1. Introduction
Famine has long been considered one of the most catastrophic disasters that threaten food security and sustainable development in human society (FAO/Food Security Analysis Unit 2006). Even today, hunger and famine continue to plague human society. According to the Food and Agriculture Organization of the United Nations (FAO), an estimated 784–945 million people worldwide, or 10.6%–14.5% of the global population, were hungry each year from 2005 to 2019. Since 2014, this number has been on the rise, and the health and socioeconomic impacts of COVID-19 in 2020 are likely to worsen the situation further (FAO et al. 2020). Climate-related shocks are a leading cause of severe food crises, but the impact of climate on famine is often intertwined with and even exceeded by that of war, conflict, and other political and economic factors. As a result, famine is often considered to be an anthropogenically caused disaster, particularly against the backdrop of a significant decrease in the prevalence of famine worldwide since the nineteenth century as a result of unprecedented socioeconomic development and globalization of relief (Sen 1981; Alfani and ÓGráda 2018).
Past experience is essential to gain a better understanding of human responses to climate change. Famine was much more prevalent and more closely tied to climatic and ecological disturbances in premodern agrarian-based society. Increasing volumes of European studies emphasize and accept climatic anomalies and weather extremes as important causes of large-scale famines in preindustrial Europe (Zhang et al. 2011; Slavin 2016; Alfani and ÓGráda 2017; Geens 2018). However, the factors leading to famines in history are still contentious. Some scholars stress the role of climate and meteorological events on famine through their effects on production (bad harvest and resource shortages) and conclude that climate or meteorological change was the proximate and even decisive cause, particularly under conditions of high population pressure on resources in preindustrial Europe (Zhang et al. 2011; Alfani and ÓGráda 2018). Other scholars insist that food resource distribution and entitlement issues were the main cause of medieval and early modern famines (Sen 1981; Slavin 2013). They believe that environmental/ecological shocks and shortages created by climate alone were incapable of creating famines without human-related determining factors such as social institutions and demographic trends (Slavin 2016). More recent scholars focus on the interaction among climate/weather, famine, and social factors in a particular area (Tian et al. 2022; Su et al. 2021; Huhtamaa 2015; Xiao et al. 2015; Zhai et al. 2020). For instance, there are spatial differences in the relationship between climate change and famine in northwest Europe since the Middle Ages (Dribe et al. 2015; Curtis and Dijkman 2017). Food system resilience and social vulnerability were also conceptualized to analyze the climate–famine interlinkage with more caution (Nelson et al. 2015).
At least two key issues contribute to the above debates, namely, the definition and identification of famine, and the scarcity of detailed famine chronology, respectively. The former remains contentious and difficult to define, even until recently. For a long time, there was no clear definition of what constitutes a famine, possibly because it is hard to measure based on limited available material in history. This question was temporarily settled by C. ÓGráda in 2009, who defined famine as “a shortage of food or purchasing power that leads directly to excess mortality from starvation or hunger-induced diseases” (ÓGráda 2009). Many subsequent studies on famine have adopted this definition, characterized by references to mass mortality and high food prices to identify the occurrence and severity of a famine (Curtis and Dijkman 2017; Alfani and ÓGráda 2018).
The scarcity of detailed famine chronology is the second key issue contributing to these debates. Unfortunately, previous studies in history and economic history were more concerned with cases or episodes of great famine. Most of these studies were qualitative descriptions of the background, causes, and social effects of famine. However, in recent years, there has been an increasing emergence of studies focused on compiling famine chronology by combining narrative sources, burial-based digital mortality data, grain prices, and even palaeoclimatological reconstructions and archaeological evidence since medieval times in Europe (Slavin 2016). This approach enables series-based investigations of the long-term climatic impact (including extreme events) on food crises and famine. More and more studies have highlighted the scale-dependent effects of past climate change on famine and the key role of famine in the transmission of climatic impact on the environment and production into higher-order social levels (Fang et al. 2015; Tian et al. 2017; Xiao 2020). However, most of these series-based studies in China focused more on great famines with large-scale areas of influence (Teng et al. 2014). The time sequence was simply reconstructed by counting the affected counties or the number of severe famine events (e.g., cannibalism; Xiao et al. 2015; Lee 2018), without incorporating both the frequency and intensity into consideration. Moreover, some studies lacked detailed criteria on the selection and treatment of original records, which could fuel criticisms of poor datasets in quantitative empirical research on linkages in the coupled climate–human system (van Bavel et al. 2019; Degroot et al. 2021), including the cause-of-effect issues that have received a lot of attention in recent years (White and Pei 2020; Degroot et al. 2022).
The purpose of this study is to create a new famine chronology for the Qing Dynasty in China (CE 1644–1911) at the county level. We use semantic differential and weighted mean methods to create composite index series of famine, which are then used to examine the frequency and severity of famines and their correlation with climate change/disasters at various time and spatial scales in the Qing Dynasty. The Qing Dynasty was chosen as the study period due to the abundance of local gazetteers and the consistency in the number of records over time. The present study focuses on eastern China as the whole study area, which can be further divided into two subregions: north China and south China, primarily based on modern provincial administrative boundaries (Fig. 1). According to modern county-level administrative divisions, eastern China as the study area comprises 1917 counties. The area of the counties is relatively uniform.
Area of the present study. The map data are from the National Geomatics Center of China (http://www.ngcc.cn/ngcc/; scale: 1:4 000 000).
Citation: Weather, Climate, and Society 16, 1; 10.1175/WCAS-D-23-0048.1
The eastern China was the center of production, population distribution, and other socioeconomic activities during the Qing Dynasty. China has been referred to as “The Land of Famine” in history (Teng et al. 2014), which is closely related to the unique climatic and socioeconomic backgrounds of eastern China. Eastern China has a marked monsoonal climate, characterized by the synchronization of rain and heat but great variability in regional seasonal and interannual distribution of precipitation (Zheng 2015). Spatially, water and heat resources gradually decrease from the south to the north in eastern China. In the south, the rainy season is very long, and the monsoon rainfall is concentrated from April to September every year. Therefore, rice in paddy fields is the main crop in south China. Most areas have at least two crops a year, usually sown in April and harvested in July or sown in July and harvested in October. In contrast, an annual wheat–maize double cropping rotation system has been developed to adapt to relatively dry conditions in northern China, where the rainy season is concentrated from July to August every year. Winter wheat, as the main crop in northern China, is commonly sown in October of the last year and harvested in June, followed by sowing of maize in July and harvest in October.
