The South China Karst region, centered on the Guizhou Plateau, is the largest and most concentrated karst ecological vulnerable zone in the world and is facing serious karst desertification (Cheng et al., 2017; Chen Q. et al., 2021). This case study was conducted in the karst mountainous area of the Guizhou Plateau. Karst landforms are typical in Guizhou Province; there are karst distribution areas in 95% of the counties of the province, and 91.7% of the cultivated land, 88.3% of the rural population, 94% of the grain output, and 95.7% of the gross national product come from counties with karst distribution. The industry, agriculture, transportation, urban construction, tourism, ecology, and other aspects of the province are directly or indirectly affected by karst (Su and Zhu, 2000). Excessive human activities, such as deforestation and rapid population growth, have contributed to the degradation of the ecosystem quality in the region (Han et al., 2020). We selected village ecosystems at different levels of KDC as research cases (two villages selected in each KDC area) to investigate vulnerability levels and influencing factors. The none-potential KDC area is located in the east of Guizhou Province, it is typical dolomite karst, and it belongs to the subtropical humid monsoon climate. The potential-mild KDC area is located in the northwest Guizhou Province. The landform type is mainly plateau mountain, and the rock type is mainly carbonate limestone. The moderate–severe KDC area is located on both sides of the Beipan River Canyon at the junction north of Zhenfeng County and south of Guanling County, Guizhou Province (Figure 1). The landform type is mainly a plateau canyon, the terrain fluctuates significantly, and the rock is mainly carbonate limestone.

The combination of human society and the environment has resulted in ecosystem vulnerability, and the factors influencing different ecosystem vulnerabilities vary (Kang et al., 2018). South China Karst has broken terrain, steep slopes, high vegetation sensitivity, low environmental carrying capacity, and poor land quality (Yang, 1990). Strongly developed underground cave systems lead to a lack of surface runoff, groundwater utilization is difficult, and engineering drought may occur (Liu C. et al., 2014; Qiu et al., 2021). According to the sensitive basic environmental characteristics, there is high system exposure and low resilience of village ecosystems in KDC. Therefore, we referred to relevant literature and selected the framework of E–S–LoR (Birkmann et al., 2013), and constructed an evaluation index system for the village ecosystem vulnerability of KDC with three dimensions of “exposure, susceptibility, and lack of resilience” and 26 indicators (Table 1). The details of the dimensions were as follows: (1) Susceptibility is the degree to which a system changes easily when disturbed, which reflects the stability of the underlying environment. Therefore, we chose the annual average temperature, annual precipitation, annual sunshine hours, altitude, average slope, terrain undulation, proportion of karst desertification area, soil erodibility K, landscape fragmentation, landscape diversity, and forest coverage as the indicators to measure susceptibility. (2) Lack of resilience is the system’s self-adaptive capacity to deal with risk stress, including pre-event risk reduction for prevention, and post-event adaptive strategies. Herefore, we chose the length of roads open to traffic, livelihood strategies, net income per inhabitant, proportion of the population with high school education or above, number of pools, food production per unit of arable land area, area of returning farmland to forest, and annual control rate of karst desertification to reflect resilience. (3) Exposure reflects the extent to which an ecosystem is exposed to human activity and the external environment. The most direct manifestation is the pressure on production and population activity in the environment. We chose population density, population dependency ratio, proportion of building area, amount of fertilizer used on farmland, amount of pesticide used on farmland, proportion of labor outflow, and per capita cultivated land area to measure exposure.

