Population history and its relationship with climate change on the Chinese Loess Plateau during the past 10,000 years

Abstract

Based on compilation of a large number of archaeological and palaeosol ¹⁴C-ages, the Summed Probability Method is used to reconstruct population history and climatic patterns on the Chinese Loess Plateau during the period 8.5–3.5 cal. ka BP. During this period, the population experienced two major expansion periods and several climatic fluctuations. The first population expansion began at around 7.8 cal. ka BP, and the second at around 5.8 cal. ka BP. During the first period, although already in combination with cultivation of broomcorn millet the population growth was supported mainly by hunting and gathering. With the switch from broomcorn millet to foxtail millet, plant cultivation became the major factor promoting a second population increase. In this paper, we demonstrate that, initially, suitable climate and agriculture both can facilitate population growth and climate events had a significant influence on demographic fluctuations. However, when reaching the threshold of land capacity, population became increasingly more sensitive towards climate fluctuations.

Keywords

agriculture Chinese Loess Plateau climate demography Neolithic radiocarbon

Introduction

Prehistoric demography dynamics and its relationship with climate variability have become a major topic during recent years. Although many different proxies (e.g. site numbers, site density, size and distribution) can be used to reconstruct population size, the basic shortcoming of all methods is due to limited time resolution (Wang et al., 2014a). In order to circumvent the dating problems, Rick (1987) was the first to use summed probability distributions (SPDs) of calibrated radiocarbon data to reconstruct trends in prehistoric populations. Because of the high achievable dating precision, in combination with the availability of continuous time-series, the method is now steadily gaining in importance (Williams, 2012). Whereas Quaternary geologists apply the method to reconstruct fluvial histories in relation to climate variability (e.g. Chiverrell et al., 2011; Howard et al., 2009; Wang et al., 2014b), in archaeology it can be used to study the relationship between population dynamics and climate (e.g. Kelly et al., 2013; Munoz et al., 2010; Shennan et al., 2013; Smith et al., 2008). The advantages and disadvantages of the method have been thoroughly discussed, and the critical discussion has helped in promoting its application over the world (Ballenger and Mabry, 2011; Bamforth and Grund, 2012; Miller and Gingerich, 2013; Steele, 2010; Surovell et al., 2009; Weninger et al., 2011; Williams, 2012).

The method has been introduced to China in recent years (Barton et al., 2007, 2009; Dong et al., 2013; Ma et al., 2012), but up to now only used to analyse individuals sites with small area. The situation changed with a recent research paper that was published online (Wang et al., 2014a) where the method is used to reconstruct the population dynamics of the whole of China during the past 50 ka. The paper also includes a discussion of climate effects. Another recent study has demonstrated that the frequency distribution of Chinese Loess Plateau (CLP) palaeosol dates can be used to reconstruct the monsoon intensity (Wang et al., 2014b). Until now, we are still lacking in regional study that discusses the relationship between population dynamics and climate change. Regional studies are important since China has complex topographic diversities and regional climate change may not follow the macroclimate change (Maher and Hu, 2006).

The CLP is a cradle of Chinese civilization with a long Neolithic history. Located in the transition regions from humid to arid of China, the CLP is sensitive to climate change and influenced strongly by the Asian monsoon. The CLP therefore provides an ideal region for environmental archaeology research (An et al., 2005). The effect of climate change on cultural evolution has been discussed in many aspects. The majority of previous studies have focused on the impact of cold/dry climate switches, events which are shown to have had dramatic effects in terms of cultural shifts, geographic site distributions, in comparison to warm/humid climate switches that facilitated population expansion and culture diffusion (An et al., 2006; Dong et al., 2013; Liu and Feng, 2012; Lv and Zhang, 2008; Mo et al., 1996; Pang and Huang, 2003; Wang et al., 1993; Wu and Liu, 2004). In our working area, the CLP, with a more complex climate pattern (Zhao et al., 2010), the main problem for discussion of climate–culture interaction is the low dating resolution of the Loess profiles used in Holocene climate reconstructions, as well as the lack of high resolution lake cores (Zhao et al., 2010). On the other hand, most of the modern discussion was limited to a relatively small area such as the Western Loess Plateau (An et al., 2004; Mo et al., 1996) or the Guanzhong Basin (Lv and Zhang, 2008; Pang and Huang, 2003). Little work has been done to discuss this relationship on the whole CLP level. This is probably on one hand because the culture types are not uniform over such large areas and the time and spatial framework of each culture type was not clear especially in the Western Loess Plateau because of insufficient archaeological investigation. The application of the Summed Probability Method provides a chance to evaluate the association between Holocene population dynamics and climate change on the CLP through correlating directly the archaeological and the palaeosol data for the past 10 ka.