The changing climate and weather patterns, despite adaptation of farming systems, had a significant impact on the temporal and spatial distribution of precipitation, leading to frequent droughts and floods during the Qing Dynasty in China. The southern regions were more prone to flooding and were also sensitive to seasonal droughts, while the northern regions experienced more frequent long droughts, especially in spring (Zheng 2015). The agrarian-based social systems, which lacked market regulation mechanisms, added to the challenge of subsistence in the face of climatic disasters. It was particularly true during periods of historically high population density. In this regard, the Qing Dynasty saw an explosive growth in China’s population. This study primarily focuses on examining the influence of natural factors, specifically climate-related disasters, on famines. As mentioned earlier, historians have extensively studied the impact of human-related factors on famines. However, there has been relatively less quantitative research on the effects of climate-related disasters on various spatiotemporal scales. Additionally, acquiring precise annual population data for each county can be challenging. Nevertheless, this study will make a best effort to integrate the interaction between climate-related disasters and human factors, such as population pressure, to elucidate the mechanism by which climate change and disasters impact famine.
2. Materials and methods
a. Reconstruction of famine series
1) Data sources
The data on famines were collected from the book A Compendium of Chinese Meteorological Records of the Last 3,000 Years (Zhang 2004). This remarkable compendium is the result of 20 years of work by a team of Chinese scientists and historians. They studied over 8000 kinds of historical documents (more than 10 000 volumes) and evaluated more than 200 000 weather-related records, including natural disasters, famines, and famine relief efforts, covering the past 3000 years (Bradley 2006). Most of the materials used for this study came from local gazetteers and official histories. The records were mainly from the Qing period, and two of the four volumes focused on the Qing Dynasty (1644–1911) (volumes III and IV). Each entry was carefully cross-checked for consistency with other reports and organized by date and location (at the county level). Furthermore, historical locations were converted to contemporary names, facilitating the straightforward frequency count of famines using modern county-level administrative divisions.
In theory, datasets relying on primary sources from historians can provide a more accurate depiction of the long-term trends and severity of famines. However, acquiring a sufficient number of samples to conduct long-term statistical analysis is a formidable challenge. Based on incomplete statistics, there are over 80 000 volumes of local gazetteers in China alone. Extracting large samples from such an extensive body of literature is not only labor-intensive but also presents a significant challenge in ensuring uniformity in data distribution (both temporally and spatially). While the abovementioned compendium may not encompass all famine-related literature, it already offers a substantial sample size. Furthermore, famines, akin to droughts and floods, were regarded as rare events and have been meticulously documented. This book currently stands as the most comprehensive compilation of records pertaining to disasters and anomalies in China. Thus, from a statistical probability standpoint, selecting a sufficient number of famine samples from these historical records is indeed reliable enough to facilitate an analysis of trends in famine frequency and severity.
2) Data selection criteria
This study focuses on compiling famine-related records at the county scale in the Qing Dynasty. However, before proceeding, it is important to establish a clear definition of famine that is applicable to ancient China. In the historical context of China, famines were typically viewed as disasters resulting from crop failures and a lack of food supply, leading to widespread starvation among the populace. Given the predominantly agrarian nature of the social and economic system, the traditional beliefs that “hunger breeds discontentment” and “grain is the foundation of the country” became deeply ingrained in the minds of ancient Chinese society. Consequently, every dynasty placed significant emphasis on understanding and addressing the occurrence and progression of famines. From this perspective, the definition proposed by ÓGráda for Europe may not be suitable for ancient China. For example, emphasis on excess mortality from starvation alone may not capture many other important hunger phenomena. Additionally, digital records of death rates were scarce in Chinese literature. Furthermore, in historical China, government-led relief played a more significant role than the market in addressing food shortages or famine. Therefore, the present study defines famine as “a shortage of food that directly leads to a subsistence crisis, including hunger, malnutrition, stress-induced behavior (e.g., eating bark and grass roots, robbery, and violence), and even excess mortality from starvation or hunger-induced diseases, which necessitate external relief.” This food-consumption perspective excludes food scarcity caused solely by bad harvests or high food prices, which are related to levels of food production or entitlement. Mortalities resulting directly from war, disease, and certain disasters (such as drowning in floods) are also excluded. This definition is more in line with the understanding of famine in Chinese traditional culture, and more useful in identifying local-scale famines based on qualitative descriptions in ancient Chinese literature.
The records that meet the above definition of famine are selected based on the genetic principle of famine. It is widely accepted that a food shortage, whether caused by natural disasters or human activity, can lead to a series of subsequent effects if timely intervention is not taken. This disaster chain typically starts with mild food scarcity and hunger, which then progress to wider damage such as malnutrition, mass migration, searching for alternative food sources, human trafficking, and social conflict, followed by death due to starvation, cannibalism, and social and economic decline in the affected areas (Ben-Noun 2018). Parker (2013) conceptualized the above process as “fatal synergy” from the perspective of the synergistic effects of various elements (Degroot et al. 2022).
Based on the definition and principle mentioned earlier, we have compiled more than 15 000 items, which can be classified into three types. The first type includes descriptions containing keywords related to food shortage, famine, hunger, starvation, and cannibalism, accounting for approximately 65% of the total items. The second type includes descriptions without using hunger-related words, but focusing on the chain effects of hunger, such as people resorting to eating barks and roots of trees, eating white clay, and mass migration/begging triggered by disasters or other causes. This type accounts for 15% of the total items. The third type is mainly related to disaster relief, and the researchers only selected records more closely related to food consumption, such as food and money relief, porridge relief, work relief, and government-led sales of grain at a lower price than the ongoing market price. This type accounts for approximately 20%. Therefore, the first two types are directly related to famine and are given higher priority in ranking the severity of a famine.
3) Famine grading
The descriptions of famine in the historical records were concise yet descriptive, providing a rich vocabulary that allowed for clear semantic distinctions. Therefore, the method of semantic differential was used to assess the severity of the famine, which has been demonstrated to be a reliable and valid approach for obtaining quantitative famine indices using qualitative descriptions (China Meteorological Administration 1981; Ge et al. 2003; Zheng et al. 2006; Wei et al. 2015; Alfani and ÓGráda 2018; Fang et al. 2019; Tian et al. 2022). It is relatively easy and feasible to classify the semantic distinctions of vocabulary into three levels, as most descriptions of key vocabulary are concentrated within three distinct levels. Each record was assigned a grade from 1 to 3 based on the severity of the famine conveyed by the words used (see Table 1), and seven key principles were proposed to facilitate the grading of the famine level:
-
We only included one instance of famine per county per year (county time). In cases where multiple famines occurred or records contained varying levels of severity for a county in a year, the final famine level was determined based on the highest-level keywords (Table 1) after a comprehensive assessment of the overall famine situation.
