Abstract
As heated debates about the origin of rice domestication and cultivation in southern and eastern China continuously attract attention of the broad scientific community, new evidence for early rice exploitation from the regions located outside the core area of domestication, the lower Yangtze, are very important. Here, we present new archaeobotanical results of plant macrofossil and phytolith analyses, including directly dated rice grains from the sites of Dongpan (4030–3820 cal. BC) and Beiqian (3700–2900 cal. BC). These results fill (at least partly) an existing gap between the c. 8000-year-old rice remains from the Early Neolithic Houli Culture (c. 6500–5500 BC) sites north of the Shandong Highlands and the Longshan Culture (c. 2600–1900 BC) sites, where intensive rice agriculture was practiced. Neither rice nor millet made substantial contribution to the plant macrofossil assemblage at Dongpan, while broomcorn (and to a lesser extent foxtail) millet contributed up to 75% to the macrofossil assemblage at Beiqian. This increase can be interpreted as a major change in regional subsistence from strongly relying on wild resources and small scale cultivation during the Beixin Cultural period to a millet-based economy during the Dawenkou Culture.
Keywords
Introduction
The eastern part of China and the Fertile Crescent in southwestern Asia expose the world’s longest histories of plant cultivation and domestication dating back to the early Holocene (e.g. Bar-Yosef, 2011; Crawford, 2009; Riehl et al., 2013; Wu et al., 2014; Zhao, 2011). Regardless of this common feature, these two macro-regions of Eurasia differ from each other in their environments and climatic conditions, the types of domesticated crops, as well as in the research methods that have been applied and the main objectives of that research. Obtaining a well-constructed record of wheat (Triticum spp.) and barley (Hordeum vulgare) domestication in southwestern Asia required decades of intensive archaeological excavations and archaeobotanical studies (e.g. Bar-Yosef, 2011; Riehl et al., 2013; and references therein). In China over much of the past few decades, archaeobotanical research has not been the focus of archaeological inquiry. However, economic development and modernization of the country during the past two decades have greatly stimulated the mingling of the natural sciences with archaeology, including archaeobotany. Furthermore, active international cooperation in the fields of archaeology and the natural sciences has facilitated the incorporation of new research methods and leads to a rapid generation of new data, ultimately helping to fill gaps in the current knowledge (e.g. Wagner et al., 2014).
These recent archaeobotanical publications primarily address questions about the origins and developmental features of Neolithic agriculture in China, including the rice-based economy that originated in the middle and lower Yangtze River and in the upper Huai River regions, the millet-based economy developed in the middle and lower Yellow River region, and the ancient roots and tubers-based agriculture developed in the tropical regions of China (e.g. Cohen, 2011; Crawford et al., 2006; Fuller et al., 2008, 2009a, 2009b; Gross and Zhao, 2014; Jones and Liu, 2009; Zhao, 2011; and references therein). Another major research topic includes questions concerning introduction, and temporal and spatial distribution patterns of different crops in China, including rice (Oryza sativa), broomcorn (Panicum miliaceum) and foxtail (Setaria italica) millet, barley, and bread wheat (Triticum aestivum) (e.g. Betts et al., 2014; Crawford et al., 2006; Liu and Chen, 2012; Wagner and Tarasov, 2014).
One of the regions where several important but challenging questions could be successfully addressed using a detailed archaeobotanical approach is the area of modern Shandong Province. Located in the eastern part of China in the lower reaches of the Yellow River and the northern part of the Huai River region (Figure 1), it represents a border area between the core regions of early rice (i.e. Xianrendong/Diaotonghuan, Shangshan, Kuahuqiao, Hemudu/Tianluoshan, Liangzhu, Jiahu, and Chengtoushan sites) and millet (i.e. Cishan, Xinglonggou, Shawoli, Peiligang, Yuezhuang, Dadiwan, and Yuhuazhai sites) agriculture (see sites in Figure 1 and Zhao (2011) for references). Furthermore, Shandong has been suggested as the area with the earliest archaeological records of wheat in China dating back to c. 5000–2270 BC (e.g. Jin et al., 2011; Li, 1984), although these early dates come from indirect dating based on associated material culture and need further verification (Betts et al., 2014; Li et al., 2007a).
