Soybean archaeobotany: multiple origins and not coincident with cereals

A important paper on Soybean domestication (subtitle "Does size matter?") was published last month in PLOSone by G-A. Lee, Gary Crawford and colleagues. This reports details on the morphometrics of more than 900 archaeological soybeans across 22 sites in (northern) China, South Korea and Japan, including 7 directly dated by AMS. This provides the first really good archaeological dataset for making inferences about soybean domestication, and includes a summary of the soybean metrics from Wangchenggang published in a Chinese monograph in 2007 by Zhao Zhijun (and in more detail his 2010 book collection) , previously suggested to indicate some size increase by the Longshan period (i.e. 2500-1900 BC). The data reported here suggests large, truly domesticated soybeans present in middle Jomon Japan Shimoyakebe (near Tokyo), from the Third Millennium BC. These are significantly larger than those from the Yellow River valley oif similar age, the Longshan period, which are also probably enlarged by selection under cultivation (at least the population from Wangchengang appear enlarged while some other Longshan samples still fall in the wild type range). In Korea a measured population form the Middle Chulmun is perhaps marginally enlarged while those of the later Mumum had clearly undergone selection for size increase. The overall impression that selection for seed size increase in soybeans was a protracted process and one that was uneven in different regions, and it may be that in some areas wild population continued to be exploited or proximity of wild populations and early cultivation methods did not lends themselves to selecting for larger seed size. This then complements data emerging from several crops for a protracted process of evolution of domestication traits (see previous, for example my articles in Annals of Botany or Evolution, and for a discussion of protracted domestication processes in the New World tropics see the recent Current Anthropology by Piperno). It also strongly points to a Jomon domestication independent from that in northern China, and one in which seed size evolved more quickly. While genetic publications present a cacophony of single versus multiple domestication models for the soybean, those favouring a single genetic origin tend to rely on genome-wide-markers and simple tree-building, or bottleneck models, e.g. Guo et al 2010 (which has been shown to be problematic, e.g. by Allaby et al. 2008 and 2010), and those favouring multiple origins, including a separate Japanese origin seem more realistic and have support from chloroplast polymorphisms, e.g. Abe et al 2002; Xu et al 2002).

For those looking for generalizations about agricultural origins, East Asian soybean has long been taken as an east Asian equivalent of the pulses and beans of Near Eastern, as the legume component of cereal-legume package of early agriculture. Proponents of a single centered origin of agriculture in the northern Fertile Crescent of the Near East, for example Abbo et al (2010), point to the overlap zone of wild wheats, barley, lentil, pea, chickpea, as the core area of domestication in part on the assumption that this package of crops must have originated together. I am among those who disagree (e.g. in World Archaeology or J. of Experimental Botany), and see this as a package that came together piecemeal and gradually. In East Asia, there is now systematic archaeobotany that points to no initial linkage between soybean and early cultivated millets of the north nor rice of the Yangtze. Soybean were a later addition, gathered wild by some and cultivated by some some, perhaps not much earlier than the third millennium BC. 

See also this news piece from the University of Oregon. See also the excellent overview on Korean archaeobotany by Lee in a recent Current Anthropology, which also has reviews on Japan by Crawford, and China by Zhao Zhijun.

International Workshop for African Archaeobotany

The 7th international workshop for African Archaeobotany, and important small gathering every 3 years, now has a website for its next meeting, next summer in Vienna. Looking forward to it.

Korean seed atlas discovered

Not sure how long this has been available but new to me is the on-line seed Atlas of Korean wild plants, which should be of use for in generally for archaeobotanical work in East Asia.

New online Phytolith reference collection

A new searchable database of l phytoliths images has been made available by the Barcelona lab GEPEG (Research Group for Palaeocological and Geoarchaeological Studies). Find it here: GEPEG phytolith database. I have not  had a chance to explore it in detail, but it will doubtless prove useful. It has about 185 images in it, so it by no means comprehensive, and it includes archaeological as well as modern reference material. (And it reminds me that our UCL phytolith on-line gallery needs updating, as I have let in languish for a few years without additions).