-
Famines that lasted until the following spring were temporally adjusted to the previous year. However, famines that occurred in the spring were not adjusted if there was no record of famine in the previous winter.
-
Without additional evidence, disaster/famine relief records were typically assigned a grade of 1 to minimize the impact of relief records. While disaster relief can reflect food shortages on one hand, it also helps alleviate food shortages on the other. If the number of people receiving relief was very high (tens of thousands), the level would be increased by one rank (grade 2).
-
Descriptions that focused on death were further analyzed to determine whether they were related to a famine based on the specific circumstances. For example, a plague outbreak in conjunction with a rise in food prices following a drought could indicate a famine. Mass deaths after a drought, cooling, or an increase in food prices were also more likely to indicate a famine.
-
According to the concept of the disaster chain/fatal synergy mentioned before, if a famine was followed by an epidemic or an increase in deaths due to cooling or other disasters, the level of the famine would be increased by one rank, usually assigned as grade 2 or grade 3.
-
If there were conflicts or discrepancies between county-level records and records at a higher administrative level (e.g., prefecture), the county-level records would be given priority in determining the famine level.
-
Records without a clear location at the county level would be discarded (e.g., “sending officials to relieve the flood and famine in counties in Zhili Prefecture”).
Grading criteria based on the semantics of key words in famine records.
4) Composite index series reconstruction
It is important to note that Eq. (1) only takes into account the weighting for the frequency and severity of famine based on the county times of occurrence. Currently, this study does not incorporate spatial weighting when computing the composite famine index for a region. It would be more comprehensive to establish specific spatial weights for each county, considering factors like population density, which is among the most crucial indicators. However, the grading of famine severity already to some extent reflects regional disparities. Furthermore, the population density for each county fluctuated over time, making it challenging to assign weighting coefficients for more than 1900 counties. This aspect warrants further investigation in future studies.
b. Climatic disasters and temperature
Drought and flood were the two most important disasters affecting human activities in historical China. Their frequency and severity were also affected by climate change. Data on droughts and floods from 1644 to 1911 were obtained from the Yearly Charts of Dryness/Wetness in China for the Last 500-Year Period (China Meteorological Administration 1981). This dataset classified dryness/wetness states into five grades ranging from very wet to very dry. The study area for the entire eastern China consisted of 90 stations, with 31 stations in the north and 59 stations in the south. The annual drought index (DI) and flood index (FI) in eastern China were calculated using the method described by Chen (1989). These indices ranged from 0 to 1, indicating the intensity of the disaster varied from slight to severe. Similarly, the drought and flood indices for north China (NDI and NFI, respectively) and south China (SDI and SFI, respectively) were also calculated.
Temperature is a major indicator of climate change. The temperature changes during the study period were determined using the temperature anomaly series provided by Ge et al. (2003), which is one of the most widely accepted and well-dated general temperature proxies with decadal resolution currently available. This series mainly reflects changes in the winter half-year (October–April) temperature, which is closely correlated with growing-season and annual temperature changes in China. As compared with other nationwide paleotemperature proxies (Yang et al. 2002), it was reconstructed from natural and farming phenological phenomena recorded in Chinese literature and thus has the advantage of being clear in explaining the effects of temperature on agriculture and famine.
c. Statistical analysis
To reveal the temporal pattern of famine in China during the years 1644–1911, six major stages can be distinguished by analyzing the decadal smooth curves of both the total number of famine counties and FAI in relation to their average values over the entire period. Additionally, within each major stage, there were episodes of frequent severe famines (EFSF), which were defined as periods with FAI (or NFAI/SFAI) higher than the average for the entire time for three consecutive years or four of five consecutive years. These episodes could reflect the persistence and clustering of severe famine prevalence, which could have a cumulative impact. An extreme famine year was defined as a year with a probability of famine occurrence lower than 10%, meaning that the FAI (or NFAI/SFAI) should be higher by at least 1.28 times the standard deviation than the average value for the entire period.
To analyze the relationship between the reconstructed famine indices and climatic variables, Pearson correlation coefficient was used. Because the temporal patterns for the disaster/climate–famine relationship were possibly scale dependent, 0.1-Hz (10 yr) FFT low-pass and high-pass filters were utilized to obtain short-term variations and long-term trends for all annual index series, respectively. Decadal-resolved series for all variables were obtained by calculating the decadal arithmetic average. Because of the small sample size, cubic polynomial fitting was used to extract the long-term trends of decadal-resolved series for all variables.
3. Results and discussion
a. Prevalence and magnitude of famine
Figure 1 reveals that there was no clear linear trend in the prevalence of famine in China from 1644 to 1911 (as seen in Figs. 2a–d). Instead, a typical pattern emerged, characterized by fluctuations occurring over multidecadal stages, punctuated by occasionally severe famines on an annual scale. Typically, after a peak, there would be periods of relative quiet, with fewer incidences or less severe famines. One possible explanation for this pattern is the significant loss of population resulting from the severe famines (Ben-Noun 2018).
Reconstructed famine series and their relationship with climatic disasters at the annual scale during the Qing Dynasty in China. Reconstructed annual (a) FAI, (b) NFAI, and (c) SFAI series. The thick blue line in (a) is a decadal FFT smoothing curve. Gray columns are EFSF, and red arrows indicate extreme famine years. (d) Total number of famines for all levels measured with county times. (e) Number of famines for each level. The red, orange, and gray bars are numbers of famine grade 3, 2, and 1, respectively. (f) DI and (g) FI (China Meteorological Administration 1981).
Citation: Weather, Climate, and Society 16, 1; 10.1175/WCAS-D-23-0048.1
From the perspective of famine occurrence frequency, it can be said that (eastern) China was a region plagued by famines. Every year had at least one county-time famine in 1644–1911. Particularly, grade 1 and 2 constituted 56.1% and 32.5% of all famines, respectively (as shown in Table 2). Furthermore, of the 268 years analyzed, 203 (76%) experienced extreme famines (grade 3) with an average return period of only 1.32 years. On average, the impact of famine affected 52 counties each year. However, if we take into account the 21 extreme famine years identified in China, the average number of affected counties per year was almost three times higher, at 150 (as shown in Fig. 2a).
Stages of famines in China during 1644–1911.
