Early agriculture in southeast Asia: phytolith evidence from the Bang Pakong Valley, Thailand.

Kealhofer, Lisa ; Piperno, Dolores R.

Phytoliths -- the microscopic opal silica bodies inside plant tissue that often survive well in archaeological deposits -- are becoming a larger part of the world of human palaeobotany. They give a new view of early rice in southeast Asia.

Introduction

In 1989 researchers working in the region of Khok Phanom Di, a 4000-year-old site in the Bang Pakong Valley, identified evidence of cultural burning and possibly early agriculture in the 5th millennium BC in sediment cores (FIGURE 1; Maloney et al. 1989). Phytolith analysis of these same cores provides a more detailed record of agriculture and grass weeds, and indicates that cultural modification of the environment may have begun even earlier.

Despite previous claims for early rice agricultural development in Thailand (Gorman 1973; 1977), abundant early evidence from southeastern China pre-dates Thai archaeological sites (An 1989; Yan 1991). Higham and co-workers investigated the coastal region of Thailand, particularly the fresh-water swamp zone, and the site of Khok Phanom Di for possible early agricultural development (Higham et al. 1992; Higham & Maloney 1989). While the site was first occupied at the beginning of the 2nd millennium BC, late relative to the beginnings of agricultural development (Higham & Bannanurag 1990), pollen cores from adjacent fields document human and natural environments back to the 6th millennium BC (Maloney 1992c). One core, KL2, showed several intense charcoal peaks pre-dating the site's occupation (5278[+ or -]420 BC (OxA-1359) at 4-95 m, 4950[+ or -]390 BC (OXA-1357) at 3.50-3.54 m). Slightly later, charcoal peaks are associated with weeds indicative of rice field cultivation (2.30 m, c. 4350[+ or -]375 BC (OxA-1356)). While rice is not directly identifiable from pollen data, the decline in mangrove species, increase in burning and increase in rice field weeds strongly suggest agriculture was practised in this region in the 5th millennium BC (Maloney et al. 1989: 367).

Methods

Two cores, KL2 and BMR2, from the Bang Pakong Valley were analysed for phytoliths. Both were located 170-200 m north of the site of Khok Phanom Di, within 30 m of each other. The pollen, spore, charcoal and sediment records are discussed in detail by Maloney (1992a; 1992b; 1992c; 1992e). The site is strategically located near the estuary of the Bang Pakong River, has access to marine, mangrove, riverine, fresh-water swamp, and alluvial plain/grassland resources. The diversity and abundance of subsistence resources made this an advantageous niche throughout the Holocene (Higham et al. 1992; Takaya 1979).

Phytolith extraction from sediment samples followed standard methods (Piperno 1988). Soils were disaggregated in [Na.sub.2] C[O.sub.3]. A 270-mesh sieve separated the sand fraction from the silts and clays. Clays were removed by gravity sedimentation, and the remaining silts were then fractionated, also by differential gravity sedimentation. The organics were wet ashed from these fractions with KCl[O.sub.3] (Schulze's solution). The phytoliths were floated (specific gravity [is less than] 2.3) on a density gradient of potassium and cadmium iodide (specific gravity 2.35). Samples were washed in water, dried in acetone, and mounted in permount. Phytoliths were counted at 400x on an Olympus photomicroscope while the permount was still fluid. Individual phytoliths were rotated to avoid confusion with similar two-dimensional forms.

Phytoliths were identified using modern reference material of over 340 species, comprising most of the southeast Asian families known to be phytolith producers (after Piperno 1988), as well as many previously untested Old World tropical taxa (Kealhofer & Piperno in press a). A major goal of this study was to retrieve and identify Oryza phytoliths from sediment cores.

Nearly a century ago German botanists studying grass morphology identified distinctive Oryza glume phytoliths (Formanek 1899; Grob 1896). Early archaeological applications include the identification of silicious remains of Oryza in Chinese Neolithic potsherds (Edman & Soderberg 1929). Studies in the 1960s also described these 'hollow swellings with acute tips' (Watanabe 1968), or short distinctive hair cells, from rice glumes.

Comparative grass phytolith studies of both wild and domesticated grasses have revealed no redundant hair shapes in over 500 species of tropical American grasses (Piperno & Pearsall unpublished data), North American grasses (Brown 1984; Twiss et al. 1969; Twiss 1992), Canadian grasses (Blackman 1971), British grasses (Parry & Smithson 1964; 1966), East African grasses (Palmer & Tucker 1981; 1983; 1985) or in over 100 species of Old World grasses most closely aligned with Oryza (Pearsall et al. in press). Morphological studies also reveal no forms mistakable for rice bilobates and glume hair cells (Clifford & Watson 1977; Metcalfe 1960; Terrell & Wergin 1981).

