Introduction

Climate change and variability on different time-scales have had profound impacts on both the natural environment and, at times, on human societies and their welfare (Ljungqvist et al. 2021; Ljungqvist et al. 2024; Pecl et al. 2017; Petraglia et al. 2020; Scheffers et al. 2016). Various external forcing agents, as well as unforced internal climate variability, have altered the climate throughout Earth’s history (Degroot et al. 2022; Steffen et al. 2020). In East Asia, extreme short-term climatic events such as droughts, floods, and the abrupt transition from drought to flood conditions have posed substantial risks and challenges to human societies (Zhang et al. 2007; Raymond et al. 2020; Degroot et al. 2021). Since the industrial revolution, and especially since the 1970s, greenhouse gas emissions have triggered considerable changes in global energy balance and, hence, the climate, and the increasing frequency and severity of certain extreme climate events (e.g., heat waves) under the anthropogenic warming have raised widespread concern within the international community (IPCC 2022; Mora et al. 2018). Integrating palaeoclimate reconstructions with climate models and incorporating political, economic, social, and natural environmental factors can facilitate an improved understanding of climate-related threats (Xu et al. 2024; Valjarević et al. 2025). For instance, during the late Ming Dynasty (1573–1644), invasions from the northern regions of the Chinese empire, internal conflicts among ruling groups, and peasant uprisings were influenced by the cold and, in this region, also predominantly dry climatic conditions during the Little Ice Age (Chen et al. 2024b). Despite numerous studies analyzing the interaction between climate change and human history, relatively little attention has been paid to the link between climate change and social instability in southwestern China compared to other regions of East Asia.

Yunnan Province, located in the southwestern border region of China, is well-known for its many ethnic minorities and its cultural diversity, and it has long served as a vital hub for cultural exchanges between East Asia and Southeast Asia (Hansen 2011). Since the Tang Dynasty (618–907 CE), the so-called Tea-Horse Ancient Road has traversed the Yunnan and Sichuan provinces, serving not only as a major commercial trade route but also as a bridge for cultural exchange and ethnic integration (Sigley 2013). Yunnan is also an important component of the contemporary Silk Road, which connected China and Southeast Asia (Wang 2015). Thus, Yunnan has been of major historical importance in both China and Asia as a whole. However, central Yunnan has frequently experienced hydrological disasters such as floods and droughts, which have severely threatened the socioeconomic stability of the region and beyond. Several studies have investigated the driving factors, frequency, trend, and characteristics of droughts and floods in the region (Li et al. 2019; Zhu et al. 2021; Lan and Yan, 2024).

Despite these advancements, the lack of detailed historical records documenting social and economic conditions in connection with winter-spring droughts in central Yunnan has limited our understanding of societal resilience to these disasters in historical times (Ji et al. 2015). Moreover, it can be expected that in the future, climate disasters in this region will intensify, due to factors including increasing urbanization and adverse effects of contemporary anthropogenic climate change (Wang et al. 2021; Dong et al. 2022). Therefore, examining the response and adaptation of social systems to climate-related disasters over centuries is crucial. While past and present societies differ significantly, studying historical responses offers unique insights into societal resilience and susceptibility to environmental stresses (Chen et al. 2023). In recent centuries, flood and drought disasters in central Yunnan have resulted from the combined effects of climatic factors and human activities. Yunnan’s topography and distinctive monsoon climate lead to substantial variations in annual dry and wet season precipitation, increasing the likelihood of both flood and drought events (Li et al. 2019).

Rapid economic development and urbanization in central Yunnan have spurred a large-scale southward population migration, resulting in a regional population surge and strong economic growth. However, this development also presents its challenges. The increased demand for arable land has led to land reclamation, water conservation projects, and flood control measures, alongside the expansion of cultivated land (Duan et al. 2020; Wang et al. 2023). These measures have, at least so far, been relatively successful in promoting sustainable economic development in the region.

In this study, we used climate proxies to understand the climatic context of important historical events to assess the resilience of local societies in the central Yunnan region since the early 19th century. We have applied a combination of spatial analysis and qualitative and quantitative methods in specific aspects of this study (see further section Literature sources). Our aims are to enhance the understanding of the complex relationship between climate change and historical events in the region. We envision that our study will also contribute to an improved knowledge of the societal aspects of past local climate disasters and their implications for the national governance in the region, offering references and potential insights for developing a sustainable socioeconomic environment in this region.