Table 2 lists six major stages characterized by different combinations of frequency (county times) and magnitude of famine (FAI). The periods of 1644–80, 1721–60, and 1811–78 had relatively higher prevalence and more severity of famines, with more frequent clusters of severe and extreme famines observed, particularly in 1644–80 and 1811–78. Nearly one-half of all of the 21 extreme famine years occurred in 1811–78, including the greatest famine in 1876–78 in north China. The relatively high frequency and severity of famine in 1721–60 were mainly due to more frequent mild famines of grade 1, which was largely related to the increasing number of relief records. The proportion of grade-1 famine evaluated based on relief records ranged from 25% in 1681–1720 to as high as 47% in 1721–60 (832 of 1759 county times for grade-1 famine), which was much higher than that of other stages. Statistics show that 62% (1706) of the collected relief records (2785) were distributed in 1681–1810. Identifying distinct time periods as mentioned above aids in gaining a clearer understanding of the phased patterns of famine changes. These patterns are often indicative of responses to phased alterations in the socioeconomic system or climate shifts. Notably, climate change frequently exhibits periodicity.
With the exception of the last stage, the changes in major stages were generally consistent with the socioeconomic development of the Qing Dynasty in eastern China (Xiao et al. 2014, 2015). In general, the prevalence and severity of famines tended to increase during periods of social upheaval and decline and decrease during periods of prosperity in the middle of the Qing Dynasty. This will be discussed in more detail in section 3c. It is worth noting that most of the prosperous eighteenth century coincided with a period of relatively warm climate (1710–70) during the Little Ice Age (Ge et al. 2003). However, socioeconomic factors alone cannot explain the temporal patterns of famines at the annual to interannual scale, nor can they explain all the differences in famines between the north and south of China, where the stages of famine differed from those in eastern China (Fig. 2). On the contrary, some of these differences seem to be better explained by extreme climatic disasters.
b. Climate/disaster–famine relationship at different spatiotemporal scales
The significance of climate change and disasters in causing famines cannot be ignored, despite the importance of socioeconomic factors (Slavin 2016; Ben-Noun 2018). In the case of eastern China, the changes in famine occurrences were closely related to temperature and disasters at the multidecadal to centennial scale (Fig. 3). The frequency and severity of famines were notably lower during the eighteenth century in the Little Ice Age, when most of the time was characterized by relatively higher temperatures and fewer disasters (Ge et al. 2003; Zheng et al. 2006). Conversely, the rise of drought and flood disasters, which began at the turn of the eighteenth and nineteenth centuries, coincided with an increase in the frequency and severity of famines. At the annual scale, while not all peaks of droughts or floods led to an increase in famines, all of the EFSF and extreme famine years were strongly associated with higher values of the drought and floods index, particularly for the droughts (Fig. 2; Table 3).
Series comparison of famine, temperature, and drought/flood disaster at the decadal scale in China in 1644–1911: (a) winter half-year temperature anomaly in eastern China, (b) DI (red curve) and FI (blue curve), and (c) FAI. All of the dotted lines are cubic polynomial fitting series.
Citation: Weather, Climate, and Society 16, 1; 10.1175/WCAS-D-23-0048.1
Coincidence between droughts/floods and typical famine periods in China during AD 1644–1911.
Correlation analysis allows for a more systematic and quantitative analysis of the multiscale linkages between climate, disasters, and famine. Results show that the long-term trends of temperature anomaly and FAI were significantly and negatively correlated (r = −0.735; p < 0.001). It is worth noting that correlation analysis based on the cubic polynomial fitting series would lead to a decrease in actual degrees of freedom, which may affect the correlation coefficients and their statistical significance. At the decadal scale, there was no correlation between famine and temperature (r = 0.013; p = 0.950). This could be due to the fact that famine’s response to temperature changes was less sensitive in south China than in north China. In fact, a significant negative correlation between famine and temperature changes at the decadal scale was found in the North China Plain during the period of 1736–1911 (Xiao 2020). Another possible explanation is that famines were commonly triggered by unexpected and abrupt temperature anomalies, which could be relatively short-lived (e.g., lasting only a season or 0.5 year). Such anomalies might not be captured in temperature reconstructions on a decadal resolution. Therefore, it will be valuable to employ annually resolved temperature data when examining the influence of temperature on famine in future.
There were two primary ways in which temperature affected famine. First, long-term cooling would diminish heat resources, shortening the growing season and lowering the altitude at which crops could be grown, particularly in higher latitude regions (Galloway 1986). Second, high climatic variability might increase the likelihood of changes in short-term weather patterns (such as droughts/floods) (Wigley 1985). Both above factors resulted in lower unit yield and total food output. In China, grain yields were reduced by 10%–25% during the cold period of 1840–90, when compared with the relatively warm period of 1730–70 (Gong et al. 1996). Temperature was a critical limiting factor for agricultural production in north China, a region mainly affected by continental monsoon climate. Therefore, on the decadal scale, temperature changes likely impacted famine primarily by triggering drought and flood disasters, especially in south China. The decadal correlation coefficients of temperature–floods and temperature–droughts in eastern China are −0.291 (p = 0.140) and −0.180 (p = 0.372), respectively. Statistical insignificance might be related to small sample or the nonlinear response relationship between temperature and drought/flood disasters. However, at least the three sharp cooling events around 1650s–60s, 1790s–1810s, and 1870s–80s were all accompanied or followed by peaks of drought or flood (Fig. 3). On the multidecadal to centennial scale, the impact of temperature on famine was to cause a reduction in food yield both by reducing heat through cooling and increasing the risk of drought/flood disasters.
On the other hand, the association between climatic disasters (drought/flood) and famine was less clear and varied depending on the scale and region of analysis. Famine showed very strong positive correlations with drought, both in short-term variations and long-term trends. This pattern was applicable at both the national (eastern China) and regional levels (Table 4). Previous studies had long recognized the positive link between drought and famine, but most of them focused primarily on northern China (Fang et al. 2013; Xiao et al. 2014). This study reveals that drought-induced famines were also prevalent in southern China, although they were less severe than those in the north.
Pearson’s correlation coefficients between famine and drought/flood at different spatiotemporal scales in 1644–1911. One, two, and three asterisks indicate statistical significance at p < 0.05, p < 0.01, and p < 0.001, respectively.
The study suggests that flood did not have a significant impact on famines on a national scale (eastern China), as the relationship between flood and famine varied greatly between the north and south regions of China. In north China, there was a statistically significant (p < 0.05) negative correlation between flood and famine at both annual and decadal scales (Table 4). This correlation was more evident in their long-term trends, with high-frequency signals excluded. This indicates that flood mainly played a positive role in alleviating famine in the long run in north China. When flood disasters caused damage, they also brought water resources and fertile soil to grain production (Will 2003). However, the effect of flood on famine could be positive or negative in the short term, depending on whether the intensity of flood exceeded the threshold of human adaptation (Fang et al. 2019).