Study area

The CLP covers an area of some 440,000 km² in the upper and middle reaches of China’s Yellow River which includes almost all of Shanxi and Shaanxi provinces, parts of Gansu province and the Ningxia Hui Autonomous Region (Liu, 1985). The loess–palaeosol sequence has accumulated to a depth of 120–300 m during the last 2.6 million years (Ding et al., 2002). The plateau generally has a semi-humid and semi-arid climate, with extensive monsoonal influence. Winters are cold and dry, while summers are warm and humid. Rainfall tends to be heavily concentrated in summer. The mean annual temperature range is 2–15°C, and the mean annual precipitation decreases from 700 to 150 mm from southeast to northwest (Li et al., 2003). The vegetation types include, in order, temperate forest, temperate forest-steppe, temperate steppe, temperate desert-steppe and temperate steppe-desert from southeast to northwest along the precipitation gradient (He et al., 2004).

It is reasonable to call the CLP the cradle of Chinese civilization. From what is presently known, the development of Neolithic cultures began ca. 8 cal. ka BP. According to the widely applied geographic partitioning and cultural terminology Partition, the Huanghe drainage basin has been divided into three different regional systems and cultural types including Gan-qing, Central Plain and Shandong cultural regions (Yan, 1987). The eastern part of the CLP (Shanxi–Shaanxi province) is a part of the Central Plain culture region, while the western part of the CLP (eastern part of Gansu province and southern part of Ningxia province) is divided into Gan-qing cultural region. The prehistoric culture sequence is broadly defined by the Laoguantai (also called Dadiwan; 8–7 cal. ka BP), Yangshao (7–5 cal. ka BP), Longshan (4.9–4 cal. ka BP) cultures on the Eastern Loess Plateau and Dadiwan (7.8–7 cal. ka BP), Yangshao (7–5 cal. ka BP), Majiayao (5.3–4 cal. ka BP), Qijia (4.3–3.6 cal. ka BP) and Siwa (3.6–2.6 cal. ka BP) cultures on the Western Loess Plateau (Xie, 2002; Yan, 2008).

Material and method

Radiocarbon dates from archaeological context are used to calculate the summed probability as a proxy for prehistoric demography since a larger population will result in more production and deposition of cultural carbon, therefore providing more determinations (Holdaway and Porch, 1995; Munoz et al., 2010; Peros et al., 2010; Surovell and Brantingham, 2007). With sufficient numbers of radiocarbon dates from large regions, numerous sites and investigators, the changes in their frequency distributions are widely approved to be a reliable indicator of the population fluctuations (Anderson et al., 2011; Kuzmin and Keates, 2005; Peros et al., 2010).

The radiocarbon dates were obtained mainly from published archaeological ¹⁴C-ages as given in datasets, reports and papers. We have referenced in particular the recently published (online) database of Wang et al. (2014a). All dates were given in years BP (Before Present) on the ¹⁴C-scale and are based on the Libby half-life of 5568 years with 1σ (68% confidence) standard deviation (SD). To avoid misunderstandings, we use the term ¹⁴C-BP.

The archaeological dates were collected according to the principles of Wang et al. (2014a): (1) the 1σ SD of the ¹⁴C-ages should be less than ±400 ¹⁴C-BP; (2) we have eliminated ¹⁴C-ages based on shells, unknown materials or materials inappropriate for dating; (3) we further eliminated ¹⁴C-ages that had weak (i.e. unreliable) association with human occupation such as ancient temples, pagodas or canoes and (4) whenever multiple dates were available for the same context at a site, the most reliable dating material was chosen. In total, 646 dates were collected from 122 sites (Figure 1, the detail is given in supplementary file, available online) and used for the following analysis.

Figure 1.

Distribution of archaeological sites and Loess profile sites on the Chinese Loess Plateau (CLP) contributing radiocarbon dates to the database.