Materials and methods

Study area

The Yunnan–Guizhou Plateau is located in southwestern China. This region has a complex, rolling terrain, primarily consisting of mountains, hills, and river valleys, and is averaging 1800–2500 meters in elevation, is a typical low-latitude plateau monsoon climate zone (Fig. 1a). The climate is warm and humid, with the South Asian Summer Monsoon bringing abundant seasonal rainfall, while winters are relatively dry. According to the climate data from Luquan meteorological station from 1960 to 2022, the annual average temperature is 11.8 °C (1960–2022), with the warmest month being July (19.7 °C, 1960–2022) and the coldest month being January (7.3 °C, 1960–2022) (Fig. 1b). The annual average precipitation is ~950 mm, with the majority falling in summer (June–August), accounting for ~59% of the total annual precipitation (Fig. 1c).

Fig. 1
Fig. 1
Full size image

Distribution of tree-ring sampling sites and the pathway of the Du Wenxiu Uprising (1856–1873) during the Qing Dynasty (a). Trends in temperature (b), and precipitation (c) during the instrumental period 1960–2022.

Literature sources

The primary sources used in this study include historical compilations and records from the Qing Dynasty, including Daoguang Yunnan Gazetteer Draft (Daoguang Yunnan Tongzhigao 道光云南通志稿, Wang and Li 1835), Continuation of Yunnan Gazetteer Draft (Xu Yunnan Tongzhigao 续云南通志稿, Wang and Tang 1901), The Newly Compiled Yunnan Gazetteer (Xinzuan Yunnan tongzhi 新纂云南通志, Zhou et al. 2007), as well as prefectural, state, and county gazetteers related to the study area, and Chinese Meteorological Disasters Ceremony (Yunnan Volume; Liu et al. 2006). These sources document various social issues in Yunnan during the Qing Dynasty, including droughts and their societal impacts. While these sources contain overlapping content, careful screening allowed us to compile a comprehensive record using sources from the official documents. The causes and development of the Du Wenxiu Uprising (杜文秀起义, 1856–1873) are also described in these documentary sources. While various textual sources mention reasons that triggered the Du Wenxiu Uprising, the related evidence may not, unfortunately, be traceable to specific dates. As documented in the literature, the reasons behind the uprising include multiple internal and external pressures taking place during the late Qing Dynasty (Orlandi et al. 2023). In particular, establishing a comprehensive and systematic causal connection between climate change and societal responses is very challenging (see further Ljungqvist et al. 2021). Given that some documents contain subjective interpretations of the causes of the Du Wenxiu Uprising, we avoid trying to establish causal relationships between climate and social change. Instead, we rely on temporal correlations to provide a climatic background and provide more diverse explanations for the causes of historical social changes.

Tree-ring data

On August 14, 2023, tree-ring cores were collected from the sampling site (ZZC, 25.71° N, 102.40° E) in Luquan, Kunming City, Yunnan Province, China. The tree species sampled was the moisture-sensitive Evelyn keteleeria (Keteleeria fortune). Two cores were extracted from each tree using an increment borer, and the core samples were then fixed in wooden trays. After basic tree-ring processing procedures, including air-drying and polishing with sandpaper, the cores were visualized using an EPSON scanner. The widths of each tree ring were measured to the nearest 0.01 mm using the CDendro 9.4 analysis system (Cybis Elektronik and Data AB). The COFECHA program (Holmes 1983) was employed to evaluate the quality of cross-dating, and the ARSTAN program was used to build the site chronology (Cook 1985). Through comparative analysis and fitting negative exponential functions to remove tree growth trends, a standard chronology was generated using double-weighted robust means. The reliable period for the chronology was determined to cover 1814–2022, based on an expressed population signal (EPS) > 0.85 and tree count > 6 (Cook 1985).

Meteorological data from the Luquan station (25.79° N, 102.46° E) were obtained from the China Meteorological Data Service (http://data.cma.cn/), including monthly average temperature and precipitation data from 1960 to 2022. The location of the meteorological station has been unchanged during the observation period, and the data are deemed reliable. Due to the lack of direct observational records for Normalized Difference Vegetation Index (NDVI) data (Pinzon et al. 2014), we obtained monthly mean NDVI and self-calibrating Palmer Drought Severity Index (scPDSI) data from the Climate Research Unit (CRU), at a resolution of 0.5° × 0.5° from 1960 to 2022 (van der Schrier et al. 2013). The spatial correlation between the meteorological data and the standard tree-ring width chronology was analyzed at the CRU, using Pearson correlation coefficients (r) to assess correlations within different monthly and seasonal windows. Since tree growth in the region is also influenced by the previous year’s climate, the monthly window was expanded from July of the previous year to October of the current year. A linear regression model was then used to establish the relationship between the standard chronology and precipitation.