In contrast, on the annual scale, there was a significant positive correlation (p < 0.001) between famine and flood in south China, for both their short-term variations and long-term trends (Table 4). However, this correlation was not significant on the decadal scale, suggesting that flood mainly had a negative impact on intensifying famine in both the short and long terms in south China. Nonetheless, the negative effects of flood were of shorter duration and easier to mitigate by human intervention on the decadal scale relative to those of drought. This was partly supported by the results of the lagging correlation analysis (Tables 5 and 6), which indicated that in north China, the effects of drought on famine could last for 2–3 years, while the effects of flood could only last for 1–2 years. In south China, the effects of both drought and flood lasted for 1–3 years, but there were some decreases in both the Pearson’s correlation coefficient and its statistical significance (Tables 5 and 6). Furthermore, the robust negative correlation with a 5-yr lag between FAI and DI may suggest a 10-yr periodicity in the famine and drought index. Regrettably, the absence of temperature data at an annual resolution makes it impossible to explore the time-lag effect of temperature on famine.
Lagging correlation between famine and drought/flood based on the original annual-resolved series at different regions in 1644–1911 for different time lags (yr). One, two, and three asterisks indicate statistical significance at p < 0.05, p < 0.01, and p < 0.001, respectively.
Lagging correlation between famine and drought/flood based on the decadal FFT low-pass-filtering series at different regions in 1644–1911 for different time lags (yr). One, two, and three asterisks indicate statistical significance at p < 0.05, p < 0.01, and p < 0.001, respectively.
Famine did not exhibit any clear delayed response to drought or flood in any of the cases examined. There are at least two plausible explanations for this. The first is related to the backward temporal corrections used to date famines that lasted into the next spring. This correction artificially weakens the lagging effects of disasters at the annual scale. The second explanation is that droughts and floods may indeed have immediate effects on famine occurrence at the seasonal scale. Studies have shown that famine-induced mortality was highest during the summer and early fall due to epidemics spreading rapidly through water and food, such as in Sweden from 1750 to 1910 (Dribe et al. 2015), and that famines mostly occurred in summer and autumn (May–August) in Jiangsu Province from 1644 to 1911 (Wei 2020). These two explanations are not mutually exclusive, and it should be acknowledged that the temporal lag in famine response to disasters is due to the buffering mechanism of human society. Original literature has extensively noted and recorded that some famines occurred in winter and the following spring, during a gap in harvest. The relatively slower development of droughts, in particular, gives human society more time to respond positively (Will 2003).
The primary concern revolves around whether the lagging impact of disasters was applicable to the seasonal or annual scale. An examination of the original annual-resolved data shows that the Pearson’s correlation coefficient experienced a minor decline in first-order lagging correlation but a substantial decline in second-order lagging correlation relative to zero lagging (Tables 4 and 5). Therefore, the lagging effects of drought/flood disasters on famine occurrences were primarily on the seasonal scale. Moreover, apart from the positive regulation mechanism from human intervention, such as disaster relief, migration, and stabilizing food prices, the strong seasonal cycle of grain harvest influenced by climate also played a significant role in minimizing the impact of disasters on famines (Fang et al. 2019).
In general, the impact of drought on famine was more severe in the northern regions of China than in the southern regions, while the impact of flood on famine was more severe in the south than the north during the Qing Dynasty. This pattern aligns with the well-known regional differences in dry conditions in the north and wet conditions in the south of China (Zheng 2015). The immediate impact of major drought or flood disasters on famines was typically felt at a seasonal or 2-yr scale. When such an event occurred, it was more likely to cause famines in the same year and the following 1–2 years. Droughts had a more lasting effect on famines than floods, especially in the north of China, where long or successive droughts could lead to a cumulative effect on famine (Wigley 1985). This was also likely due to the predominance of one crop farming systems that increased the risk of drought. Thus, these factors can help to explain the strong positive correlation between drought and famine in north China.
In comparison with the north, the impact of flood on famines in the south appeared to be better managed at longer time scales ranging from 3 years to a decade, and thus did not exhibit significant correlation at the decadal scale. In general, the short-term impacts of drought and flood disasters on famines in south China were less severe. These differences could be attributed to variations in geographical factors and regional buffering capacity in the south, including milder drought conditions, shorter and more localized flood disasters, greater diversity in subsistence resources such as crops, and a more flexible crop system that allowed for at least two crops per year (Zheng 2015). In summary, the effects of drought and flood on famine were complex and varied across temporal and spatial scales, highlighting the importance of considering regional characteristics and adaptation measures when studying the relationship between climate disasters and famine.
c. Dynamic relationships between famine and disasters
Previous studies have consistently shown that the relationship between climate and human society is dynamic and ever changing (Wei et al. 2015; Tian et al. 2017; Lee 2018; Fang et al. 2019). The occurrence of famines, therefore, is the result of intricate and multifaceted interactions between climate change, disasters, and human agency within a specific region and time period.
In eastern China as a whole, a strong positive correlation between drought and famine was observed in almost all intervals from 1644 to 1911, except for certain periods disrupted by floods (Fig. 4a). These positive correlations were primarily due to the contribution of north China (Fig. 4b). It was also confirmed that the impact of flood disasters on famines was double-edged, with its influence becoming negative when flood intensity exceeded the threshold of human adaptation for short-term periods, and vice versa.
Moving correlation analysis between famine indices and drought/flood disasters using a 20-yr moving window centered at the x-axis values. The orange lines are pairs of famine indices and drought, and black lines are pairs of famine indices and flood. Shown are (a) FAI–DI and FAI–FI, (b) NFAI–NDI and NFAI–NFI, and (c) SFAI–SDI and SFAI–SFI. Yellow boxes and gray boxes indicate times of intensified droughts and floods, respectively.
Citation: Weather, Climate, and Society 16, 1; 10.1175/WCAS-D-23-0048.1
The first major stage of 1644–80 (Table 2) coincided with the early chaos stage of the Qing Dynasty, which was still facing frequent wars in reunifying eastern China (mainly the south) prior to the pacification of the Three Feudatories’ revolt (1673–81). The prevalence of famines in eastern China during this period was largely attributed to the contribution from south China (Figs. 2a,c). The social upheaval caused by wars may have worsened the impact of intensified drought on famines in south China during the early Qing Dynasty (Fig. 4c).