In order to reduce the over-representation of sites (or site-phases) for which scientists had conducted intensive archaeological investigations and dating efforts on one site, we calculate the occupation events using Ward and Wilson’s χ² test (Ward and Wilson, 1978). Following this statistical test, we use the multiple dates to obtain a weighted average value for single sites (or site-phases). This procedure was implemented using the ‘test sample significance’ and ‘create pooled mean’ functions in the CALIB 7.0.2 program (Stuiver and Reimer, 1993): the averaged dates were then calibrated (95.4% confidence) and used to generate SPDs (applying the ‘sum probability’ option in CALIB 7.0.2) based on the IntCal13 calibration curve (Reimer et al., 2013). We also applied the empirical model put forward by Surovell et al. (2009) to correct for taphonomic bias by taphonomic processes (e.g. erosion and weathering) thereby causing over-representation of recent events relative to older events. Following correction, the data were standardized by Xi/Xmax, with Xi representing all single values for N dates (i = 1,…, N) and Xmax calculated as the oldest age-value in the series. Although Williams (2012) recommended secondary smoothing of the SPDs in an effort to remove the artificial effects of plateaus and steep-regions of the calibration curve on the SPD-shape for the last 11 ka, this smoothing may also change the shape of any real peaks in the SPD. Because the use of a 500-year floating window, as proposed by Williams (2012), will strongly impact the shape of any SPD, for our data we applied both a 200- and a 500-year smoothing procedure, and compared the differences in shape between the uncorrected and the smoothed SPD. As an additional check on the existence of artificial peaks or troughs (Williams, 2012), we also plotted the number of radiocarbon dates (using their weighted averages as described above) as well as the corresponding number of calibrated dates on both time-scales (¹⁴C and calendric). This test was deemed important since many authors take the peaks and troughs of SPDs as direct evidence for population fluctuations (Bamforth and Grund, 2012; Gamble et al., 2005; Shennan et al., 2013).

For purposes of comparison with the archaeological ¹⁴C-database, we collected a second and independent ¹⁴C-database for the palaeosols on the CLP. The palaeosols formed when the climate was warm and humid, and their development may therefore be expected to reflect prevailing climate change on a regional scale (Wang et al., 2014b). In order to date individual soil samples, as well as to generate an extended chronology for the overall soil development for the Holocene, in most cases bulk organic material is used for radiocarbon dating. The palaeosol dates were assembled according to the principles proposed by Wang et al. (2014b). In total, 151 ¹⁴C-ages that cover the past 10 ka were collected from 52 sites (Figure 1, supplementary file, available online). The palaeosol ¹⁴C-ages were calibrated and processed by essentially the same procedures as for the archaeological ¹⁴C-ages. In both cases, we applied CALIB 7.0.2 software (Stuiver and Reimer, 1993) to generate the SPDs, with the application of the IntCal13 calibration curve (Reimer et al., 2013). As mentioned above, the distributions were then smoothed using a 200- and a 500-year floating window. Following smoothing, the data were standardized by the Xi/Xmax-method and are plotted on the cal BP-scale.

Results

According to the previous studies, the sample size will affect the reliability of the ¹⁴C-age, and therefore also the shape of the SPD. Michczynska and Pazdur (2004) suggest that, for the time interval 0–12 cal. ka BP, a minimum number of 200 ¹⁴C-ages with ΔT (mean SD) = 115 years is necessary, with statistically reliable results only achieved for 780 dates (when referenced to ΔT = 115 years). Similarly, Williams (2012) points out that summed probability diagrams based on less than 200–500 radiocarbon ages should be treated as provisional. The present studies are based on a total of 646 radiocarbon ages, with ΔT = 74 years. As can be taken from Figure 2, more than 90% of the ¹⁴C-ages have errors better than 125 years. Since the number of radiocarbon dates required to produce a robust SPD decreases with SD, we expect that our database is capable of producing results that are robust against chance statistical fluctuations.

Figure 2.

Histogram (25-year bin) showing the cumulative proportions of the conventional ¹⁴C-ages arranged according to increasing standard deviations.

As shown in Figure 3d, when the data are evaluated using a 500-year moving average, the development of the CLP population can be divided into three phases. At around 8 cal. ka BP, we observe an initial population increase that corresponds to the introduction of Neolithic agriculture. After 7.8 cal. ka BP, the population strongly expands and then reaches a near-stable level which lasts until ~5.9 cal. ka BP. There follows a second large population expansion which lasts until ca 2.7 cal. ka BP, at which time the population size (as measured by the number of ¹⁴C-ages in the database) is twice as high in comparison to the previous phases. As deviations from this general development, there are two fluctuations at ages ca. 4.5 and 3.6 cal. ka BP. Finally, after ca. 3 cal. ka BP, the curve declines rapidly to very low levels. The 200-year moving average curve (Figure 3c) generally has the same trend as for 500-year smoothing, although the amplitude of the minor peak at 5.4 cal. ka BP (Figure 3d) is now attenuated (Figure 3c). According to the frequency distribution of the number of radiocarbon dates (using their mean; Figure 3a) and the number of calibrated dates (using their mean; Figure 3b) by 200-year time interval, three distinct fluctuations respectively existed at around 3.5, 4.1 and 4.7 ka BP (Figure 3a) and 3.7, 4.5 and 5.4 cal. ka BP (Figure 3b), respectively. As such, the 500-year floating window appears to (significantly) change the shape of the distribution for the 4.7 ka/5.4 cal. ka BP peak. We take this as an indication that the corresponding climate deterioration at this time was of shorter duration than the subsequent two climate events (cf. below). The initial population expansion of the CLP began about 1000 years later than the whole China (Wang et al., 2014a) and the fluctuations were not consistent with each other, reflecting a different growth trajectory probably because of regional environmental diversity and different adaptive strategies (Wang et al., 2014a).