In identifying rice phytoliths, we relied in part on recent work by Pearsall and co-workers (in press), describing the glume and leaf phytoliths of the genus Oryza. We also independently analysed 50 species of Thai grasses, as well as multiple species and replicates of Oryza. Our analyses confirm that the phytoliths thought diagnostic of Oryza are valid markers of rice. Additional Oryza specific (i.e. genus specific) phytoliths were revealed, including distinctive bilobates and flat lobed bodies from leaves, and 'bottle'-shapes and 2-celled hairs from glumes. Domesticated Oryza sativa and its putative wild ancestor Oryza rufipogon/nivara can be statistically differentiated in large samples (Pearsall et al. in press), but this is problematic in soil contexts.

Comparisons of phytoliths from the two series of Oryza, Latifoliae and Sativae (Shastry & Sharma 1974), show they can be easily differentiated by the same generic indicators. The distinctive bilobate shapes vary in size, shape and proportion between species, as does the presence of additional morphologically distinctive bilobates. The Sativae series (A genome) of Oryzeae includes the African and Asian domesticated rices and their progenitors, as well as one New World species. Within this series, the African and southeast Asian rices can be distinguished, by glume short hair morphology, leaf bulliform shape, and bilobate size and diversity. The rice phytoliths identified here all fall within the Sativae series of Oryza.

Twenty-two samples from the KL2 core were analysed for phytoliths. All but one of these samples (from a depth of 2[center dot]28 m) contained quantifiable phytolith concentrations. Fourteen samples were analysed from the BMR2 core, and 13 of these contained phytoliths in quantifiable concentrations. Detailed summaries of both of these sequences are published elsewhere (Kealhofer & Piperno in press b), and the focus here is on KL2 and its relevance for early agriculture in Thailand. The chronology is based on dates cited in Maloney (1992: 61) for KL2 samples.

Pollen and phytolith quantification procedures often rely on influx, or pollen or phytolith concentrations per cubic centimetre of sediment. Terrestrial and alluvial environments are not continuously deposited, including not only periods of erosion, fut episodes of variable deposition; these factors preclude a determination of influx, so the phytolith counts are presented as percentages per sample.

The phytolith record

KL2 samples from 6.0 m to 4.35 m show forest grasses (Bambuseae), a few intrusive grasses (Panicoideae), palms, tress, and a small percentage of sedges (Cyperaceae) and Compositae. Fragments of Oryza glumes occur at 5.0-5.1 m and at 4.5 m. Early in this period a shift in the arboreal assemblage (c. 5.45-5.50 m: decline in large smooth spheres, increase in large rough spheres) is correlated with a decline in charcoal, and most grass types (some Bambuseae increase). The mangrove forest dynamics of this shift are not clear from the phytoliths; arboreal pollen displays a peak in Rhizophora (65%), and a decline in Bruguiera (25%), suggesting a shift toward a slightly more coastal and saline mangrove forest. After this short-lived (c. 100 years?) perturbation, conditions soon return to the slightly drier Bruguiera/Ceriops forest. Charcoal, not abundant in these phytolith samples, is most common in this phase. Maloney (1992) found the largest peak in burning at 4.95 m.

At 4.45-4.35 m (c. 7000 BP) there is an abrupt change. The arboreal indicators virtually disappear, while Bambusoideae (and some Chloridoideae) grasses peak; then forest indicators recover immediately. At the same time a sharp increase in grasses is associated with disturbance and agricultural fields (Panicoideae, including Andropogoneae). Oryza glume fragments and leaf phytoliths present in this and the subsequent level (4.17 m) are associated with the weedy agricultural grasses. The small sample size precludes the possibility of differentiating their status as wild or domesticated Sativae. The association of Oryza (Sativae series) with field weeds constitutes strong circumstantial evidence that rice was cultivated. The presence of freshwater, necessary for rice, is confirmed by Podostemaceae, a moss found in submerged riverine habitats.

The Bambusoideae grass phytoliths (regular saddles), associated with the earlier more saline phase, peak during the arboreal decline, and just prior to a major shift. This may represent natural secondary growth after forest destruction, as bamboos are common in many forest regrowth successions. Carbon is absent from both phytolith and pollen profiles at the onset of this weedy interval. Evidence for human impact lasts until about 4.0 m, when Oryza disappears and the weedy indicators of human activity decline.

