Rice finds in Thailand
During the course of my research, I have identified 29 sites in Thailand (Fig. 1) which report rice finds dating from the Hoabinhian to the Late Prehistoric period (Fuller et al. 2010 online supplement). There are differences in the data quality dependent on the accuracy of interpretation and these were taken into account. From these sites, only nine have evidence of rice resulting from flotation and an additional eight from phytolith analysis. In the case of rice, the presence of rice spikelet bases allows for the identification of the domestication status of rice.
The earliest sites where rice has been found show rice may have possibly been cultivated as early as the mid-Holocene in north and central Thailand (Kealhofer 2002). These inferences have been based on phytolith studies. The phytoliths from these sites were taken from sediment sequences in lake cores and alluvial deposits and the rice cannot be considered domesticated. Evaluating the status of domestication using phytoliths remains problematic (Fuller and Qin 2009; Fuller et al. 2010), though there are several scholars who believe it is possible (Saxena et al. 2006; Zhao et al. 1998). In order to assess the domestication status of rice, archaeobotanists examine the abscission scars found in rice spikelet bases, which can only be done with macroremains (Thompson 1997; Fuller et al. 2009). If one were to strictly adhere to the examination of rice spikelet bases, there would only be a handful of sites in Thailand that would positively yield evidence for domesticated rice. Most rice reports come from rice temper or impressions in pottery consisting mainly of husks and, like phytoliths, these data do not provide information on the domesticated status of the cereal.
The first evidence of domesticated rice in Thailand using macroremains dates to 2000–1500 BCE from the Neolithic period in the coastal site of KPD (Thompson 1996). KPD provides rice finds in the form of domesticated-type rice spikelet bases and weeds of cultivation. Higham (2002) originally proposed that rice agricultural expansion followed major riverine routes and would be archaeologically visible in interior sites, an idea previously put forth for Austroasiatic language expansion by Blust (1996). However, Ban Tha Kae and Ban Chiang are the earliest interior sites dating to the Neolithic and are reported to have rice cultivation, but the evidence is based on rice-tempered pottery, so it may be open to doubt. The first inland sites that provide reliable domesticated rice finds are NPW, NKH and NML in Lopburi. The rice finds at these sites date to the first millennium BCE and not earlier. Interestingly, these sites provide evidence that millets were cultivated before rice. The primary crop found in all three sites in the second millennium BCE was foxtail millet (Setaria italica) and the mode of cultivation was dryland farming (Weber et al. 2010). NPW has evidence of Setaria during the third millennium BCE signifying the introduction of millet cultivation at least a thousand years before rice at this site. This cereal originates from the north of China, though in the third millennium BCE, it was also evident in south China bordering Vietnam together with rice remains (Fuller et al. 2010).
In the Late Prehistoric period, more evidence for domesticated rice comes from samples from the Iron Age site BNW. These were floated and rice grain, spikelet bases, husk and weeds of cultivation have been identified. The Metal Age (400–200 BCE) sites KSK and PKT in the southern Peninsula have also yielded a large number of rice remains and associated weeds, as well as the Indian pulses Vigna radiata and Macrotyloma uniflorum. All three sites have domesticated japonica-type rice and possibly dryland and rainfed rice cultivation systems (Castillo and Fuller 2010).
Origins of rice
The widely held view is that rice in Southeast Asia came from China and that it was Oryza sativa spp. japonica. The linguistic evidence indicates that the original domesticators of rice, depending on the author of the hypothesis, were the Miao-Yao coming from south and central China (Blench 2005), the Austroasiatic speakers (Sagart 2005) or Austric speakers (Blust 1996; Higham 1996) coming from the Yangzi Valley. Movements of agriculturalists have also been proposed such as the Austronesians from Taiwan to the Philippines and further southwards (Bellwood 2007) and the Tibeto-Burman into northern China (van Driem 1998). The archaeological evidence consistently points to the Yangzi valley as the area where rice was first domesticated (Fuller et al. 2007, 2010; Nakamura 2010; Zhao 2010). However, which group of people brought rice cultivation to Thailand remains a matter of debate. Unfortunately, archaeological work in the region does not assist due to the lack of archaeobotanical sampling (Castillo and Fuller 2010). There are not enough rice finds to permit geographic and chronological resolution for a clear picture of the diffusion of rice cultivation to emerge.
Genetic studies remain divided as to whether rice domestication had a single origin or multiple origins (He et al. 2011; Molina et al. 2011; Sang and Ge 2007). The multiple-origins model proposes two centres of domestication, one in China ca. 4000 BCE and the other in South Asia ca. 2000 BCE (Fuller 2007). The single-origin model considers indica to be a hybrid of japonica rice and therefore, the origin lies in China even though its development or expansion occurred in India (Molina et al. 2011). The archaeobotanical evidence in Thailand does not corroborate one or the other. It does, however, point towards a largely japonica-type variety in prehistory and using either model ultimately shows that rice in prehistoric Thailand until at least the Iron Age has its origins in China.