Recognizing immature millets

Recently published on-line is an highly significant, but rather unassuming paper, about variation in millet grains due to immaturity. Motuzaite-Matuzeviciute et al from the Cambridge archaeobotany lab report on "Experimental approaches to understanding variation in grain size in Panicum miliaceum" in Vegetation History and Archaeobotany.  One of the major conclusions is that immature grains are likely to preserve archaeologically and contribute small grains to samples, and their recognition is important for archaeobotanical interpretation. This study vindicates the recognition and quantification of immature millet grains, indeed both immature Setaria italica and Panicum miliaceum have been recognized and reported from recent work in China (starting from Fuller and Zhang (2007)'s report on Ying valley survey samples [pdf from here], although it did not provide adequate illustrations). Immature grains have often is overlooked or lumped with other small grasses, such as Digitaria sp. as indeterminate "panicoids". Examples can be found in illustrated reports, such as the the two at the right, in which the modern grain at the top is a 8-day-old grain from the new Motuzaite-Matuzeviciute paper and the other two are from Huizui and the Yiluo reports (published by Lee et al in Indo-Pacific Prehistory Bulletin and PNAS in 2007). I do not mean here to single out any lab as worse than any other, all or most archaeobotanists were failing to deal adequately with highly immature millet grains-- indeed I suspect I need to go back through samples from Neolithic South India sorted during my PhD to check for mis-counted immature Brachiaria ramosa grains. The important thing is for practice to change. Immature grains are important for the recognition of crop-processing stages in millets (as discussed in Fuller and Zhang 2007). In addition, a shift from more to less immature grains harvested might be expected to take place with domestication, much as was the case with rice, i.e. morphologically wild panicles needed to be targeted on average more green to avoid grain loss to shattering. (This issue I raised in relation to rice domestication a few years ago, for example in Antiquity 2007).  

Early Agriculture & Anthropogenic Climate

A quick note on the publication of the updated rice archaeology database and a model effort based on it examining the spread of rice with an attempt to test its hypothesized contribution to rising global methane levels between 3000 BC and 1000 AD. This was a team effort involving students and post-docs with Early Rice Project (Kingwell-Banham; Castillo; Weisskop; Qin Ling) collaborators from abroad (Sato (Kyoto); Hijmans (David)) and some nice GIS modelling work by Jacob van Etten (Madrid). Some maps wet rice distirbution in selected  time-slices are shown left; but for more details read the paper.  This article "The contribution of rice agriculture and livestock pastoralism to prehistoric methane levels" is available ahead-of-print on-line from The Holocene.  Our key conclusion is that by 1000 AD perhaps 80% of 'anomalous' methane could be attributed to rice cultivation but also at 2000 BC rice is still not that significant and other sources should be sought such as the rapid spread of pastoralism around this period especially through the savannas of Africa and South Asia.

I have previously blogged the rice-part of the Ruddiman (Early Anthropogenic) hypothesis i.e. that early rice framing (and its spread) contributed enough extra methane to the atmosphere to make a global impact from sometime just after 3000 BC. I did raise some questions about the quality of the data: how many early rice finds actually represent cultivated rice (not wild) and how many represent flooded or paddyfield rice rather than upland rainfed rice? Also what role did the spread of cattle pastoralism over the Old World play in contributing to Mid-Holocene methane levels? So this new paper is an attempt to address some of these questions. We make a first stab at mapping the areas over which pastoralism spread in addition to modelling the spread of rice and the land area under wet rice cultivation. For rice at least we are able to estimate methane output but further work is needed for an equivalent calculation from cattle. I think we also would all admit that our estimate from rice remains imperfect and there is a lot of additional work to do! Collecting better quality archaeobotany (and more of); more zooarchaeology; and more sophisticated modelling...

A recent summary of the Early Anthropogenic Greenhouse Gas hypothesis has been published on-line by Bill Ruddiman at realclimate.org. It has attracted a lot of discussion. Essentially this provides a preview to some of the results due out in the August issue of the journal The Holocene (although many of the paper are already available on-line). This issue has also received attention in Nature in their News section (in March) and was one of the issue debated at the AGU Chapman conference in Santa Fe in March.

