2021 年 129 巻 2 号 p. 203-222
The Yangtze River Delta is the best-known homeland of wet-rice agriculture. From the Middle Neolithic, rice farming expanded from the Yangtze region to both the north and the south. However, poor preservation of ancient human skeletal remains in the region has meant that the population history of these expansions has not been fully understood. In order to clarify the ancestry of early wet-rice farmers in East Asia, we conducted a cranial morphometric analysis and comparison of a Middle Neolithic skeletal assemblage excavated from the Guangfulin site, Shanghai. The results of bivariate and multivariate analyses showed that: (1) Neolithic wet-rice farmers from the lower Yangtze retained local morphological characteristics, but were nevertheless morphologically more similar to Neolithic and later populations in northern China, which was home to early millet farmers, than to Neolithic populations in south China; and (2) Neolithic and later agricultural populations in East Asia were morphologically homogeneous compared to pre-Neolithic hunter-gatherer groups even though the area occupied by both was equally vast. These results suggest, respectively, that: (1) Middle Neolithic wet-rice farmers in the Yangtze Delta experienced significant gene flow from regions of northern China such as the Central Plains and Shandong even though there is currently no evidence that millet cultivation itself had yet reached the delta region; and (2) Neolithic populations resulting from interaction between the Yangtze Delta and northern China dispersed widely across much of East Asia including the Japanese archipelago together with the spread of wet-rice agricultural technologies. These two proposals are paralleled by recent stable isotope analyses using tooth enamel and bone collagen, as well as archaeological evidence from Shandong. Finally, a facial approximation was conducted using a cranium (M252) excavated from Guangfulin for the purpose of visually expressing the results of this study.
The Neolithic revolution was one of the most significant milestones in human history, especially from the perspective of demography and the Neolithic demographic transition (Bellwood, 2005; Bocquet-Appel and Naji, 2006; Bellwood and Oxenham, 2008; Bocquet-Appel and Bar-Yosef, 2008). Neolithic agriculture often increased birth rates and led to population growth and diffusion. Although many researchers have focused on the Neolithic demographic transition, its impact on early farmers in the homeland zone of wet-rice domestication has rarely been discussed. In East Asia, the geographical line along the Qinling Mountains and Huai River climatically divides the Chinese mainland into northern and southern regions (e.g. Jiang et al., 2018). In the Neolithic, foxtail and broomcorn millet agriculture provided the main living subsistence in northern China (north of the Qinling–Huai line) with wet-rice farming in southern China (south of the line), although both territories changed and mixed over time (e.g. Fuller, 2011; Guedes et al., 2015; Zhang and Hung, 2015; Lu, 2017; Miyamoto, 2017; Stevens and Fuller, 2017; Leipe et al., 2019; Liu et al., 2020). Five separate centers of Neolithic cultures developed millet agriculture in northern China based on current archaeological evidence (Peiligang in northern Henan, Cishan in southern Hebei, Houli in western Shandong, Xinglongwa in Inner Mongolia, Dadiwan in Gansu) (e.g. Stevens and Fuller, 2017). In the south, the Yangtze River Delta can be considered one of the homeland zones of wet-rice domestication according to archaeological evidence excavated from Early Neolithic sites such as Shangshan, Kuahuqiao, Hemudu, and Tianluoshan (Zong et al., 2007; Fuller et al., 2009). Furthermore, urbanization developed in the Yangtze at Middle Neolithic sites such as the Liangzhu (良渚) complex (Renfrew and Liu, 2018). Taking account of the fact that wet-rice agriculture was later established in the Korean peninsula, the Japanese archipelago, and parts of Southeast Asia, it is necessary to study population movements and diffusion together with rice farming to understand the population history of East Asia (Bellwood, 2005; Bellwood and Oxenham, 2008). Revealing the skeletal traits of the early rice farmers thus could be one of the clues for understanding the roots and origins of East Asian groups including the modern Japanese.
This study focuses on cranial morphology and examines the relationship of the early farmers among both the homeland and spread zones of wet-rice agriculture in East Asia. Few well-preserved prehistoric skeletal collections have been reported from the Yangtze River Delta. Only two sets of Early Neolithic skeletal materials have been reported from this region: the Hemudu (河姆渡) site (c. 5000–4500 BCE) from Zhejiang province and the Weidun (圩墩) site (c. 5000–4000 BCE) from Jiangsu province. The cranial morphology of Hemudu shows individuals with relatively long cranial length, low facial height, wide nasal breadth, and low orbital height who have been categorized as exhibiting an ‘ancient south China’ morphology by Han and Pan (1983). By contrast, the cranial morphology of Weidun indicates a different trend and has been categorized as possessing an ‘ancient Central Plains’ morphology (Zhu, 2004). However, a phylogenetic tree based on 16 cranial measurements found that both Hemudu and Weidun are separately grouped into neighboring clusters (Matsumura et al., 2019). Thus, the results of the previous studies need to be re-examined, especially in light of the small sample size of the Hemudu materials. Additional materials should be studied to evaluate the skeletal features of the Neolithic people in the Yangtze Delta. This study evaluates the Middle Neolithic skeletal remains unearthed from the Guangfulin (広富林) (c. 3900–2400 BCE) site in the Yangtze Delta. A morphometric analysis is performed to examine whether the early wet-rice farmers from Guangfulin were similar to north (i.e. Hebei, Shanxi, Inner Mongolia) or south (i.e. Fujian, Guangdong, Guangxi) China groups. The present study also aims to understand whether or not the Neolithic farmers of the Yangtze Delta and their successors who introduced wet-rice agriculture to surrounding regions shared the same ancestry.
Guangfulin SiteThe site is located on the flatlands of Songjiang District of Shanghai City. The site is approximately 60 km east of one of the five major lakes in China, Lake Tai, and nearly 100 km east-northeast of the Liangzhu site complex mentioned above (Figure 1). The excavations were carried out between 1999 and 2015, ultimately resulting in the recovery of more than 300 human skeletal individuals (Archaeological Department of Shanghai Museum, 2002, 2008). These skeletal remains date to the Songze (c. 3900–3200 BCE) and Liangzhu (c. 3200–2400 BCE) cultural periods, based on a combination of archaeological evidence, including stratigraphy and analysis of associated burial goods, as well as radiocarbon dating (5305 ± 130 and 4690 ± 150 calBP) of carbonized material recovered from the Songze and Liangzhu cultural layers, respectively (Chen et al., 2007; Okazaki et al., 2019).
Locations of the Guangfulin site and other major Neolithic sites in the Yangtze River Delta. Early Neolithic evidence for wet-rice domestication is found in Zhejiang Province at the following sites: Shangshan, Kuahuqiao, Hemudu, and Tianluoshan; evidence for urbanization is found at the Liangzhu site. The photograph shows the 2010 excavation of buried individuals at the Guangfulin site.
The sample size of the Guangfulin skeletal assemblage is the largest of all archaeological sites in the Yangtze River Delta. Paleodemographic data indicate that 77.5% of the total individuals (N = 182) were adults (over 20 years old, N = 141) (Okazaki, unpublished data). Among the 105 adults who could be aged or sexed, young adults (20–40 years old) were most prevalent (N = 73, 23 males and 46 females), followed by middle-aged adults (40–60 years) (N = 31, 19 males and 12 females), and old adults (over 60 years old) (N = 1, male). The disparity in the frequency of females relative to males, particularly in the young adult demographic, suggests a heightened risk of morbidity and mortality, possibly due to obstetric-related complications (e.g. Šlaus, 2000).
