Dietary shift and social hierarchy from the Proto-Shang to Zhou Dynasty in the Central Plains of China

In this paper, the Central Plains of China include the Guanzhong Plain of Shaanxi Province, southern Shanxi and Hebei Province, western Shandong and northern Henan Province (figure 1). This region witnessed the origin and development of ancient civilization, including the Proto-Shang period (∼2000–1600 BCE), the Shang Dynasty (1600–1046 BCE), the Western Zhou Dynasty (1046–771 BCE) and the Eastern Zhou Dynasty (770–221 BCE), as well as the rise of the Qin-Han Empire (221 BCE–220 CE). A strict hierarchical structure in society was well-established during these periods. For instance, the burial pattern, such as the grave sizes and the type and quality of graves including the burial goods can reflect the socio-economic status of individuals in these ancient societies.

Zoom In Zoom Out Reset image size

Xinancheng cemetery, situated in the southeast margin of the Loess Plateau, is a part of the territory of Zhangzi County, Shanxi Province (figure 1). Between 2012 and 2015, a salvage excavation project for this cemetery was carried out by the Shanxi Provincial Institute of Archaeology. Pottery ware, bronzes, jade and stone artifacts, cowries and other funerary objects were unearthed. According to the characteristics of the mortuary ritual, all the graves were dated as belonging to the mid-late Western Zhou Dynasty (Han et al 2016, Li 2017).

A total of 62 human and three animal bones (two from horses and one from a dog) were collected from Xinancheng cemetery for isotopic analysis. Published isotopic data of human samples (supplementary table S1) were also scrutinized to trace human dietary shifts in the Central Plains from ∼2000 to 221 BCE.

Furthermore, isotopic data for human samples from 13 of these sites (including Xinancheng cemetery) were selected for tracing socio-economic status differences in dietary signatures in the Central Plains for the period ∼2000–221 BCE (supplementary table S1). The status level of individuals was defined as being either upper or lower level based on several criteria: (1) burials with no grave goods were classed as lower-status and burials with jade, bronze, shells, etc, were upper-status; (2) occupants of burials classed as human victims including those sacrificed were classed as lower-status otherwise upper-status; (3) burials with no container were classed as lower-status, while burials with containers as upper-status; (4) burials in small graves were classed as lower-status, while those in large/medium graves as upper-status.

Carbon and Nitrogen stable isotope ratios in bone collagen reflect the long-term isotopic composition of an individual's diet, which has been successfully applied to study past human diets and subsistence strategies (e.g. Hedges and Reynard 2007, Hu et al 2008, Ma et al 2016). Feeding experiments (e.g. Ambrose and Norr 1993) show that the collagen δ¹⁵N values are closely associated with the trophic levels of individuals, while the δ¹³C values are influenced by the ratios of C₃ and C₄ plants in their diet. In this study, we extracted bone collagen samples according to the method of Richards and Hedges (1999) with some modifications (Ma et al 2016). Stable carbon and nitrogen isotopic analyses were conducted in the Key Laboratory of Western China's Environmental System (MOE), Lanzhou University. Collagen samples were processed in an automated carbon and nitrogen analyser linked to a Thermo Finnigan Flash DELTAplus XL mass spectrometer. All carbon and nitrogen isotope ratio analyses were measured relative to V-PDB and AIR standard samples. The measurement analytical precision was better than 0.2‰ for both carbon and nitrogen isotopic ratios.

We selected eight bone samples with high collagen yields from both upper- and lower-status individuals for accelerator mass spectrometry (AMS) radiocarbon dating at Peking University, Beijing. Conventional ages from ¹⁴C dating values were calibrated to the calendar age based on the IntCal13 calibration curve (Reimer et al 2013) using OxCal v.4.3.2 (Bronk Ramsey 2017). All ages were reported as 'cal BCE'.

The independent samples t-test was used to detect differences between two independent groups of samples. Statistical analysis was performed using SPSS 22.0 software.

The ages of the eight human collagen samples for age ranges with confidence intervals (CIs) of 68.2% and 95.4%, based on ¹⁴C dating, are shown in table 1. The dating results range from 1013 to 811 cal BCE (95.4%), the values being coincident with typological studies on cultural relics (Han et al 2016, Li 2017) and typically reflecting a period in the mid-late Western Zhou Dynasty (1046–771 BCE).