Short rice: another domestication trait (for Early Japonica)

New genetics work on rice from a Japanese team indicates that the mutations SD-EQ1, which makes rice plants shorter, was strongly selected early on in the process of rice domestication, at least within the japonica sub-species, which is presumed to be that domesticated in the  Neolithic Yangtze. This article bu Matsuoka et al, published in PNAS, has been highlighted in Science magazine's on-line news. A transcript of my full comments and initial thoughts, are provided here.

This study by Asano et al on the SD1 gene, which shows strong selection for shorter rice plants in domesticated japonica, adds to a growing list of genetic evidence for the different origins of indica and japonica rices and for the differing cultural ecologies in which these crops were first cultivated. Subsequent hybridizations transferred some domestication-selected genes, but selectively from japonica to indica, probably around 3800-4000 years ago (see, e.g. Fuller & al. 2010, Archaeological & Anthropological Sciences). Deductions from the modern ecology of wild rice varieties and archaeobotanical evidence (e.g.  Fuller & Qin 2009; p. 147 in  Fuller & Qin 2010) points to a important role for human manipulation of the water conditions of early japonica rice cultivated in China with necessary adaptations in the growth habit of rice and shift towards a more annual seasonal pattern as opposed to the wild-type perennial pattern. Wild O. rufipogon is a perennial, which prefers growing in more less permanent water or areas that only dry up for short periods. It can produce extremely tall, long culms in order to grow in deeper water, While deeper water reduced competition from other plants it also reduced productivity since growth is focused on vegetation tissues (leaves and culms), and as long as water conditions are fairly staple plants with reproduce vegetatively and produce few seeds. Early cultivation of rice in the Yangtze region of China clearly focused on wetlands margins (for example at Kuahuqiao, 6000-5400 BC, or Tianluoshan, 5000-4300 BC) in Zhejiang. The pollen and microcharcoal data from Kuahuqiao, published already by Zong et al 2007 (for further details see this 2009 paper) already indicated the manipulation of wetland margin environments by 5700-5400 BC. While these sites, especially the evidence from Tianluoshan (which we published in Science in 2009) indicates selection for non-shattering rices over this period, it also suggests a change in the ecology in which the rice was growing. Accompanying weed seeds indicate a shift away from a predominance of perennial, and taller, sedges (Cyperaceae) towards a wider diversity of shorter annual grasses and dicot weeds (summarized inFuller & Qin 2010, but only published in detail in Chinese this past month in a Tianluoshan monograph; English papers still to come!). This suggests that there were also developments for how (and where) rice was grown. This would have involved manipulating the soils and water depth in which rice was grown. One reason for this is that in order to promote annuality and higher seed production the rice plants need to water-starved, i.e. subjected to drought like conditions, when they are starting flowering: the drought conditions  lead to increased grain output, a strategy that would be suitable for true drought but also would increase yields for early farmers. Thus more productive early rice required humans to create seasonal drought like conditions; genetic adaptations to these conditions would then be selected for.

Subsequent to Tianluoshan, at around 4200-3800 BC, other sites in the Lower Yangtze regions, such as Caoxieshan and Chuodun (see discussion in Fuller and Qin 2009), show increased human efforts at managing water levels in very small paddy fields some with adjacent channels and water 'storage pits', which would allow these small fields to be drained. Such systems would have strongly selected for changed in rice plant architecture towards less spreading grow habits (based on the gene Prog1, published in by Tan et al in Nature Genetics 2008; some discussion for the selection of this which is parallels in other cereals is provided in Fuller, Allaby and Stevens "Entanglements...", in World Archaeology). Shorter rice plants might also have been favoured here, if not already at the early wetland margin cultivation of Tianluoshan, since human manipulation of water would remove the need for taller plants, which would be prone to falling over (i.e. lodging) and would produce fewer grains (since metobolism was being invested in more culm rather than more seeds). An interesting question would be to know whether shorter plants (SD1) or less branching (Prog1) was selected first.

Once japonica was introdouced into areas with proto-indica, selection for these traits would be well-finished, and shorter growth habits may have been unnecessary and unattractive to South Asian cultivators. Thus in early India there was a selection process of crossing japonica to indica to acquire some, but not all, domestication related traits.