Guangfulin is a cemetery site and associated residential or agricultural field remains have not yet been detected. However, given the location and date of the site, it is probable that the Guangfulin population was engaged in rice farming. The archaeological site report for Guangfulin is currently in preparation under the supervision of one of the present authors (J.C.). Soil flotation has confirmed the presence of carbonized rice, gourd, melon, prickly waterlily, water chestnut, peach, and Japanese plum, and evidence for hunting of deer, wild boar or pig, crocodile, crane, wild goose, and elephant is also present. Although detailed analyses have yet to be completed, these plants and animals may also have been used to provision the larger Liangzhu site complex (Liu et al., 2017; Nakamura, 2015), which researchers suggest may represent the beginnings of an early state. The Liangzhu site complex shows evidence of urbanization including: ramparts surrounding the area and waterways connecting to river networks; the construction of dams, evidenced by the presence of sandbags; apparent specialization of labor dedicated to the production of elaborate jade goods; and remains of a granary that could have contained up to 15 tonnes of rice grain.
The skeletal materials considered in this study stem from the human skeletal remains excavated at the Guangfulin site in 2010. It was possible to measure the crania of 17 out of 43 adult male individuals because they were relatively well-preserved. The materials of five other skeletal assemblages were used for detailed comparisons using individual data not only because of their geographical proximity to Guangfulin but also their large sample sizes: the Neolithic groups of (1) Weidun in Jiangsu province, (2) Jiangjialiang (姜家梁) in Hebei province, and the Eastern Zhou groups of (3) Jiangsu, (4) Henan (Xinghong (興弘) and Zhouzhuang (周庄)), and (5) Inner Mongolia (Tuchengzi (土城子)) (Table 1, Table 2). Furthermore, the materials of 35 skeletal assemblages were added for global comparison in East Asia using mean data of cranial measurements (Table 1). Age at death and sex of the Guangfulin individuals were estimated using standard morphological observations of the cranium and os coxae (Buikstra and Ubelaker, 1994; White and Folkens, 2005). Biological sex was assessed using direct examination of diagnostic features of the sexually dimorphic hip bone (Phenice, 1969; Bruzek, 2002; Walker, 2005; Takahashi, 2006). Age at death was estimated based on age-related changes in the pubic symphysis (Todd, 1920; Hanihara, 1952; Sakaue, 2006) and the auricular surface morphology (Lovejoy et al., 1985; Buckberry and Chamberlain, 2002; Igarashi et al., 2005). When the hip bone was unavailable, assessment of biological sex followed metric assessment of postcranial bones (Nakahashi and Nagai, 1986) and cranial features (Buikstra and Ubelaker, 1994). Estimation of age at death was based on evaluation of the degree of obliteration of cranial sutures if the hip bone was not preserved (Meindl and Lovejoy, 1985; Sakaue, 2015).
| Groups | Regions | Date | Periods1 | Cereal2 | References | |
|---|---|---|---|---|---|---|
| Guangfulin | 広富林 | Shanghai | c. 5287–4619 cal BP | Songze—Liangzhu culture | rice | This study |
| Weidun | 圩墩 | Jiangsu | c. BCE 5000–4000 | Majiabang culture | rice | Nakahashi et al. (2002) |
| Hemudu | 河姆渡 | Zhejiang | c. BCE 5000–4500 | Hemudu culture | rice | Han and Pan (1983) |
| Jiangsu | 江蘇 | Jiangsu | c. BCE 771–8 CE | Eastern Zhou—western Han Dynasty | rice | Nakahashi et al. (2002) |
| Tanshishan | 昙 石山文化 | Fujian | c. 3905 BP | Tanshishan culture | rice | Han et al. (1976) |
| Hedang | 河宕 | Guangdong | c. 3800–3600 cal BP | rice | Han and Pan (1982) | |
| Zengpiyan | 甑皮岩 | Guangxi | c. 6600 BP | Mesolithic | Zhang et al. (1977), Okazaki (unpublished data) | |
| Liujiang | 柳江 | Guangxi | c. ~68000 BP | Upper Paleolithic | Woo (1959), Wu and Zhang (1985) | |
| Gaomiao | 高廟 | Hunan | c. 6500 cal BP | Mesolithic | Matsumura et al. (2017) | |
| Hoabinhian | Vietnam | c. 10000–8000 cal BP | Hoabinhian culture | Matsumura et al. (2011) | ||
| Man Bac 1 | Ninh Bình | c. 3800–3500 cal BP | Phùng Nguyên culture | rice | Matsumura (2011) | |
| Man Bac 2 | Ninh Bình | c. 3800–3500 cal BP | Phùng Nguyên culture | rice | Matsumura (2011) | |
| Shandingdong 101 | 山頂洞 101号 | Beijing | c. 33200–10175 BP | Upper Paleolithic | Wu (1961), Wu and Zhang (1985) | |
| Dawenkou | 大汶口 | Shandong | c. BCE 4300–2600 | Dawenkou culture | millet | Yan (1972) |
| Beiqian | 北阡 | Shandong | c. BCE 4300–4000 | Early Dawenkou culture | millet | Nakahashi et al. (2013b) |
| Dinggong | 丁公 | Shandong | c. BCE 3000–2000 | Longshan culture | millet and rice | Nakahashi and Luan (2008) |
| Yedian | 野店 | Shandong | c. 6170–4640 cal BP | Dawenkou culture | millet | Zhang (1980) |
| Xixiahou | 西夏侯 | Shandong | Dawenkou culture | millet | Yan (1973) | |
| Wangyin | 王因 | Shandong | c. 6020–5225 cal BP | Dawenkou culture | millet | Han (2000) |
| Banbo | 半坡 | Shanxi | Yangshao culture | millet | Yan et al. (1960b) | |
| Xiawanggang | 下王崗 | Henan | Yangshao culture | millet | Zhang and Chen (1984) | |
| Miaodigou | 廟底沟 | Henan | c. BCE 3900–2780 | Yangshao—Early Longshan culture | millet | Han and Pan (1979) |
| Baoji | 宝鶏 | Shanxi | c. 6075–4885 cal BP | Yangshao culture | millet | Yan et al. (1960a) |
| Hengzhen | 横陣 | Shanxi | Yangshao—Longshan culture | millet? | Institute of Archaeology, LPA (1977) | |
| Jiangjialiang | 姜家梁 | Hebei | c. 4156–4112 cal BP | Longshan culture | millet | Okazaki et al. (unpublished data, 2013) |
| Yinxu (noble) | 殷墟 | Henan | c. BCE 1400–1100 | Late Shang Dynasty period | Han and Pan (1985) | |
| Yinxu (citizen) | 殷墟 | Henan | c. BCE 1400–1100 | Late Shang Dynasty period | Han and Pan (1985) | |
| Yinxu (slave) | 殷墟 | Henan | c. BCE 1400–1100 | Late Shang Dynasty period | Han and Pan (1985) | |
| Xinghong | 興弘 | Henan | c. BCE 771–221 | Eastern Zhou Dynasty period | Nakahashi (2014), Okazaki et al. (2016a) | |
| Zhouzhuang | 周庄 | Henan | c. BCE 771–221 | Eastern Zhou Dynasty period | Nakahashi (2014) | |
| Tuchengzi | 土城子 | Inner Mongolia | c. BCE 771–206 | Eastern Zhou and Qin Dynasty period | Nakahashi (unpublished data), Okazaki et al. (2016a) | |
| Minghegetaozhuang | 民和核桃庄 | Qinghai | c. BCE 1500–1000 | Xindian culture | Wang and Zhu (2004) | |
| Liuwan | 柳湾 | Qinghai | c. 4350–4050 cal BP | Machang culture | Pan and Han (1984) | |
| Chuanxi highlands | 川西高原 | Sichuan | c. BCE 1500–800 | Shang Dynasty period | Nakahashi et al., (2013a) | |
| Outer Mongolia | 外蒙古 | Outer Mongolia | c. BCE 1392–812 | Pre-Xiongnu period | Okazaki et al. (2016b), Okazaki and Yonemoto (2017, 2018) | |
| Minatogawa 1 | 港川 1号 | Okinawa | c. 21800–19900 cal BP | Upper Paleolithic | Suzuki (1982) | |
| Shiraho 4 | 白保竿根田原 4号 | Okinawa | c. 27759–27433 cal BP | Upper Paleolithic | Kono et al. (2018) | |
| Tsukumo Jomon | 津雲縄文 | Okayama | Middle-late Jomon culture | Kawakubo (personal communication), Ikeda (1988) | ||
| TY Jomon | 津雲・吉胡縄文 | Aichi/Okayama | Middle-late Jomon culture | Kiyono and Miyamoto (1926), Kintaka (1928) | ||
| NK-Y Yayoi | 北部九州・山口弥生 | Fukuoka/Yamaguchi | c. BCE 400–300 CE | Middle-late Yayoi culture | Nakahashi and Nagai (1989) | |
| Aoya Yayoi | 青谷上寺地 | Tottori | c. 0–200 CE | Late Yayoi culture | Okazaki (unpublished data) |
Abbreviations: TY, Tsukumo and Yoshigo; NK-Y, north Kyushu and Yamaguchi.