Basic natural and socio-economic data were used in this study. Basic natural data included meteorological, topographic, land-use type, and soil texture data. Socioeconomic data included demographic, economic income, production and population, and ecological governance-related data. Meteorological data were obtained from The China Meteorological Data Service Center. ¹ We downloaded the 30 m resolution digital elevation model from the geospatial data cloud ² and then used ArcGIS10.2 to extract the elevation, slope, and topographic relief. The land-use type data were interpreted using 30 m resolution remote sensing image data downloaded from the geospatial data cloud. Based on land-use data, we calculated landscape diversity and landscape fragmentation using Fragstats 4.2. We referred to the classification standard of karst desertification (Xiong et al., 2002) to extract karst desertification data in the study area. Soil type data were obtained from the Resource and Environment Science and Data Center of the Chinese Academy of Sciences. ³ Demographic, economic income, production, and ecological governance-related data were obtained through village group statistics and interviews with farmers. All the data were obtained in December 2020.

The vulnerability values of the three KDC areas (six villages) are listed in Table 2. The average village ecosystem vulnerability values was 0.468, with a minimum value of 0.29, and the maximum value was 0.646. We referred to the relevant research (Zhang et al., 2017), and divided the vulnerability into five levels according to the vulnerability values: slight (0–0.2), mild (0.2–0.4), moderate (0.4–0.6), high (0.6–0.8), and extreme vulnerability (0.8–1). The vulnerability values of the Baiduo and Yuntai Villages in the none-potential KDC area were 0.318 and 0.29, respectively, indicating mild vulnerability. The vulnerability values of Chaoying and Chongfeng Villages in the potential-mild KDC area were 0.494 and 0.532, respectively, indicating moderate vulnerability. The vulnerability values of Chaeryan and Xiagu Villages in the moderate–severe KDC area were 0.527 and 0.646, respectively, indicating moderate and high vulnerability. Overall, the village ecosystem vulnerability of the KDC area was mild, moderate, and high.

Based on statistical results (Figure 2A), the susceptibility values of Baiduo and Yuntai Villages in the none-potential KDC area were 0.167 and 0.105, respectively, whereas those of Chaoying and Chongfeng Villages in the potential-mild KDC area were 0.266 and 0.314, respectively. The susceptibility values of Chaeryan and Xiagu Villages in the moderate–severe KDC area were 0.204 and 0.289, respectively. We found that the susceptibility of the village ecosystem in the none-potential KDC area was the smallest, whereas that of the mild-potential KDC area was the highest. The village ecosystem susceptibility contributions in the none-potential karst desertification areas were 52.5 and 36.3%, and in the potential-mild karst desertification areas were 53.9 and 58.9%, respectively. The contributions in the moderate–severe karst desertification areas were 38.7 and 44.8%, respectively (Figure 2B).

The lack of resilience values for village ecosystems in the different KDC areas differed (Figure 3A). The lack of resilience values were smallest for Baiduo and Yuntai Villages in the none-potential KDC area at 0.072 and 0.078, respectively. The lack of resilience values were largest in Chaoying and Chongfeng Villages in the potential-mild KDC area, at 0.154 and 0.171, respectively. The lack of resilience values of Chaeryan and Xiagu Villages in the moderate–severe KDC area were 0.121 and 0.145, respectively, which fall between the other two grades of karst desertification areas. The contributions of the lack of resilience of the village ecosystems of the none-potential KDC areas were 22.7 and 27%, respectively, in the potential-mild KDC area were 31.1 and 32.1%, respectively, and 23 and 22.4% in the moderate–severe KDC area, respectively (Figure 3B).

Through the analysis of the exposure of the study area (Figure 4A), we found that the exposure values between the village ecosystems of the KDC displayed large differences, with a minimum exposure value of 0.048 and a maximum value of 0.211. The exposure values of the Baiduo and Yuntai Villages in the none-potential KDC area were 0.079 and 0.106, respectively. The exposure values of the village ecosystems in the potential-mild karst desertification area were the lowest, and the exposure values of the Chaoying and Chongfeng Villages were 0.074 and 0.048, respectively. The exposure values of the village ecosystem in the moderate–severe KDC area were the largest, and the exposure values of the Chaeryan and Xiagu Villages were 0.202 and 0.211, respectively. In general, there was little difference in the exposure values of village ecosystems in the same KDC area. The exposure values of village ecosystems in different KDC areas were in the order of moderate–severe KDC areas > none-potential KDC areas > potential-mild KDC areas. The contribution of exposure differs from that of vulnerability. The exposure contribution of the village ecosystems in the none-potential KDC area was 24.8 and 36.7%, respectively, in the potential-mild KDC area, it was 15 and 9%, respectively, and in the moderate–severe KDC areas, it was 38.3 and 32.7%, respectively (Figure 4B).