Isotopic data and collagen quality indicators for the individual samples are presented in table 2 and figure 2, and isotopic data for the two status categories are summarized in table 3. The collagen of all samples was well preserved with C:N ratios (atomic basis) of 3.1 to 3.4. The human and animal isotopic data for samples from Xinancheng cemetery are plotted in figure 2(a). Given that the natural vegetation of northern China is composed of C₃ plants (Gu et al 2003, Wang et al 2003, Auerswald et al 2009, Liu et al 2011), C₄ food recourses (i.e. millet) are always associated with human settlements. The δ¹³C value for the dog implies a C₄ food-based diet, reflecting a relatively large intake of millet and/or human waste and scraps. One horse (sample SX65) with a relatively enriched δ¹³C value (−9.2‰) is likely to have been sustained with C₄ plants (e.g. millet). However, data for another horse sample (sample SX68) suggests it ranged freely and consumed both wild (e.g. grasses and fruits) and domesticated plants (e.g. millet).

Zoom In Zoom Out Reset image size

The δ¹³C and δ¹⁵N values for all individuals ranged from −10.8 to −7.4‰ (mean = −8.4‰ ± 0.7‰) and from 7.6 to 10.9‰ (mean = 9.0 ± 0.7‰), respectively. These values suggested that all individuals buried in Xinancheng relied on C₄ foods (millet and/or millet-based animal protein). Furthermore, the δ¹³C values for the upper-status individuals were significantly more positive than those of the lower-status individuals (mean = −9.1‰ versus −8.2‰, p = 0.003, figures 2(b), (c) and table 3), indicating a little more reliance on C₃ foods for upper-status individuals. Upper-status individuals also had higher δ¹⁵N values than lower-status individuals (mean = 9.5‰ versus 8.8‰, p = 0.001, figures 2(b), (c) and table 3), indicating a higher protein intake for upper-status individuals.

The carbon and nitrogen isotopic values for human bones in the Central Plains reveal that there were no significant differences between the upper (n = 47) and lower (n = 102) socio-economic groups during Proto-Shang (supplementary table S3), which implies that there were no dietary differences between the upper and lower-status groups. Significant differences, however, were observed in the δ¹³C and δ¹⁵N values for humans during the subsequent period, that is, from Shang to Zhou (supplementary table S3). The δ¹³C values of the upper-status individuals (n = 64) were significantly more negative than those of the lower-status individuals (n = 121) (supplementary table S3) from Shang to Western Zhou, suggesting a little more reliance on C₃ foods for the upper-status group. Furthermore, the upper-status individuals had higher δ¹⁵N values than the lower-status individuals from Shang to Western Zhou, indicating there was more consumption of animal protein. However, the δ¹³C values for the Eastern Zhou lower-status individuals (n = 45) were significantly more negative than those of the upper-status individuals (n = 112) (supplementary table S3), indicating more reliance on C₃ foods for the lower-status group. The eastern Zhou lower-status group also had lower δ¹⁵N values than the upper-status group, suggesting those peoples ate less animal protein.

For human bone collagen, the ranges of δ¹³C values for C₄-based, C₄/C₃ mixed and C₃-based food sources are −6‰∼−12‰, −12‰∼−18‰ and −18‰∼−23‰, respectively (Pechenkina et al 2005, Barton et al 2009, Ma et al 2014). According to the δ¹³C values for all individuals (−10.8∼−7.4‰) from Xinancheng cemetery, the Western Zhou individuals in the cemetery had a C₄-based diet. Given that the vegetation in the Loess Plateau consisted mainly of C₃ plants (Wang et al 2003), individuals from Xinancheng consumed millet as their staple crop and were heavily reliant on millet-based agriculture. Wheat (Triticum aestivum) was first introduced to China around 2500 BCE, probably through one of the three possible routes Zhao (2009) hypothesized, the Eurasian Steppe route, the Ancient Silk Road, and the Sea route. Both the wheat researchers retrieved from archaeological contexts and those traditionally cultivated by farmers are exclusively hexaploid bread wheat in China (Fuller and Lucas 2014, Stevens et al 2016). The important role of millet and the limited role of wheat and rice in the human diets for the Western Zhou period are consistent with archaeobotanical data. Archaeobotanical studies reveal that millet was the staple crop cultivated in Shanxi Province in the Neolithic period (Song et al 2019). Although there was an increase in crop species from Proto-Shang to Western Zhou, including wheat, rice and soybean (Lee et al 2007, Song et al 2019), these new plant species were supplementary crops for the overall agriculture system with millet being the staple crop in the Central Plains of China (Zhao and Fang 2007, Wu et al 2014, Zhou and Garvie-Lok 2015, Lu et al 2019).