Calibration and validation statistics were applied to the reconstruction model to assess its performance and differences, with evaluation metrics including Pearson correlation coefficients (R), explained variance (R2), reduction of error (RE), product mean test (PMT), sign test (ST), and relative root mean square error (RMSE). The relative RMSE was calculated as RMSE divided by the mean value of precipitation over the observation period. To display low-frequency precipitation variability, an 11-year low-pass Gaussian filter was utilized. We defined an extremely dry year as having rainfall lower than the mean − σ, and an extremely wet year as having precipitation higher than the mean + σ, while wet/dry periods were defined as continuous 10-year periods of above/below average precipitation changes in the 11-year low-pass filtered values. The joint probability method of copula was used to estimate the return time by assessing the joint probability of the duration and severity of drought (Nelsen 1999). Superimposed epoch analysis (SEA; Rao et al. 2019) was used to explore the association between extreme low precipitation and outbreaks of specific plague events. The method of significance testing (random sampling) was used to assess significance in the presence of autocorrelation. The SEA is performed using the precipitation reconstruction by the average of the pre-event years by subtracting, from each event window precipitation series, the mean of the values for the years before the event corresponding to that window.

Results

Precipitation reconstruction

The standard chronology (ZZC) shows a negative correlation with the period from the previous July to the current January, with significant correlations (p < 0.05) observed only in the months of September and October (Fig. 2a). The correlation in May of the current year had a negative impact on tree growth, while the relationship between summer temperatures and tree growth was not significant. Precipitation from the previous October to the current June promoted tree growth, but this positive effect was offset from July to September (Fig. 2a). Specifically, the highest correlation between precipitation and the chronology is occurred in April and May (p < 0.05), confirming the significant correlation with the scPDSI for the moisture-sensitive Chinese fir. The sampling sites are located in high-altitude plains with shallow soils with a poor water retention capacity. The potential evapotranspiration is much higher than the actual evaporation, and tree growth has a higher demand for moisture than the favorable conditions provided by the temperature regime (Jung et al. 2010; King et al. 2015). Spring is a crucial period for the earlywood formation of coniferous trees, and a warm and humid environment has a stronger impact on tree growth (Zhao et al. 2023, 2024 Zhirnova et al. 2021).

Fig. 2
Fig. 2
Full size image

Correlation analysis of the tree ring width standard chronology (SC) and instrumental climate factors (1960–2022) (a). Comparison of precipitation and reconstructed results for the instrumental period (b). Reconstruction of precipitation changes on the central Yunnan–Guizhou Plateau over the period 1814–2022, with the dark red shaded zones denoting persistent drought prior to the Du Wenxiu Uprising (1856–1873), and red triangles indicating the 10 driest years (c). Runs analysis and estimated return time of droughts (d), and wet periods (e). Contours of the return time are derived from copula functions for the joint probability of the severity and duration of wet and drought events.

The correlation between tree growth and moisture decreases in summer, possibly due to abundant rainfall during the rainy season, which increases the soil moisture content and alleviates the effects of anomalous drought caused by high temperatures (Littell et al. 2016; Reich et al. 2018). Furthermore, we analyzed the relationship between climatic factors and tree growth using different seasonal combinations, with seasonal combinations being more representative of regional climate changes (Ruane et al. 2017; Zhao et al. 2023). The mean temperature in the study area is above 0 °C throughout the entire year, with the period from the previous December to the current May considered the dry season. During this period, there was a significant positive correlation between precipitation and tree growth (r = 0.65, p < 0.01). Compared to changes in moisture during the rain season, precipitation during the dry season better reflects the sensitivity of tree growth to hydroclimate conditions.

Based on the sensitivity of tree growth to moisture during specific seasons, we reconstructed the precipitation variations from the previous December to the current May using the standard chronology (Pre = 172.911 × ZZC – 11.804). The linear regression model explained ~42% of the variance in the observed precipitation records (Fig. 2b). During the calibration and validation process, the first-order difference correlation coefficient and sign test were 0.71 (p < 0.01) and 43 + /19 − (p < 0.01), respectively, indicating that the model can capture both high-frequency and low-frequency precipitation signals well. The RE and PMT values were 0.42 and 7.03, respectively, passing the significance test (p < 0.05) and indicating that the model is relatively dependable and accurate. The relative RMSE is 0.30, indicating that minor differences between the observed data and the reconstructed precipitation. Overall, the good performance of the model means that the reconstruction can adequately represent the observed data for the exploration of precipitation changes over the period 1814–2022.