Figure 3.

Histograms and summed probability curve: (a) histogram of uncalibrated radiocarbon dates, (b) histogram of calibrated radiocarbon dates, (c) 200-year moving average summed probability curve and (d) 500-year moving average summed probability curve.

Turning now to the SPD of the palaeosol dates (Figure 4), we observe a (relative) increase in soil development that begins around 9 cal. ka BP, and which remains high until 3.2 cal. ka BP, with peaks at 8.2, 7.5, 6.5, 5.8 and 3.4 cal. ka BP and troughs at 7.8, 6.9, 5.9, 4.8 and 3.8 cal. ka BP. The curve shows an abrupt decline after 3.4 cal. ka BP. The general climate pattern reflected by the Summed Probability is consistent with that of other loess proxies for Holocene climate reconstruction (e.g. magnetic susceptibility, CaCO₃ content; Huang et al., 2004), and the fluctuations correlate well with the result of arboreal pollen percentage from Lake Daihai located in the northern part of the CLP (Xiao et al., 2004).

Figure 4.

Summed probability curve of ¹⁴C-ages on Holocene palaeosol layers: (a) 200-year moving average and (b) 500-year moving average.

Discussion

Population dynamics and correlation with climate change

Many authors have identified correlations between climatic and demographic change, and this in a wide variety of regions (e.g. Kelly et al., 2013; Munoz et al., 2010; Smith et al., 2008), including Wang et al. (2014a) for China, but their conclusions are by no means unanimously accepted (e.g. Buchanan et al., 2008; Maher et al., 2011; Shennan et al., 2013). On the CLP, most scholars working in this region apply large attention to the effects of climate change on cultural and in particular on economic evolution.The basic argument is that, on the environmentally sensitive CLP, climate amelioration affords favourable environments for agricultural production, thus promoting culture development and/or population expansion, whereas harsher climate conditions cause cultural decline in combination with demographic shift and/or movements (An et al., 2006; Lv and Zhang, 2008; Mo et al., 1996; Pang and Huang, 2003).

Our radiocarbon records suggest the existence of a correlation between population dynamics and climate change (Figure 5). Following the 8.2 cal. ka BP climate event, climate amelioration and population expansion began at the same time. Around 6.9 cal. ka BP, a cold–dry event occurred and population expansion was restrained and remained at a stable level until 6 cal. ka BP when the climate was humid and warm again during 6.8–6 cal. ka BP. From 5.8 to 3 cal. ka BP, there was a general warm and humid climate with three climate deterioration events, and we observe a second population increase which was maintained at a scale twice as high as during the previous period. Three climate deterioration events correspond well with three population fluctuations at around 5.4, 4.5 and 3.7 cal. ka BP. Of course, many factors may contribute to the decline in amplitude of calibrated age distribution (e.g. research focus, site visibility). In this particular case, as goes for the historic period (especially the past 2 ka), many archaeological sites are accurately dated by information given in ancient Chinese literature. This is visible in the decline of the probability distribution for these younger ages.

Figure 5.

Comparison between summed probability of archaeological dates and palaeosol dates: (a) 200-year moving average summed probability curve of archaeological dates and (b) 200-year moving average summed probability curve of palaeosol dates.

Although the climate evidence generally correlates well with population size, there are some exceptions. From 8.5 to 3.5 cal. ka BP, during which the climate was consistently warm and humid, the population experienced two phases of growth (first: 7.8–5.9 cal. ka BP; second: 5.9–2.6 cal. ka BP) with large amplitude in the second phase than in the first. The fluctuation of the second phase appears to be more sensitive to climate change than the previous phase. From 5 cal. ka BP on, climate shows a trend towards cold and dry, but the population size remained high until at least 2.7 cal. ka BP. We conclude that climate is not the only factor that influences population dynamics.