Little evidence for human disturbance is found between 3.5 m and 1.9 m. Arboreal indicators show little variation. Grass species' composition fluctuates, but without clear trends. A small percentage of Oryzeae and Panicoideae types are present, as in the deeper samples. These may mark the presence of human activity in the general area, or they may be part of the riverine grass assemblage, where some low level disturbance would be expected (Maloney 1992d). Maloney's (1992d: 66) carbonized particle count, relatively high during this phase, is intermediate between the early and later peaks.

At 1.9 m (c. 5000 BP) agricultural weeds once again appear. This time they dominate the assemblage, slightly after the increase in Gramineae in the pollen profile (c. 2[center dot]25 m). The pollen are likely to be sampling a larger region than the phytoliths, suggesting a disturbance which intensifies in the immediate area around 5000 BP. Arboreal indicators decline simultaneously, subsequently contributing only a very small percentage. A small peak in arboreal types, including palms, near 90 cm, correlates well with pollen evidence for a brief recovery in Rhizophora. Oscillations in grass subfamilies (cf. crosses) possibly indicate shifts in the crops grown, as they do not correlate with changes in the mangroves. The grass type (regular saddle) previously associated with arboreal shifts declines through to the present. Other Bambusoideae species appear just when crosses decrease, in the 50 cm sample. Carbonized particles in the pollen samples were very high 2[center dot]0-1[center dot]0 m (25-50K; Maloney 1992d: 66). From 1[center dot]0 m, to the end of the sequence, Podostemaceae again confirm freshwater in the immediate locale.

Oryza phytoliths, common in the 1[center dot]25 m sample (before 1650 BC, c. 2000 BC?), reach their highest percentage above 90 cm (after 1650 BC; cf. 1670[+ or -]240 BC (OxA-1354)). In the greater quantity of remains, and the size range of glume hair cells, these phytoliths are consistent with those in domesticated Oryza sativa glume samples.

The adjacent five-hectare site of Khok Phanom Di was first occupied around 2000 BC (Higham & Bannanurag 1990). Rice remains are found throughout the deposits (Higham 1989: 28; Thompson in press) as chaff in cultural layers, as chaff impressions on pottery, and in faecal matter. According to the phytolith sequence, agriculture was practised extensively in the region prior to the occupation of Khok Phanom Di. With the reappearance of Oryza, intensification of rice cultivation correlates with site settlement. In the sediment cores, agricultural activity seems to begin about 1000 years before the occupation of Khok Phanom Di. The location of earlier habitation may have been affected by shifting sea levels, or simply related to demographic in-filling eta geographically limited but productive zone.

A characteristic of this assemblage is the apparent inverse correlation of agricultural activities with burning in the phytolith samples. Charcoal peaks identified by Maloney (1992) were not as evident in the phytolith samples. The peaks correlate with higher percentages of bamboos and lower percentages of agricultural weeds and rice, as if burning was related not to agricultural techniques, but to other cultural activities (such as mangrove charcoal production?).

The evidence for rice agriculture is not abundant until about 2000 BC. The earlier Oryza phytoliths may be cultural, or may represent a natural occurrence of wild rice in fresh-water backswamps, associated with the river's flood regime. The contemporary increase in Panicoideae phytoliths may represent either other cultigens (e.g. millet or Job's Tears) or field weeds, but strongly suggests agricultural activities. No Panicoideae grasses are present within mangrove habitats; however they are found in back mangrove swamps (e.g. Eragrostis sp.). With increasing definition of grass species in the phytolith type collections, individual species may he recognizable in the future.

Discussion

The phytolith record enlarges on the pollen and geomorphological study of these same cores and supports Maloney and co-workers' (1989) interpretation: agriculture in coastal Thailand, where the back mangroves meet the alluvial plain, is indicated by field weeds in samples from the end of the 6th millennium BC. While we cannot, at present, identify domesticated rice, the complement of weedy species mirrors those found in the present rice agricultural regime.

The phytolith record documents a pattern of vegetation changes, more detailed and localized than in the pollen. The fingerprint of human activities, both from burning and agricultural weeds, is evident from near the beginning of this 8000-year-old sequence. The burning is more likely related to fuel production and use of the mangroves than to early rice agriculture.

Despite a suggested sea level transgression from 5000-4000 years ago (Geyh et al. 1979; Maloney 1992b), with an expansion of Rhizophora mangrove near the site, both cores reveal a shift toward intensive cultivation and a decline in mangrove indicators. Tectonic uplift, suggested by Pramojanee & Hastings (1983), perhaps raised the coastal plain. A recent geemorphological study of the site indicates it was c. 20 km inland on a stream levee in the 3rd millennium BC (Aitken 1992).