After regaining a peaceful environment in 1681, the population in China experienced explosive growth, increasing from approximately 76 million in 1661 to 330 million in 1812 (Sun and Zhang 1979; Cao and Chen 2002). During the first period from 1681 to 1780, drought was relatively less severe in both northern and southern China, despite some extreme events at the annual scale and temporary aggravation at the decadal scale (e.g., 1716–25) (Figs. 4b,c; 5). Although flood disasters increased to some extent at the decadal to multidecadal scale (e.g., 1725–60 in north China and 1700–10 in south China), their influence on famine was relatively short-lived and less significant (Figs. 4b,c). The “Kong and Qian Eras,” known as the golden era of the Qing Dynasty (approximately 1681–1796), may have played a significant role in reducing the occurrence of serious famines in this period, as evidenced by the relative increase in famine relief records.
Drought and flood index series in north and south China. The red lines are 10-point adjacent-averaging smoothing series. The dashed lines are the average value in 1644–1911.
Citation: Weather, Climate, and Society 16, 1; 10.1175/WCAS-D-23-0048.1
During the second period, 1781–1810, drought became more severe in both north and south China, particularly during 1775–1816 in the north and 1775–87 in the south (Figs. 4a,b). This resulted in a linear increase in the Pearson’s correlation coefficients between famine and drought, reaching a high level for the entire eastern China. Consequently, two major famines occurred in 1778 and 1786, respectively, with relatively less severe social impacts than the famines of 1876–78. This was possibly due to the remaining financial stability and effective relief during the later years of the “Kong and Qian Eras.” Despite this, the Qing Dynasty began its decline from prosperity to decline due to the emergence of a sharp contradiction between humans and the land (Fang et al. 2013).
The stages of 1811–78 marked a further decline for the Qing Dynasty. This period can be best explained by Malthusian population theory, and the heavy prevalence of famines played a role in checking population growth (Malthus 1798). It is generally accepted that the population and cropland reached its peak during the reign of the Emperor Daoguang (1821–50) for the Qing Dynasty (Sun and Zhang 1979; Cao and Chen 2002). However, the crisis of high population pressure on resources had appeared since the nineteenth century or even earlier (Feuerwerker 1990). The cropland per capita had decreased from approximately 6–7 mou before 1734 to 1.6–3.69 mou after 1753 (Sun and Zhang 1979; Wu 1985). Based on the unit yield of 183.5 kg mou−1 in the mid–Qing Dynasty, the grain output per capita dropped from 576 kg in 1753 to 292 kg in 1822 (Wu 1985). This was lower than the critical food security share of 300 kg per capita and left the population vulnerable to the impact of climatic and disaster disturbances on famines.
The decline of the Qing Dynasty continued in the stages of 1811–78, which can be explained by the Malthusian population theory, as the heavy prevalence of famines acted as a check on population growth. The increase of population and cropland had already reached its peak during the reign of Emperor Daoguang, but the crisis of high population pressure on resources had appeared much earlier. The decrease in cropland per capita and grain output per capita had given opportunities for the influence of climatic and disaster disturbances on famines, especially in north China where the impact of drought on famine had stepwise increased until the great famine of 1876–78. Although the changes in drought showed no obvious linear trends, they still fluctuated with cycles. Flood disasters had a long rising trend during 1825–60 in south China (Fig. 5d), which undermined grain production and led to serious increases in famine. The Taiping Rebellion was one of the serious responses to the long-term failure of the grain harvest (Ge and Wang 1995). The economic depression caused by floods in south China during the early nineteenth century was accepted as a major driving factor of the “Daoguang Depression” (Li 2007; Cao and Yang 2012). The great famine of 1876–78 was triggered by an extreme drought event, but it was a result of interactions among intensifying human–land contradictions since the early nineteenth century, periodic influence of drought trends in the north, and weakening of the regional buffering mechanism due to the flood-induced decline of south China.
In the final stage of 1879–1911, the incidence of famine decreased significantly, except for the famine of 1900, which was caused by droughts in both the north and south of China. This could be attributed to the fact that the north became wetter while the droughts/floods in the south declined. Moreover, the correlation between famine and drought appeared to decrease, particularly in the south, which could be related to a reduction in the human–land contradiction in the late Qing Dynasty. Studies indicate that the Taiping Rebellion alone caused the deaths of 70 million people, while the great famine of 1876–78 resulted in another 23 million deaths. If the population loss during the Shan Gan Muslim Rebellion was considered, then at least 118 million people were lost between 1850 and 1877. Furthermore, the development of modern industrialization helped to reduce the incidence of famine, especially in southern China.
4. Conclusions
In traditional agrarian-based societies, such as Qing Dynasty China, human beings were vulnerable to droughts and floods. This study uses a reconstructed famine index series to perform a quantitative analysis of famine prevalence and magnitude, as well as its relationship with climate and disasters at different spatiotemporal scales, and the dynamic mechanism of drought/flood-induced famine. The following conclusions were drawn:
-
Famines occurred frequently in China during the Qing Dynasty, with multidecadal-scale stages interrupted by intermittent occurrences of great famines at an interannual scale. The most prevalent and severe famines clustered in 1811–78, followed by 1644–80.
-
At the seasonal to interannual scale, drought/flood events were the main triggers of great famines. The effects could last for at least 1–2 years and were usually inevitable, even in periods of social and economic prosperity. The influence of drought on famine was most prominent, in both short-term variations and long-term trends. The effect of floods was relatively shorter in duration and easier to regulate by human agency at the decadal scale.
-
Flood disasters had a double-edged influence on famine. In north China, it was dominantly positive, but negative in south China. It was usually negative when the magnitude of floods exceeded the threshold of human adaptation and vice versa.
-
The long-term trends of climate change, represented by temperature, showed good coincidence and negative correlation with famines in eastern China. However, on the decadal scale, there is no significant correlation between temperature and famine.
-
Drought-induced famine in the north was much more serious than in the south, while flood-induced famine in the south was more serious than in the north during the Qing Dynasty. Overall, the impacts of drought and flood disasters on famines in south China were less serious than in the north, which may be related to differences in regional buffering capacity.
-
The main clusters of severe famines occurring in 1811–78 were related to both climatic cooling and the rise of drought/flood events under a situation of increasing population pressure on resources. For instance, the great famine in 1876–78, triggered by an extreme drought event in north China, was a result of long interactions among intensifying human–land contradiction since the early nineteenth century, periodic influence of drought in north China, and weakening of regional buffering mechanisms due to the flood-induced decline in south China.
While the risk of famine due to food shortages has significantly decreased in current society, food security remains a major threat to human health and sustainable development. The mechanisms behind famine prevalence are much more complex today. Understanding the influence of climate change and extreme events on food security is crucial, particularly in the context of global warming. Numerous studies have already demonstrated that global warming has led to an increase in the frequency and severity of extreme events, posing a significant threat to food security.