It is very important to reveal the level and influencing factors of village ecosystem vulnerability in KDC to further restore vulnerable karst ecosystems. The differences in the natural and socio-economic conditions of village ecosystems in different KDC areas lead to significant spatial differences in vulnerability. According to our research, the vulnerability level of the village ecosystems of KDC in South China Karst showed mild and moderate vulnerability. The unique geology and lithology are the basis of ecosystem vulnerability, and the unreasonable human social and economic activities are the external pressure factors of ecosystem vulnerability in the South China Karst (Li et al., 2002).

We calculated the contribution rate of each indicator to the vulnerability of the village ecosystem and screened factors with a large contribution rate. In the none-potential KDC area, the maximum contribution rates of landscape diversity, landscape crushing, soil corrosive factor, slope (Figure 2B), and labor loss factors were 14.3, 11.8, 8.4, 7.1, and 14.1%, respectively. The maximum contribution rates of pesticide and fertilizer use were 8 and 9.7%, respectively (Figure 4B). In the potential-mild KDC area, the maximum contribution rates of the annual average temperature, altitude factor, maximum soil corrosive factor (Figure 2B), livelihood strategies, per capita net income (Figure 3B), and population density were 14.1, 15.2, 5.9, 9.2, 8, and 5.8%, respectively (Figure 4B). In the moderate–severe KDC area, the maximum contribution rates of precipitation, topography, karst desertification, landscape diversity, forest coverage (Figure 2B), construction land, and pesticides were 8.3, 6.9, 8.3, 6.3, 8.7, 15, and 7.9%, respectively (Figure 4B). The terrain significantly affects the stability of the slope material and conditions for agricultural farming. Rainfall and temperature have an important impact on ecosystem stability. The landscape pattern index, degree of karst desertification, and forest coverage were the main factors that causeing ecosystem sensitivity. Soil erodibility is an important factor affecting soil and water loss. The use of chemical fertilizers and pesticides affects the quality of cultivated land. Livelihood strategies, population density, and labor loss have caused pressure on local residents, increasing their exposure. A low per capita income leads to a lower adaptability of farmers.

In none-potential KDC areas, local residents have a low level of education. To increase agricultural output, farmers use more pesticides and fertilizers during cultivation. There are few employment opportunities in the countryside; many young people choose to go away for work, leaving behind many older adults and children who are exposed to uncertain risks. The potential-mild KDC area is high in altitude, the average annual temperature is low, and the annual precipitation is insufficient. Serious soil erosion, low labor efficiency, and low economic income mainly rely on labor exports and small-scale farming, and the incidence of poverty is extremely high (Ji et al., 2020; Ren et al., 2020). Our survey found that most local farmers’ livelihoods came from labor output acquisition income, and small-scale planting and breeding industries, while the population pressure was high. The moderate–severe KDC area surface is crushed, the terrain of the area is steep, the surface soil and soil reservation capacity is insufficient (Mu et al., 2021), and annual precipitation is low. Karst in the region has a strong role; the soil formation rate is low, and a large amount of unreasonable reclamation in the early days has led to poor soil and discontinuous soil cover (Ren et al., 2020). This results in poor landscape diversity and forest cover. This shows serious karst desertification, which is difficult to control. To facilitate water access and agricultural cultivation, many people are concentrated in the valley area, accompanied by many construction facilities. Farmers use many pesticides to increase the output of Chinese red pepper and dragon fruit. According to the analysis of the difference in indicator vulnerability values (Figure 5), we found that topography, climate, forest coverage, landscape pattern, karst desertification degree, soil erodibility, KDC rate, production and construction activity, livelihood, and per capita net income were the key factors leading to the differences in village ecosystem vulnerability in the three different levels of KDC areas.