Two areas were potential centres for the development of rice agriculture in Thailand based on early dates, ecology, and the presence of rice: the Khorat Plateau (White in press) and the coastal fresh-water swamp zone (Higham et al. 1992). The Khorat Plateau, in northeastern Thailand, falls within the region rice geneticists have suggested as the 'homeland' of domesticated rice (Chang 1976) from the distribution of genetic diversity and ecological factors. This 'homeland' runs from coastal southeastern China to eastern India, in the river basins that dissect the Himalayas.

The larger region of central Thailand, outside the 'homeland', does not provide abundant habitats for annual Oryza species, although the confluence of alluvial plain with fresh-water swamp near Khok Phanom Di would have been suitable for early Oryza annuals (Takaya 1979). In this zone, rivers commonly overflow in the rainy season and then dry up. Importantly, the presence of rice 7000-8000 years ago, even if not domesticated, shows wild precursors were present, and available for human manipulation, in this region in the early Holocene.

References

AITKEN, J.J. 1992. Archaeological sediments as artefacts. Unpublished Ph.D thesis, University of Otago, Dunedin.

AN, Z. 1989. Prehistoric agriculture in China, in Harris & Hillman 1989: 643-9.

BLACKMAN, E. 1971. Opaline silica in the range grasses of southern Alberta, Canadian Journal of Botany 49: 769-81.

BOZARTH, S. 1986. Diagnostic opal phytoliths from rinds of selected Cucurbita species, American Antiquity 52: 60715.

BROWN, D. 1984. Prospects and limits of a phytolith key for grasses in the central United States, Journal of Archaeological Science 11: 221-43.

CHANG, T.T. 1976. The origin, evolution, cultivation, dissemination, and diversification of Asian and African rices, Euphytica 25: 425-41.

CLIFFORD, H.T. & L. WATSON. 1977. Identifying grosses: data, methods, and illustrations. Canberra: University of Queensland Press.

EDMAN, G. & E. SODERBERG. 1929. Auffindung von Reis in einer Tonscherbe aus einer etwas funftausendjahrigen Chinesischen Siedlung, Bulletin of the Geological Society of China 8: 363-5.

FORMANEK, J. 1899. Uber die Erkennung der in den Nahrungs-und Futtermitteln Vorkommenden Spelzen, Zeitschrift fur Untersuchung der Narhrungs-und Genussmittel 11: 833-43.

GEYH, M.A., H.-R. KUDRASS & H. STREIF. 1979. Sea level changes during the Late Pleistocene and Holocene, Nature 278: 441-3.

GORMAN, C. 1973. Excavations at Spirit Cave, North Thailand, Asian Perspectives 13: 79-107.

1977. A priori models and Thai prehistory: a reconsideration of the beginnings of agriculture in southeast Asia, in C.A. Reed (ed.), Origins of agriculture: 321-55. The Hague: Mouton.

GROB, A. 1896. Beitrage zur Anatomie der Epidermis der Gramineenblater, Bibliotheca Botanica 7:36

HARRIS, D.R. & G.C. HILLMAN (ed.). 1989. Foraging and farming: the evolution of tlant exploitation. London: Unwin Hyman.

HIGHAM, C. 1989. Social organization at Khok Phanom Di, central Thailand, 2000-1500 BC, Arts Asiatiques 44: 25-43.

HIGHAM, C. & R. BANNANURAG. 1990. The excavation of Khok Phanom Di, a prehistoric site in central Thailand 1: The excavation, chronology, and human burials. London: Society of Antiquaries. Reports of the Research Committee of the Society of Antiquaries of London 47.

1992. The excavation of Khok Phanom Di, a prehistoric site in central Thailand 2 (Part I): The biological remains. London: Society of Antiquaries. Reports of the Research Committee of the Society of Antiquaries of London 48.

HIGHAM, C., R. BANNANURAG, G. MASON & N. TAYLES. 1991. Human biology, environment, and ritual at Khok Phanom Di, World Archaeology 24(1): 35-54.

HIGHAM, C. & B. MALONEY. 1989. Coastal adaptation, sedentism, and domestication: a model for socio-economic intensification in prehistoric south-east Asia, in Harris & Hillman 1989: 650-66.

KEALHOFER, L. & D.R. PIPERNO. In press a. Phytoliths from modern south-east Asian flora, Review of Paleobotany and Palynology.

