How global change impacted the rise and fall of the Guge Kingdom

Climate changes and its influences on human society are of increasing concern in science communities. Based on the reconstructed climate data and CENTURY model, we simulated net primary productivity of grassland and the grain yield of highland barley during the pre-industrial millennium in Ali Prefecture, Tibet Autonomous Region. It showed that the variation of precipitation and temperature together affected the fluctuation of land productivity. Wavelet analysis results showed that the land productivity in Ali fluctuated within main periodic bands of 180 yr. We found that the rise and falls of the Guge Kingdom in Ali was synchronic with the fluctuation of land productivity, and the collapse of the Guge Kingdom was obviously related to the sudden change of climate in the 17th century, which turned dry and cold. By combining with historical studies, this study further proposes the mechanism of land productivity fluctuation under climate change on Guge Kingdom. Our findings are helpful to understand the relationship between climate change and social vulnerability, especially providing a typical case study of ancient plateau countries.


Introduction
The impact of climate change on human society has a long and complex history, and its action mechanism varies with the change of space and time, which is closely related to social vulnerability (Zheng et al 2014). More and more studies show that climate has a significant impact on the development of civilizations and ancient countries in the pre-industrial era (Haug et al 2003, Hsiang et al 2011, 2013Chambru 2020, Hao et al 2020, Felzer et al 2020, Degroot et al 2021, Han and Yang 2021. Analysis on the impact of climate change on historical development will favor to understand the relationship between human society and natural environment, and help human beings adapt to future natural environment changes and achieve sustainable development. At present, there have been a large number of studies on the impact of climate change on social development in the pre-industrial era. These studies suggested links between climate change and social conflict, civilisational decline, and population (Zhang et al 2010, Wei et al 2015, Nooren et al 2018, Liu and Yan 2020, Ljungqvist et al 2021, Kennett et al 2022. Because of the sensitivity of plateau ecosystem to climate change (Yu et al 2010), plateau civilizations might be more significantly affected by climate change. For example, the Mayan civilization withstood the environmental change better in lowland than highland (Lee and Zhang 2013). Pollen evidence from lake sediments suggested that periodic climatic changes may have contributed to the rise and fall of the Tubo civilization in the southern Tibetan Plateau (Li et al 2021).
Guge Kingdom was an ancient country on the Tibetan Plateau from the 10th century to the 17th century. It was roughly located in today's Ali Prefecture, and its core area was in Zanda County. In recent years, paleoclimatologists found that there may be a relationship between climate change and the collapse of the Guge Kingdom (Sinha et al 2015, Li et al 2019, Liang et al 2022. Stalagmites in northern India was pointed out that the Indian monsoon weakened in the first half of the 17th century, which brought serious drought to the western Qinghai Tibet Plateau, and the researchers speculated that drought was an important reason for the demise of Guge Kingdom (Sinha et al 2015). The evidence of multi-proxy indicators of sediment in Xiada Co in the west of Tibetan Plateau showed that the runoff in the catchment area was low during 320 yr BP and intense evaporation process occurred, so it was speculated that drought triggered war and conflict and led to the collapse of Guge Kingdom (Li et al 2019). A lake sediment suggested that a decline in temperature led to a decreased barley's crop yield that may have factored into the disappearance of the Guge Kingdom. (Liang et al 2022). However, these studies only made assumptions about the relationship between climate change and the rise and fall of the Guge Kingdom, and had not analyzed how climate changes affected social mechanisms and eventually promoted or led to the collapse of the Guge Kingdom.
Based on the historical climate reconstruction data, we used CENTURY model to reconstruct the change process of grassland net primary productivity (NPP) and highland barley yield in Ali from 850 to 1849, so as to analyze how climate change affected the vegetation dynamics in Ali, and then affected the society of Guge Kingdom and promoted the development of historical process. By referring to historical documents neglected by previous researchers and combining natural evidence with historical records, we outlined the impact of climate change on the ancient civilizations. This will provide a historical case to help researchers better understand the relationship among climate change, ecosystem and human history, and improve their understanding of the nature of climate vegetation society interaction and the vulnerability of plateau society under climate change.

