COMMENTARY

Agricultural origins in pushed back to the Pleistocene–Holocene boundary

Gary W. Crawford1 Department of Anthropology, University of Toronto Mississauga, 3359 Mississauga Road, Mississauga, ON, Canada L5L 1C6

wo grains, common (proso or which the dates and the crops were ob- broomcorn) millet (Panicum tained. Lu et al. (5) obtained 46 new miliaceum) (Fig. 1) and foxtail samples of plant remains and 9 accelera- millet (Setaria italica), were tor mass spectrometry (AMS) radiocar- Tfundamental to the development of agri- bon dates directly on the millet. Two of cultural societies that eventually evolved the AMS dates are younger and 7 are into the first urban societies of China older than the non-AMS radiocarbon between 4500 and 3800 calibrated years dates on charcoal from the original ex- (cal.) B.P. (1). Today, these grains are cavations. The structure and contents of important mainly in parts of Russia, the site appear to be consistent with the South Asia, and East Asia. How, when, extended chronology. Archaeologists and in what settings these millets ini- suspected that the site has a longer his- tially evolved is not well known (2). One tory, and in 2008 Lu et al.’s team re- hypothesis holds that common millet viewed the stratigraphy at Cishan and was domesticated rapidly in the central learned that the opening of pit CS-I was Wei river basin shortly after ca. 8000 just at the bottom of the Holocene soil cal. B.P. (3). Another hypothesis pro- layer, while the mouth of the younger poses that common millet was domesti- CS-II pit was excavated into slightly cated in the Northeast China Liao river shallower Holocene soil. The stratigra- basin around the same time (4). In real- phy structures of the site appear to be ity, archaeological data have simply not consistent with the extended chronology been adequate to resolve the issues sur- (Lu et al., personal communication). rounding the domestication of millet Three of the older dates extend the oc- and the development of the first agricul- Fig. 1. Modern broomcorn millet (P. miliaceum). cupation to 10,400–10,100 cal. B.P. The tural communities in North China. Com- dates also show for the first time that the plicating the problem, common millet is community spanned a period of Ϸ3,000 also present in Europe ca. 8000–7500 lar in plan view. Some of the large, oval years (ca. 10,400 to 7500 cal. B.P.). cal. B.P. (2)., so this timing opens the pits are houses. Most of the pits belong How were the crops identified? The possibility that the crop was domesti- to period II (7). Many of these charac- remains are extremely fragmentary, be- cated more than once. Otherwise, its teristics are still unique to the Cishan ing the remnants of chaff. Identification origins must predate 8000 cal. B.P. The culture but it was the large quantity of is relatively straightforward when the Early Holocene Cishan site in North silicified millet remnants in 88 of the actual grains are present but in this case China is one of several sites considered pits, nearly 80% from period I, that was Lu et al. (5) needed some ingenuity to key to understanding millet domestica- so surprising. When the millet was first identify the specimens. They used 3 tion and the origin of dry-land agricul- exposed, it decomposed rapidly, leaving techniques to clarify that broomcorn ture in China, yet the dating and identity some grains that were too fragile to be millet was the only crop in period I and of the crops recovered there have never recovered and the tough chaff whose the primary crop in period II. A small been adequately documented. The study cells are composed mainly of silica. At quantity of is present in published in this issue of PNAS (5) re- first, the grain was thought to be foxtail period II, indicating that it was a later visits Cishan, located on a terrace on millet (S. italica) but details of the mor- addition to the crop assemblage and was the western edge of the North China phology were never reported and the not a significant crop there. The first Plain Ϸ9 km from where the Nanming identification has not been universally identification method, an examination of river emerges from the Taihang moun- accepted. The age of the site was also the epidermal cells of the husks, in- tains. Two outstanding issues regarding surprising considering its complexity: volved measuring the amplitude of the the early archaeological record of millet Ϸ8000 cal. B.P. Archaeologists were un- undulations of the long cells and statisti- at Cishan first reported nearly 30 years certain how long the site was occupied, cally testing for significant differences. ago (6, 7), their dating and identifica- if the millet was as old as the charcoal The measurements have a bimodal dis- tion, are resolved in the new study. that provided the radiocarbon dates, or tribution, indicating that, indeed, the Cishan, a village at least if, indeed, the site was as old as they epidermal cell measurements form 2 400 m square, has an exceptionally-rich surmised. Lu et al. (5) report that some clusters, 1 for common millet and 1 for pottery, stone, plant, and animal bone of the pits are at least 2,000 years older foxtail millet. Furthermore, the mor- assemblage. The pottery consists of than previously thought and the plant phology of silica bodies (phytoliths) in plates, bowls, and pots decorated by remains are composed primarily of com- chaff epidermal cells have been used to cord-wrapped paddle and incising/ mon millet rather than foxtail millet. trailing; none of it is painted. No spe- To make an incontrovertible case for cific precursors are known. The houses the age and association of crops, archae- Author contributions: G.W.C. wrote the paper. and pits are divided into 2 occupations, ologists require secure dating, well- The author declares no conflict of interest. periods I and II. Excavations revealed documented identifications, and proper See companion article on page 7367. 476 pits, among which 376 are rectangu- documentation of the context from 1E-mail: [email protected].