In press b. Two phytolith sequences from the Bang Pakong Valley, Thailand: Cores KL2 and BMR2, in Thompson (in press).

MALONEY, B. 1992a. Palaeoenvironments of Khok Phanom Di: the pollen, Pteridophyte spore, and microscopic charcoal record, in Higham & Bannanurag 1992 Part 1: 7-20.

1992b. Khok Phanom Di: the physical environment, in Higham & Bannanurag 1992 Part 1: 21-44.

1992c. Sedimentary stratigraphy and dating, in Higham & Bannanurag 1992 Part 1: 53-62.

1992d. Pollen analysis: results and discussion, in Higham & Bannanurag 1992 Part 1: 63-90.

1992e. The pollen, Pteridophyte spore, and microscopic charcoal: conclusions, in Higham & Bannanurag 1992 Part 1: 91-4.

MALONEY, B., C. HIGHAM & R. BANNANURAG. 1989. Early rice cultivation in southeast Asia: archaeological and palynological evidence from the Bang Pakong Valley, Thailand, Antiquity 63: 363-70.

METCALFE, C.R. 1960. Anatomy of the Monocotyledons I: Gramineae. London: Oxford University Press.

PALMER, P.G. & A. TUCKER. 1981. A scanning electron Microscope survey of the epidermis of East African grasses I. Washington (DC): Smithsonian Institution Press. Smithsonian Contributions to Botany 49.

1983. A scanning electron Microscope survey of the epidermis of East African grasses II. Washington (DC): Smithsonian Institution Press. Smithsonian Contributions to Botany 53.

PALMER, P.G., S. GERBETH-JONES & S. HUTCHINSON. 1985. A scanning electron Microscope survey of the epidermis of East African grasses III Washington (DC): Smithsonian Institution Press. Smithsonian Contributions to Botany 55.

PARRY, D.W. & F. SMITHSON. 1964. Types of opaline silica depositions in the leaves of British grasses, Annals of Botany 28: 169-85.

1966. Opaline silica in the inflorescences of some British grasses and cereals, Annals of Botany 30: 525-38.

PEARSALL, D., D.R. PIPERNO, E.H. DINAN, M. UMLAUF, Z. ZHAO & R.A. BENFER, JR. In press. Identifying rice, Oryza sativa, Poaceae, through phytolith analysis, Journal of Archaeological Science.

PIPERNO, D. 1988. Phytolith analysis: an archaeological and geological perspective. San Diego (CA): Academic Press.

PRAMOJANEE, P. & P.J. HASTINGS. 1983. Geomorphological and palynological investigation of sea level changes in Chanthaburi, south-east Thailand, in N. Thiramongkol & V. Pisutha-Arnond (ed.), Proceedings of the First Symposium on the Geomorphology and Quaternary Geology of Thailand: 35-51. Bangkok: Chulalongkorn University.

SHASTRY, S.V.S. & S.D. SHARMA. 1974. Rice, in J. Hutchinson (ed.), Evolutionary studies in world crops: 55-61. London: Cambridge University Press.

TAKAYA, Y. 1979. Agricultural development of a tropical delta: a study of the Chao Phraya Delta. Honolulu: University of Hawaii Press.

TERRELL, E.E. & W.P. WERGIN. 1981. Epidermal features and silica deposition in lemmas and owns of Zizania (Gramineae), American Journal of Botany 68(5): 697-707.

THOMPSON, G.B. In press. Excavation of Khok Phanom Di 4: The biological remains (Part II): Environment and subsistence: the botanical evidence. London: Society of Antiquaries.

TJIA, H.D., S. FUJII & K. KIGOSHI. 1983. Holocene shorelines of Tioman Island in the South China Sea, Geologie en Mijnbouw 62(4): 599-604.

TWISS, P.C. 1992. Predicted world distribution of [C.sub.3] and [C.sub.4] grass phytoliths, in G. Rapp & S. Mulholland (ed.), Phytolith systematics: 113-28. New York (NY): Plenum.

TWISS, P.C., E. SUESS & R.M. SMITH. 1969. Morphological classification of grass phytoliths, Soil Science Society of America Proceedings 33: 109-15.

WATANABE, N. 1968. Spodographic evidence of rice from prehistoric Japan, Journal of the Faculty of Science, University of Tokyo, (section V) 3(3): 217-35.

WHITE, J. In press. Modeling the development of early rice agriculture: ethnoecological perspectives from northeast Thailand, Asian Perspectives.

YAN, WEN-MING. 1991. China's earliest rice agricultural remains, Bulletin of the Indo-Pacific Prehistory Association 10: 118-26.