Agriculture development and its relationship with population expansion

The emergence of agriculture is often considered to be a revolution in human history. It represents the period when humans adapted their cultural and economic behaviour to the environment. Northern China is thought to be the origin centre of rain-fed agriculture. Agriculture provides a stable source of food for prehistoric humans, which enables them to cope with seasonal food shortages and therefore promotes culture expansion (Jia et al., 2013). Flotation results (Figure 6) and stable isotope analysis of bones can show us the development process and character of agriculture on the CLP.

Figure 6.

Abundance ratio of crops from the floatation results in the study area: Dadiwan (Liu et al., 2004), Qinan and Lixian (An et al., 2010), Xinglefang and Xiahe (Liu et al., 2013), Zhouyuan (The Zhouyuan Archaeology Team, 2004), Buziping (Jia et al., 2013), Taosi (Zhao and He, 2006), Qiaocun (Zhou et al., 2011), Lajia (The Team Study on Agriculture, 2011). The culture type and age of each site can be seen in Appendix 1 (available online).

After 8 cal. ka BP, broomcorn millet was gradually cultivated in many parts of northern China (Liu et al., 2004; Lv et al., 2009; Zhao, 2011), and the earliest (presently known) broomcorn millet remains in the Western Loess Plateau date to 7.8–7.3 cal. ka BP at the Dadiwan site (An et al., 2010; Barton et al., 2009; Liu et al., 2004). However, the primary human subsistence strategy at Dadiwan between 7.9–7.2 cal. ka BP was hunting, not farming. This is indicated by a brief and non-intensive agriculture (Barton et al., 2009). During the early Yangshao period, broomcorn millet was still the main crop rather than foxtail millet in the Western Loess Plateau (Liu et al., 2004). It was only later that agriculture replaced the hunter-gatherer economy, for example, around 6.5 cal. ka BP in Guanzhong Basin (Zhao, 2014). After 6 cal. ka BP, broomcorn millet was replaced by foxtail millet as the main crop (An et al., 2010; Liu et al., 2004; Zhao, 2014; Zhou et al., 2011). The most likely reason for this is that foxtail millet has a higher productivity, requires less water and is more disease-resistant than broomcorn millet (Zhao, 2014; Zhou et al., 2011). Stable isotopes measured on bones from the western CLP and the Guanzhong Basin indicate that humans, pigs and dogs were all heavily reliant on millet after 6 cal. ka BP (Barton et al., 2009; Pechenkia et al., 2005), and this is indicative of an increasingly intensive agricultural economy. At around the same time, millet agriculture expanded from northern China towards the Changjiang river, Chengdu Plain and even as far as the Qinghai-Tibetan Plateau (Guedes, 2010; Nasu et al., 2012; Zhao and Chen, 2011). Although not an equally important food, rice appeared in many sites of Shaanxi and Gansu province along the Wei River during the Holocene (An et al., 2010; Li et al., 2007b; Lv and Zhang, 2008) providing a complementary food resource for periods with suitable climate. After 4 cal. ka BP, wheat was introduced into the CLP, with steadily increasing proportion (Lee et al., 2007; Zhao, 2014). The increasingly diverse agricultural system is accompanied with an enhanced human resilience towards instable climate in late Holocene (An et al., 2014; Li et al., 2007a, 2007c; Zhao, 2014).

To conclude, in parallel with the cultivation of millet we observe an initial population boom. During this period (~8–6 cal. ka BP), climatic amelioration and agriculture both have important roles in the development of food resources. Nonetheless, hunting and gathering remain important and actually provide many of the main food resources. During this period, the climate generally supported a stable environment that provided favourable conditions both for wild plants and animals, as well as for plant cultivation.

After 6 cal. ka BP, an increasingly intensive agriculture with livestock breeding was established, with foxtail millet replacing broomcorn millet as major crop. Agriculture afforded the main food resource rather than hunting and gathering. New food production practices enabled people to establish permanent sites and expand their settlements, thus facilitating pronounced human population growth (Wang et al., 2014a). The increasingly diverse agricultural system also offered a certain degree of protection in the late Holocene, when the climate became colder and drier.Finally, climate was replaced by agriculture as the dominant factor in controlling population stability and growth.

During the second period, population fluctuations correspond well with climate deterioration events. However, the population size did not fluctuate significantly when climate deteriorated at around 7 and 5.9 cal. ka BP in the first period. As explanation, we propose that – when humans were becoming increasingly reliant on agriculture – the population size became more sensitive to climate change as shown in Figure 5. The main underlying reason would therefore be that, once the agricultural technology had reached a certain threshold, in combination with an increasingly stable demography, the economic system becomes more sensitive towards climatic fluctuations.