The long-term experience uncovered in this study indicates that the impact of extreme droughts and floods on food production and consumption in the short term is largely inevitable. However, positive human responses on an annual to decadal scale are crucial in preventing the exacerbation of famine and the escalation of food security issues. Additionally, it is imperative not to overlook the complementary and buffering roles between regions. Droughts and floods often disrupt food entitlement through damage to transport facilities and disruptions in market order. Given the backdrop of population growth, shrinking arable land, and the commercialization and globalization of food production, achieving a balance in coordinating large-scale food production with basic regional food production is essential in mitigating the short-term effects of extreme disasters and ensuring regional food security. Furthermore, in recent years, global warming appears to have started affecting precipitation patterns in China, with an increase in the frequency of summer heavy rain and floods in the north, while the threat of compound extreme events of high temperature and drought in the south impacting crop production has been observed. If a pattern of drought in the south and floods in the north emerges in the future, the government or policy makers will need to adjust regional adaptation strategies to proactively mitigate the impacts of disasters.
Acknowledgments.
We express many thanks for the helpful and constructive suggestions from three anonymous reviewers. The study is under the auspices of the National Natural Science Foundation of China (41701219; 41972193; 72061137072), the Dutch Research Council (482.19.607), and the National Key Research and Development Program of China (2019YFC0507801). The authors of the study have no conflicts of interest.
Data availability statement.
Data of the temperature anomaly sequence are included in Ge et al. (2003). Data of the dry–wet sequence are included in China Meteorological Administration (1981). Data of the sequence of famine instances are collected by the authors from Zhang (2004).
REFERENCES
Alfani, G., and C. ÓGráda, 2017: Famine in European History. Cambridge University Press, 325 pp.
Alfani, G., and C. ÓGráda, 2018: The timing and causes of famines in Europe. Nat. Sustainability, 1, 283–288, https://doi.org/10.1038/s41893-018-0078-0.
Ben-Noun, L., 2018: Characteristics of Famine from Ancient to the Modern Times. B.N. Publication House, 120 pp.
Bradley, R. S., 2006: Book Review: A compendium of Chinese meteorological records of the last 3,000 years. Holocene, 16, 621–622, https://doi.org/10.1177/095968360601600415.
Cao, S., and Y. Chen, 2002: Malthusian theory and Chinese population since the Qing Dynasty—A review of recent studies by American scholars (in Chinese). Hist. Res., 1, 41–54.
Cao, S., and Y. L. Yang, 2012: Mt. Tambora, climatic changes, and China’s decline in the nineteenth century. J. World Hist., 23, 587–607, https://doi.org/10.1353/jwh.2012.0066.
Chen, Y., 1989: A study of hazard indexes of drought and flood (in Chinese). Zaihaixue, 4, 10–13.
China Meteorological Administration, 1981: Yearly Charts of Dryness/Wetness in China for the Last 500-Year Period (in Chinese). Sinomaps Press, 332 pp.
Curtis, D. R., and J. Dijkman, 2017: The escape from famine in the northern Netherlands: A reconsideration using the 1690s harvest failures and a broader northwest European perspective. Seventeenth Century, 34, 229–258, https://doi.org/10.1080/0268117X.2017.1410494.
Degroot, D., and Coauthors, 2021: Towards a rigorous understanding of societal responses to climate change. Nature, 591, 539–550, https://doi.org/10.1038/s41586-021-03190-2.
Degroot, D., K. J. Anchukaitis, J. E. Tierney, F. Riede, A. Manica, E. Moesswilde, and N. Gauthier, 2022: The history of climate and society: A review of the influence of climate change on the human past. Environ. Res. Lett., 17, 103001, https://doi.org/10.1088/1748-9326/ac8faa.
Dribe, M., M. Olsson, and P. Svensson, 2015: Famines in the Nordic countries, AD 536–1875. Lund Papers in Economic History 138, Lund University Dept. of Economic History, 46 pp., https://lucris.lub.lu.se/ws/portalfiles/portal/5987932/8032075.pdf.
Fang, X., Y. Su, J. Zheng, L. Xiao, Z. Wei, and J. Yin, 2019: The Social and Economic Impact of Historical Climate Change in China (in Chinese). Science Press Beijing, 275 pp.
Fang, X. Q., L. B. Xiao, and Z. D. Wei, 2013: Social impacts of the climatic shift around the turn of the 19th century on the North China Plain. Sci. China Earth Sci., 56, 1044–1058, https://doi.org/10.1007/s11430-012-4487-z.
Fang, X. Q., Y. Su, J. Yin, and J. C. Teng, 2015: Transmission of climate change impacts from temperature change to grain harvests, famines and peasant uprisings in the historical China. Sci. China Earth Sci., 58, 1427–1439, https://doi.org/10.1007/s11430-015-5075-9.
FAO, IFAD, UNICEF, WFP, and WHO, 2020: The State of Food Security and Nutrition in the World 2020: Transforming food systems for affordable healthy diets. FAO, 320 pp., https://www.fao.org/3/ca9692en/ca9692en.pdf.
FAO/Food Security Analysis Unit, 2006: Integrated Food Security and Humanitarian Phase Classification (IPC) framework: Technical manual version 1. FAO/FSAU Tech. Series IV, 47 pp., https://www.fao.org/3/a0748e/a0748e.pdf.
Feuerwerker, A., 1990: Chinese economic history in comparative perspective. Heritage of China: Contemporary Perspectives on Chinese Civilization, P. Ropp, Ed., University of California Press, 224–241.
Galloway, P. R., 1986: Long-term fluctuations in climate and population in the preindustrial era. Popul. Dev. Rev., 12 (1), 1–24, https://doi.org/10.2307/1973349.
Ge, Q., and W.-Q. Wang, 1995: Population pressure, climate change and Taiping Rebellion (in Chinese). Geogr. Res., 14, 32–41.
Ge, Q., J. Zheng, X. Fang, Z. Man, X. Zhang, P. Zhang, and W.-C. Wang, 2003: Winter half-year temperature reconstruction for the middle and lower reaches of the Yellow River and Yangtze River, China, during the past 2000 years. Holocene, 13, 933–940, https://doi.org/10.1191/0959683603hl680rr.
Geens, S., 2018: The Great Famine in the county of Flanders (1315–17): The complex interaction between weather, warfare, and property rights. Econ. Hist. Rev., 71, 1048–1072, https://doi.org/10.1111/ehr.12605.
Gong, G., Q. Ge, and K. Xu, 1996: Influence of climatic changes on agriculture. Historical Climatic Changes in China, P. Y. Zhang, Ed., Shandong Science and Technology Press, 406–425.
Huhtamaa, H., 2015: Climatic anomalies, food systems, and subsistence crises in medieval Novgorod and Ladoga. Scand. J. Hist., 40, 562–590, https://doi.org/10.1080/03468755.2015.1063081.
