Major leaps forward in understanding rice both in genetics and archaeology have taken place in the past decade or so-with the publication of full draft genomes for indica and japonica rice, on the one hand, and with the spread of systematic flotation and increased recovery of archaeological spikelet bases and other rice remains on early sites in China, India and Southeast Asia. This paper will sketch a framework that coherently integrates the evidence from these burgeoning fields. This framework implies a reticulate framework in the phylogeny of early cultivated rice, with multiple starts of cultivation (two is perhaps not enough) but with the key consolidations of adaptations that must have been spread through hybridisation and therefore long-distance cultural contacts.Archaeobotanical evidence allows us to document the gradual evolutionary process of domestication through rice spikelet bases and grain size change. Separate trends in grain size change can be identified in India and China. The earliest centre of rice domestication was in the Yangtze basin of China, but a largely separate trajectory into rice cultivation can be traced in the Ganges plains of India. Intriguingly, contact-induced hybridisation is indicated for the early development of indica in northern India, ca. 2000 BC. An updated synthesis of the interwoven patterns of the spread of various rice varieties throughout Asia and to Madagascar can be suggested in which rice reached most of its historical range of important cultivation by the Iron Age. The distribution of wild rice and genetic diversity in domesticated ricesRice is a highly diversified crop, being grown from the equator to over 40°N, from sea-level ca. 2,700 m in parts of the Himalayas and in a wide ecological range of cultivation systems. Although there is much less ecological variation found within its wild progenitor complex (Oryza rufipogon and Oryza nivara), these are nonetheless distributed over a wide geographical range and a spectrum of ecological niches from permanent to seasonal wetlands. As the origins of cultivation must have developed in places where hunter-gatherers were utilising wild populations, the distribution of the wild progenitor, in the past when cultivation began, is a key element in identifying the origins of rice. There are three lines of approach to inferring this without archaeological evidence, including (1) the Electronic supplementary material The online version of this article (

We review the origins and dispersal of rice in Asia based on a data base of 443 archaeobotanical reports. Evidence is considered in terms of quality, and especially whether there are data indicating the mode of cultivation, in flooded ('paddy' or 'wet') or non-flooded ('dry') fields. At present it appears that early rice cultivation in the Yangtze region and southern China was based on wet, paddy-field systems from early on, before 4000 bc, whereas early rice in northern India and Thailand was predominantly dry rice at 2000 bc, with a transition to flooded rice documented for India at c. 1000 bc. On the basis of these data we have developed a GIS spatial model of the spread of rice and the growth of land area under paddy rice. This is then compared with a review of the spread of ungulate livestock (cattle, water buffalo, sheep, goat) throughout the Old World. After the initial dispersal through Europe and around the Mediterranean (7000-4000 bc), the major period of livestock expansion is after 3000 bc, into the Sub-Saharan savannas, through monsoonal India and into central China. Further expansion, to southern Africa and Southeast Asia dates mostly after 1000 bc. Based on these two data sets we provide a quantitative model of the land area under irrigated rice, and its likely methane output, through the mid to late Holocene, for comparison to a more preliminary estimate of the expansion of methane-producing livestock. Both data sets are congruent with an anthropogenic source of later Holocene methane after 3000 bc, although it may be that increase in methane input from livestock was most significant in the 3000-1000 bc period, whereas rice paddies become an increasingly significant source especially after 2000 bc.

Archaeological surveys and excavations in the Jebel Oraf palaeolake basin, northwestern Saudi Arabia, have identified a well-preserved early-to mid-Holocene landscape. Two types of occupation site can be distinguished: nine small and ephemeral scatters from single occupation phases on the slopes of sand dunes and three hearth sites indicative of repeated occupation on palaeolake shorelines. In Highlights  Well preserved Neolithic landscape in the Nefud Desert of northern Saudi Arabia  Evidence for changes in landscape use across the transition to herding  Two lake high stands show seasonal use of the area by early pastoralists  Rock art documents the presence of wild camel in the Iron Age  Lithic industry indicative of contacts with the Levant over several millennia

No abstract