Study area and Guge Kingdom
Our study area was located in Ali Prefecture, Tibet Autonomous Region of China, which was the main ruling area of Guge Kingdom in historical period (figure 1). Ali is located in the northwest of the Tibetan Plateau and is known as the 'roof of the roof ' . From north to south, there are mainly three basin areas, including Rutog, Gar and Zanda-Purang, which were called 'Ali Sanwei' in historical period. The Shiquanhe river flows through Rutog and Gar regions in the north, with an average altitude of more than 4000 m. Rutog and Gar regions are mainly alpine grassland and alpine desert grassland vegetation types, with an average annual temperature of 0.2 • C, and the annual precipitation is about 70 mm.
The Xiangquanhe river flows through Zanda and Purang regions in the south, with an average altitude of more than 3900 m. Zanda and Purang regions are mainly alpine grasslands, with an average annual temperature of 3.1 • C, and the annual precipitation is about 160 mm. Ali is a semi-grassland and semi-farmland region. Agriculture is mainly concentrated in the southern Zanda and Purang river valleys, where mainly plant highland barley and other crops.
The Guge Kingdom was founded in the 10th century by Skyid lde nyima mgon, a descendant of Zanpu who were the kings of Tubo Kingdom. After more than a hundred years, the Guge Kingdom developed rapidly and reached its peak in the 11th century. During the reign of Yuan Dynasty, Ali was once controlled by the emperor of Yuan Dynasty. At the end of the 14th century, the Guge Kingdom was restored again. Since the late 16th century, the Guge Kingdom gradually declined and was conquered by the Ladakh Kingdom in the 1630's. The rise and fall of the Guge Kingdom left many questions about the history of Tibet. According to the textual research of Tibetan history researchers, the Guge Kingdom destroyed the country during the period of Mongolian rule in the 13th-14th century, but then restored the country again (Huang 2016a). After 1637, Guge was completely destroyed, and its population dissipated in a short time. In addition, the destruction of the Guge Kingdom almost coincided with the Ming Dynasty: more than ten years after the fall of Tsaparang, the capital of the Guge Kingdom, the Ming Dynasty also perished in internal and external difficulties (Lee and Zhang 2013). This period happened to be the '17th century crisis' period concerned by researchers of history and paleoclimatology (Zhang et al 2011).

Data analysis 2.2.1. CENTURY model
We used CENTURY_47 model to estimate the carbon cycle change process in the study area in the historical period of 850 AD-1850. The CENTURY model was originally a comprehensive model established from the study of C (carbon), N (nitrogen) and other element cycles in grassland ecosystem. It is one of the widely used dynamic models of soil organic matter (Parton et al 1987). The model adopts the theory that soil organic matter is divided into three reservoirs: active, slow and passive. By simulating the accumulation or loss process of soil organic matter in each reservoir, the dynamic simulation of soil C, N and other elements is realized. The process based on CENTURY model has been verified and widely used in various grassland and farmland ecosystems. Some scholars had effectively explained the process and mechanism of climate change affecting grassland/farmland ecosystem by using the simulation results of the model (Parton et al 1987(Parton et al , 1988  Cong et al 2014, Baldassini and Paruelo 2020). More information about the CENTURY model could be found at www.nrel.colostate.edu/projects/century/ index.php.