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0903375106 PNAS ͉ May 5, 2009 ͉ vol. 106 ͉ no. 18 ͉ 7271–7272 Downloaded by guest on September 25, 2021 distinguish plant taxa and evaluate their that included at least 23 species of age pits. In South America some of the domesticated status. In a related study mammals, tortoise, birds, fish, and shell- oldest traces of agriculture date between Lu et al. (8) demonstrate that phytoliths fish (6). The location likely mitigated 11,000 and 6000 cal. B.P (15). Mexican can, indeed, distinguish foxtail and some of the effects of the early occupa- agricultural origins appear not to be broomcorn millet. The phytoliths are tion period that was cooler and drier associated with villages or pottery consistent with the identifications based than it was after 8700 cal. B.P. How- (16, 17). A key exception to agriculture on epidermal cell morphology. Further- ever, rhesus monkey and masked civet appearing during the Pleistocene– more, the analysis adds to the under- were hunted and are evidence that the Holocene transition is eastern North standing of the sophisticated storage at climate was warmer than it was in the America where native plants were being Cishan. Leaves of a wild grass, Phrag- domesticated between 5000 and 3800 mites australis, along with millet chaff, cal. B.P (18). However, like many other were layered on the bottom of the pits. Cishan now has the regions, the setting was relatively re- Finally, to alleviate any potential doubt source rich; populations were not strug- that the majority of the Cishan millet is oldest clear evidence gling for food. Cishan adds another broomcorn millet, Lu et al. sought bi- variant to the comparative mix of devel- omarkers that could distinguish the mil- for a significant degree opmental puzzles that are the origins of lets. They discovered that 5 biomarkers agriculture (19). distinguish broomcorn and foxtail millet. of food production The new study (5) indicates that we The biomarker analysis confirms the com- still do not have quality data pertaining mon millet identification. in China. to the onset of millet domestication and Why are the new Cishan site data sig- agriculture in North China because food nificant? Understanding developments production was already established by in North China has been frustrated by late 1900s, but it is possible that these 10,400–10,100 cal. B.P. at Cishan. If our the seemingly sudden appearance of animals are associated with period II. understanding of Cishan is being signifi- millet production ca. 8000–7500 cal. Plant exploitation and anthropogenic cantly revised, what about the other B.P. in an area Ϸ1,000 km east–west habitats must have been extremely im- Early Neolithic sites in North China re- (Yuezhang to Dadiwan) and the same portant at Cishan but neither has been ported in the 1970s and 1980s? An issue distance north–south (Xinglonggou to thoroughly investigated. that is still unresolved is foxtail millet Peiligang) (3, 4, 9, 10). Cishan now has The events at Cishan for the first time domestication. The new data are consis- the oldest clear evidence for a signifi- place the timing of agricultural origins tent with other reports that it developed cant degree of food production in China in North China at the Pleistocene– later than broomcorn millet, but we still instead of being one of several sites in Holocene transition, similar to its timing need to be careful about Cishan because North China where millet suddenly ap- in Southwest Asia, Mexico, and South so few of the plant remains have been peared over an exceptionally large area. America. How the circumstances of ag- examined. We may yet learn that other Furthermore, the 3,000-year occupation ricultural developments compare are pits contained more foxtail millet or by a food-producing community is a lon- still open to question. Despite the rela- other crops. Furthermore, so few plant gevity rivaled elsewhere at the time only tively continuous Upper Palaeolithic remains assemblages have been recov- by sites such as Abu Hureyra in Syria through Neolithic record in Southwest ered from the period between 10,000 (11). We still do not know how continu- Asia (12) the earliest stages of agricul- and 7500 B.P. in North China that we ous the occupation was, but even if Cis- ture there are not yet clear. What we do may still learn that the 2 millets were han was occupied intermittently, it was know is that the shift to agriculture is domesticated independently. Crop com- still clearly an important location for evident only by the Pre-Pottery Neo- plexes may form well after individual early hunter–gatherer–fisher–farmers. lithic B (10,500–8200 cal. B.P.) (13, 14), taxa were separately domesticated (18). Cishan represents an extremely success- the earliest dates of which correspond to We still await a comprehensive study of ful adaptation to an area bounded on the oldest dates at Cishan. However, in subsistence and human–environment the west by mountains and the loess pla- contrast to Southwest Asia, pottery had interaction at Cishan but with this new teau and situated in a riverine habitat developed in China long before obvious study (5) the development of agriculture that provided a rich array of resources traces of agriculture, villages, and stor- in North China is becoming clearer.