Topographic map of Shandong Province and surrounding regions, showing location of the studied sites Dongpan (D) and Beiqian (B) indicated with a red square. Black circles indicate locations of the other sites discussed in the text: 1: Zhangmatun; 2: Xihe; 3: Yuezhuang; 4: Xinglonggou; 5: Cishan; 6: Bianbiandong Cave; 7: Shawoli; 8: Peiligang; 9: Yuhuazhai; 10: Dadiwan; 11: Jiahu; 12: Liangzhu; 13: Hemudu/Tianluoshan; 14: Kuahuqiao; 15: Shangshan; 16: Xianrendong/Diaotonghuan; 17: Chengtoushan; 18: Longqiuzhuang; 19: Zhoubeidun; 20: Shuangduan; 21: Houjiazhai; 22: Hongdunsi; 23: Xiaosungang; 24: Tanghu; 25: Liangmatai; and 26: Laomuwa.
Being peripheral to the so-called ‘core area of Chinese civilization’, centered in the area of the modern provinces of Henan, Shanxi, and Shaanxi, the area of Shandong, however, is extremely rich in archaeological sites (Hosner et al., 2009; Wagner et al., 2013) and reveals a very long and rich sequence of Neolithic and Bronze Age archaeological cultures (e.g. Luan, 1997a, 1997b, 1997c, 2013; Wagner et al., 2009). This sequence dates back to the Early Neolithic, that is, c. 7000 cal. BC, as demonstrated by recently obtained AMS dates from the Zhangmatun site (Figure 1, Wu et al., 2014). The latter study and several earlier publications (e.g. Crawford et al., 2006; Kong, 1996, 2000; Wagner et al., 2009) discussing archaeological data and plant and animal remains suggest that the Neolithic Houli Culture (c. 6500–5500 BC) inhabitants of Shandong already lived in permanent settlements and had domesticated dogs and pigs, and a knowledge of millet (and possibly rice) agriculture. A systematic archaeobotanical study at the Xihe site (Figure 1) recovered a few foxtail millet grains and numerous remains of rice grains, showing that rice might have been an important food resource for the Xihe inhabitants as early as 8000 years ago, despite their strong reliance on fishing, hunting, and gathering (Jin et al., 2014). The Early Neolithic rice exploitation record from Xihe corroborates other early evidence for rice in the region between the Yellow River and the Huai River (Figure 1), that is, at the Yuezhuang site (Figure 1), where both millet and rice remains dated to c. 6000–5850 cal. BC were excavated (Crawford et al., 2006); at the Tanghu site, where rice and millets phytolith were dated (though, not directly) to c. 5800–2500 cal. BC (Zhang et al., 2012); and at the Liangmatai and Laomuwa sites with rice remains of the Yangshao Culture, c. 5000–3000 cal. BC (Kong et al., 2003). This surprisingly early use of rice outside the commonly accepted area of rice domestication in the Yangtze River basin (e.g. Gross and Zhao, 2014) is of particular importance for the understanding of the agricultural history of Neolithic China. The key role of these early to middle Holocene archaeological sites from Shandong in deciphering this puzzling story of early rice cultivation is even more evident when we consider the results of a detailed archaeological excavation project carried on in the Yiluo River valley, Henan province (Lee et al., 2007). The radiocarbon-dated archaeobotanical data from 26 archaeological sites revealed the introduction of rice to this region only after c. 3000 cal. BC. Li et al. (2007b) reported the radiocarbon-dated evidence of rice (although predominantly millet) cultivation at the Xishanping site (34°33′50″N, 105°32′41″E) from southern Gansu back to c. 3120 cal. BC, thus providing one of the oldest record of cultivated rice in the northwestern regions of Neolithic China. It is generally accepted that intensive rice agriculture around the Shandong Highlands developed during the Longshan Period c. 2600–1900 BC (Luan and Wagner, 2009; Jin, 2013). Furthermore, the absence of rice remains in the Middle Neolithic archaeological records from the region (Fuller et al., 2009b) led to the conclusion that the Early Neolithic rice found in Xihe and Yuezhuang belongs to a harvested wild species, although domestication was not fully excluded (Zhang, 2011).
Our current paper presents plant macrofossil and phytolith records and AMS radiocarbon dating results from the recently excavated sites of Dongpan and Beiqian, situated in southern and eastern parts of Shandong (Figure 1). The archaeological excavation reports and typological dating assign the cultural layers to the Middle and Late Neolithic, respectively (Department of Archaeology, 2011; Jin et al., 2011; Wang and Jin, 2013; Wang et al., 2011).