In recent years, scholars have conducted relevant research on karst ecological vulnerability. These studies included vulnerability and impact-factor analyses. Chen (2007) pointed out that owing to the deep soil layer and continuous soil cover in the karst trough area, there is a slight vulnerability. This is consistent with the results of our study of the karst plateau trough area (none-potential KDC area). Wang et al. (2021) studied the ecological vulnerability of karst areas in Yunnan Province, China, and revealed high vulnerability and extreme vulnerability in moderate-to-severe karst desertification areas. However, our results show that the moderate-to-severe karst desertification areas are moderately vulnerable and highly vulnerable. This may be due to the differences in KDC measures, leading to different ecological restoration effects and different degrees of vulnerability. Guo et al. (2017) analyzed the vulnerability level and influencing factors were analyzed of mountain ecosystem in Southwest China Karst using the remote sensing method, and found that the vulnerability of regions with strong karst development, low vegetation coverage, and high bedrock exposure rate was higher than that of regions with high vegetation coverage, low karst desertification and better ecological environment in karst mountain areas in southwest China. The results of this study are consistent with the actual vulnerability of the village ecosystems in the three KDC areas. Many studies demonstrated that vegetation cover factors, precipitation, topography, soil erosion factors, and the degree of karst desertification on the impact of karst ecosystem vulnerability is more significant (Wang and Yu, 2005; Chen, 2007; Wang et al., 2021). However, the KDC village ecosystem was characterized by karst ecosystem vulnerability. Owing to the differences in spatial scale, data accuracy, and measurement indicators, large-scale studies cannot reflect the characteristics of small-scale ecosystem vulnerability. For example, the influence mechanism of farmers’ production activities, living activities, and socio-economic development on the vulnerability of karst ecosystems is a problem that has not been investigated in current large-scale research. The study of vulnerability at the village ecosystem scale can accurately reveal the factors influencing vulnerability. This has important significance for providing guidance to the government in formulating planning policies, which is also the significance and necessity of small-scale research.

Various ecosystem problems caused by karst desertification seriously affect the lives of local residents and hinder the coordinated development of the local socioeconomic and ecological environments (Xiong and Chi, 2015). Over time, humans have attempted to control the deterioration of karst desertification in karst areas. For example, the Italian government restricts the cutting of firewood and prohibits goat breeding (Ford and Williams, 2013). KDC mainly adopts the measures of water storage, land management, returning farmland to forest and grassland, afforestation, three-dimensional ecological agriculture, and agricultural and forestry management development in the South China Karst (Xiong et al., 2006; Jiang et al., 2009). However, different levels of vulnerability still exist in rural areas of the KDC environment. Therefore, in view of the current situation of the village ecosystems of KDC, we should concentrate on both ecological management and socio-economic development, focusing on the sustainability of village development and proposing feasible adaptive management measures. In none-potential KDC areas, existing vegetation coverage should be maintained, the population scale should be controlled, population quality should be improved, and the rural labor force should be retained. Organic fertilizers and non-residual pesticides should be popularized and the use of stereoscopic agriculture in mountains should be developed. Forestry should be developed on mountain tops, the middle area of the mountain should be used for fruit industry development, and crop cultivation and livestock and poultry breeding should be carried out at the foot of the mountain and in low flat areas. The development of eco-tourism and the promotion of the sale of ecological products could promote economic development of the eco-industry. In mild-potential and moderate–severe KDC areas, the population size should be controlled, population quality should be improved, and population skills training should be strengthened along with the implementation of ecological and water storage irrigation engineering measures. The selection of an economic fruit forest with drought tolerance, calcium preference, and developed root systems for planting, implementing mountain closures for afforestation, and returning farmland to forests to increase vegetation coverage and reduce water and soil loss are recommended while popularizing the use of organic fertilizers and non-residual pesticides and developing stereoscopic agriculture on mountains. Projects for transforming slopes into terraces and building of water-saving, intensive agricultural production systems should be implemented. In addition, unreasonable human activities, such as grazing and felling of trees, should be prohibited in stone and semi-stone mountainous areas. Local scientific research departments should increase their investments in scientific and technological research and investigate new processed agricultural products to increase their added value. Local governments should take advantage of local natural conditions to build research and tourism bases with karst characteristics and promote local employment while simultaneously promoting the coordinated development of socio-economic and ecological factors.