1. Liu L, Chen X (2003) State Formation in Early China 7. Tong WH (1984) The remains of primitive agriculture at 14. Willcox G (2005) The distribution, natural habitats, and (Duckworth, London). the Cishan site and related issues. Agric Archaeol availability of wild cereals in relation to their domesti- 2. Hunt H, et al. (2008) Millets across Eurasia: Chronology 1:194–207. cation in the Near East: Multiple events, multiple cen- and context of early records of the genera Panicum and 8. Lu H, et al. (2009) Phytoliths analysis for the discrimi- ters. Vegetation Hist Archaeobot 14:534–541. Setaria from archaeological sites in the Old World. nation of foxtail millet (Setaria italica) and common 15. Piperno DR, Dillehay TD (2008) Starch grains on human Vegetation Hist Archaeobot 17:5–18. millet (Panicum miliaceum). PLoS ONE 4:e4448. teeth reveal early broad crop diet in northern Peru. 3. Bettinger RL, et al. (2007) The transition to agriculture 9. Barton L, et al. (2009) Agricultural origins and the Proc Natl Acad Sci USA 105:19622–19627. in northwestern China. Late Quaternary Climate isotopic identity of domestication in northern China. 16. Piperno DR, Ranere AJ, Holst I, Iriarte J, Dickau R Proc Natl Acad Sci USA 106:5523–5528. Change and Human Adaptation in Arid China, eds (2009) Starch grain and phytolith evidence for early 10. Crawford GW, Chen X, Wang J (2006) rice Madsen DB, Gao X, Chen FH (Elsevier, Amsterdam), pp ninth millennium B.P. maize from the Central Balsas from the Yuezhuang Site, Jinan. East Asia Archaeol 83–101. River Valley, Mexico. Proc Natl Acad Sci USA 3:247–251. 4. Zhao Z (2005) Discussion of the Xinglonggou site flo- 106:5019–5024. 11. Moore AMT, Hillman GC, Legge AJ (2000) Village on tation results and the origin of dry farming in northern 17. Smith BD (1997) The initial domestication of Cucurbita the Euphrates: From Foraging to Farming at Abu pepo in the Americas 10,000 years ago. Science China. Antiquities Eastern Asia A:188–199. Hureyra (Oxford Univ Press, Oxford). 5. Lu H, et al. (2009) Earliest domestication of common 12. Bar-Yosef O (1998) The Natufian culture in the Levant, 276:932–934. millet (Panicum miliaceum) in East Asia extended to threshold to the origins of agriculture. Evol Anthropol 18. Smith BD, Yarnell RA (2009) Initial formation of an 10,000 years ago. Proc Natl Acad Sci USA 106:7367– 6:159–177. indigenous crop complex in eastern North America at 7372. 13. Nesbitt, M (2002) When and where did domesticated 3800 B.P. Proc Natl Acad Sci USA 106:6561–6566. 6. Zhou B (1981) Animal remains discovered at Cishan cereals first occur in southwest Asia? The Dawn of 19. Smith BD (2001) Documenting plant domestication: Village, Wu’an, Hebein Province. Kaogu Xuebao Farming in the Near East, eds Cappers R, Bottema S (Ex The consilience of biological and archaeological ap- 3:339–347. Oriente, Berlin), pp 113–132. proaches. Proc Natl Acad Sci USA 98:1324–1326.

7272 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0903375106 Crawford Downloaded by guest on September 25, 2021