The charred plant remains from the Dongpan (excavation 2008) and Beiqian (excavations 2007 and 2009) sites have been partly reported by Jin and Wang (2011), Wang et al. (2011), and Wang and Jin (2013) in the local Chinese journals. However, no comprehensive paper is available to date. In order to fill this gap, the earlier results are presented here together with the new results of phytolith analysis from the Dongpan site and charred plant remains and phytolith analyses obtained from the Beiqian site excavation of 2011. Another aim of the archaeobotanical research that we present here was to screen the sediment samples for rice, millet, and wheat remains and to date these remains directly by using the AMS radiocarbon dating method (Table 1), in order to address the problems mentioned above.
AMS dates of charred plant remains from Beiqian and Dongpan. Radiocarbon ages were converted into calendar dates (68% range) using CalPal online calibration program (http://www.calpal-online.de) quickcal2007 ver.1.5 and the calibration curve CalPal2007_HULU; 95.4% probability ranges are calculated using OxCal v4.1.5 Bronk Ramsey (2010); r:5 and atmospheric data from Reimer et al. (2013).
Study sites
All archaeobotanical samples presented in this study were collected from the archaeological sites of Dongpan and Beiqian (Figure 1). The Dongpan site is located on the terrace of the Dongpan River in Linshu County, Shandong Province. The modern vegetation is strongly modified by urban sprawl and rice cultivation (both paddy-field and dry-soil farming are practiced); wheat and corn (Zea mays ssp. mays) fields are common too.
The Beiqian site is situated in the northern part of the Beiqian village, northeast of the town of Jimo, Shandong Province (Department of Archaeology, 2011). The modern agricultural fields there are corn- and wheat-based, as well as some vegetable and fruit plantations. The coastal location of the site also provides a variety of aquatic foods for the local population.
Materials and methods
A rescue excavation of the Dongpan site, preceding the construction of a highway, was run by an archaeological team from the Shandong Provincial Institute of Cultural Relics and Archaeology from August to November of 2008. The excavation focused on archaeological layers and artifacts typologically assigned to the Beixin Culture (c. 5000–4100 BC), Longshan Culture (c. 2600–1900 BC), and Zhou Dynasty (c. 1046–256 BC). Sediment samples collected from the excavated Beixin Culture layers and pits were screened for seeds, fruits (altogether 31 samples), and phytolith (11 samples) remains. The results of carpological and phytolith analyses from the Dongpan site are presented in Tables 2 and 3, respectively.
Results of carpological analysis from the Beixin Culture layers at Dongpan.
Selected results of phytolith analysis (phytolith counts) of the Beixin Culture layers at Dongpan.
During the excavation campaigns in 2007, 2009, and 2011, pits, houses, tombs, ditches, pottery, and stone tools of the Early Dawenkou Culture (c. 4100–3500 BC) were unearthed at the Beiqian site. Altogether 796 sediment samples representing these features were used for archaeobotanical research (Table 4) and 11 samples were used for phytolith study (Table 5).
Summary of the 2007, 2009, and 2011 excavation results of the carpological analysis from the Dawenkou Culture layers at Beiqian.
Selected results of the phytolith analysis (counts) from the Dawenkou Culture layers at Beiqian.
The archaeobotanical results from East China published in recent years (e.g. Sun et al., 2014; Wu et al., 2014; Zhao, 2011 and references therein) reveal a relatively high abundance and diversity of wild plant macroremains at the Early Neolithic sites, but remains of possibly cultivated plants are rare in comparison with assemblages reported for the Late Neolithic. Combining and contrasting plant macrofossil studies and phytolith analysis is becoming a more common practice in China, especially for the identification of cultivated rice, both millets, and their wild ancestors (Fujiwara, 1976; Lu et al., 2002, 2005, 2009; Wang and Lu, 1993; Zhang et al., 2011; Zhao, 1998; Zhao et al., 1998).
We followed standard flotation and phytolith methods (see Jin et al., 2014; Pearsall, 1989, 2000; Piperno, 2006; Wang and Lu, 1993 for further details). Flotation sample sizes varied between the archaeological features and sites, although 5–10 L was the target volume. A froth-flotation device with a 0.2-mm mesh screen for light fraction samples and 1.0 mm for heavy fraction was used. Sediment samples were cleaned with hydrogen peroxide and 30% hydrochloric acid to remove the organic matter and then centrifuged for phytolith extraction.