| Martin | 遺跡 | 広富林 | 圩墩 1 | 姜家梁 2 | 江蘇東周・前漢 1 | 河南東周 3 | 土城子 4 | One-way ANOVA | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Sites | Guangfulin | Weidun | Jiangjialiang | Jiangsu complex | Henan complex | Tuchengzi | |||||||||||||||
| Regions | Shanghai | Jiangsu | Hebei | Jiangsu | Henan | Inner Mongolia | |||||||||||||||
| Cultural periods | Songze-Liangzhu | Majiabang | Yangshao-Longshan | Eastern Zhou-Han | Eastern Zhou | Eastern Zhou | |||||||||||||||
| 文化期 | 崧沢・良渚 | 馬家浜 | 仰韶―龍山過渡期 | 東周・前漢 | 東周 | 東周 | |||||||||||||||
| Estimated ages | c. 5900–4200 BP | c. 5000–4000 BCE | c. 4800–4500 BP | ||||||||||||||||||
| N | Mean | SD | N | Mean | SD | N | Mean | SD | N | Mean | SD | N | Mean | SD | N | Mean | SD | F-value | P-value5 | ||
| 1 | Maximum cranial length | 11 | 186.1 | 10.55 | 17 | 178.7 | 7.93 | 25 | 178.0 | 7.94 | 18 | 177.4 | 5.49 | 54 | 182.6 | 6.22 | 56 | 182.2 | 6.42 | 4.3 | 0.005 |
| 5 | Basal length | 4 | 105.0 | 5.77 | 8 | 101.9 | 4.97 | 18 | 102.4 | 5.09 | 13 | 100.0 | 5.55 | 37 | 102.7 | 4.43 | 40 | 103.6 | 3.85 | 1.5 | 0.202 |
| 8 | Maximum cranial breadth | 12 | 144.0 | 4.84 | 18 | 141.3 | 4.91 | 24 | 136.2 | 7.51** | 18 | 140.7 | 5.12 | 58 | 144.1 | 4.49 | 59 | 141.3 | 3.82 | 9.6 | 0.000 |
| 8:1 | Cranial length-breadth index | 9 | 76.9 | 3.07 | 15 | 79.5 | 4.95 | 24 | 76.8 | 6.77 | 17 | 79.5 | 3.31 | 54 | 78.9 | 3.67 | 54 | 77.4 | 3.05 | 2.1 | 0.067 |
| 9 | Minimum frontal breadth | 11 | 97.6 | 4.93 | 19 | 96.7 | 4.40 | 24 | 90.5 | 3.99** | 17 | 92.9 | 3.22* | 54 | 94.3 | 4.08 | 56 | 93.1 | 4.97** | 6.6 | 0.000 |
| 17 | Basibregmatic height | 6 | 145.3 | 3.78 | 11 | 141.6 | 2.80 | 19 | 139.4 | 4.00* | 14 | 137.4 | 6.15** | 43 | 143.0 | 5.31 | 48 | 141.1 | 4.77 | 4.3 | 0.001 |
| 17:1 | Cranial length-height index | 6 | 80.6 | 1.92 | 10 | 79.4 | 4.18 | 19 | 78.7 | 3.44 | 14 | 78.0 | 3.72 | 42 | 78.2 | 3.30 | 43 | 77.6 | 2.98 | 1.3 | 0.272 |
| 17:8 | Cranial breadth-height index | 6 | 102.7 | 3.34 | 11 | 99.9 | 5.23 | 19 | 103.1 | 6.82 | 14 | 97.4 | 4.75 | 42 | 98.9 | 4.58 | 46 | 99.8 | 4.26 | 3.1 | 0.011 |
| 40 | Facial profile length | 2 | 97.0 | 7.07 | 5 | 101.4 | 2.51 | 17 | 100.5 | 4.94 | 13 | 95.8 | 3.44 | 28 | 99.9 | 6.00 | 38 | 97.4 | 4.27 | 2.6 | 0.030 |
| 45 | Bizygomatic breadth | 3 | 141.2 | 3.75 | 9 | 134.3 | 3.43 | 21 | 136.1 | 5.03 | 13 | 138.3 | 4.37 | 44 | 137.2 | 6.20 | 42 | 135.3 | 4.42 | 1.8 | 0.125 |
| 48 (sd) | Upper facial height | 10 | 71.9 | 2.59 | 19 | 71.7 | 3.71 | 30 | 76.8 | 4.07** | 17 | 75.2 | 3.71 | 56 | 76.1 | 4.09** | 53 | 75.9 | 3.75* | 6.4 | 0.000 |
| 48:45 | Upper facial index | 3 | 52.3 | 2.73 | 8 | 53.1 | 3.30 | 19 | 56.1 | 3.55 | 13 | 54.5 | 2.93 | 40 | 55.8 | 3.47 | 42 | 56.2 | 2.58 | 2.5 | 0.037 |
| 51 | Orbital breadth | 9 | 44.1 | 2.60 | 19 | 43.2 | 1.36 | 25 | 43.3 | 1.73 | 17 | 42.4 | 1.33 | 52 | 42.8 | 1.72 | 52 | 42.3 | 1.90* | 2.3 | 0.044 |
| 52 | Orbital height | 10 | 35.5 | 2.37 | 19 | 34.0 | 1.73 | 25 | 33.7 | 1.68 | 17 | 35.1 | 1.97 | 57 | 34.2 | 2.06 | 51 | 34.0 | 2.16 | 1.7 | 0.130 |
| 52:51 | Orbital index | 9 | 81.7 | 4.63 | 18 | 79.0 | 5.05 | 25 | 78.1 | 4.57 | 17 | 82.6 | 4.75 | 51 | 79.9 | 4.37 | 51 | 80.5 | 5.28 | 2.3 | 0.045 |
| 54 | Nasal breadth | 9 | 27.5 | 1.87 | 21 | 28.4 | 2.13 | 27 | 26.8 | 2.32 | 17 | 28.1 | 2.61 | 57 | 27.1 | 1.76 | 54 | 26.4 | 1.74 | 4.2 | 0.001 |
| 55 | Nasal height | 10 | 53.3 | 2.51 | 19 | 52.8 | 2.15 | 28 | 53.8 | 2.86 | 17 | 53.5 | 1.91 | 56 | 53.4 | 2.76 | 53 | 53.4 | 2.78 | 0.4 | 0.877 |
| 54:55 | Nasal index | 9 | 51.8 | 3.72 | 18 | 54.6 | 5.08 | 27 | 49.9 | 5.12 | 17 | 52.4 | 4.73 | 54 | 50.9 | 3.97 | 53 | 49.5 | 4.09 | 4.4 | 0.001 |
| FC | Frontal chord | 8 | 99.4 | 4.50 | 14 | 97.7 | 3.83 | 23 | 97.0 | 3.66 | 15 | 97.4 | 2.95 | 37 | 98.8 | 3.50 | 44 | 96.5 | 3.12 | 2.5 | 0.037 |
| FS | Frontal subtense | 7 | 16.6 | 2.79 | 14 | 15.4 | 2.78 | 23 | 14.8 | 2.53 | 15 | 16.3 | 3.02 | 37 | 14.5 | 2.10 | 44 | 15.0 | 2.56 | 1.7 | 0.130 |
| FS:FC | Frontal index of flatness | 7 | 16.6 | 2.83 | 14 | 15.7 | 2.62 | 23 | 15.3 | 2.43 | 15 | 16.7 | 2.93 | 37 | 14.7 | 2.13 | 44 | 15.6 | 2.50 | 1.9 | 0.102 |
| SC | Simotic chord | 10 | 8.8 | 2.01 | 18 | 9.0 | 2.41 | 22 | 7.1 | 2.46 | 16 | 8.4 | 0.89 | 42 | 7.6 | 1.66 | 45 | 7.9 | 2.16 | 2.6 | 0.039 |
| SS | Simotic subtense | 10 | 2.3 | 0.93 | 16 | 2.4 | 0.83 | 15 | 2.3 | 0.81 | 16 | 2.9 | 0.58 | 39 | 2.2 | 0.71 | 41 | 2.4 | 0.89 | 1.6 | 0.175 |
| SS:SC | Simotic index of flatness | 10 | 26.3 | 8.68 | 16 | 26.8 | 9.50 | 15 | 28.3 | 7.12 | 16 | 34.3 | 7.77 | 39 | 28.7 | 9.61 | 41 | 30.3 | 11.85 | 1.4 | 0.245 |
| ZC | Zygomaxillary chord | 6 | 104.3 | 4.24 | 12 | 104.7 | 4.49 | 20 | 106.5 | 4.84 | 15 | 103.8 | 7.25 | 24 | 101.8 | 4.52 | 39 | 99.3 | 5.40 | 5.9 | 0.000 |
| ZS | Zygomaxillary subtense | 6 | 22.9 | 3.86 | 10 | 21.8 | 4.19 | 20 | 22.7 | 3.15 | 15 | 23.2 | 3.34 | 24 | 21.9 | 2.83 | 39 | 22.2 | 3.26 | 0.4 | 0.819 |
| ZS:ZC | Zygomaxillary index of flatness | 6 | 21.9 | 3.34 | 10 | 20.9 | 3.99 | 20 | 21.3 | 3.09 | 15 | 22.4 | 2.94 | 24 | 21.6 | 2.84 | 39 | 22.4 | 3.01 | 0.7 | 0.662 |
Dunnett’s test was used for multiple comparison between each pair of Guangfulin and the others. The Games–Howell test was used when the homogeneity of variance was not guaranteed by Levene’s test.
