We were thus able to segregate about 70% of all the Cucurbita phytoliths from Unit E8-9, 110- to 120-cm level. Carbon isolated from approximately 550 domesticated-sized Cucurbita phytoliths and 3200 large grass phytoliths concentrated from the sand yielded a date of 10,130 ± 40 14C yr B.P. (Table 1, box) (8).

Another date on an isolate of coarse silt phytoliths from this sample, in which the remainder of the Cucurbita occur, was 9320 ± 250 yr B.P. Hence, an age of between ∼10,100 and 9300 yr B.P. can be assigned to the appearance of a domesticatedCucurbita. A new date of 7990 ± 40 yr B.P. on the fine silt (5- to 20-μm) phytolith fraction (Table 1, box), in which no Cucurbita phytoliths occur (these were never observed in any fine silt fractions), is younger than that of the coarse silt and sand isolates, thus revealing some variability in the micromovement of phytoliths of different sizes in sediments. Because a previous age determination of 9740 ± 60 yr B.P. on phytoliths from Unit F8-9, 110- to 120-cm level, was carried out on fine and coarse silt phytoliths (here 90% of Cucurbita phytoliths occur in the coarse silt), the true age of the Cucurbita in this level near the base of the midden is probably older, an inference which is suggested by their size as well (Table 1).

The results from site 67 are similar. In the Unit 1 column samples, the earliest Cucurbita phytoliths that are identifiable as derived from a domesticated species occur from 60 to 70 cm b.s. Those from this and stratigraphically higher levels segregate almost entirely in the sand, but they are less densely distributed than in site 80, and separate sand isolations from individual levels could not be dated. Coarse silt phytoliths combined from 50 to 60 and 60 to 70 cm b.s. of Unit 1 have an age of 8240 ± 170 yr B.P. An earlier Holocene age is therefore likely for the 60- to 70-cm b.s.Cucurbita remains, a scenario supported by a 8980 ± 40 yr B.P. age on a phytolith assemblage containing domesticatedCucurbita from Unit E4, 110- to 120-cm level.

The 5900 ± 40 yr B.P. phytolith age from a combination of equal amounts of sand fraction phytoliths from 3 to 10 and 20 to 30 cm b.s. of Unit 1, where coarse silt phytolith ages were 3810 ± 40 and 7250 ± 190 B.P., respectively, accurately reflects a mixture of phytoliths from these two time periods. This finding further indicates that downward displacement of larger-sized phytoliths is negligible, because it should have resulted in a much younger age determination for the phytoliths combined from 3 to 10 and 20 to 30 cm.

The morphologies of the Cucurbita phytoliths documented at sites 80 and 67 closely conform to those from modern wild and semidomesticated C. ecuadorensis, and the earliest Holocene domestication probably resulted from the husbandry of a local wildC. ecuadorensis. Some land races of C. moschata, including those collected in southwest Ecuador (10), contribute phytoliths whose size and morphological characteristics make them indistinguishable from semi-domesticated C. ecuadorensis. Therefore, contribution from both C. ecuadorensis and C. moschata, with postulated origins in tropical lowland Colombia (16), should not be ruled out, particularly during the later stages of the Las Vegas preceramic occupations. In addition to Cucurbita, phytoliths from the bottle gourd (Lagenaria siceraria) andCalathea allouia, a root crop used casually today in indigenous neotropical horticulture, are present at sites 80 and 67 as early as ∼9300 yr B.P. (7, 9).

Our findings indicate that an intensification of plant use in broad-spectrum foraging economies, leading to plant food production, arose independently in tropical northern South America shortly after the termination of the Pleistocene. Seeds of Cucurbita spp. are oil- and protein-rich and were favored by some of the latest foragers and earliest farmers in Mesoamerica (1,2) and South America. Our data also indicate that Las Vegas horticulture prepared the way for the intensification of agriculture based on major seed and root crops during the early ceramic Valdivia period (beginning ∼5500 yr B.P.) in coastal Ecuador (5,17, 18). The domestication of local varieties ofCucurbita in southwest Ecuador supports views (19) that in South America there was no single center of agricultural origins.

Supporting Online Material

Materials and Methods

Fig. S1

Table S1


View Abstract

Navigate This Article