The discovered carbonized plant remains and phytoliths were identified at the Archaeobotanical Laboratory at Shandong University using a Nikon SMZ1000 and Nikon E800 microscopes, respectively. Seeds were studied under a dissecting stereomicroscope Nikon SMZ645. Seed size, external features, and internal anatomy were used for identification by comparison with our modern plant reference collection and published identification keys (e.g. Liu et al., 2008). Botanical nomenclature follows that laid out in the Flora of China (Editorial Committee of Flora Republicae Popularis Sinicae, 2001). Shape and texture of the surface of the phytoliths were used for identification by comparison with the modern rice and millet phytolith reference collection and published identification keys (e.g. Fujiwara, 1993; Lu et al., 2009; Zhang et al., 2011, 2012; Zhao et al., 1998).
For the presentation and following discussion of the flotation results, only the identified remains were used and the non-identifiable plant fragments were left out of the total counted sum, which was taken as 100%. Each of the identified seeds or fruits, either intact or broken, was counted as an individual when the area of the embryo was preserved; we excluded from the total sum the fragments without this feature in an attempt to avoid using minimum number of individual estimates.
For the accurate chronological assignment of the identified plant remains and the respective cultural layers, 15 carbonized plant samples were sent for direct AMS dating to Poznan Radiocarbon Laboratory (Poland) and to Beta Analytic Inc. (USA). Thus, 10 samples of cultivated crop grains from Beiqian, representing broomcorn millet, wheat, and rice, and one sample from Dongpan, representing rice, have been directly dated.
Results and interpretations
Chronology
The results of the AMS radiocarbon dating, including 14C and calibrated calendar ages (cal. BC) are shown in Table 1. The AMS dates from Dongpan, one on the rice grain and one on a herbaceous wild plant seed, fell between 4030 and 3820 cal. BC (95% range). These dates are slightly younger than the Beixin Culture (c. 5000–4100 BC), which had previously been assigned to the layer by the excavating team using regional pottery typology. Pottery typology in northern China, particularly in the Shandong area, is commonly used to date archaeological sites (Luan, 1997a; Wagner et al., 2009). However, based on these radiocarbon dates, it seems plausible that the age of the Beixin/Dawenkou boundary in this area needs more careful verification and possibly revision. However, other scholars have already suggested that there may be regional issues with the ceramic chronologies of this period (e.g. Sun et al., 2014; Wagner et al., 2009; Wu et al., 2014).
Eleven of the 13 dates obtained on the plant remains from the Beiqian site (Table 1) fell within the age range between c. 3700 and 2900 cal. BC (95% range). These dates are within the range of the Dawenkou Culture (c. 4100–2600 BC), specifically representing the middle Dawenkou phase (e.g. 3500–3000 BC: Luan, 1997a; Luan and Wagner, 2009). However, excavators, again based on pottery typology, proposed that the site was dated to the early Dawenkou Cultural phase (Department of Archaeology, 2011).
While pottery typology is commonly used in China as a dating tool, it is clear that its accuracy is limited. It becomes less accurate when applied to finer units, such as phases and sub-phases. Furthermore, pottery vessel types are rather conservative cultural features, which are often held onto in certain regions longer than in other regions, hence blurring the boundaries of such seriations. Even in the modern quick-changing world, putting into service a new car or computer system does not imply the immediate disappearance of the old models. These discrepancies illustrate the necessity of supplementing typological dating of the archaeological features and cultural layers with direct radiocarbon dating of short-lived plant macrofossils, for example, seeds.
The two dates obtained on the wheat grains from Beiqian are much younger (Table 1) than the archaeologically defined cultural period of the excavated cultural layer, that is, one originating from the Zhou Dynasty and the other is modern. These dates better fit accepted dates for the spread of this crop, arguing that wheat in the eastern part of China appeared rather late, for example, after c. 2000 cal. BC (e.g. Betts et al., 2014; Flad et al., 2010; Lee et al., 2007; Li et al., 2007). Therefore, excavated cultural layers at Dongpan and Beiqian represent the interval between c. 4030 and c. 2900 cal. BC, which is – in more general terms – the transition from the Middle to Late Neolithic in East China.
Plant remains from Dongpan
Table 2 shows the plant macrofossil remains (i.e. seeds) from the Dongpan site reported by Wang et al. (2011). The number of identified seeds (i.e. 52) recovered from the 31 samples including 433 L of sediment (for a density of 0.12 seeds/L) from the studied cultural layer is rather small. The total number of rice (n = 2), foxtail (n = 1), and broomcorn (n = 1) millet seeds (Figure 2a–c) is even smaller, only 7.7% of the total seed assemblage from c. 4030–3820 cal. BC. By comparison, sediment samples from younger archaeological periods at the site (i.e. Longshan, Western Zhou, Spring and Autumn, and Eastern Han time periods) revealed a 5-, 17-, 10-, and 8-time higher concentration of recovered fossil seeds, respectively (Wang et al., 2011).