The cranial measurements were selected from those used in previous Japanese–Chinese joint research conducted in Jiangsu, Henan, and Shandong (Nakahashi, 2014; Nakahashi et al., 2002, 2013b) to enable comparisons between Guangfulin and the comparative assemblages (Table 2). Measurements were recorded following the methods of Martin and Saller (1957), which are coordinated and summarized in Baba (1991). Measures of facial flatness follow Yamaguchi (1973).
The difference of mean values was examined among groups by a one-way analysis of variance (ANOVA). Dunnett’s test was then applied for multiple comparisons between each pair of Guangfulin and the other samples. Welch’s test (a robust test of the equality of means) and the Games–Howell test were used instead of one-way ANOVA and Dunnett’s test, respectively, in cases where the homogeneity of variance was not guaranteed by Levene’s test.
Discriminant function analysis was used to detect morphometric differences between Neolithic wet-rice and millet farmers using individual data without missing values. The function was calculated based on 12 cranial measurements. It was then applied to Eastern Zhou–Han Dynasty individuals to confirm whether the morphometric difference seen in the Neolithic period was valid in the later period.
The Mahalanobis generalized distances were calculated using 12 cranial measurements for global comparison in East Asia. Generalized distance provides a quantitative scaling of dissimilarity between groups using several variables (measurements) while removing the correlation between the variables (Mahalanobis, 1936). The variance–covariance matrix to evaluate the correlation was derived from individual data of the Eastern Zhou Dynasty in northern China (N = 49, pooled from Tuchengzi, Xinghong, and Zhouzhuang) because the numbers of samples without missing values were relatively large in these skeletal series. This procedure allows use of the mean values of skeletal samples published in previous studies without the corresponding individual data. In this study, mean values for each group are applied to the Mahalanobis generalized distances assuming that the mean value of all the measurable specimens for each measurement is the representative value of the group. Although the reliability of the results in this analysis is lower than analysis based on individual data only, this was determined to be the best way to analyze the present data set given limitations in data availability. The results of the Mahalanobis generalized distances were expressed through multidimensional scaling.
These analyses were performed using Excel (Microsoft) and the computer program package SPSS Statistics 24 (IBM).
Table 2 and Figure 2 and Figure 3 show the comparison of cranial measurements between Guangfulin and the five comparative assemblages. Table 2 indicates the descriptive statistics values and Figure 2 and Figure 3 visualize the distribution of the individual data. In the neurocranium, Guangfulin had the largest mean in cranial length (186.1 mm, Figure 2A), and the smallest in cranial length–breadth index (76.9, Figure 2F), and was thus characterized by a relatively long and narrow head. Guangfulin also had the largest mean in basibregmatic height (145.3 mm, Figure 2E) and cranial length–height index (80.6, Figure 2G), and the second largest mean in cranial breadth–height index (102.7, Figure 2H). In the facial cranium, the Guangfulin sample had the largest mean in bizygomatic breadth (141.2 mm, Figure 3A) as well as in minimum frontal breadth (97.6 mm, Figure 2D) and the second smallest mean in upper facial height (71.9 mm, Figure 3C). As a result, Guangfulin had the smallest mean upper facial index (52.3, Figure 3E), meaning that the site sample was characterized by a relatively low and wide face. In other features, Guangfulin had the smallest mean in the simotic index of flatness (26.3, Figure 3H), marked by a relatively flat nasal root. The results of these multiple comparisons indicate that statistically significant differences exist in maximum cranial breadth, minimum frontal breadth, basibregmatic height, upper facial height, and orbital breadth between all pairs of Guangfulin and the other samples except for Weidun (Table 2).
Comparison of neurocranial measurements and indices. A, maximum cranial length; B, basal length; C, maximum cranial breadth; D, minimum frontal breadth; E, basibregmatic height; F, cranial length–breadth index; G, cranial length–height index; H, cranial breadth–height index. GFL, Guangfulin; WD, Weidun; JJL, Jiangjialiang; JSC, Eastern Zhou assemblage of Jiangsu; HNC, Eastern Zhou assemblage of Henan (Xinghong and Zhouzhuang); TCZ, Eastern Zhou assemblage of Inner Mongolia (Tuchengzi).
Comparison of facial cranial measurements and indices. A, bizygomatic breadth; B, bimaxillary breadth; C, upper facial height; D, least nasal breadth; E, upper facial index; F, orbital index; G, nasal index; H, simotic index of flatness.
Table 3 indicates the discriminant function for classifying between both assemblages of Neolithic northern China (Jiangjialiang) and the Neolithic Yangtze Delta region (Guangfulin, Weidun, and Hemudu) using individual data without the missing values of 12 cranial measurements (M1, M5, M8, M9, M17, M40, M45, M48av, M51, M52, M54, M55). This function detects an almost statistically significant difference between both assemblages (Wilks’s λ = 0.271, P = 0.052). The difference was driven by upper facial height, nasal height, nasal breadth, minimum frontal breadth, and basibregmatic height in descending order of contribution according to the standardized coefficients. The rate of correct discrimination is 95.8% (N = 24), and a wrong case occurred in an individual of the Neolithic Yangtze Delta group (Weidun). The mean discriminant score of Neolithic northern China was –1.007 while the Neolithic Yangtze Delta score was 2.445. The mean discriminant scores of all Eastern Zhou groups (not only the Henan complex and Tuchengzi but also the Jiangsu complex) are inclined toward negative values.
| Martin | Standardized coefficients | Unstandardized coefficients | |
|---|---|---|---|
| M48 (SD) | −1.340 | −0.355 | |
| M55 | 1.234 | 0.470 | |
| M9 | 1.185 | 0.286 | |
| M17 | 0.799 | 0.183 | |
| M51 | −0.600 | −0.332 | |
| M8 | −0.537 | −0.083 | |
| M54 | 0.501 | 0.244 | |
| M52 | 0.419 | 0.223 | |
| M5 | −0.302 | −0.062 | |
| M45 | −0.238 | −0.047 | |
| M40 | −0.053 | −0.010 | |
| M1 | 0.008 | 0.001 | |
| Constant | −25.499 | ||
| Rate of correct discrimination | % | N | |
| Neolithic northern China1 | 100.0 | 17 | |
| Neolithic Yangtze Delta2 | 85.7 | 7 | |
| Total | 95.8 | 24 | |
| Discriminant score | Mean | SD | |
| Neolithic | |||
| Northern China | −1.007 | 0.80 | |
| Yangtze Delta | 2.445 | 1.41 | |
| Eastern Zhou | |||
| Jiangsu complex | −0.128 | 1.63 | |
| Henan complex | −0.231 | 2.23 | |
| Tuchengzi | −0.081 | 1.99 | |
Wilks’s l = 0.271 (P = 0.052).