Conclusion

The population experienced two major expansion periods and several fluctuations. The first population expansion began at around 7.8 cal. ka BP, and the second at around 5.8 cal. ka BP. In addition, we observe three distinct population fluctuations at around 5.4, 4.5 and 3.7 cal. ka BP.

Suitable climate between 8.5 and 3.5 cal. ka BP facilitated demographic growth especially for the first period, during which hunting and gathering were the major economic factors. Climate variability had significant impact on population size, and caused a number of population fluctuations.

The development of agriculture finally became the major factor, promoting a second population boom when the increasingly intense and diverse agricultural system provided the necessary nutrition to support further population growth. Once the population scale had reached a certain threshold of land capacity, population became more sensitive and fragile to climate fluctuations.

Footnotes

Funding

This work was funded by National Key Technology R&D Program of China (Grant No. 2013BAK08B02).

References

Chen

. (2010) Evolution of prehistoric agriculture in central Gansu Province, China: A case study in Qin’an and Li County. Chinese Science Bulletin 55(18): 1925–1930.

Dong

. (2014) How prehistoric humans use plant resources to adapt to environmental change: A case study in the western Chinese Loess Plateau during Qijia Period. The Holocene 24(4): 512–517.

Tang

Barton

. (2005) Climate change and cultural response around 4000 cal yr B.P. in the western part of Chinese Loess Plateau. Quaternary Research 63: 347–352.

Wang

. (2006) The temporal and spatial changes of neolithic cultures in Gansu-Qinghai region and possible environmental forcing. Quaternary Sciences 26(6): 923–927.

Anderson

Goodyear

Kennett

. (2011) Multiple lines of evidence for possible Human population decline/settlement reorganization during the early Younger Dryas. Quaternary International 242: 570–583.

Ballenger

JAM

Mabry

(2011) Temporal frequency distributions of alluvium in the American Southwest: Taphonomic, paleohydraulic, and demographic implications. Journal of Archaeological Science 38: 1314–1325.

Bamforth

Grund

(2012) Radiocarbon calibration curves, summed probability distributions, and early Paleoindian population trends in North America. Journal of Archaeological Science 39: 1768–1774.

Barton

Brantingham

(2007) Late Pleistocene climate change and Paleolithic cultural evolution in northern China: Implications from the Last Glacial Maximum. Developments in Quaternary Science 9: 105–128.

Barton

Newsome

Chen

. (2009) Agricultural origins and the isotopic identity of domestication in northern China. Proceedings of the National Academy of Sciences of the United States of America 106: 5523–5528.

10.

Buchanan

Collard

Edinborough

(2008) Paleoindian demography and the extraterrestrial impact hypothesis. Proceedings of the National Academy of Sciences of the United States of America 105: 11651–11654.

11.

Chiverrell

Thorndycraft

Hoffmann

(2011) Cumulative probability functions and their role in evaluating the chronology of geomorphological events during the Holocene. Journal of Quaternary Science 26(1): 76–85.

12.

Ding

Derbyshire

Yang

. (2002) Stacked 2.6-Ma grain size record from the Chinese loess based on five sections and correlation with the deep-sea δ¹⁸O record. Paleoceanography 17: 1–21.

13.

Dong

Wang

Cui

. (2013) The spatiotemporal pattern of the Majiayao cultural evolution and its relation to climate change and variety of subsistence strategy during late Neolithic period in Gansu and Qinghai Provinces, northwest China. Quaternary International 316: 155–161.

14.

Gamble

Davies

Pettitt

. (2005) The archaeological and genetic foundations of the European population during the Late Glacial: Implications for ‘agricultural thinking’. Cambridge Archaeological Journal 15: 193–223.

15.

Guedes

(2010) Millets, rice, social complexity, and the spread of agriculture to the Chengdu Plain and southwest China. Rice 4(3–4): 104–113.

16.

Tian

Tang

. (2004) Bio-climatic imprints on a Holocene loess palaeosol from China. Journal of Asian Earth Sciences 22: 455–464.

17.

Holdaway

Porch

(1995) Cyclical patterns in the Pleistocene human occupation of Southwest Tasmania. Archaeology in Oceania 30: 74–82.

18.

Howard

Gearey

Hill

. (2009) Fluvial sediments, correlations and palaeoenvironmental reconstruction: The development of robust radiocarbon chronologies. Journal of Archaeological Science 36: 2680–2688.

19.

Huang

Pang

Zhou

. (2004) Holocene pedogenic change and the emergence and decline of rain-fed cereal agriculture on the Chinese Loess Plateau. Quaternary Science Reviews 23: 2525–2535.