Photos of charred grains of (a) rice, (b) broomcorn millet, and (c) foxtail millet recovered from the Beixin Culture layers at Dongpan; and (d) rice, (e) broomcorn millet, and (f) foxtail millet grains recovered from the Dawenkou Culture layers at Beiqian.
Apart from rice and millet grains, seeds of various wild plants were identified in the macrofossil assemblage (Table 2). Some of them (e.g. Amaranthaceae, Fabaceae, Perilla frutescens, Portulaca, Polygonaceae, small wild grasses) are edible and could be collected by people, while others (i.e. Xanthium, Cyperaceae) are not edible, or at least not worth eating. Chenopodium seeds, although potentially edible, are among the most commonly found kinds of seeds in any archaeobotanical assemblage in the entire Northern Hemisphere (e.g. Spengler et al., 2013, 2014, and references therein).
Phytolith analysis was performed on 16 sediment samples representing the Beixian cultural layer. Results of the analysis summarized in Table 3 demonstrate abundant rice, foxtail millet, broomcorn millet, and reed grass (Phragmites sp.). Differing from seeds of grain crops, which were found only in one or two of 31 analyzed samples, rice (c. 27%) and millets (c. 13%) phytoliths are rather abundant in the assemblage, representing 16 analyzed sediment samples (Table 3) and appear in 80% and 100% of the analyzed samples, respectively. This result is important as it points to a local cultivation of rice and two types of millet (suggested by husk and stem phytoliths) and to a possibly more significant role of rice and millet agriculture from c. 4030–3820 cal. BC than could be concluded on the solely macrofossil data.
Plant remains from Beiqian
Macroscopic plant remains recovered by the excavations in 2007 and 2009 have been reported in Jin et al. (2011) and Wang et al. (2013). They suggest that broomcorn millet was the main crop at the study site and no rice was found. However, in sediment samples from the 2011 excavation, we recovered two rice grains, together with abundant broomcorn millet, and some foxtail millet grains (Figure 2d–f), and a variety of other wild plant seeds. Furthermore, a rice husk impression has been identified in one excavated piece of potsherd (Figure 3). This find provides a good argument for local cultivation, as the impression has its husk still articulated and the rachis attached. When people transport grain, they do so post-threshing.
Impression of rice rachis and glume on a potsherd excavated from the Dawenkou Culture layers at Beiqian. Surface structuring is typical of a rice husk and rachis impression is not well-enough preserved to determine the nature or the break and cannot speak to issues of domestication verses wild status of the grain.
Table 4 shows the combined results of plant macrofossil analysis based on all three excavation campaigns. The seed concentration in the analyzed sediment samples from Beiqian (0.69 grain/L on average) is about 6-time higher than in the samples from the Beixin Culture layers of the Dongpan site. Among the 44 identified plant taxa, presented in Table 4, broomcorn millet dominates the assemblage, that is, comprising about 71% of all recovered seeds and being present in 62% of all analyzed samples. Foxtail millet comprises 3.7% of the seeds and appears in 10.2% of all samples, and rice was represented by two grains, one being directly dated.
There are a few fruits and other possible wild edible plants (including the wild relative for the soy bean, wild grapes, Prunus, acorns, Amaranthus, Actinidia, Physalis, etc.) in the assemblage from the Beiqian site, so it would be easy to make an argument for wild foraging at that site.
Phytolith analysis results (Table 5), once again, show rather good representation of millets (16.5%), but also rice (10.2%), in the phytoliths assemblage, which represents 11 analyzed sediment samples. Rice phytoliths are identified in six samples, foxtail in four and broomcorn millet in 10 samples.
Discussion
The current research contributes to the history of plant use and agricultural development in the Shandong region (and in eastern China as a whole), and in this article, we compare these new findings with research results and hypotheses presented in the earlier publications.