Characteristics of precipitation changes

The new central Yunnan precipitation reconstruction spans the period 1814–2022. The average reconstructed precipitation (mean) from the previous December to the current May in the study area was ~153 mm and the standard deviation (σ) was ~42 mm (Fig. 2c). In precipitation reconstruction, 34 years were defined as “extremely dry” and 34 years as “extremely wet”. Throughout the 209 years 1814–2022, 10 dry events occurred (1827–1832, 1849–1855, 1865–1871, 1876–1880, 1917–1923, 1930–1936, 1965–1972, 1991–1996, 1975–1979, and 2010–2014) with precipitation below the average precipitation for at least five consecutive years (Fig. 2 d); and four consecutive winter-spring extreme droughts occurred (1850–1855, 1876–1878, 1917–1919, and 1921–1923) lasting for more than three years. It is noteworthy that during the dry period of 1846–1856, a prolonged drought in central Yunnan coincided with the onset of the Du Wenxiu Uprising, with the reconstructed precipitation (~123 mm) being below the long-term average (~–20%). Although a wet period occurring 1857–1864 coincides in time with the peak of the Du Wenxiu Uprising, drought conditions resumed again in 1865 and then lasted until the 1880s.

The reconstructed precipitation, and measured precipitation during the instrumental period 1960–2022, spatially represent moisture and hydrological changes on the southeastern Tibetan Plateau and Yunnan–Guizhou Plateau (Fig. 3a). In addition, evaluating the relationship between changes in vegetation productivity in the Asian monsoon region and climatic factors with human history may provide valuable additional information (Fig. 3c, d). The cumulative distribution function of the long-term reconstructed precipitation data closely matches the probability inferred from the instrumental period mean (54%). However, during the decade preceding the Du Wenxiu Uprising, the probability of precipitation not reaching the long-term average increased to as much as 80%. Although this probability decreased to ~44% during the actual Du Wenxiu uprising, if future climate conditions resemble those before the Du Wenxiu Uprising, eight out of ten years would not meet the normal requirements for vegetation growth and demand for water resources (Fig. 3b). Independent hydroclimatic records from the Chayu area, the Monsoon Asia Drought Atlas, the northwestern Yunnan area, and the central Hengduan Mountains all reflect abnormally dry climatic conditions in the decade prior to the Du Wenxiu Uprising (1856–1873) (Fig. 4). In addition, they agree with our new reconstruction on the existence of dry periods in the 1870s, 1920s, and 1980s and wet periods in the 1860s, 1930s, and early 21st century (Fig. 4). Moreover, the independent hydroclimatic records are significantly and positively correlated with our reconstructed precipitation (p < 0.05), enhancing the reliability of our assessment of the link between climatic anomalies and historical events (Fig. 4).

Fig. 3
Fig. 3
Full size image

a Spatial correlation between reconstructed and observed precipitation over the instrumental period (1960–2022). b Distribution of extreme low/high (±1 SD) precipitation years and drought/wet periods. c Spatial correlation between reconstructed and observed runoff over the instrumental period (1960–2020). d Spatial correlation between reconstructed and observed NDVI over the instrumental period (1981–2019).

Fig. 4
Fig. 4
Full size image

Comparison between (a) reconstructed precipitation in this study (b), April–May scPDSI reconstruction in the Chayu area (Zhao et al. 2021), c June–August scPDSI reconstruction from the Monsoon Asia Drought Atlas (Cook et al. 2010), d November–February precipitation reconstruction in the northwestern Yunnan area (Keyimu et al. 2021), and e April–May relative humidity reconstruction in the central Hengduan Mountains (Cui et al. 2024). All reconstructions were standardized. Red and blue shaded zones indicate common dry and wet periods, respectively.