20.

Jia

Dong

. (2013) The development of agriculture and its impact on cultural expansion during the late Neolithic in the western Loess Plateau, China. The Holocene 23: 83–90.

21.

Kelly

Surovell

Shuman

. (2013) A continuous climatic impact on Holocene human population in the Rocky Mountains. Proceedings of the National Academy of Sciences of the United States of America 110: 443–447.

22.

Kuzmin

Keates

(2005) Dates are not just data: Paleolithic settlement patterns in Siberia derived from radiocarbon records. American Antiquity 70: 773–789.

23.

Dodson

Zhou

. (2007a) Early cultivated wheat and broadening of agriculture in Neolithic China. The Holocene 17(5): 555–560.

24.

Zhou

Dodson

(2003) The vegetation characteristics of the ‘Yuan’ area at Yaoxian on the Loess Plateau in China over the last 12000 years. Review of Palaeobotany and Palynology 124: 1–7.

25.

Zhou

Zhang

. (2007b) The record of cultivated rice from archaeobiological evidence in northwestern China 5000 years ago. Chinese Science Bulletin 52(10): 1372–1378.

26.

Zhou

. (2007c) The earliest archaeobiological evidence of the broadening agriculture in China recorded at Xishanping site in Gansu Province. Science in China Series D: Earth Science 50: 1707–1714.

27.

Liu

Kong

Lang

(2004) On crop relics in Dadiwan site and human habitat. Cultural Relics of Central China 4: 26–30.

28.

Liu

Feng

(2012) A dramatic climatic transition at ~4000 cal. yr BP and its cultural responses in Chinese cultural domains. The Holocene 22(10): 1181–1197.

29.

Liu

Zhang

. (2013) Floatation results and comparative analysis from two Yangshao culture sites of Shaanxi Province. Archaeology and Cultural Relics 4: 106–112.

30.

Liu

(1985) Loess and the Environment. Beijing: China Ocean Press.

31.

Zhang

(2008) Neolithic cultural evolution and Holocene climate change in the Guanzhong Basin, Shaan Xi, China. Quaternary Sciences 28(6): 1050–1059.

32.

Zhang

Liu

K-B

. (2009) Earliest domestication of common millet (Panicum miliaceum) in East Asia extended to 10,000 years ago. Proceedings of the National Academy of Sciences of the United States of America 106: 7367–7372.

33.

Dong

Jia

. (2012) Analysis of settlement patterns during Neolithic and Bronze period and its influencing factors in Hualong County, Qinghai Province, China. Quaternary Sciences 32: 209–218.

34.

Maher

(2006) A high-resolution record of Holocene rainfall variations from the western Chinese loess plateau: Antiphase behavior of the African/Indian and East Asian summer monsoons. The Holocene 16(3): 309–319.

35.

Maher

Banning

Chazan

(2011) Oasis or mirage? Assessing the role of abrupt climate change in the prehistory of the southern Levant. Cambridge Archaeological Journal 21: 1–29.

36.

Michczynska

Pazdur

(2004) Shape analysis of cumulative probability density function of radiocarbon dates set in the study of climate change in Late Glacial and Holocene. Radiocarbon 46(2): 733–744.

37.

Miller

Gingerich

JAM

(2013) Regional variation in the terminal Pleistocene and early Holocene radiocarbon record of eastern North America. Quaternary Research 79: 175–188.

38.

(1996) A preliminary study on the palaeoenvironment of the middle Holocene in the Hulu River area in Gansu Province and its effects on human activity. Acta Geographica Sinica 51(1): 59–69.

39.

Munoz

Gajewski

Peros

(2010) Synchronous environmental and cultural change in the prehistory of the northeastern United States. Proceedings of the National Academy of Sciences of the United States of America 107: 22008–22013.

40.

Nasu

Momohara

. (2012) Land-use change for rice and foxtail millet cultivation in the Chengtoushan site, central China, reconstructed from weed seed assemblages. Archaeological and Anthropological Sciences 4(1): 1–14.

41.

Pang

Huang

(2003) A discussion on the relationship between environment change and the Neolithic culture Guanzhong area, China. Scientia Geographica Sinica 23(4): 448–453.

42.

Pechenkina

Ambrose

. (2005) Reconstructing northern Chinese Neolithic subsistence practices by isotopic analysis. Journal of Archaeological Sciences 32: 1176–1189.

43.

Peros

Munoz

Gajewski

. (2010) Prehistoric demography of North America inferred from radiocarbon data. Journal of Archaeological Science 37: 656–664.

44.