Chronological issues
The area of Shandong has the highest number of archaeological sites dated from the Early Neolithic to the Iron Age among all provinces and autonomous regions of China (Wagner et al., 2013). Shandong Volume of the Atlas of Chinese Cultural Relics (Wenwuju, 2007) documented altogether 7134 sites for the interval between c. 6500 and c. 500 BC. This number is substantially higher than that reported for the three central provinces, that is, Henan (2159 sites), Shanxi (4611 sites), and Shaanxi (6267 sites), regarded as the core area of Chinese civilization (Wagner et al., 2013). Alongside the highest number and density of archaeological sites, Shandong reveals one of the longest and most detailed sequences of prehistoric archaeological cultures, starting from the Early Neolithic Houli Culture (c. 6500–5500 BC) to the Bronze Age Yueshi Culture (c. 1900–1500 BC) (Luan and Wagner, 2009). Definition and dating of this cultural sequence relies primarily on well-developed pottery typology and other associated objects (e.g. Wagner et al., 2009, for details and references), while radiocarbon dates have played a secondary role. The introduction of the AMS radiocarbon dating technique to the routine archaeological work has revealed some weaknesses in the existing chronology and called into question the absolute age boundaries of certain archaeologically defined cultures (e.g. Sun et al., 2014; Wagner et al., 2009; Wu et al., 2014).
For example, a series of AMS radiocarbon dates from the Zhangmatun site, situated at the northern edge of the Shandong Highlands (Figure 1), and typologically assigned to the Houli Culture, were up to 500 years older than conventional dates for the Houli culture (Wu et al., 2014). Another study presenting archaeobotanical and radiocarbon dating results from the Bianbiandong cave in central Shandong also reports pottery shards with typological features of the Houli Culture, but predates it by about 1000 years (Sun et al., 2014). Furthermore, the AMS dating places Phase 3 (i.e. cultural layer 2 in the cave) between interval from c. 5400–4900 cal. BC, which corresponds to the cultural hiatus between the Houli and Beixin Cultures. This indicates that the Houli-Beixin chronological gap is likely an artifact and might soon be overcome when more finds are systematically radiocarbon dated (Sun et al., 2014).
The Dawenkou Culture (c. 4100–2600 BC), investigated in the current study, is by far the longest continual prehistoric cultural phase designated by archaeologists in the Shandong region, lasting about 1500 years. The published radiocarbon dates (Wagner et al., 2009) would allow an even longer time span, reaching into the preceding Beixin Culture and subsequent Longshan Culture. This highlights several important issues for future research (Wagner et al., 2009). First, the transitions from one cultural complex to another is a blurred boundary and ultimately just a construct of the archaeologists. Moreover, micro-regions express cultural developments of their own. While many areas in the world are moving past rigid cultural historic frameworks, strict chronological borderlines are still an important component of Chinese archaeology (e.g. Lee et al., 2007; Liu and Chen, 2012; Ruddiman et al., 2008; Wagner et al., 2013). However, systematic AMS dating of the key archaeological sequences and newly excavated sites is absolutely necessary for verification of the main cultural boundaries. In particular, robust reconstruction of the leads/lags and spatial/temporal patterns of crop cultivation and agricultural activities is impossible without direct dating of the investigated plant remains. The two wheat grains recovered from the Dawenkou cultural layer at Beiqian are a good example. Without direct AMS dating, these wheat grains would suggest the introduction of wheat to China as early as 6000–5000 years ago, thus predating all earlier published records.
Pattern of rice exploitation in Shandong
In the modern world, rice is one of most important crop species in the world today. Rice agriculture was important for the development of East and South Asian civilization. Therefore, it is not surprising that scientists are so interested in the origin and early development of rice. Molina et al. (2011) reported that archaeological and genetic studies basically agree that all forms of Asian rice are the result of a single domestication from the wild progenitor (Oryza rufipogon) that occurred sometime between 11,500 and 6200 BC in the southeastern part of China. However, this is an oversimplification of the results and the story is likely much more complicated. Geneticists argue that the first cultivation of rice occurred in the middle Pearl River valley, Guangxi province (Huang et al., 2012), although archaeological records point toward the lower Yangtze region as one of the earliest centers of rice cultivation (Fuller et al., 2009; Lee et al., 2007). This picture is complicated by the Jiahu site (c. 7000–5800 cal. BC) (Figure 1), located in the upper Huai River valley, on the watershed between the Yellow and Yangtze River, clearly exhibiting characteristics of rice agriculture as early, if not earlier than sites on the Yangtze River (Zhao, 2011). However, some authors downplay the Jiahu data noting that it is not clear that the rice was grown in human-prepared paddy-fields, as can be shown at the Yangtze River sites c. 4500 cal. BC (Cohen, 2014 and references therein).