Drought-inducing ocean–atmosphere circulation mechanisms

During the instrumental period, and for the entire time-span of the reconstruction, changes in precipitation were found to be strongly influenced by global sea surface temperatures (SSTs), in particular to the SST in the North Pacific and Indian Oceans (Fig. 5a). This indicates that changes in ocean circulation and atmospheric pressure modify monsoon intensity (Fig. 5b). The water vapor fluxes in drought and wet years demonstrate the moderating role of the ocean–atmosphere coupled system (Fig. 5c, d). A positive Indian Ocean Dipole (IOD) phase promotes a strong Southwest Monsoon, with anomalous Indian Ocean SST increasing the land–sea thermal contrast (Dong et al. 2023; Fig. 5e). This enhanced South Asian Summer Monsoon (SASM) brings abundant precipitation to the study area, while a weak IOD phase with reduced water vapor and southward movement of the western Pacific subtropical high pressure, instead leads to regional drought development (Buckley et al. 2014). During the positive phase of the overall PDO, a weakened Walker circulation and anomalous westward water vapor transport in the central equatorial Pacific result in drought conditions in southern China, with the rain belt shifting southward (Tan et al. 2017; Liu et al. 2023; Zhang et al. 2022; Fig. 5f).

Fig. 5
Fig. 5
Full size image

Spatial regression distributions of global SST and precipitation for 1870–2022 (a). Spatial regression distributions of 200 hPa geopotential height and precipitation for 1961–2022 (b). Water vapor flux anomalies for the ten wettest years (c), and the ten driest years (d). Eleven-year running correlations of observed IOD (red), ENSO (blue) (e) and AMO (brown), and PDO (green) indices (f) with reconstructed precipitation.

In addition, AMO throughput is a teleconnection that influences water vapor transport and regulates regional moisture changes (Fig. 5f). During the positive phase of the AMO, it triggered negative pressure anomalies over the North Atlantic region and extended through the westerlies into Eurasia, weakening the Siberian high pressure and augmenting the Asian monsoon to replenish regional precipitation (Song et al. 2024; Tang et al. 2023). Atmospheric circulation undoubtedly exerts a crucial influence in influencing precipitation changes in our study area, and our findings provide background information on climate–ocean circulation changes during the period of societal turbulence and uprisings in the Yunnan Province.

Discussion

Possible societal impacts of extreme drought in Yunnan

Climate change at different times may have contributed to the ethnic exchanges and societal development in areas around the Tibetan Plateau (Su et al. 2021; Chen et al. 2022, 2023). Changes of political boundaries during the historical period often coincided with climate change. Previous studies, such as the expansion of the Lijiang Chiefdom supported by the Ming Dynasty, which developed and established new areas and military bases influenced by a warm and humid environment (Liang et al. 2022). The exceptional drought 1849–1855 was characterized by its large spatial extent and intensity (Chen et al. 2022; Gao et al. 2021; Yin et al. 2023) (Fig. 3), and the close relationship (r = 0.44, p < 0.01) of precipitation with NDVI provides background information on the biomass productivity and climatic conditions, particularly in regions where human development relies on the cultivation in the valleys and pasture in the high-altitude grasslands (Fig. 3 c, d). This period of drought coincided with political instability in Yunnan, exacerbated by ethnic cultural exchanges and economic developments, including increased British and French colonial activities and the influx of missionaries, which intensified cultural tensions and societal conflicts (Yang 2020).

Yunnan’s financial burdens were primarily due to military expenses and copper transportation costs, which far exceeded local revenues (Bai 1952). Due to Yunnan’s location in the southwestern border area of China, the Qing government stationed over 40,000 troops in Yunnan, costing more than 730,000 taels of silver annually in military pay, a sum the local economy could not sustain without central government subsidies. “兵额四万七百余名, 奉饷等银七十三万六千余两”, “本省地丁等银, 所入不敷, 豫先咨部请拨, 或四五十万两, 或六七十万两不等, 此所以养云南之兵也。” (奕诉.钦定平定云南回匪方略. 铅印本, 1896, 卷13, 第365页). Historically, Yunnan has been a region rich in mineral resources, attracting merchants from various regions who engaged in mining activities (Yang, 2014). However, the transportation of copper from Yunnan to other provinces required an annual cost of over 800,000 taels of silver, which needed to be allocated by the Qing government from other provinces. “户部按年指拨各省解滇, 贫民借以谋生, 此所以养云南之民也。” (奕诉.钦定平定云南回匪方略. 铅印本, 1896, 卷13, 第365页). The Ministry of Revenue allocated funds annually to each province to financially relieve Yunnan, so that the poor could make a living and support the people of Yunnan. Following the Taiping Uprising (1850–1864) in southern China, the military expenditures by the Qing Dynasty increased substantially, contributing to the economic downturns in Yunnan due to inadequate local military funding (Lee, 2004). The Qing government increased economic extraction, imposing over 404,200 taels of silver annually in taxes on land and resources (Zhou et al. 2007:《新纂云南通志》卷一百五十《财政考一·岁人一》, 1948年, 铅印本), which became an unbearable burden for the local population. More importantly, the escalating ethnic conflicts stemming from the competition for land and mineral resources led to oppression by the Qing Dynasty and landlord class (constituting of a small elite, gentry and some retired officials) to protect their own interests, exacerbating pre-existing tensions among various ethnic groups (Jing 1991; Lin 2006), and a large number of armed conflicts and casualties among the civilian population, during this period fueled strong public discontent (Notar 2001; Atwill 2003).