The morphometric analyses of rice grains from four Thai archaeological sites (KSK, PKT, BNW and Noen U-Loke [NUL]) suggest that rice in prehistoric Thailand was Oryza sativa japonica (Fig. 2). The length–width (L/W) ratios of these rice grains were compared with those of modern populations of domesticated and wild rice. According to Ahn (1993), L/W ratios are not affected by charring so ancient and modern rice should therefore be comparable. Indica rice normally has a L/W ratio above 2.5, whereas japonica rice is below 2.3 (Fuller et al. 2009). The prehistoric rice measured all come from Iron and Late Metal Age sites from the NE region and southern Peninsula of Thailand (map 1) indicating that ca. 400–200 BCE, it was the Chinese rice subspecies japonica that was being consumed and cultivated. However, genetic studies are needed to confirm this view as morphometrics is just a first step towards identification. At present, there has been no DNA fingerprinting of prehistoric rice in Thailand.
The first rice samples from the sites of BNW, NUL and KSK were sent to Japan in July 2011 and will be analysed using DNA chloroplast and nuclear genome markers. The analysis will, hopefully, provide information on rice variety (indica, tropical japonica, temperate japonica) and whether it was the waxy or sticky type of rice.
The morphometric analysis above indicates that at the four sites (KSK, BNW, PKT and NUL) during the late prehistoric period, japonica was the type of rice found across Thailand. But several questions arise. When did indica become the dominant rice variety; what agricultural regime was practised in prehistory (wetland vs. dryland cultivation); was the cultivation technique also brought in with the introduction of rice or was it a local innovation?
Cultivation systems and weeds of cultivation
Today, indica is the dominant rice type and rainfed cultivation is the main agricultural system practised in Thailand. Cultivation today uses bunded fields inundated by retaining rainwater and allowed to dry naturally. Rainfed systems of cultivation were most likely practised during prehistoric times as well. White (1995) proposes that in Thailand, both wetland and dryland rice cultivation evolved from inundated rice cultivation, a less labour demanding technique than irrigated cultivation. This is true in the case of the low-lying coastal site KPD during the Neolithic where rice cultivation is believed to have been dependent on natural flooding at a nearby swamp (Thompson 1996).
To define systems of land use and cultivation practices, archaeobotanists have relied on the weed flora associated with economic crops because weed species occur in certain ecological zones, are displaced travelling with particular crop packages and help identify crop processing stages (Bogaard et al. 1999; Colledge 1994; Colledge et al. 2005; Fuller and Qin 2009; Jones 2002; Kealhofer and Piperno 1994). Unfortunately, macroremains from the Khao Wong Prachan valley sites (NKH, NPW and NML) dating to the Bronze Age do not contain sufficient numbers of weed seeds to define the rice agricultural regime and such weed seeds as are found provide ambiguous results. Sedges normally associated with wetland rice as well as dryland weeds such as chenopods are found in the samples (Weber et al. 2010). Furthermore, prior to rice cultivation, millets were being cultivated in dryland conditions, potentially signifying a continuum in the cultivation practice for rice in the area.
During the Metal Age at KSK, the majority of the weed assemblage comes from dryland habitats. The predominant weed is Spilanthes acmella belonging to the Asteraceae family (Fig. 3). It is significant in that 94% of the samples with rice contained this weed representing a high level of co-occurrence. Furthermore, S. acmella is reported to be a weed of rice throughout Indonesia, Bangladesh, India, Philippines, Sri Lanka and Thailand (Moody 1989) found in rainfed and upland fields (Soerjani et al. 1987). The other crops found at KSK, such as foxtail millet and the mungbean, are also indicative of dryland cultivation systems and are drought-resistant and found mainly in upland cultivation systems. So it is clear from the weeds and other cultivars that the rice cultivation system at KSK was dryland.
A similar assemblage of pulses and weeds associated with rice at KSK is also found at PKT, another Metal Age site in the southern Peninsula. The weed assemblage at the Iron Age site BNW in northeast Thailand also indicates dryland cultivation. It appears that rice cultivation in Thailand during the Metal Age was rainfed and upland. As a point of comparison, a geomorphological study in Kedah situated in the Thai–Malay Peninsula hypotheses communities in the first millennium CE being dependent on dryland cereal cultivation and not irrigated rice agriculture (Allen 1991).
The cultivation practices inferred from prehistoric sites in Thailand and one in the Thai–Malay Peninsula discussed above differ from the lowland paddy field agricultural system that was in place at the centre of origin in the Lower Yangtze when rice spread outwards to other regions ca. 4000 BCE (Fuller and Qin 2009). This difference may be because wetland paddy field agriculture in Thailand developed later. Although the earliest paddy field agriculture is found in China, it is during the first millenium CE that indica together with wetland systems of cultivation may have been introduced into Southeast Asia from India as a result of exchange networks. In India, the expansion of rice agriculture occurs during the Iron Age and is linked to labour-intensive irrigated rice cultivation (Castillo and Fuller 2010; Fuller and Qin 2009; Shaw et al. 2007). It seems likely that during the early contact period with South Asia (300 BCE onwards), Thailand already had an established rice agricultural regime primarily focused on dry cropping in low-lying areas and the rice grown was japonica. KSK attests to this conclusion. It was after continuous contact with India that wetland systems of agriculture were developed.