Climate data
The input climate data of CENTURY model are monthly precipitation, maximum monthly temperature and minimum monthly temperature. The paleoclimate reconstruction data in Ali mainly came from lake sediments with low accuracy, which was not suitable to be used as the input data of the model. Scholars reconstructed the annual resolution climate data of the Tibetan Plateau in the past 22 000 yr based on multi-generation data and climate models (Shi et al 2021). The accuracy and reliability of the data had been verified, so they were used as our model input climate data. Since the reconstructed climate data was the average anomaly in summer and the result of the whole Tibetan Plateau, the data needed to be processed according to the conditions of the study area. Since the raw data is the average temperature for the summer (May-July), it needs to be converted into standard CENTURY model input data. We calculated the ratio of summer temperature to monthly mean temperature according to the measured climate data of modern Purang meteorological station from 1974 to 2013, and converted the original data into monthly mean temperature data. Then, the ratio of monthly maximum and minimum values to the average temperature was calculated respectively. Finally, the monthly maximum and minimum temperature data were calculated from the annual average temperature data. At the same time, because the reconstructed climate data was smoothed, the fluctuation trend was less than the actual situation, so the change process needed to be amplified according to the actual fluctuation range. The processing method of annual precipitation data was similar. Finally, the climate data from 850 to 1849 were obtained.

Model calibration
In addition to the monthly scale climate data, the inputs of CENTURY model include information on the biophysical (soil physical and chemical properties, vegetation characteristics) and management (land use, crop sequence, tillage, fertilizer, stocking rate, combustion frequency and intensity) characteristics of the sample site. Vegetation features were from Grassland Resources of Tibet Autonomous Region (Land Administration Bureau Of Tibet Autonomous Region, 1994). Over the study period, the land use Therefore we designed four land use modes for simulation, namely natural grassland, low-intensity grazing grassland, irrigated highland barley farmland and non-irrigated highland barley farmland. Therefore, this study used modern parameters to deduce the change process of sample ecosystem and yield in the historical period.
The CENTURY model divides the soil texture according to the content of sand, silt and clay. The parameter information we used was from the Harmonized World Soil Database version 1 (Fischer 2008). The wilting point, soil bulk density, saturated hydraulic conductivity and field water capacity were obtained by the soil calculator provided by the CEN-TURY model. The initial values of vegetation parameters provided by the model were used, and several trial runs were carried out to obtain the optimal values of vegetation parameters (table S1 of the electronic supplementary material, ESM). Existing studies and our simulation experiments showed that the Century model can better simulate the process of vegetation carbon cycle in the Tibetan Plateau (Zhang et al 2007, Chang et al 2014 (figure 2). The model simulation was divided into two stages: (a) before the soil organic carbon is stable, the model was preheated for 5000 years using the average climate data to make the simulated ecosystem reach equilibrium; (b) the model was formally run using real climate data, and the sample plot was simulated by grassland/farmland ecosystem.

Wavelet analysis
We used wavelet analysis to explore the correlation and periodic resonance characteristics of grassland and farmland productivity. Wavelet analysis is a non-stationary time series analysis tool developed in recent decades, and has been widely used in climate change and other fields (Grinsted et al 2004, Rouyer et al 2008, Baddoo et al 2015. Wavelet analysis can perform two-dimensional local transformation in both time domain and frequency domain simultaneously to track the time change characteristics of non-stationary periods (Torrence and Compo 1998). This study mainly used continuous wavelet transform to analyze the periodicity of time series. The real part of the wavelet coefficient reflects the distribution of high-value period and low-value period on different time scales. The larger the wavelet spectrum value is, the more common the corresponding wavelet is and the more obvious the periodic signal is. The value of wavelet spectrum can reflect the strength of periodic signal (Torrence and Compo 1998). In order to remove the interference of high frequency signals, we use the 20 yr period moving average low-pass filter sequence for wavelet analysis.