Figure 4, Figure 5, and Figure 6 compare mean cranial measurements between Guangfulin and the reference data from China, Outer Mongolia, Vietnam, and Japan. Figure 4 is a bivariate plot of the cranial index and the upper facial index. The Neolithic Shandong groups of Dawenkou and Beiqian are located on the right side of the figure because they include the individuals who had relatively short and wide heads. By contrast, the pre-Neolithic and Neolithic south China groups of Hedang, Zengpiyan, and Tanshishan are located on the left side of the figure because they had relatively long and narrow heads. They also had relatively low and wide faces, although not so pronounced as Late Paleolithic Minatogawa 1 and pre-Neolithic Tsukumo and Yoshigo Jomon. Guangfulin is located at the middle of the figure near to the Yayoi groups of the Japanese Archipelago. Figure 5 is a bivariate plot of the orbital and nasal index. Guangfulin is located on the right side together with Eastern Zhou Jiangsu and Beiqian, which were characterized by a relatively high orbital cavity. Figure 6 is a bivariate plot of the minimum nasal breadth (or simotic chord) and the simotic subtense. The northern or western groups such as Outer Mongolia, Jiangjialiang, Minghegetaozhuang, and Liuwan are located on the left and upper corner of the figure and were marked by three-dimensional nasal roots, whereas southern groups such as the Hoabinhian, Hedang, and Gaomiao are on the right and lower corner with relatively flat nasal roots. The Guangfulin sample had relatively flat nasal roots which approximated with Weidun and North Kyushu-Yamaguchi Yayoi.
Plots using the two indexes of neuro/facial cranial measurements among groups. Cross plot, Guangfulin; open circle plots, the Late Paleolithic to Neolithic assemblages; filled circle plots, early Dynastic assemblages.
Plots using the two indexes of orbital/nasal measurements among groups.
Plots using the two measurements on nasal root among groups. The dotted line shows the result from the regression formula using all the data.
Table 4 indicates the Mahalanobis generalized distance matrix using 12 cranial measurements (M1, M5, M8, M9, M17, M40, M45, M48av, M51, M52, M54, M55). Figure 7A shows the generalized distances from Guangfulin and Figure 7B shows the two-dimensional expression of the generalized distance matrix by multidimensional scaling. The pre-Neolithic hunting-gathering groups (Minatogawa 1, Shiraho 4, Tsukumo-Jomon, Gaomiao) and the Neolithic groups of south China (Tanshishan, Hedang) are remote from Guangfulin (Figure 7A) and distributed extensively in the figure (Figure 7B), which means that their regional morphological variation was relatively significant. In contrast, the agricultural groups of the Neolithic and Bronze Ages, except for the south China (Hedang, Tanshishan), Shandong (Dawenkou, Beiqian), west China (Minghegetaozhuang, Banbo, Chuanxi highlands) and Outer Mongolia groups, are close to Guangfulin (Figure 7A) and converge into the middle of the figure (Figure 7B). This means that these agricultural groups shared a similar cranial morphology even though they are located geographically apart from each other. Guangfulin is located at the middle of the figure and morphologically resembled the groups of northern China and the Japanese archipelago which were not geographically close to Guangfulin.
| Yangtze Delta region | South China and Northern Vietnam | Shandong | Northern China | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Guangfulin | Weidun | Hemudu | Jiangsu | Tanshishan | Hedang | Gaomiao | Man Bac 1 | Man Bac 2 | Dawenkou | Beiqian | Dinggong | Yedian | Xixiahou | Wangyin | Banbo | Xiawanggang | |
| Guangfulin | |||||||||||||||||
| Weidun | 4.695 | ||||||||||||||||
| Hemudu | 19.658 | 20.100 | |||||||||||||||
| Jiangsu | 7.669 | 3.796 | 28.973 | ||||||||||||||
| Tanshishan | 11.087 | 5.117 | 13.773 | 9.271 | |||||||||||||
| Hedang | 13.009 | 11.627 | 19.732 | 15.073 | 12.834 | ||||||||||||
| Gaomiao | 13.365 | 28.133 | 23.936 | 23.897 | 43.993 | 30.820 | |||||||||||
| Man Bac 1 | 4.886 | 3.852 | 11.936 | 8.946 | 5.839 | 13.092 | 20.082 | ||||||||||
| Man Bac 2 | 5.160 | 3.419 | 17.342 | 4.657 | 7.238 | 18.873 | 22.248 | 4.425 | |||||||||
| Dawenkou | 19.449 | 9.216 | 34.363 | 12.753 | 23.021 | 29.455 | 38.711 | 18.219 | 19.947 | ||||||||
| Beiqian | 10.218 | 9.958 | 25.105 | 6.919 | 23.638 | 33.832 | 31.145 | 11.307 | 6.350 | 7.122 | |||||||
| Dinggong | 11.676 | 4.694 | 27.570 | 10.079 | 14.028 | 12.445 | 18.148 | 13.006 | 9.340 | 12.433 | 17.312 | ||||||
| Yedian | 11.062 | 8.598 | 28.501 | 11.369 | 13.546 | 15.243 | 18.257 | 13.236 | 10.473 | 11.297 | 7.935 | 5.525 | |||||
| Xixiahou | 6.287 | 6.269 | 25.577 | 16.421 | 7.900 | 9.523 | 16.805 | 8.383 | 6.618 | 4.133 | 14.637 | 17.984 | 14.260 | ||||
| Wangyin | 3.277 | 5.899 | 20.997 | 11.289 | 6.417 | 15.250 | 9.756 | 5.174 | 11.952 | 8.023 | 9.756 | 12.844 | 15.596 | 2.242 | |||
| Banbo | 16.386 | 12.534 | 28.212 | 9.170 | 12.613 | 32.748 | 50.295 | 18.754 | 13.139 | 30.335 | 10.151 | 23.704 | 27.710 | 23.203 | 25.155 | ||
| Xiawanggang | 14.134 | 7.776 | 21.300 | 22.438 | 10.113 | 11.990 | 19.674 | 6.888 | 14.277 | 11.863 | 10.512 | 4.465 | 10.796 | 7.951 | 13.221 | 21.246 | |
| Miaodigou | 7.139 | 7.558 | 20.378 | 15.157 | 8.955 | 12.381 | 15.486 | 6.457 | 18.113 | 10.823 | 20.838 | 2.111 | 2.812 | 5.481 | 9.183 | 39.044 | 1.836 |
| Baoji | 2.921 | 2.941 | 22.292 | 5.820 | 10.024 | 15.949 | 16.425 | 11.449 | 3.404 | 5.996 | 7.423 | 2.573 | 9.278 | 9.704 | 11.291 | 15.686 | 6.412 |
