Saturday 9 December 2017

Buckwheat origins remain elusive

Harriet Hunt and colleagues have provided a new critical assessment of data and potential data on origins of the buckwheats (Fagopyrum esculentum and F. tartaricum) in a Vegetation History and Archaeobotany article. Buckwheat is an important carbohydrate crop at high elevations in Asia, as well as parts of Japan and Europe, but it has remained quite elusive archaeobotanically.

It is absent from the many large charred seed assemblages in central China, or the charred and Fagopyrum identifications as distinct from many other Polygonaceae. Even if we accept all identifications of Fagopyrum, there are several wild taxa in this genus that will have nothing to do with the cultivtion of the crop. They consider how reliable stratigraphic dating controls are for many of pollen sequences, but even so, pollen never allows for the direct dating nor direct association with human activities that archaeobotany does.
waterlogged assemblages of the Lower Yangtze. In the Indian Himalayas where it is traditionally an important crop, finds have been few, restricted to later First Millennium BC and medieval finds in Nepal. The new review by Hunt et al has compiled evidence from archaeobotanical macro-remains, a few reported based on apparent archaeological starch remains, and the many more reports from pollen diagrams. They take a threshold of fairly high quantities in pollen diagrams, but less clear is whether one can always rely on

Distribution of wild Fagopyrum species (Campbell 1997, IPGRI)
One of their key conclusions is that the past distribution of wild Fagopyrum species, including the wild progenitor of F. esculentum, was more widespread. Extending further north, even to the north of Sichuan. This certainly seems plausible and could support a domestication in Sichuan north of where modern wild populations (in NW Yunnan) have tended to suggest domestication. They point to a few pollen cores from Shaanxi and Gansu apparently 5000 years old or more, as perhaps relating to early cultivation-- although the absence of grain finds in these regions which have had considerable archaeobotanical sampling in recent years surely calls into question the paper's tentative conclusion that cultivation had begun before 5000 BP.  Another problem with many of these pollen cores is the reliability of dating. For example, the pollen sequence at Xishanping, which was collected through an archaeological sequence, has a number of inverted radiocarbon dates, suggesting reworked residual materials, but the short (and old) chronology followed by Hunt et al. removes the out of sequence dates-- which would make sense if this were a lake core with constant sedimentation, rather than a sequence 5 varied archaeological layers. A safer, and archaeologically logical reading of the original stratigraphy (see raw data in Li et al 2007) date makes the buckwheat pollen occurrence only slightly older than 3000 BP. (The short chronology also implies that wheat was present at this Gansu site before 2600 BC, which does not fit with the accumulated evidence on wheat's arrival in Gansu (as noted already in a previous blog), especially AMS dates (see Stevens et al 2006).

A more critical reading of the dates in the sequence of the earlier pollen cores find little support for any substantial quantities of Fagopyrum pollen before around 4000 years ago, so I stand by previous inferences of domestication taking place around this period. Nevertheless from the Second Millennium BC onwards, some archaeological seeds of Fagopyrum, possible supported by starch finds points to cultivation of this crop, with a focus on west Central China and southwest China, consistent with early dispersal around the eastern front of the Tibetan plateau. Nevertheless with central and eastern China, the lower reaches of the Yellow and Yangtze basins buckwheat appears to have been absent, from macro-remains (and supported by early Chinese written sources). In this regard some of the apparent pollen reports from natural cores in the Lower Yangtze seem unlikely to represent cultivation. Despite being clearly present among the crops known in early Tibetan languages (and many related Burmic languages), and having likely been loaned from a Tibetan language into Chinese since the Han dynasty period (see Bradley 2011), buckwheat remains elusive in Asian archaeology.

This new paper by Hunt et al. provides a solid starting point for new research on buckwheat origins, with a thorough compilation of pollen and archaeobotanical evidence (in China)long with some critical thinking on the rather limited genetic data.