The number (n = 3) and species (n = 2) of animal remains at Xinancheng (table 2) was too small to predict how much animal protein contributed to human diets. However, previous research has indicated that human δ¹⁵N values could be correlated with social status and/or gender (e.g. Ambrose et al 2003, Linderholm et al 2008, Zhang et al 2012, Zhang 2015, Dong et al 2017b). The range of human δ¹⁵N values was relatively wide (7.7‰–10.9‰) at Xinancheng cemetery, indicating that humans consumed varying amounts of protein. The δ¹⁵N values for the upper-status individuals were significantly higher than that of the lower-status individuals (p = 0.001, table 3 and figure 2(b)), indicating that the upper-status individuals consumed slightly more meat and/or milk. The human δ¹³C values for the upper-status individuals were also significantly more negative than that of the lower-status individuals (p = 0.003, table 3 and figure 2(c)), implying that upper-status individuals had more C₃-food intake (most likely wheat and/or C₃-based animal protein). The same phenomenon was also reported for sites at Shang and Western Zhou in the adjacent area of Henan Province (Cheung et al 2017b, Dong et al 2017b, Zhang et al 2017). This similarity in dietary consumption implies that upper-status individuals in the Central Plains might have consumed slightly more exotic C₃ staples (i.e. wheat) in the period ∼1600–771 BCE, given that wheat was difficult to procure and/or required higher labor input to produce when it was introduced in China at an early stage (Chen 2016, Long et al 2018).

To investigate potential dietary changes in the Central Plains from Proto-Shang to Eastern Zhou, the present results were compared with other stable carbon isotope ratio data for human bones unearthed from sites dating to the period between ∼2000 and 221 BCE (figure 3). In general, humans had a C₄-dominant diet (including millet and C₄-based animal protein) and were heavily reliant on millet-based agriculture during the period from Proto-Shang to Western Zhou, a fact supported by archaeobotanical evidence and historical documentary records. Archaeobotanical and documentary records show that millet was the most important crop in the Central Plains from Proto-Shang to Western Zhou, while other crops, including wheat and rice, were supplementary crops (Zhao and Fang 2007, Wu et al 2014, Yang et al 2017, Zhong et al 2018). Furthermore, ancient peoples used millet and millet by-products to feed livestock (Barton et al 2009). For instance, not only did pigs and dogs eat a large amount of millet and millet by-products, but sheep, goats and cattle also consumed some millet and millet by-products based on the results of stable carbon isotope analysis (Hou et al 2013, Dai et al 2016, Ma et al 2016), giving further evidence to the fact that human societies were heavily reliant on millet-based agriculture in the Central Plains from Proto-Shang to Western Zhou.

Zoom In Zoom Out Reset image size

It can be seen that the range of human δ¹³C values expanded and shifted towards C₃ plants during Eastern Zhou (p = 0, supplementary table S2 and figure 3(a)), which probably implies an increased consumption of wheat during the Eastern Zhou. An increased consumption of wheat in the Eastern Zhou (Dong et al 2017b, Zhou et al 2017) diet is consistent with the archaeobotanical data, which shows a rapid increase in the frequency and/or ubiquity of wheat in the crop assemblages during this period (e.g. Liu et al 2017, Ma 2017, Deng et al 2019). And historical documentary records also support this increase. For example, wheat cultivation was supported by governments during the Eastern Zhou period according to ancient Chinese books such as 'Zuo Zhuan' (左传) and 'Zhan Guo Ce (战国策).' Human dietary patterns and subsistence strategies, however, were inconsistent in different regions of the Central Plains from Proto-Shang to Eastern Zhou (figure 3). Humans shifted their diets from predominantly C₄ to mixed C₃ and C₄ in Henan Province during Eastern Zhou (figure 3(b)), while individuals of Shanxi and Shaanxi Provinces were sustained by a predominantly C₄ diet during Eastern Zhou (figures 3(c), (d)). This suggests that human diets changed in the Central Plains during Eastern Zhou, but the change was more apparent in the core region (i.e. northern Henan Province) of the Central Plains.

The subsistence patterns of humans could shift as a result of changes in climate and society. Climate change has been identified as an important factor in influencing the evolution of human subsistence patterns in the past (Ma et al 2016, Gong et al 2019). When natural resources become unavailable and induced by climate deterioration and/or population pressures, humans may change their subsistence strategies to increase social resilience and adaptive capacity.

Foxtail millet and broomcorn millet, which were domesticated in northern China, were the staple crops in the Central Plains before Eastern Zhou (figure 3). Furthermore, wheat, an exotic crop, played a limited role in the overall subsistence system before Eastern Zhou in the Central Plains (Chen 2016), although wheat, first domesticated in West Asia, was introduced into the Central Plains around 2000 BCE (Lee et al 2007, Zhao 2014, 2015), when trans-Eurasia cultural exchanges gradually intensified. Subsequently, the peoples of the Central Plains consumed more wheat during Eastern Zhou, and spatial differentiation of human dietary changes was observed for different regions of the Central Plains, according to stable carbon and nitrogen isotopic data (figure 3).