New discoveries of c. 8000-year-old rice remains at the Houli Culture sites of Yuezhuang (Crawford et al., 2006) and Xihe (Jin et al., 2014) in the lower Yellow River region complicated the above debates even more, as both of these sites are situated far from the traditionally suggested source areas of rice agriculture. Keeping in mind that it is still unclear whether the Yuezhuang and Xihe rice was gathered from the wild or cultivated, Jin et al. (2014) put forward several questions and proposed future research directions. One of the key issues remains an apparent discontinuity of the rice exploitation in the Shandong region and lack of rice remains in the cultural layers between c. 5500 and 3000 cal. BC (Fuller et al., 2009b).
The presence of rice remains (i.e. grains and phytoliths) dated to 4000–3800 cal. BC at Dongpan, and c. 3700–2900 cal. BC at Beiqian closes the later part of the existing gap in the record of rice cultivation. The data from these sites may point toward a semi-continuous use of rice in Shandong, possibly spanning back to the beginning of the Middle Neolithic. However, the data presented here are insufficient to answer questions regarding the importance of any of these grains, including rice, in the economy, or even whether they were locally cultivated or exchange goods.
It is generally accepted that millet cultivation started in the Early Neolithic somewhere to the north of the Yellow River or in the Yellow River valley (Figure 1). Millet cultivation continued through the Middle and Late Neolithic in this region, with millet being the main staple food source (e.g. Lee et al., 2007; Zhao, 2011). At approximately the same time in the Yangtze and Huai River region, archaeological evidence reveals the beginnings of rice cultivation/domestication, with rice becoming the most important agricultural product (e.g. Fuller et al., 2009; Ruddiman et al., 2008; Zhao, 2011). Lee et al. (2007) claim that rice was not brought to the Yiluo River area, in the middle of the Yellow River, until c. 3000 cal. BC (or slightly earlier, as suggested by Li et al., 2007b), and rice was not an important crop there, likely because of the relatively dry climate. While northern China in general has a relatively dry environment, the area of Shandong is well suited for both millet and rice cultivation, with a warm (i.e. mean July temperatures are above 25°C) climate (Domrös and Peng, 1988; Tarasov et al., 2009), fertile soils, an abundant river water supply, and heavy precipitation during the summer monsoon season, which continues from May to September. Some evidence (Luo, 2014) suggests that morphologically wild forms of rice may have been grown in the Huai River valley during the middle Holocene, that is, that this area was within the natural range for wild rice. The presence of the wild rice in the region would explain rice finds at the Houli Culture sites of Xihe and Yuezhuang. It is plausible that the Early Neolithic population of the north, being familiar with millet, collected grains from wild stands of Oryza rufipogon. The archaeobotanical record from Xihe shows that rice contributed c. 36% to the macrobotanical assemblage dating to 6070–5900 cal. BC, while foxtail millet contributed about 1% (Jin et al., 2014). The data from Yuezhuang (Crawford et al., 2006) demonstrate more equal proportions of millet (n = 41 grains) to rice (n = 26 grains). The archaeobotanical record from the Zhangmatun site, dated to 7080–6460 cal. BC, contains no rice, and millets represent less than 1% of the total assemblage, which is dominated by the wild plant seeds (Wu et al., 2014).
The data at hand are not enough to make any major conclusions about the importance of rice and millet in the economy of local people by the end of the Beixin Culture period. We may speculate that the relatively high values of rice and millet phytoliths indicate local production of these crops; however, it is unlikely that they made a major contribution to the diet. Such interpretation basically agrees with the low number of archaeological sites assigned to Beixin (i.e. 26) in the provincial archaeological atlas (Wenwuju, 2007; Wagner et al., 2013). This number is only slightly higher than during the Houli Culture period (11 registered sites), which is known for its predominantly hunting–fishing–gathering subsistence strategy (Jin et al., 2014; Wagner et al., 2009).
The figure provided by the research results from Beiqian is very different. Although rice remains a minor component of the macrofossil assemblage (Table 4), broomcorn (71%) and foxtail (3.7%) millets do not leave any doubts about their contribution to the diet between c. 3700 and 2900 cal. BC. Change in subsistence toward intensive millet agriculture results in a major increase in archaeological sites from 26 to 563 during the Dawenkou period (Wenwuju, 2007; Wagner et al., 2013) and their spread from the foothills and mountains of central Shandong, favored by the Neolithic hunter-gatherers, across the whole region suitable for agriculture (Wagner et al., 2009). A major intensification of land use and agricultural activities after c. 3700 cal. BC has been reported for the millet-oriented middle Yellow River region (Lee et al., 2007) and in the southern rice-cultivating areas (Ruddiman et al., 2008).