In addition, the rugged terrain and low soil fertility of this region adversely affected an economy that largely depended on agriculture (Yang et al. 2022). Heavy taxation by the Qing Dynasty, levied in silver, and a harsh land and business levies regime inflicted by local governments, imposed a heavy burden on the population of Yunnan (Hu 2023; Wang 1973). Agriculture and livestock production are closely linked to the prevailing climate conditions, and self-sufficiency in food and materials was the foundation of regional economic and social stability (Clift et al. 2021; Du et al. 2021; Liang et al. 2023; Simelton 2011). Our new tree-ring evidence reveals the occurrence of a prolonged drought in Yunnan just prior to the onset of the Du Wenxiu Uprising. Precipitation 1846–1856 was far below (around –20%) the long-term average, including the dry event 1849–1855, with an estimated return time of more than 100 years, with historical literature also indicating that ~73% of the years during this period experienced severe drought. For instance, according to historical records, there were two consecutive years of poor harvests from 1848 to 1849 in Kunming, resulting in expensive rice prices, a shortage of food, and limited rice production in the surrounding areas. “昆明于道光二十八年（1848年）有记载曰：“去冬既少雪，今春又少雨。宿麦满平野，芃芃那可睹。市米价日增，穷檐多疾苦。”因夏旱，“已是兼旬鲜米到。遂使米价日日增，何惜倾囊问斗升。时还几处无米卖，街头奔走贫人惫。”([清]戴淳,晚翠轩诗钞. 昆明: 云南教育出版社, 2019, 第328页；第332页). During 1849–1850, crop failures occurred in cultivated areas like Yaoan, Xuanwei, Songming, Jianshui. “1849年, 正月至六月下旬, 姚安不雨, 大旱。1850年, 春, 宣威雨少旱灾, 禾苗高低不齐; 五、六月, 昆明、嵩明、江川、河西、建水天旱不雨, 华宁半年天苦旱, 禾苗难栽, 河西概种玉黍、荞、豆等类, 收农家苦, 岁大饥。” (Liu and Wen, 2006). During 1855–1856, due to drought, grain prices rose substantially in central and northern Yunnan. 哀牢山区有记载曰：“乙卯（1855年）天旱, 五谷不收, 庄主逼租, 皇差索粮。”“久旱无雨, 杂粮不获。夷众皆闭门拒赋, 往山野掘蕨根制粉为食。” (李文海等. 近代中国灾荒纪年. 长沙: 湖南教育出版社, 1990, 第166页). “1856年四月, 宜良大旱; 夏, 昆明、晋宁、昆阳、曲靖、武定、元谋、禄劝大旱, 改种荞, 歉。”(Liu and Wen, 2006). The food shortages severely impacted the mining industry, exacerbating ethnic tensions and existing conflicts. While climatic conditions and biomass productivity were not the sole determinants, they created a favorable environment for the Du Wenxiu Uprising amidst pre-existing social and ethnic tensions. Amidst the exacerbation of economic and political problems and ethnic conflicts during the late Qing Dynasty, in 1856, the population of Yunnan revolted against imperial hegemony and bureaucratic oppression (Lin et al. 1991; Giersch 2006). The rebel forces established strongholds in the eastern, western, and southern parts of Yunnan, prompting the Qing Dynasty to deploy troops to suppress the rebellion (清实录·文宗实录, 卷205, 第4-5页; 卷208第9-10页). During the peak of the Du Wenxiu Uprising from 1857 to 1865, central Yunnan experienced a relatively wet period, which then rapidly shifted to drought conditions, reoccurring in 1866. The wet climate during the actual rebellion provided favorable conditions for the rise of the rebel forces, increasing the availability of food and resources (Liang et al. 2022). Although some areas experienced local drought during this peak period, the overall moist environment laid the foundation for the rebel forces. Due to Yunnan’s border location (Fig. 1a), its unique political status led to intensive suppression by the Qing Dynasty in an attempt to maintain regional stability (Ma, 2014). Therefore, the shifts between wet and dry environments driven by climate change were one of several potential factors influencing the Du Wenxiu Uprising, especially in the border regions dependent on arable agriculture and animal husbandry.

