MicroCT reveals domesticated rice (Oryza sativa) within pottery sherds from early Neolithic sites (4150–3265 cal BP) in Southeast Asia

Rice (Oryza sativa) was domesticated in the Yangtze Valley region at least 6000–8000 years ago, yet the timing of dispersal of domesticated rice to Southeast Asia is contentious. Often rice is not well-preserved in archaeobotanical assemblages at early Neolithic sites in the wet tropics of Southeast Asia and consequently rice impressions in pottery have been used as a proxy for rice cultivation despite their uncertain taxonomic and domestication status. In this research, we use microCT technology to determine the 3D microscale morphology of rice husk and spikelet base inclusions within pottery sherds from early Neolithic sites in Vietnam. In contrast to surface impressions, microCT provides images of the entire husk and spikelet base preserved within the pottery, including the abscission scar characteristic of domesticated rice. This research demonstrates the potential of microCT to be a new, non-destructive method for the identification of domesticated plant remains within pottery sherds, especially in contexts where archaeobotanical preservation is poor and chaff-tempered sherds are rare and unavailable for destructive analysis. The method has the potential to greatly advance the understanding of crop domestication and agricultural dispersal for ceramic cultures in different parts of the world.

southern Vietnam were subjected to microCT analysis to determine if organic inclusions, or combusted impressions of organic inclusions, of domesticated rice (Oryza sativa) were present.
The archaeological sites date to the southern Vietnam Neolithic ( Fig. 1): An Son to c.4200-3150 cal BP 15 , Loc Giang from at least 4000-3300 cal BP 16 and Rach Nui to 3555-3265 cal BP 17 . Each site is an artificial mound averaging one hectare in extent, which rises 4-6 m above the surrounding alluvial or estuarine landscape, and contains artificially-laid floors with post-holes representing former timber constructions [15][16][17][18][19][20] . Habitation layers contain bones of domesticated pigs and dogs, together with decorated pottery, stone adzes and bone tools. An Son and Loc Giang have riverine locations, whereas Rach Nui is located in a mangrove-flanked estuary.
An Son and Loc Giang are currently the oldest excavated Neolithic sites in the greater Mekong region of southern Vietnam, dating to between c.4200 and 3200 cal BP. The stamped and incised pottery suggests widespread relationships with other regions of the Greater Mekong Basin, especially in Thailand, and ultimately with southern China [18][19][20] . Rach Nui is a slightly younger site, with cultural materials sufficiently different from those of An Son and Loc Giang (which have identical material cultures) to suggest occupation by a group with a different ethnic and economic orientation.
Sparse and often indeterminate evidence of rice has previously been reported for archaeobotanical and pottery assemblages from An Son, Loc Giang and Rach Nui. At An Son, a single rice husk was extracted from a poorly fired pottery sherd, which genetic analysis suggests is the domesticated species Oryza sativa ssp. japonica 15 . At Loc Giang, archaeobotanical research identified one domestic-type spikelet base, as well as multiple lines of evidence for rice of indeterminate domestication status, including husk impressions in pottery, two phytoliths and a charred rice husk. At Rach Nui, two domesticated spikelet bases and indeterminate rice husks were extracted from sediment samples 17,21 .
Here, multiple imaging methods of varying scales and resolution were employed to detect inclusions or impressions within pottery sherds with the intention of identifying rice (Oryza spp.) and to positively discriminate between wild and domesticated rice (Oryza sativa) (Fig. 2). Sherds were described visually at the macro-scale provided detailed images of husk impressions within the pottery surface, yet these were not diagnostic of domesticated rice because these types of images are unable to capture the 3D morphological characteristics resulting from human selection, namely the recessed abscission scar 22 . Husk morphologies are insufficient to accurately differentiate potential wild rices growing in the natural environment from domesticated rice cultivated by people. Problematically, most current identifications of putatively domesticated rice in pottery are based on surficial impressions of husks in Southeast Asia 23,24 and their domestication status is thereby unsubstantiated.