Land productivity reconstruction
Under the action of climate change, the simulation results of grassland and farmland productivity of different land uses had changed significantly during the study period. The NPP simulation value of natural grassland fluctuated from 11.55 g C m −2 -56.47 g C m −2 , the multiyear average value is 25.19 g C m −2 , and the coefficient of variation (CV) was 42%. The NPP simulation value of grazing grassland fluctuated from 9.96 gC m −2 -56.32 g C m −2 , the multi-year average value was 24.06 g C m −2 , and the CV was 47%. The grain yield of irrigated farmland fluctuated from 1.83 gC m −2 -6.96 g C m −2 , with a multiyear average of 3 g C m −2 and the CV of 17%. The grain yield of non-irrigated farmland fluctuated from 0.63 g C m −2 -4.04 g C m −2 , with an annual average of 1.64 g C m −2 and the CV of 22%.
In order to preliminarily verify the reconstruction results, this study compared the pollen data with the reconstruction results. Pollen data were obtained from lake sediment of Angrenjin Co in the southwestern Tibetan Plateau (Li et al 2021). The comparative results showed that the pollen data and reconstructed results were coincident with the low values in ∼1100, ∼1280, ∼1500 and ∼1630 AD and the high values in ∼1020, ∼1180, ∼1380 and ∼1560 AD (figure 3). This preliminarily confirmed the reliability of the reconstruction results.
This study compared reconstructed land productivity results with climate data. The results showed that the effect of precipitation on grassland NPP was significant, which was obviously greater than the grain yield of farmland. Temperature had a significant effect on the grain yield of farmland, but had a weak correlation with grassland NPP. This indicated that changes in precipitation and temperature together affected land productivity in the study area. This is consistent with the analysis results from lake sediments, pollen and South Asian monsoon sequences, further confirming the reconstruction results of this study (Sinha et al 2015, Li et al 2021, Liang et al 2022.

Periodic characteristics of land productivity change
The wavelet analysis results of land productivity series under different land uses are shown in figure 4. After low-pass filtering with a 20 yr moving average, the long-period vibration was more obvious, which was more continuous in time distribution and more statistically significant (

The synchronicity between the change of land productivity and the rise and fall of the Guge Kingdom
Due to the scarcity of historical documents, especially the lack of economic and demographic data that could be used for quantitative analysis, this study extracted religious and war information from historical documents to analyze the development progress of Guge Kingdom. Historians had confirmed that the history of the Guge Kingdom was not like the past understanding of 'sudden demise after 700 yr of prosperity' , but there were many ups and downs of   Taking the rises and falls of various Tibetan Buddhist sects as the context, supplemented by the records of invasion by foreign enemies, this study found that the religious development of the Guge Kingdom had a cycle of 150-200 yr. The periodicity of foreign invasion was not prominent, but it was related to the historical process of religious decline. By comparing the periodic fluctuations of the wavelet real part value in figure 5, the religious prosperity events such as revival of Buddhism, the spread of Drigung Kargyu Sect and the spread of Gelug Sect corresponded to the wave crest, and the decline events such as the division of the Guge Kingdom, the invasion and conquest of the Mongols and the collapse of the Guge Kingdom corresponded to the wave trough. This meant that there was an obvious synchronic phenomenon between the rise and fall of the Guge Kingdom and the fluctuation of land productivity (figure 7).
According to the real part fluctuation of the wavelet analysis results, this study divided the four fluctuation periods between 1000 and 1720 into eight stages, including four crest periods and four trough periods. By comparing the historical events of Guge Kingdom summarized in tables S2 and S3, this study could conclude the main events of each stage.
Comparing the rise and fall process of Guge Kingdom with the simulation results of land productivity indicated by grassland NPP and farmland grain yield, we further analyzed the synchronic phenomenon (table 1). During the four crest periods, Guge Kingdom experienced historical events such as the territory expansion and Buddhism dissemination. During the four trough periods, Guge Kingdom experienced historical events such as division, nomadic invasion, and kingdom collapse. It could be found that in different historical periods, there was a significant synchronic phenomenon between the simulation results of land productivity in Ali and the rises and falls of Guge Kingdom. Unfortunately, due to the lack of historical documents, this study was unable to obtain quantitative economic and social data to further analyze the mathematical relationship between them.