Friday 8 December 2017

Neolithic wine drinkers in Georgia or wishful thinking

Strong inference I was always taught comes from thinking through multiple working hypotheses and assessing which hypothesis is best supported by available evidence and trying to falsify alternatives. It is unfortunate if the quest for headlines and high profile publication gets in the way of clear thinking and an scientific approach. Of course sometimes evidence and conclusions are only partly certain, but when that is the case it should not be carefully hidden in online supplementary text and onreference claims of background fact as is the case in this study. I have to conclude weak inference reigns in the recent headline grabbing claim that the first grape wine makers anddrinkers were to be found in the Neolithic Georgia. While McGovern et al present a compelling read in their PNAS paper, and some apparently very technical scientific support, their presentation seems to me aimed to grab headlines and appeal to journalist or generalist and not really to convince scientifically the specialist. The fine print in the supplement raises many unanswered questions that undermine their conclusion. The failure to reject alternative plausible hypotheses for their result, the lack of reference to scientific names, regional flora inventories or vegetation surveys, as these would so clearly support alternative hypotheses…

Here is the claim: some interesting liquid storage vessels that could well be for wine storage have produced tartaric acid residues. This would indeed constitute part of an evidential base to argue for early wine, but on its own this is necessary but not sufficient evidence for the claim. Tartaric acid occurs as background in the soil—thus one learns in the supplement that some 11 sherds were rejected as having levels not sufficiently above background soil levels. What is more tartaric acid occurs in many fruits, not only grapes. Oddly they reject other sources with a flippant line in their supplement, “Other plants with high tartaric acid–e.g., hawthorn fruit and star fruit from east Asia, tamarind from the Indian sub-continent, and yellow plum from the New World—can be ruled out”- notable for being without any Latin names, without any citation of botanical sources on  these taxa, their chemistry, or on the regional flora from which to based claims about their distribution. What is worrying here is the inclusion of hawthorns (Crataegus spp.), and plums in this list. They claim that “yellow plum” is exclusively American—and while this is ture if what they mean is the species Prunus americana, but the broader Prunus genus, including numerous Cerasus cherries and Padus bird cherries, has high endemic diversity in the Caucusus regions, as well as numerous Crataegus species. There are ~70 species of Prunus native to Eurasia: are we really to believe that none contains tartaric acid in contrast their common American cousin? If so, how have these been excluded. Contrary to the dismissive statement in the supplement there are half a dozen Crataegus reported form the Caucasus region in the old Flora of the USSR, and several more to the South or to the north in Ukraine! (Flora of the USSR Vol. IX. Rosales and Sarraceniales, by Botisova et al 1939, English Language 1971 from Jerusalem, online here)

Grape pip with measurements (from
Bouby & al 2013 PLOSone
One might support an argument for grape wine production on archaeobotanical grounds if flotation samples were rife with grape pips and no other fruits, but in fact we learn that all AMS dates run grape pips turn out to be intrusive, Bronze Age and later. This does not support major use of grapes for wine in the Neolithic but quite the opposite! For Neolithic grape finds one must go to the Fertile Crescent, or indeed to parts of Mediterranean Europe, like Greece or Italy! While some of the co-authors have have done some cutting age work on the geometric morphometrics on grapes, e.g. Laurent Bouby, whose work on archaeological grape diversity in France is indeed cutting edge (e.g. the Vegetation History and Archaeobotany paper by Bacilieri, Bouby et al earlier this year), the deployment of these techniques on Georgian grapes from the Bronze Age and later (based on direct dates) does little to support a sequence of grape domestication in the Neolithic Caucausus.

Some sort of fermented fruit wine—the conclusion is plausible. But grape wine? That still seems wishful thinking. At best the sloppy and journalistic presentation of the evidence might be attributed to weak editing and inadequate peer reviewing combined with authors’ excitement, but at worst it represents obfuscation of the science to claim headlines and citations indices.