| Jiangjialiang | 10.123 | 9.226 | 20.597 | 5.853 | 12.185 | 9.243 | 25.752 | 15.434 | 12.117 | 18.435 | 17.448 | 4.810 | 13.468 | 10.210 | 10.058 | 12.095 | 21.958 |
| Yinxu (citizen) | 4.775 | 4.030 | 21.022 | 2.484 | 4.426 | 7.855 | 22.071 | 9.931 | 6.318 | 15.014 | 10.268 | 3.847 | 6.631 | 5.923 | 5.997 | 10.620 | 10.256 |
| Yinxu (slave) | 7.560 | 7.099 | 22.804 | 2.489 | 8.577 | 6.235 | 23.208 | 12.760 | 6.627 | 16.087 | 12.919 | 5.025 | 8.218 | 9.924 | 10.911 | 8.180 | 12.345 |
| Xinghong | 2.180 | 3.140 | 18.855 | 4.796 | 9.601 | 17.615 | 14.884 | 7.561 | 2.502 | 5.727 | 4.851 | 3.009 | 5.672 | 8.833 | 8.349 | 12.957 | 4.551 |
| Zhouzhuang | 4.191 | 4.358 | 17.819 | 3.927 | 7.832 | 15.329 | 16.410 | 8.872 | 8.819 | 8.221 | 8.454 | 3.896 | 6.326 | 3.405 | 3.293 | 17.759 | 11.804 |
| Tuchengzi | 5.850 | 6.498 | 25.150 | 3.734 | 9.974 | 12.271 | 15.441 | 11.638 | 7.755 | 8.965 | 13.053 | 4.414 | 5.601 | 7.363 | 7.667 | 15.846 | 12.578 |
| Minghegetaozhuang | 11.122 | 16.673 | 39.222 | 3.976 | 17.026 | 12.796 | 23.463 | 19.625 | 14.764 | 16.248 | 10.537 | 9.763 | 14.205 | 20.670 | 17.777 | 7.663 | 29.993 |
| Liuwan | 8.991 | 8.549 | 18.551 | 3.203 | 8.297 | 13.405 | 27.183 | 14.538 | 12.850 | 18.404 | 18.251 | 8.105 | 15.424 | 7.723 | 7.725 | 19.471 | 21.716 |
| Chuanxi highlands | 14.365 | 13.244 | 34.147 | 3.698 | 21.386 | 10.537 | 28.875 | 20.706 | 5.066 | 20.145 | 17.715 | 7.643 | 10.118 | 19.851 | 19.134 | 5.966 | 19.782 |
| Outer Mongolia | 17.313 | 11.377 | 35.975 | 13.025 | 15.555 | 29.744 | 29.748 | 22.289 | 21.738 | 21.080 | 9.812 | 15.952 | 20.695 | 19.387 | 16.398 | 22.687 | 24.540 |
| Minatogawa 1 | 18.309 | 11.224 | 31.727 | 15.842 | 14.026 | 20.268 | 19.765 | 20.255 | 21.287 | 25.552 | 13.121 | 10.631 | 19.140 | 18.905 | 21.357 | 20.365 | 17.133 |
| Shiraho 4 | 17.603 | 9.744 | 25.480 | 19.935 | 19.479 | 14.502 | 25.994 | 12.868 | 11.936 | 34.565 | 15.784 | 14.540 | 20.635 | 23.975 | 20.651 | 23.303 | 21.160 |
| Tsukumo Jomon | 15.070 | 8.378 | 32.611 | 7.773 | 8.814 | 13.302 | 24.839 | 11.518 | 7.267 | 26.304 | 10.076 | 10.184 | 19.373 | 24.550 | 23.869 | 26.638 | 17.477 |
| NK-Y Yayoi | 4.668 | 4.502 | 22.474 | 2.246 | 6.933 | 11.351 | 18.606 | 8.411 | 2.521 | 12.206 | 6.255 | 3.984 | 7.462 | 10.690 | 9.112 | 11.839 | 10.269 |
| Aoya | 14.538 | 15.258 | 39.358 | 6.623 | 23.944 | 18.508 | 24.926 | 22.515 | 10.096 | 21.547 | 16.483 | 9.025 | 17.206 | 22.122 | 19.146 | 8.629 | 27.697 |
| Northern China | West China and Mongolia | Japanese Archipelago | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Miaodigou | Baoji | Jiangjialiang | Yinxu (citizen) | Yinxu (slave) | Xinghong | Zhouzhuang | Tuchengzi | Minghegetaozhuang | Liuwan | Chuanxi highlands | Outer Mongolia | Minatogawa 1 | Shiraho 4 | Tsukumo Jomon | NK-Y Yayoi | Aoya | |
| Guangfulin | |||||||||||||||||
| Weidun | |||||||||||||||||
| Hemudu | |||||||||||||||||
| Jiangsu | |||||||||||||||||
| Tanshishan | |||||||||||||||||
| Hedang | |||||||||||||||||
| Gaomiao | |||||||||||||||||
| Man Bac 1 | |||||||||||||||||
| Man Bac 2 | |||||||||||||||||
| Dawenkou | |||||||||||||||||
| Beiqian | |||||||||||||||||
| Dinggong | |||||||||||||||||
| Yedian | |||||||||||||||||
| Xixiahou | |||||||||||||||||
| Wangyin | |||||||||||||||||
| Banbo | |||||||||||||||||
| Xiawanggang | |||||||||||||||||
| Miaodigou | |||||||||||||||||
| Baoji | 4.050 | ||||||||||||||||
| Jiangjialiang | 10.816 | 4.576 | |||||||||||||||
| Yinxu (citizen) | 6.043 | 1.860 | 4.705 | ||||||||||||||
| Yinxu (slave) | 8.993 | 5.206 | 4.895 | 1.337 | |||||||||||||
| Xinghong | 7.711 | 1.093 | 6.228 | 2.127 | 4.731 | ||||||||||||
| Zhouzhuang | 4.195 | 1.087 | 6.828 | 2.968 | 6.127 | 1.264 | |||||||||||
| Tuchengzi | 5.691 | 2.705 | 3.415 | 1.568 | 3.905 | 1.726 | 1.841 | ||||||||||
| Minghegetaozhuang | 20.570 | 12.883 | 6.801 | 3.983 | 4.244 | 10.116 | 10.167 | 3.770 | |||||||||
| Liuwan | 4.864 | 4.291 | 2.865 | 2.795 | 4.816 | 6.880 | 4.000 | 2.913 | 4.553 | ||||||||
| Chuanxi highlands | 14.488 | 11.556 | 8.053 | 7.373 | 2.415 | 8.878 | 12.047 | 5.349 | 4.896 | 9.845 | |||||||
| Outer Mongolia | 21.807 | 12.818 | 17.301 | 10.591 | 16.288 | 13.038 | 13.117 | 16.478 | 16.061 | 14.944 | 15.202 | ||||||
| Minatogawa 1 | 15.399 | 13.667 | 17.235 | 9.602 | 12.746 | 13.316 | 16.929 | 13.044 | 13.833 | 20.229 | 12.665 | 13.009 | |||||
| Shiraho 4 | 18.918 | 17.887 | 21.466 | 16.957 | 13.388 | 14.627 | 23.204 | 14.513 | 22.969 | 25.628 | 10.651 | 31.298 | 24.207 | ||||
| Tsukumo Jomon | 8.916 | 11.735 | 13.416 | 9.143 | 6.381 | 9.672 | 17.349 | 13.871 | 10.169 | 17.491 | 7.374 | 11.293 | 7.638 | 12.170 | |||
| NK-Y Yayoi | 6.248 | 1.464 | 3.833 | 2.414 | 3.427 | 1.760 | 2.856 | 2.150 | 6.895 | 3.937 | 3.682 | 8.949 | 11.819 | 14.143 | 5.649 | ||
| Aoya | 18.915 | 11.629 | 7.728 | 9.058 | 9.800 | 12.161 | 15.101 | 7.795 | 4.495 | 9.398 | 3.912 | 14.679 | 12.367 | 13.803 | 8.085 | 5.895 | |
The Mahalanobis generalized distance using 12 cranial measurements (M1, M5, M8, M9, M17, M40, M45, M48av, M51, M52, M54, M55). The data set pooled from the Tuchengzi, Xinghong, and Zhouzhuang assemblages was applied to calculate the variance–covariance matrix for the procedure. (A) The distances from the Guangfulin assemblage. Gray bars indicate the Late Paleolithic to Neolithic assemblages; black bars show early Dynastic assemblages. (B) The two-dimensional expression of the shape distance matrix by multidimensional scaling. Most of the farming assemblages of the Neolithic and early Dynastic periods converge inside the circle.