The transmission process of climate change in the society of Guge Kingdom
In order to further explore the mechanism of climate change induced land productivity fluctuation on society, we searched more historical documents to obtain possible relevant information. During Guge Dynasty, the output of land was controlled by the ruler according to the feudal system, so it is necessary to understand the feudal production relations.
According to the research of historians, the political system of the Guge Kingdom was different from that of the U-Tsang region. The Guge Kingdom implemented the political system of enfeoffment and garrison (Huang 2016b). During the rule of Mongolia and Yuan Dynasty, a feudal manor system called 'Xika' , which replaced the original traditional slavery economy, was widely popularized in the western region of the Tibetan Plateau (Ran 1994). In addition, due to the 'unity of politics and religion' nature of the civilization on the Tibetan Plateau, feudal lords and Buddhist temples controlled a large amount of land and population, and the king did not have centralized political power (Luczanits 2007). Therefore, the political and economic system of the Guge Kingdom could be summarized as follows: the supreme authority was the king, and the religious leaders were slightly weaker than the king, followed by the leaders of monks and customs at all levels, manor lords and temple Living Buddhas who were enfeoffed by the Royal Palace, and the serfs at the bottom (Huang 2013). This formed a pyramid like feudal social structure like today's Guge Palace site. The manor lords collected the agricultural and animal husbandry productivity from the bottom serfs and handed it over to the king layer by layer. When the climate changed from dry and cold to wet and warm, land productivity in Ali recovered and grew. The kingdom of Guge, with a larger population and greater strength, was revived. On the one hand, the king expanded outward, fighting the nomads and controlling the surrounding valleys and plateaus; On the other hand, Buddhism was promoted and temples were built to gain the support of religious forces (figure 8; Luciano and Zhang 2012, Huang 2016a; the detailed historical documents are given in Text S1). With the periodic change of climate, Guge kingdom also experienced constant cycles of rise and falls.
Due to the sensitivity and vulnerability of plateau vegetation to climate change, grassland degradation and farmland yield declined when the climate turn to cold and dry. This led to widespread famine and the escape of serfs. The loss of the bottom population and the decline of land productivity brought about the feudal economic crisis of the manor, which shook the ruling foundation of the Guge Kingdom (figure 8; Huang 2012; Huang 2016a; a detailed historical is given in Text S2). A similar mechanism also occurred in the collapse of the Tubo Kingdom in the late 9th century (Li et al 2021). The religious reforms were measures taken to respond a similar situation in which the ruling foundation was shaken, further intensified social contradictions and maybe accelerated the collapse of the kingdom. The shaking of the ruling foundation also made the number of troops that the royal family could recruit seriously insufficient, unable to resist the invasion of the Ladakh Kingdom army as in the past. This historical process of decline due to domestic and foreign aggression was very similar to the collapse of the Ming Dynasty in the same period (Zheng et al 2014, Liu et al 2018.
Therefore, this study summarized the process of climate change transmission in the society of Guge as 'climate-land productivity-manor economy-Kingdom rule' . Guge showed the social resilience of a strong feudal kingdom in the land productivity cycle changes. Despite the decline of society, kingdoms were able to maintain their rule or complete their revival. But during the the Little Ice Age (LIA) period, low temperatures and drought caused land productivity to decline more than twice as much. This eventually broke through social resilience and brought about the collapse of the Guge Kingdom. Many studies had confirmed that similar processes played a role in the rise and fall of civilization in the pre-industrial era (Zhang et al 2010, Douglas et al 2015, Beach et al 2019, Benati and Guerriero 2021. As a highland kingdom, the lower land productivity and lower population density made the Guge Kingdom located at high altitude more sensitive to climate change.