Sub-samples of each sherd were dissolved and processed to yield phytoliths of domesticated rice. However, no diagnostic rice phytoliths were identified in any sherd. Further, the discrimination of domesticated from wild rice using phytoliths is contentious, although increasingly used in East Asian research [25][26][27] .
A 1 cm off-cut from each pottery sherd was scanned at the 'National Laboratory for X-ray Micro Computed Tomography (CTLab)' based at the Australian National University (ANU) using a HeliScan MicroCT system to yield images at a resolution of 5-8 μm 28,29 . MicroCT data for each sherd was rendered using Drishti v2.3.2 and Drishti Paint v.2.6 software 30 to differentiate clay matrix from mineral and organic inclusions (Fig. 2D-F and Supplementary Animations S1-4) and to analyse morphological typography. Following digital isolation of the organic fraction within each sherd, individual organic inclusions of potential rice spikelet bases and husks were targeted for higher resolution processing. The resultant visualisations of individual rice husks (Supplementary Animations S5 and 6) and non-shattering spikelet bases (Supplementary Animations S7-9) were characteristic of domesticated rice in sherds from An Son and Loc Giang. Rice inclusions were not identified in sherds from Rach Nui, which contained mostly low-density non-organic inclusions, possibly reflecting mineral dissolution, and a seed-like inclusion with organic detail (Supplementary Animation S10).
The distinctive checkerboard patterning of the external surfaces of rice husks was clearly discernible in the optical (Fig. 2B) and SEM images (Fig. 2C), as well as the tomographs, of sherds from An Son and Loc Giang. MicroCT enabled high resolution visualisation of near-complete rice husks (Fig. 2G) and spikelet bases (Fig. 2H) within pottery sherds from these two sites, including a near-complete spikelet base with attached husk from Loc Giang (Fig. 3 & Supplmentary Animation S8). Spikelet base inclusions were compared to SEM reference images of domesticated, wild and immature rice spikelet bases (Fig. 4). All abscission scars on inclusions within pottery corresponded to the domesticated type, primarily due to irregular shape, concavity and size (Supplementary Animations 5-7).
The asynchronies between previous archaeobotanical data and the microCT analyses at these three sites may represent the vagaries of archaeological preservation of plant remains in the wet tropics. However, they may also indicate differences between communities cultivating and those importing domesticated rice. For instance, the absence of domestic rice inclusions within pottery made locally at Rach Nui, yet its reported presence in archaeobotanical assemblages there, may indicate people imported rice for consumption rather than growing it locally 30 . If grown locally, domesticated rice chaff, husks and spikelet bases would be expected to occur as temper within pottery made at the site. Anecdotally, local villagers there today suggest the estuarine environment was too saline to grow rice until recently. However, caution is needed as the limited application of modern tropical archaeobotanical methods in mainland Southeast Asia and the preliminary application of microCT to pottery assemblages precludes more refined interpretation.
In this study, microCT has been applied to positively discriminate domesticated rice inclusions within pottery sherds from two early Neolithic sites in Southeast Asia. The accurate study of rice inclusions and impressions within pottery opens up new possibilities for understanding the spread of domesticated rice from China to Vietnam and Thailand on mainland Southeast Asia, as well as Taiwan, the Philippines and Borneo in Island Southeast Asia. The dispersal of domesticated rice in Southeast Asia has previously been inferred using rice husk impressions in pottery or from phytoliths; although the accuracy of both methods for differentiating domesticated from wild rice is unclear. MicroCT is a reliable diagnostic technique for the discrimination of domesticated rice within pottery sherds and its application may be especially significant in the wet tropics where botanical preservation in archaeological contexts is often poor. The technique is non-destructive, thereby enabling the analysis of crop remains, as well as potentially other materials, within pottery at sites where only a few key sherds have been preserved, or for regions where archaeobotanical methods have not been systematically applied. Further, microCT offers broad potential for tracking the domestication of major crops in other regions, such as cereals and legumes in Africa and Southwest Asia, through the analysis of chaff-tempered pottery, as well as the subsequent dispersal of these crops.