This study compared cranial morphometric data between the early wet-rice farmers of the Guangfulin site and five assemblages of the Neolithic and Metal Age from the neighboring area that includes the Yangtze River Delta, the Central Plains, and north China (i.e. Hebei, Shanxi, Inner Mongolia). The results of the univariate/bivariate analysis showed that the Guangfulin sample was characterized by a relatively long and high neurocranium but broad forehead, low facial but high orbital height, and flat nasal roots compared to these comparative assemblages (Figure 2, Figure 3, Figure 4, Figure 5, Table 2). Most of these differences were statistically significant between each pair of Guangfulin and the northern China groups (Jiangjialiang, Jiangsu complex, Henan complex, and Tuchengzi) except for flat nasal roots, while no significant difference was seen between Guangfulin (Shanghai) and Weidun (Jiangsu) (Table 2). Thus, these traits could have been common among the Neolithic people in the Yangtze Delta. Next, a statistically more powerful analysis was tentatively conducted by discriminant function using individual data without missing values of cranial measurements. In this analysis, the samples were classified into two groups, i.e. Neolithic Yangtze Delta (Guangfulin, Weidun, and Hemudu) and Neolithic northern China (Jiangjialiang), because individual data were very limited in the Yangtze Delta region. The result of the discriminant function analysis mostly confirmed the results of the univariate/bivariate analyses and multiple comparisons. Upper facial height (not including nasal height), forehead breadth, and neurocranial height were determined to be the key factors discriminating individuals of the Neolithic Yangtze Delta from Neolithic northern China (Table 3) although the probability did not quite reach statistical significance. Accordingly, the morphometric differences between both the Yangtze Delta and northern China groups could have existed in the Neolithic period, but were relatively small compared to the southern China and pre-Neolithic groups as discussed below.
The Mahalanobis generalized statistic was applied to mean cranial measurements recorded in Guangfulin and other assemblages from East Asia (Table 1). The results of the Mahalanobis generalized statistic showed that Guangfulin was close to the Neolithic, to early Dynastic assemblages of northern China, especially the Central Plains, such as Xinghong/Zhouzhuang, Yinxu, and Baoji, and to the northern Kyushu/Yamaguchi Yayoi of the Japanese Archipelago (Figure 7A). In contrast, Guangfulin was distant from the Neolithic assemblages of south China such as Hedang and Tanshishan (Figure 7A), even though they are geographically and culturally located relatively close to the Yangtze River Delta (Tanshishan culture as expressed in pottery and jade ornaments is influenced by the Liangzhu culture) (Zhang and Hung, 2010). Thus, the Guangfulin group can be hypothesized to have received significant gene flow from northern China. Most Neolithic Shandong skeletal assemblages are quite different from the other farming assemblages, although this could be caused by cranial deformation during infancy (Han, 2000; Nakahashi et al., 2013b). The assemblages from Outer Mongolia and Qinghai province were also separated from the others due to western Eurasian gene flow into both regions (Okazaki and Yonemoto, 2017, 2018). Excluding these Outer Mongolian, Qinghai and hunter-gatherer assemblages from the pre-Neolithic, the cranial traits of the farming assemblages from the Neolithic or later time periods become significantly homogenized over an extensive area (Figure 7B). This suggests that a large-scale population diffusion had occurred in the past, which is consistent with the ‘two layer model’ for explaining the population history of East and Southeast Asia based on morphometric analyses of teeth and crania (Matsumura et al., 2008, 2011, 2017, 2019). In this model, the ‘first layer’ consists of the people who had resided in East Asia since the Late Paleolithic, whereas the ‘second layer’ represents populations who drastically extended their territory in the Neolithic and following Bronze or early Metal Age. The ‘first layer’ includes the human fossils of the Late Paleolithic, the Hoabinhian of Southeast Asia, and the Jomon of the Japanese archipelago, whereas the ‘second layer’ is mainly composed of farmers and is thought to have originated from regions of Northeast Asia such as north China and the Central Plains. Recent ancient DNA analyses also show a certain amount of gene flow between northern and southern regions of East Asia: in coastal areas, the northern influence was stronger during the Neolithic (Yang et al., 2020), whereas inland, the genetic change could be correlated with the intensification of rice farming in the Central Plains (Ning et al., 2020). It is, however, not certain if rice farmers immigrated into the Central Plains due to the lack of information on the first rice farmers. Nevertheless, this significant population diffusion, which began by the Middle Neolithic at the latest, could have impacted the Guangfulin site in the Yangtze River Delta. The results of the discriminant function analysis conducted in this study suggest that the impact of this population diffusion gradually increased in later periods, and most Eastern Zhou Dynasty individuals of the Yangtze Delta were classified not into the Neolithic group of the same region but into that of northern China (75.0%, N = 12, Table 3).
The results of our cranial morphometric analysis are supported by a preliminary isotope study using tooth enamel from the Guangfulin assemblage. A strontium isotope analysis estimated 12 out of a sample of 119 individuals to be non-local, although the place of origin of these individuals is still under discussion (Gakuhari, 2019). At the Liangzhu site complex, located approximately 100 km west-southwest of Guangfulin, individuals who had consumed a high quantity of C4 plants, presumably millets, have been detected by carbon isotope analysis using bone collagen (Yoneda, 2019). These individuals could have immigrated from millet agricultural societies in northern China because little millet has been found in the Neolithic sites of the Yangtze River Delta. Furthermore, spinal tuberculosis is present in both the Yangtze Delta and the Central Plains during Neolithic times, probably representing the earliest evidence of tuberculosis in East Asia (Pechenkina et al., 2007; Okazaki et al., 2019). This finding parallels the results of this study because Mycobacterium tuberculosis thrives in specific social environments such as those with relatively high population density and active intergroup movements such as migration (Roberts, 2012: 442–443). Accordingly, early wet-rice agricultural societies might have been highly susceptible to disease introduced by immigrants from millet agricultural societies in northern China.
The present interpretation based on the osteological evidence (cranial morphometric and paleopathological analyses, isotope analyses using tooth enamel and bone collagen) might be unexpected because archaeological evidence suggests that local agriculture focused exclusively on rice without millet in the Neolithic sites of the Yangtze River Delta, presumably including Guangfulin (e.g. Stevens and Fuller, 2017). The integration of rice and millet agriculture was seen in other regions, such as the Middle Yangtze, Upper Han River (Hubei, southernmost part of Shanxi, and Henan), Central Plains, and Shandong, associated with the expansion of the Yangshao and following Longshan cultures (e.g. Zhang and Hung, 2015; Stevens and Fuller, 2017). It is possible that immigrants from northern China to the Yangtze Delta may have been recruited into wet-rice rather than millet cultivation, perhaps because of the high productivity and labor demands of wet-rice in paddy fields.
The Neolithic skeletal materials reported in the Yangtze River Delta have so far been limited to the two assemblages of Weidun and Hemudu. Of the two, Weidun approximates Guangfulin whereas Hemudu is different and retains characteristics of the ‘ancient south China’ morphology in this study (Figure 7A, B). In previous studies, the skeletal material of the Hemudu site was regarded as representative of the early wet-rice farmers in the Yangtze delta because the site is widely known for the paddy field remains unearthed there. The present study, however, shows that the morphological traits at Hemudu cannot necessarily be considered as representative of Neolithic people in the Yangtze Delta.