The collapse of the Guge Kingdom and the global crisis in the 17th century
In the 17th century, the world experienced an extremely cold and dry period. There were wars, regime collapse, revolution and population reduction in the world at the same time. Geoffrey Parker called it a 'global crisis' (Geoffrey 2021). By analyzing the quantitative response of agricultural, socio-economic and population disasters in Europe from 1500 to 1800 to climate change, scholars found that since the 1610s, drought, severe cold and extreme weather had continued to appear in the historical records of all continents (Zhang et al 2011, Geoffrey 2021. The grain productivity, harvest index and labor wage in the northern hemisphere had decreased with the decline of average temperature, while the population height (Komlos 2003), war, grain price and immigrants (Zhang et al 2011) were negatively correlated with temperature.
Current studies generally pointed out that the weakest monsoon event in China in recent 500 yr occurred in the first half of the 17th century, which brought serious drought in East Asia. It was an important reason for the peasant uprising in the late Ming Dynasty (Zhao et al 2021). The South Asian monsoon also decayed abnormally simultaneously. The South Asian monsoon in the past 2000 displayed by oxygen isotopes in Indian caves showed that the Indian monsoon weakened significantly in the early 17th century and brought disaster to India (Sinha et al 2015). Gujarat, India, suffered a rare drought from 1627 to 1630, killing more than one million people (Geoffrey 2021). The same description of decline appeared in the historical records of the Guge Kingdom. Andod, who arrived in Guge and preached in 1625, once wrote that 'there are many monasteries with about 500 monks' , only five small halls and many ruins remain, and there were only two managers and about 30 families in the flat land. He believed that 'it is not even one tenth of the estimated residential population in medieval times' (Michael 2007). Other researchers pointed out that the North Atlantic Oscillation and the El Nino Southern Oscillation promoted the occurrence of the Malthusian trap on a global scale, resulting in the simultaneity of global disasters and social upheavals (Lee and Yue 2020). This was consistent with our research results in Ali.
Through the collation of historical documents, modern historians generally believed that the Guge Kingdom finally perished in the crisis brought by the religious reform and the invasion of neighboring countries of Ladakh (Luciano and Zhang 2012; a detailed historical is given in Text S3). According to analysis of Sect. 4.2, this kind of internal and external troubles was not only the spontaneous outbreak of contradictions within the society. Compared with the disasters and crises in other regions in the 17th century, this study believed that when climate change exceeded the threshold of human society, it would undoubtedly bring about unbearable upheaval, which was global. However, due to the differences among culture, productivity development level, political characteristics and other factors, this upheaval was regional and special. The case of the collapse of the Guge Kingdom provided a unique example of plateau civilization for climate affecting civilization, and the change of plateau civilization was often ignored by history and climate researchers in past research. As pointed out by historical scholars, some regimes ushered in peace, innovation and development after the upheaval (Geoffrey 2021), while the Guge Kingdom collapsed and completely incorporated into the ruling system of the Qing government, and there was no independent regime or vassal state that owned autonomy after being canonized by the Chinese emperor.
Moreover, it should be noted that due to uncertainties in the paleoclimate data, model outputs, as well as the reliability and difficulty in quantifying historical documents, the study still had the regret of not being able to recover historical scenes more accurately. We look forward to further work incorporating additional historical documents and archaeological evidence. At the same time, quantitative studies will undoubtedly help researchers explore the dynamic relationship between climate change and social resilience.

Conclusion
This study found that the farmland crop yield and grassland NPP in Ali change dramatically with the climate change in precipitation and temperature, and this change presents a period of 180 yr. This was similar to the rise and falls of religion and the war cycles of invasion in the Guge Kingdom. At the same time, the change of land productivity in Ali and the rise and fall of the Guge Kingdom had a synchronic phenomenon: when the land productivity rose, the social stability and religious development of the Guge Kingdom; when land productivity declined, the Guge Kingdom was invaded by foreign enemies and religion declined. Based on the investigation of historical documents, this study believed that the periodic decline of land productivity had affected the population of the Guge Kingdom, thus shaking the manor and temple economy, weakening the national strength of the Guge Kingdom and bringing about the contradiction between royal and religious forces, resulting in the decline of the national strength of the Guge Kingdom and foreign invasion. As a plateau Kingdom, the Guge Kingdom was more sensitive to climate change, which led to its final collapse in the global crisis of the 17th century.

Data availability statement
The data generated and/or analysed during the current study are not publicly available for legal/ethical reasons but are available from the corresponding author on reasonable request.