Lee, H. F., 2018: Internal wars in history: Triggered by natural disasters or socio-ecological catastrophes? Holocene, 28, 1071–1081, https://doi.org/10.1177/0959683618761549.
Li, B. Z., 2007: The Daoguang depression and the 1823 flood—Economic decline, climatic cataclysm and the nineteenth-century crisis in Songjiang (in Chinese). J. Soc. Sci., 22, 173–178.
Malthus, T. R., 1798: An Essay on the Principle of Population, as It Affects the Future Improvement of Society. Johnson, 126 pp.
Nelson, M. C., and Coauthors, 2015: Climate challenges, vulnerabilities, and food security. Proc. Natl. Acad. Sci. USA, 113, 298–303, https://doi.org/10.1073/pnas.1506494113.
ÓGráda, C., 2009: Famine: A Short History. Princeton University Press, 327 pp.
Parker, G., 2013: Global Crisis: War, Climate Change and Catastrophe in the Seventeenth Century. Yale University Press, 904 pp.
Sen, A., 1981: Poverty and Famines: An Essay on Entitlement and Deprivation. Clarendon Press, 257 pp.
Slavin, P., 2013: Market failure during the Great Famine in England and Wales (1315-1317). Past Present, 222, 9–49, https://doi.org/10.1093/pastj/gtt025.
Slavin, P., 2016: Climate and famines: A historical reassessment. Wiley Interdiscip. Rev.: Climate Change, 7, 433–447, https://doi.org/10.1002/wcc.395.
Su, Y., Y. Kang, X. Zhai, and X. Fang, 2021: The relationship between temperature changes and peacemaking events between farming and nomadic groups in northern China over the past 2000 years. Wea. Climate Soc., 13, 327–339, https://doi.org/10.1175/WCAS-D-20-0153.1.
Sun, Y., and J. Zhang, 1979: The records of reclamation and Dingkou in the Qing Dynasty (in Chinese). Essays Hist. Qing Dynasty, 1, 117–120.
Teng, J., Y. Su, and X. Fang, 2014: The reconstruction and analysis of famine sequence form the western Han to the Qing Dynasty (206BC-1911AD) (in Chinese). J. Chin. Hist. Geo., 29, 26–32.
Tian, F., X. Chen, and Y. Su, 2022: Comparative analysis of extreme drought events and social impacts in Henan Province during the Middle Ming Dynasty. Wea. Climate Soc., 14, 1009–1021, https://doi.org/10.1175/WCAS-D-22-0015.1.
Tian, H., C. Yan, L. Xu, U. Büntgen, and Z. Zhang, 2017: Scale-dependent climatic drivers of human epidemics in ancient China. Proc. Natl. Acad. Sci. USA, 114, 12 970–12 975, https://doi.org/10.1073/pnas.1706470114.
van Bavel, B. J. P., D. R. Curtis, M. J. Hannaford, M. Moatsos, and T. Soens, 2019: Climate and society in long‐term perspective: Opportunities and pitfalls in the use of historical datasets. Wiley Interdiscip. Rev.: Climate Change, 10, e611, https://doi.org/10.1002/wcc.611.
Wei, Z., 2020: Spatio-temporal characteristics of famine and its environmental causes in the Jiangsu-Shanghai region during the Qing Dynasty (1644-1911) (in Chinese). Prog. Geogr., 39, 1333–1344, https://doi.org/10.18306/dlkxjz.2020.08.008.
Wei, Z., A. M. Rosen, X. Fang, Y. Su, and X. Zhang, 2015: Macro-economic cycles related to climate change in dynastic China. Quat. Res., 83, 13–23, https://doi.org/10.1016/j.yqres.2014.11.001.
White, S., and Q. Pei, 2020: Attribution of historical societal impacts and adaptations to climate and extreme events: Integrating quantitative and qualitative perspectives. Past Global Changes Mag., 28, 44–45, https://doi.org/10.22498/pages.28.2.44.
Wigley, T. M. L., 1985: Climatology: Impact of extreme events. Nature, 316, 106–107, https://doi.org/10.1038/316106a0.
Will, P. E., 2003: Bureaucracy and Famine in 18th Century China (in Chinese). Jiangsu People’s Publishing House, 364 pp.
Wu, H., 1985: Study on Grain Yield per Mu in China (in Chinese). Agricultural Publishing House, 246 pp.
Xiao, L., 2020: Spatio-temporal distribution of famine and its relationship with climate, disaster, harvest in North China during 1736-1911 (in Chinese). Adv. Earth Sci., 35, 478–487, https://doi.org/10.11867/j.issn.1001-8166.2020.33.
Xiao, L., X. Fang, Y. Zhang, Y. Ye, and H. Huang, 2014: Multi-stage evolution of social response to flood/drought in the North China Plain during 1644–1911. Reg. Environ. Change, 14, 583–595, https://doi.org/10.1007/s10113-013-0516-5.
Xiao, L., X. Fang, J. Zheng, and W. Zhao, 2015: Famine, migration and war: Comparison of climate change impacts and social responses in North China between the late Ming and late Qing dynasties. Holocene, 25, 900–910, https://doi.org/10.1177/0959683615572851.
Yang, B., A. Braeuning, K. R. Johnson, and S. Yafeng, 2002: General characteristics of temperature variation in China during the last two millennia. Geophys. Res. Lett., 29, 1324, https://doi.org/10.1029/2001GL014485.
Zhai, X., X. Fang, and Y. Su, 2020: Regional interactions in social responses to extreme climate events: A case study of the North China famine of 1876–1879. Atmosphere, 11, 393, https://doi.org/10.3390/atmos11040393.
Zhang, D. D., H. F. Lee, C. Wang, B. S. Li, Q. Pei, J. Zhang, and Y. L. An, 2011: The causality analysis of climate change and large-scale human crisis. Proc. Natl. Acad. Sci. USA, 108, 17 296–17 301, https://doi.org/10.1073/pnas.1104268108.
Zhang, D. E., 2004: A Compendium of Chinese Meteorological Records of the Last 3,000 Years (in Chinese). Phoenix Press, 3666 pp.
Zheng, D., 2015: General Introduction to Natural Geography of China (in Chinese). Science Press, 767 pp.
Zheng, J. Y., W. C. Wang, Q. S. Ge, Z. M. Man, and P. Y. Zhang, 2006: Precipitation variability and extreme events in eastern China during the past 1500 years. Terr. Atmos. Oceanic Sci., 17, 579–592, https://doi.org/10.3319/TAO.2006.17.3.579(A).