Linking early wet-rice farmers with migration to the Japanese archipelagoAs the local expression of the ‘two-layer model’ in the Japanese archipelago, it has been proposed that the modern Japanese were formed through a process of mixture between ‘native’ residents, who had arrived in the Late Paleolithic or Jomon period (c. 14000–800 BCE), and immigrants from the continent from the Yayoi period (c. 800 BCE–300 CE) and later (e.g. Yamaguchi, 1982; Hanihara, 1991; Nakahashi, 1993; Matsumura, 1994; Kaifu, 1997; Kawakubo, 2007; Okazaki and Nakahashi, 2011; Takamuku, 2019; cf. Hudson et al., 2020). The difference in the ratio of the mixture of these two layers produced the geographical variation in anatomical traits among people in the Japanese archipelago, although it does not fully explain the whole situation especially in Hokkaido (e.g. Hanihara et al., 1982; Matsumura, 1998; Pietrusewsky, 1999; Dodo and Kawakubo, 2002; Hanihara et al., 2008; Kawakubo et al., 2009). The time lag in the diffusion of the ‘second layer’ between the continent and the Japanese archipelago is sometimes explained by the maritime strait between the two regions.
The results of this study provide important data on the formation process of the Yayoi period population of Japan. The northern Kyushu/Yamaguchi Yayoi samples were morphologically the fifth most similar to Guangfulin among the 33 groups even though both areas are geographically separated by a natural barrier (Figure 1, Figure 7A). The Guangfulin sample had lower and wider faces than the millet agricultural groups in northern China, but the upper facial index was still similar to the northern Kyushu/Yamaguchi Yayoi (Figure 4). Furthermore, the results of dental morphometric analysis conducted by the present first author demonstrate that the crown size of the northern Kyushu/Yamaguchi Yayoi is on average the largest in the history of East Asia, but is approximated by crown sizes from the Yangtze Delta Neolithic (Guangfulin and Weidun) (Okazaki, 2021). This could mean that the dispersal of the early wet-rice farmers was of sufficient scale to reach the eastern end of Asia across the sea.
In previous studies, the relationship between the early wet-rice farmers of the continent and the immigrant Yayoi people who are thought to have introduced rice farming to the Japanese archipelago has been poorly understood because of the lack of skeletal materials in the homeland zone of wet-rice domestication. As a result, skeletal materials from early millet farmers, which are generally better preserved, have been used as alternative ancestral groups for the immigrant Yayoi-period people in previous studies. So far, the following skeletal materials have been reported as possible ancestral groups of the Yayoi based on cranial or dental morphological analysis: the Han Dynasty assemblage of the Linzi site (Matsushita, 2000), the Dawenkou culture assemblage of the Beiqian site (Nakahashi et al., 2013b), the Longshan culture assemblage of the Dinggong site (Nakahashi and Luan 2008) all in Shandong, the Eastern Zhou Dynasty assemblage of the Xinghong/Zhouzhuang sites in the Central Plains (Nakahashi, 2014), the Majiabang culture assemblage of the Weidun site, and Eastern Zhou–Han Dynasty assemblages in the Yangtze River Delta (Nakahashi et al., 2002). Of these, three assemblages were engaged in wet-rice farming: the two Yangtze Delta (Weidun and the Eastern Zhou–Han Dynasty assemblage, mainly rice) and the one Shandong sample (Dinggong, mainly foxtail millet but including a certain quantity of rice) (Wu et al., 2018). Given the significant role of wet-rice farming in many Yayoi sites with immigrant populations, additional skeletal assemblages of wet-rice farmers need to be studied in order to understand the roots of Yayoi culture and peoples. In the present study, the early wet-rice farmers of the Guangfulin site show significant proximity to the immigrant populations of Yayoi Japan, suggesting that population growth in the Yangtze River Delta may have been one trigger for the dispersal of rice farming in East Asia.
By which routes did rice farming travel to reach the Japanese archipelago? The strongest hypothesis based on recent archaeological evidence suggests that rice farming moved north from the Yangtze River Delta to the Shandong peninsula during the Late Neolithic period, crossed the sea to the Liaodong peninsula, reached the Korean peninsula not later than 1300 BCE, and then crossed the water again to northern Kyushu and Yamaguchi at the beginning of the Yayoi period (Guedes et al., 2015; Stevens and Fuller, 2017; Miyamoto, 2017). This archaeological hypothesis is not contradicted by the results of the cranial morphometric analysis in this study, which shows clear resemblances between the Guangfulin, Dinggong, and northern Kyushu/Yamaguchi Yayoi populations (Figure 7B). The result of archaeological flotation at the Dinggong site suggests that the main subsistence was millet farming but a substantial amount of rice was also present amongst the identified cereals (46% foxtail millet, 29% rice, 6% broomcorn millet, and 0.16% wheat) (Wu et al., 2018).
In conclusion, the cranial morphometric analysis conducted in this study suggests that by the fourth millennium BC, wet-rice farmers in the Yangtze Delta contained significant numbers of individuals who were migrants from the millet agricultural societies of northern China or else were people who had received genetic influences from those societies. The impact of the gene flow from northern China gradually increased over time as shown by the fact that the morphometric differences seen in the Neolithic had disappeared by the Eastern Zhou–Han Dynasty. These people participated in the population diffusion and movement associated with the spread of rice farming to surrounding areas including the Japanese archipelago. The Neolithic revolution and Neolithic demographic transition in East Asia could be characterized by the existence of population flows between societies initially based on two different subsistence patterns (rice and millet). This explanation is supported by a preliminary stable isotope analysis of tooth enamel and bones at Gunagfulin and matches the evidence for the spread of tuberculosis between the Yangtze River Delta and the Central Plains. The explanation also matches recent archaeological models for the spread of rice and millet farming, except that little millet has so far been found from the Neolithic sites of the Yangtze River Delta. Rice paddy farming may have been preferred there due to the climate and the high productivity of wet rice.
The results of this study need to be confirmed in future research using geometric morphometric methods, as well as through ancient DNA. Although the results of ancient DNA analysis in China have recently been published, they do not yet include samples from sites in the Yangtze River Delta (Ning et al., 2020; Yang et al., 2020). Furthermore, the missing links of skeletal material on the route of rice farming transmission between the Eurasian continent and the Japanese archipelago need to be filled. At present, human skeletal remains of the Early Neolithic and Mumun pottery period (early Metal Age) are respectively reported from the two islands of Gadeokdo and Neukdo in Busan, South Korea, but the sample size is unfortunately not yet sufficient for morphometric analysis.
Lastly, Figure 8 shows the facial approximation of an early wet-rice farmer using a well-preserved cranium (M252) unearthed from Guangfulin to supplement the cranial morphometric analysis with a concrete image. The facial approximation was performed using the Manchester method by Y.K. (Kawakubo et al., 2020), using the average values of facial soft tissue depth of modern Japanese whose body mass indexes are less than 21.75 kg/m2 (Kimura and Okazaki, 2018). The M252 cranium shows a slightly higher facial height than the average but is a very typical case among the individuals at Guangfulin (discriminant score 2.125; Table 3). The facial features of this individual would seem to be very common among East Asian people today, including the modern Japanese, suggesting further support for an extensive population dispersal which began in the Neolithic period.
Facial approximation of an early wet-rice farmer in the Yangtze River Delta. (A) M252 cranium of the Guangfulin site, which shows a typical example of the morphological traits generally shown in the ‘second layer’. (B) The result of forensic facial reconstruction by the Manchester method on the basis of this cranium.
We thank Dr. S. Yonemoto, Dr. K. Ohno, Ms. Y. Muramatsu, Mr. H. Tomita, Mr. M. Matsuura, and Mr. B. Zhen for their assistance in cleaning and arranging the skeletal remains, Dr. T. Nakahashi for providing his unpublished data, the fellow researchers of the JSPS KAKENHI grant project (Rice Farming and Chinese Civilization) for their advice regarding Chinese archaeology, CINQ Art Co., Ltd for their working-out of details of the facial approximation, and the three anonymous reviewers for their advice. This work was supported by JSPS KAKENHI grant numbers JP26440259, JP15H05969, and JP18K06443 and by funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 646612).