Yunnan experienced frequent extreme droughts from the late Qing Dynasty into the early Republic of China (formally established on 1 January 1912), with the most obvious being the so-called Dingwu Great Famine (1877–1878, during the so-called late Victorian Great Drought) and the new dry period, 1917–1923. Proxy data from the surrounding region recorded the extent and severity of these two extreme drought events (Cook et al. 2010; Shi et al. 2017; Zhao et al. 2021). The frequency of the failure to achieve the normal yearly precipitation during the two extreme droughts exceeded ~67%, implying that the impacts of successive inter-annual droughts were substantial (Table 1). In addition, frequent locust plagues and earthquakes occurring at the same time as the extremely dry conditions prevailed had significant compound effects on the natural environment and societal stability in this region (Liu et al. 2006).

Table 1 Historical records of plague outbreaks and extreme drought events range from 1814 to 1900 in Yunnan Province.
Full size table

The Dingwu Great Famine not only caused over an estimated 10 million deaths in northern China, but it also had serious impacts in the low-latitude Asian regions (Hao et al. 2021; Ren et al. 2021). This extreme drought event during 1876–1880, with the estimated return time of ~50 years, also affected the Yunnan region, causing serious economic losses and rising food prices (Edgerton-Tarpley et al. 2017). During this period, low precipitation causing extreme drought in Yunnan was the worst during the 200-year study period, while it was accompanied by intense conflicts between warlords in the context of the unstable political situation in China during the early years of the Republic of China (Phillips, 1996). A large proportion of local fiscal expenditure was used for military purposes, rendering it impossible for local governments to provide sufficient disaster relief-resulting in a deterioration of the living standards among the Yunnan population (Bello et al. 2021).

Moreover, the unstable economic environment adversely affected regional ethnic and cultural integration and triggered social unrest, weakening the national government’s ability to respond to disasters and its own stability (Han et al. 2017). The unusually extreme drought triggered large-scale population movements, which not only increased the economic burden on other regions, but also made the already fragile social structure even more unstable (Zhang et al. 2019). Even though the national government undertook various mitigation measures, including tax reductions and resettlement schemes, they have had only a limited effect and have made it difficult to address the primary causes of the problem (Lin et al. 2020).

Possible relationship between plague transmission and climate change in Yunnan

Historically, Yunnan has been a major source area for plague (Yersinia pestis) in the East Asian region. The spread of plague in western Yunnan has been influenced by the region’s unique geographical conditions and multicultural society (Atwill 2023, 2005; Li and Cao 2003). Commercial exchanges and the movement of troops and refugees during the war also promoted the spread of plague (Benedict, 1996). However, the potential role of climate change and variability in the spread of plague in Yunnan is poorly understood. Plague outbreaks in Yunnan reached a peak starting from the Du Wenxiu Uprising (Table 1). The frequent wars in the 19th century, coupled with plague, led to the devastation of the Yunnan economy and a drastic reduction of its population (Atwill 2005; Athni et al. 2021;). The SEA results indicated a positive effect on the plague outbreak during three and two years of severe drought (Fig. 6).

Fig. 6
Fig. 6
Full size image

Possible mechanisms of the impact of extreme drought on socioeconomic and ecological development in Yunnan during the late Qing Dynasty. Superposed epoch analysis (SEA) indicates the relationship between plague outbreaks and extreme drought years. The cumulative distribution function indicates the probability that precipitation does not satisfy the long-term mean during the Du Wenxiu Uprising and the Dingwu Great Famine (1876–1879), and the distribution of plague outbreaks recorded in historical documentary data in Yunnan during the 1814–1900 period.