During the Dawenkou Culture period, intensified rice agriculture is reported for the settlements in the Huai and lower Yangtze River region, such as Longqiuzhuang (Tang et al., 1996), Zhoubeidun (Xiao, 1997), Shuangduan, Houjiazhai, Hongdunsi (Zhang et al., 2004), and Xiaosungang (Luo, 2014). In all these cases, settlement densities indicate that the growing population size during the Dawenkou period correlates with an intensification of agriculture.
Conclusions
While historically rice, broomcorn, and foxtail millet were grown in the Shandong region, none of these crops are thought to have been domesticated there. Therefore, in this paper, we explore the question of their spread into this area and the local adoption of these grains into the economy. Recent archaeobotanical studies have shown that early rice grains (likely wild) were present, c. 6000 cal. BC, at the sites of Xihe and Yuezhuang. These new data have created a chronological gap between rice remains from these two sites, assigned to Houli Culture (c. 6500–5500 BC), north of the Shandong Highlands, and sites of the Longshan Culture (c. 2600–1900 BC), which express solidly confirmed rice agriculture.
In the current article, we present new results from macrobotanical and phytolith analyses, including directly dated rice grains from the sites of Dongpan and Beiqian. Although the dataset is insufficient to answer questions about the importance of rice and millet in the local economy, it allows to draw some conclusions:
The AMS dates from Dongpan fell between 4030 and 3820 cal. BC (95% range), placing occupation at the site during the Early Dawenkou (c. 4100–3500 BC) cultural period. Eleven of the 13 dates from Beiqian fell within the age range of c. 3700 and 2900 cal. BC (95% range), that is, the middle phase of the Dawenkou Culture (e.g. 3500–3000 BC). Interestingly, the identification and direct dating of rice remains from about 4000–3800 cal. BC at Dongpan, and c. 3700–2900 cal. BC at Beiqian, partly fills a dearth of data on early rice cultivation in the Shandong area.
At first glance, the low density of rice and millet grains compared with the other seeds recovered at the Dongpan site from the layer dated to c. 4030–3900 cal. BC does not provide much support for these crops being important food resource in the local economy. Earlier publications from the region suggest strong reliance on wild resources (i.e. hunting–fishing–gathering) during the Houli and Beixin Culture periods, but also during the early phase of the Dawenkou Culture. Higher values obtained for the rice and millet associated phytoliths, and particularly husk phytoliths, suggest local processing of these grains. Based on available data, we assume that rice and millet cultivation was low-investment and only a small part of the overall economy at Dongpan during the end of the Beixin Culture period. However, the truth is also that the site clearly has low densities and/or poor preservation of seeds. This prevents us from making further claims about what parts of the diet were important at the Dongpan site.
At Beiqian, the archaeobotanical assemblage is dominated by broomcorn millet (71%); foxtail millet (3.7%) and rice are present but less prevalent. These data support theories that the Middle Dawenkou period was characterized by a greater investment in agriculture, although wild foraging was common practice.
Systematic archaeological and archaeobotanical research on Houli, Beixin, and Dawenkou Culture sites and the potential farming areas surrounding the settlements will be necessary in the future to address questions concerning the contribution of a specific food source to the diet of any particular community. Studying the ratios of millets and rice in a diachronic perspective and in comparison between northern and southern areas of the Shandong Highlands will illuminate the cultural contacts between the inhabitants of Neolithic Shandong and the areas of the Huai and Yangtze Rivers to the south and will eventually help to understand early rice cultivation and its spread and intensification across the East China Plain.
Footnotes
Acknowledgements
The manuscript was discussed when JGY was attending the Wenner Gren Foundation funded Conference ‘Introduction and Intensification of Agriculture in Eurasia’ (Grant CONF-673). The authors are grateful to both anonymous reviewers for their constructive suggestions and to Dr R. Spengler for the language polishing and first critical reading of the manuscript.
Funding
The work in China was supported by the CAS Strategic Priority Research Program (Grant No. XDA05130603-B), the National Natural Science Foundation of China (Grant No. 41072135), and the National Social Science Foundation of China (Grant No. 11AZD116). The work of PET and MW is a contribution to the ‘Bridging Eurasia’ research initiative supported by the German Archaeological Institute (DAI), Center for International Cooperation of the Free University Berlin and German Research Foundation (DFG TA 540/5).