Reimer

Bard

Bayliss

. (2013) IntCal13 and Marine13 radiocarbon age calibration curves 0–50,000 years cal BP. Radiocarbon 55: 1869–1887.

45.

Rick

(1987) Dates as data: An examination of the Peruvian preceramic radiocarbon record. American Antiquity 52: 55–73.

46.

Shennan

Downey

Timpson

. (2013) Regional population collapse followed initial agriculture booms in mid-Holocene Europe. Nature Communications 4: 2486.

47.

Smith

Williams

Turney

. (2008) Human-environment interactions in Australian drylands: Exploratory time-series analysis of archaeological records. The Holocene 18(3): 389–401.

48.

Steele

(2010) Radiocarbon dates as data: Quantitative strategies for estimating colonisation front speeds and event densities. Journal of Archaeological Science 37: 2017–2030.

49.

Stuiver

Reimer

(1993) Extended 14C database and revised CALIB radiocarbon calibration program. Radiocarbon 35: 215–230.

50.

Surovell

Brantingham

(2007) A note on the use of temporal frequency distributions in studies of prehistoric demography. Journal of Archaeological Science 34: 1868–1877.

51.

Surovell

Byrd Finley

Smith

. (2009) Correcting temporal frequency distributions for taphonomic bias. Journal of Archaeological Science 36: 1715–1724.

52.

The Team Study on Agriculture (2011) Formative period of Chinese civilization’s agriculture economics characteristics. In: CASS Technology Archaeology Center (ed.) Technology Archaeology. Beijing: Science Press, pp. 1–35.

53.

The Zhouyuan Archaeology Team (2004) Results and preliminary analysis of the trial floatation at Zhouyuan site. Culture Relics 10: 89–96.

54.

Wang

Zhang

. (2014a) Prehistoric demographic fluctuations in China inferred from radiocarbon data and their linkage with climate change over the past 50,000 years. Quaternary Science Reviews 98: 45–59.

55.

Wang

Chen

Zhang

. (2014b) Palaeosol development in the Chinese Loess Plateau as an indicator of the strength of the East Asian summer monsoon: Evidence for a mid-Holocene maximum. Quaternary International 334–335: 155–164.

56.

Wang

Chen

Cao

(1993) The probable relationship between environmental change and Neolithic Culture in the loess plateau of Longxi. Journal of Lanzhou University(Natural Sciences) 29(4): 284–290.

57.

Ward

Wilson

(1978) Procedures for comparing and combining radiocarbon age determinations: A critique. Archaeometry 20: 19–31.

58.

Weninger

Edinborough

Clare

. (2011) Concepts of probability in radiocarbon analysis. Documenta Praehistorica 38: 1–20.

59.

Williams

(2012) The use of summed radiocarbon probability distributions in archaeology: A review of methods. Journal of Archaeological Science 39: 578–589.

60.

Liu

(2004) Variations in the East Asia Monsoon around 4000a B.P. and the collapse of Neolithic around Central Plain. Quaternary Sciences 24(3): 278–284.

61.

Xiao

Nakamura

. (2004) Holocene vegetation variation in the Daihai Lake region of north-central China: A direct indication of the Asian monsoon climatic history. Quaternary Science Reviews 23: 1669–1679.

62.

Xie

(2002) Prehistoric Archaeology of Gansu and Qinghai Province. Beijing: Cultural Relics Press.

63.

Yan

(1987) The unity and diversity of Chinese prehistoric culture. Antiquity 3: 38–50.

64.

Yan

(2008) Review of the Chinese Archaeology Research on the 20th Century: Neolithic Archaeology. Beijing: Science Press.

65.

Zhao

Chen

Zhou

. (2010) Vegetation history, climate change and human activities over the last 6200 years on the Liupan Mountains in the southwestern Loess Plateau in central China. Palaeogeography, Palaeoclimatology, Palaeoecology 293: 197–205.

66.

Zhao

(2011) New archaeobotanic data for the study of the origins of agriculture in China. Current Anthropology 52: S295–S306.

67.

Zhao

(2014) The process of origin of agriculture in China: Archaeological evidence from flotation results. Quaternary Sciences 34(1): 73–84.

68.

Zhao

Chen

(2011) The flotation results and analysis at Yingpanshan site of Maoxian in Sichuan province. Cultural Relics in Southern China 3: 60–67.

69.

Zhao

(2006) Floatation results and analysis at the Taosi site in 2002. Archaeology 5: 78–87.

70.

Zhou

Zhao

. (2011) Early agricultural development and environmental effects in the Neolithic Longdong basin (East Gansu). Chinese Science Bulletin 56: 762–771.