The dietary shift toward more wheat consumption in the Central Plains most likely was associated with the deteriorating climate from the late Shang to the early Eastern Zhou dynasties. An oxygen isotope record of Greenland suggests that globally, the climate was cooling during this period (figure 4(g)). Wang (2011) also shows that dry events increased during 2.8–2.6 ka BP along with a decrease in temperature. Although climate patterns in China were more complex (Wanner et al 2011), its general trend was also similar. The weakening Asian summer monsoon recorded by speleothem δ¹⁸O from Dongge Cave appears to be consistent with a general cooling and drying climate condition from the late Shang to the early Eastern Zhou dynasties (figure 4(f)). Regionally, three paleoclimatic records from the Gonghai and Daihai Lakes all suggest a declining trend in temperature and precipitation in the Central Plains during this period (figures 4(c)–(e)). Two high-resolution pollen records from the Gonghai and Daihai Lakes further illustrate that an extremely cold episode occurred at the turn of the Western and Eastern Zhou Dynasties, which is consistent with the weakest episode of the Asian summer monsoon (figure 4(f)). The Warring States period (475–221 BCE) chronicle 'Zhushu Jinian' (竹书纪年) recorded that 'Peaches and apricots ripen in September (circa October in Gregorian Calendar)' (九月, 桃杏实) and 'Frost occurred in June (circa July in Gregorian Calendar)' (夏六月, 陨霜) in the Central Plains, also indicating an unusually cold climate during the reign of King You (795–771 BC), the last king of the Western Zhou.

Zoom In Zoom Out Reset image size

Given that millet was drought-tolerant but not cold-resistant (Wang 1994), the grain yield of millet-based agriculture might have started to decrease with the deterioration of climate since the late Shang Dynasty (figure 4). This form of agriculture probably eventually lost its viability during the extremely cold and dry episode that lasted from the late Western Zhou to early Eastern Zhou dynasties. The food supply would have been at risk and could not feed the large population (20 million) because an agricultural system based on only one crop would have been vulnerable to a changing environment. Responding to the climatic pressure, the society of the Central Plains showed its resilience and adaptive capacity by incorporating cold-resistant wheat in the traditional millet-based agricultural system to supplement and vary the agricultural system. Furthermore, wheat is much more high-yielding than millet (Peng 2010), probably facilitating, in part, the rapid increase of population from 10 million to 20 million (Lu and Teng 2000) in the Central Plains from 1046 to 771 BCE, which might in turn have promoted a dependency by society on wheat production, especially in Henan, the core area of the Eastern Zhou government. Even in the Han Dynasty (202 BCE–220 CE), the importance of wheat in the food assemblage slightly exceeded that of millet (Hou et al 2012, Deng et al 2019).

To investigate potential dietary changes between the upper- and lower-status individuals, the δ¹³C and δ¹⁵N values are plotted for the respective status categories (figures 4(a), (b)). It can be seen that the upper-status individuals ate more C₃ foods than the lower-status individuals from Shang to Western Zhou (figure 4(b), supplementary table S3), including wheat and C₃-based animal protein. Given that wheat would have been difficult to procure and/or would have required higher labor input to produce when it was introduced in China at an early stage, upper-status individuals might have consumed more exotic crops (i.e. wheat) from Shang to Western Zhou. With the adoption and expansion of wheat cultivation in the Central Plains (Ma 2017, Tang et al 2018, Zhong et al 2018, Deng et al 2019), wheat would have gradually increased in the daily diet and would have assumed increasing agricultural importance for all individuals during Eastern Zhou, but especially for the lower-status individuals (supplementary table S3, figures 4(a), (b)). This could have occurred because wheat was not scarce, and although it tasted bad in comparison to millet, the people in the Central Plains before the end of the Han Dynasty (Zeng 2005) would have opted to steam or boil the whole grain.

Human societies adopted a wide variety of strategies to adapt to environmental changes and extreme climatic events. Migration and subsistence strategy changes are two of them that have received the most attention (Kuper and Kröpelin 2006, Flohr et al 2016, Lespez et al 2016, Dong et al 2019). However, along with the developing civilization, the adaptive capacity of human communities in response to the environmental pressures also developed. Our analysis reveals that the agricultural communities in the Central Plains responded to the environmental stress during the Zhou Dynasty by changing their crop structures rather than radically changing their subsistence strategies. Specifically, they started to plant more wheat to adapt to environmental and population pressure. Researchers have also reported similar phenomena in northwestern China (Ma et al 2016, Yang et al 2019b), India (Pokharia et al 2017), and Italy (Primavera et al 2017). This implies that collapse of a society was not the only way for ancient societies to respond to climate change, thus, in some instances humans could increase their resilience and adaptive capacity through alteration of their agricultural practices and strategies in response to the environmental pressures. Since then, the rise and fall of civilizations is likely to have been dominated by human society itself instead of climate change as a result of further improvements in the capacity of humans to adapt to environmental change.