The recent article in PNAS by Loukas Barton et al., on stable isotopes (of people, dogs and pigs) from the site of Dadiwan has provided new data and a new methodology for looking at the establishment of millet agriculture in China. This article attracted a nice news summary by Michael Balter on the ScienceNOw site. My first reactions below:
It remains the case that the transition to agriculture in most parts of China is poorly understood and documented with limited data. This is an important new study for a couple of reasons: it provides new evidence for agriculture in one of the north Chinese early millet cultures; also, it provide a novel methodology for thinking about the development and intensification of agriculture in terms of shifts in how people fit into the food chain.
Most clearly it shows that when Dadiwan is reoccupied in the Middle Yangshao period (from 4500 BC) that a fully developed millet dependence was developed in which humans and domesticated animals (pigs and dogs) were reliant on a millet-based staple plant food diet. As the authors argue the data suggest that this intensive, and sedentary, Yangshao millet-pig economy spread into the Dadiwan region, presumably from the East. This is significant because it highlights the fact that it was the Yangshao, which starts closer to 5000 BC in its core region, was the first expansive Neolithic agricultural tradition, in China. In other words, Yangshao spread represents the result of demographic transition towards higher levels of population growth and spread. This is happening at a period in which early rice cultivators in the Yangtze are still just collectors, settling into cultivation, and divided into smaller, regional and non-expansive cultural transitions (the Chenbeixi, Hemudu, Majiabang, Longqiuzhuang cultures). This is a smoking gun for millet agriculture developing earlier than full-fledged rice agriculture.
The earlier data from the Dadiwan culture is thinner and harder to be certain about, but is provocative. Significant are its differences from the better understood later samples: a lack of millet-fed pigs, and only a single millet-eating dog—thus implying the pigs had not been domesticated, although I would express some caution on this point: if Dadiwan people were low level food producers, then by definition millet would not have been a staple food, and their other wild staples would have been C3, and could have been shared with pigs without leaving a signature. Also, thinner and smaller occupation layers, by contrast to the impressive architectural remains from the Yangshao period (including building F901 which was a large public building)—thus implying that Dadiwan culture was less sedentary and more seasonally mobile.
It suggests that there was some millet-reliance, but the level remains problematic. This relies on two dogs having eaten a lot of millet, but we can not be sure either that millet was a regular staple for the people nor that the millet was domesticated. They infer this to be cultivation of millet, although in the absence of large archaeobotanical assemblages and morphological domestication traits, this is hard to be certain about. This difficulty is compounded by our lack of understanding of the ecology, and potential wild availability, of the wild progenitor of broomcorn millet. While the weedy Panicum ruderale provides a good model of wild morphology, what we do not have is a good model of its habitat and distribution. If it were like Panicum sumatrense, the domesticated little millet of India (e.g. Harappan Gujarat), then the wild progenitor would be a patchily distributed clump-forming grass, and large scale reliance probably would imply cultivation for bringing large stands together. On the other hand if we take Panicum laetum, an important West Africa food that was never domesticated, as a model then we could envision extensive monospecific stands (more like wild barley or wild rice) on which hunter-gatherers could easily subsist in quantity without cultivation. How they can then conclude that millet cultivation must have started earlier somewhere else, seems something of a stretch. Nevertheless, their argued scenario seems entirely plausible, and what’s more it provides a novel food-chain perspective for looking at agricultural origins in North China. While carbon isotopes have certainly been measured on pigs and dogs before, the integration of these data as a way of getting at shifts in human and dependent animal food chains involving possible crop resources is new and important. What we need now is more of this kind of data from across North China, which is in the lucky position of being a region low in natural C-4 plants, but with major C-4 crops. [The same approach would not work in most world regions which lack such conditions or such crops!