This implies that the regional ecosystem was damaged by drought, and the drastic reduction in vegetation productivity and lack of moisture prevented rodents from obtaining sufficient food and triggered migration, which helped spread the plague. This perspective was confirmed in previous Eurasian plague studies, where fragile ecology and poor climate forced rodents to migrate and bacteria to spread (Büntgen et al. 2013; Slavin et al. 2023; Schmid et al. 2015; Spyrou et al. 2022). The relationship between drought and plague has also been recorded in many relevant historical records. For example, in 1808, there was a severe drought in Kunming, Songming, Shunning, Dayao, and the surrounding areas, followed by a plague in 1809. In 1817, Ding Chou,; Kunming, Songming, Shunning, and Dayao are experiencing hunger and epidemics (Wang and Tang, 1901:《续云南通志稿》卷二《祥异》, 光绪二十七年四川岳池县刻本).

Between 1856 and 1872 numerous deaths occurred in Yunnan due to plague, resulting in a 70% decline of Yunnan’s population, a total of over 1.5 million people, with the remaining 30% mainly due to conflict and other causes, and the most seriously affected areas included Chuxiong, Dali, Menghua, Puer, with the numbers of deaths ranging from 56,000 to 597,000 (Li and Cao 2001, 2003). Especially between 1871 and 1872, the plague outbreaks in the Yunnan Province were the most extensive (Table 1). The results of the cumulative distribution function showed that the probability of precipitation not reaching the long-term average was ~73% in the 1870s, while the probability of drought prior to the Du Wenxiu uprising reached 80 % (Fig. 6). A frequently strong historical relationship between wars and epidemics have been noted by numerous scholars (Cooter, 2003; Herring et al. 2020; Parker, 2013). Wars increased the demand for food at the same time it disrupted normal agricultural activity, leading to an increase in food shortages, and potentially famines and refugees, and causing the human population to attempt to exploit natural food resources, while increasing the spread of plague and other diseases (Li and Cao, 2001; Chen et al. 2022).

Moreover, when it comes to plague, the destruction of the people’s living environment led to the deterioration of public health and living conditions, which had an important impact on the production of plague bacteria (Y. pestis) (McNeill 2010). There were some cultures of religious worship during the epidemic, in response to a belief in its supernatural origin and the view that it represented divine punishment (Li and Gu 2012). Hence, there was a tendency to resort to attitudes and responses that proved completely ineffective in combating the plague. However, increased understanding of plague causes led to improvements in regional epidemic prevention measures, including changes in economic activities, lifestyles, animal rearing practices, and waste disposal (Li 2022). With the founding of the Republic of China in 1912, plague still existed, but the frequency of large-scale plague epidemics had significantly decreased (Xu et al. 2011). Therefore, the decline in vegetation productivity caused by drought was not necessarily the principal agent of the plague outbreaks. Indeed, it is likely that the late Qing Dynasty Yunnan plague epidemics were the result of a combination of factors.

Natural disasters induced by climate change not only had profound socioeconomic impacts on Yunnan, but they also resulted in increased communication and integration among the region’s many ethnic groups (Liang et al. 2022). Yunnan’s hydrological conditions are influenced mainly by the Tibetan Plateau, with numerous rivers flowing through it to the Pacific Ocean. Drought-induced reduction in water resources affects the entire Southeast Asian region, affecting political and social stability that may then amplify the drought effects (Chen et al. 2023; Yong et al. 2024). If temperatures continue to rise under future climate change scenarios, droughts could pose significant challenges to ecological and economic development in the region. Therefore, it is important to understand the interactions between the natural environment and the economy and society, and to develop strategies for mitigating future drought-induced risks.

Conclusions

We reconstructed precipitation changes in the central Yunnan–Guizhou Plateau over the 1814–2022 period using tree-ring width data, and provided a climatic background for certain social events and changes during the 19th century. Firstly, we clarified the severity and duration of some extreme drought events and attempted to explore possible socio-economic responses to drought in Yunnan during the 19th century. Particular emphasis was placed on the prolonged drought in the mid-19th century that created a favorable context for the outbreak of the Du Wenxiu uprising. Drought had a negative impact on vegetation productivity, and the lower vegetation productivity led to food shortages, rising food prices, and exacerbated social unrest. Secondly, we also established the relationship between climate and plague outbreaks, and found that drought conditions promoted plague epidemics, especially a significant increase in plague outbreaks after the onset of severe drought. Frequent wars indirectly promoted population mobility, increasing the possibility of epidemic transmission. Finally, it should be noted that this study does not aim to establish causal relationships between climate change and any social changes. The factors influencing historical societal changes are complex and include, in this context, fiscal taxation, military affairs, and a series of other factors. Therefore, the impacts of extreme drought on Yunnan were diverse and interacting, and drought-induced socioeconomic problems presented challenges to cultural exchanges and the relationship between ethnic groups.