In general, it is necessary to change the current development mode and concurrently improve the ecological and socio-economic benefits to promote high-quality and sustainable development of rural areas in karst desertification areas. Talents, technology, capital, and superior management modes are necessary to achieve high-quality and sustainable development, enhancing the management ability of grassroots leaders, strengthening the training of farmers’ knowledge and skills, and comprehensively enhancing the production skills and environmental awareness of residents. Based on the advantages of rural resource endowment, we should optimize the allocation of resources, build a sustainable production system, integrate various industries, strengthen the construction of rural industrial chains, and promote the transformation and upgrading of the industrial structure and technological innovation. A reciprocal mechanism between rural industry development and farmers’ interests should be established to promote the integration and development of the rural industry. We should vigorously develop the ecological industry and promote the specialization and integration of production, processing, storage, transportation, and sales of ecological products. Furthermore, amendments to the quality requirements of ecological products should focus on improvements in product quality and economic benefits. We should rationally plan land-use patterns and optimize the spatial structure of rural production, life, and ecology. The quality of the ecological environment and the service function of the system should be improved. We should abandon the development model of destroying the environment for economic benefit, build a virtuous rural system, and form a local sustainable and high-quality development model.

Current vulnerability research has been applied to ecological, natural, and societal subsystems, and coupled socio-ecological systems (Gallopín, 2006). Research on the vulnerability of coupled socio-ecological systems has not yet resulted in the formulation of a perfect theoretical system, and is not unified in terms of concept connotation, research framework, and evaluation methods. Analysis of the process and mechanism of human–environment coupling is still an unresolved issue in research on ecosystem vulnerability (Tian and Chang, 2012). Currently, empirical research is mostly quantitative, the research methods are immature, existing models are used to build the index system, and there is a lack of innovation and lack of pertinence (Tang et al., 2022). Moreover, there is a lack of analysis on the formation process and internal mechanisms of ecosystem vulnerability. The research mainly focuses on a particular spatial and temporal scale and lacks a dynamic comparative analysis of vulnerability at different spatial and temporal scales (Huang et al., 2014). In future research, it is necessary to improve the indicator system, research models, and innovative research methods. We should pay attention to the process of the comprehensive action of human and natural factors and analyze the impact of the human–environment coupling mechanism on the formation mechanism of ecosystem vulnerability. We need to extend the spatial and temporal scales of research and use the 3S technology to reveal the spatial and temporal dynamic change processes of ecosystem vulnerability to realize dynamic monitoring and prediction. We should analyze the interaction mechanisms of material flow, energy flow, information flow, and ecological processes in the social-ecological system, and the relationship between stakeholders and ecological processes, to reveal the coupling mode of social economic factors and ecological environment factors. We should explore the breakthrough point of the social-ecological system from one steady state to another to reveal the threshold of vulnerability of the social-ecological system. The mutual feedback mechanism of the relationship between social activities and ecological environment degradation or restoration should be studied to explore the mode of balance and coordination between human production, living activities, and ecological restoration to combine theoretical research and practical applications and provide a decision-making basis for promoting sustainable governance of fragile ecosystems.