Vulnerability-based analysis of traditional village distribution in Northwest Yunnan as an ecologically fragile area

Abstract

Traditional villages in Northwest Yunnan represent complex socio-ecological systems navigating the dual challenges of ecological fragility and rapid socio-economic transformation. This study examines how vulnerability patterns shape the spatial distribution of 248 traditional villages through an integrated assessment framework. We developed a Sensitivity-Resilience (S–R) model where sensitivity captures environmental constraints including topographic factors (elevation, slope, terrain roughness) and climatic variables (temperature, precipitation), while resilience encompasses ecological buffers (vegetation coverage, water network density) and socio-economic adaptive capacity (road network density, population density, GDP, nighttime lights)—combining 11 indicators from remote sensing data (DEM, MODIS climate products), national census records, and geographic databases (2000–2020). The geographic detector method, a spatial statistical technique that quantifies factor contributions without assuming linear relationships, revealed both individual and interactive effects on village distribution. Results show 81% of villages concentrate in low-to-moderate vulnerability zones, forming linear clusters along river valleys and lake peripheries. Natural factors impose primary spatial constraints (slope q = 0.72, elevation q = 0.69), while socio-economic factors exhibit stronger influence overall (GDP q = 0.89, road density q = 0.66). Critically, all factor interactions show enhanced effects, especially GDP interactions with road density and nightlight that reveal how economic development, accessibility, and urbanization jointly determine village vulnerability patterns. These findings directly inform differentiated conservation strategies for regional planners and heritage policymakers. High-vulnerability villages (17.7%) require adaptive infrastructure and hazard mitigation; moderate-vulnerability clusters (70.2%) can serve as development nodes with controlled tourism; low-vulnerability settlements (10.9%) need protection from urbanization pressure. The vulnerability framework provides transferable methods for understanding how traditional villages persist within, rather than despite, ecological constraints in fragile mountain regions globally.

Introduction

Traditional villages, as defined by the Ministry of Housing and Urban–Rural Development of China along with three other national ministries, are those villages formed earlier in history, rich in traditional resources, and currently well-preserved. They carry profound historical, cultural, scientific, artistic, social, and economic values, and retain ancient historical and local wisdom¹. However, with rapid urbanization and socioeconomic transformation, these irreplaceable heritage sites face unprecedented threats. According to data released at the Third China Ancient Villages and Towns Conference in 2017, the number of traditional villages has decreased by approximately 920,000 over the past 15 years and continues to decline at 1.6 villages per day. This alarming trend is particularly pronounced in ecologically fragile areas, where environmental constraints compound socio-economic pressures.

Research on traditional villages has evolved significantly from heritage preservation to comprehensive socio-ecological analysis. While “traditional villages” as a formal designation is unique to China’s protection system established in 2012, international research addresses similar settlements under various terms. Initially, studies predominantly focused on cultural preservation and architectural heritage, examining morphological evolution^2,3 and transformation strategies^4–6. However, scholars increasingly recognize that traditional villages represent complex socio-ecological systems, necessitating integrated perspectives^7,8. Traditional multifunctional landscapes across European regions—from Spanish dehesas to Alpine transhumance systems—integrate agriculture, forestry, and livestock management through complex socio-ecological systems^9,10. These landscapes embody centuries of accumulated local knowledge and adaptive practices^10,11. However, post-1950s socio-economic transformations have driven significant landscape changes⁹, with traditional ecological knowledge and management practices at risk of disappearing within a single generation¹⁰, threatening both cultural heritage and landscape sustainability¹².

Chinese scholarship has developed distinct research priorities for traditional villages, emphasizing protection and development challenges. Li advocates protecting cultural and landscape genes as fundamental to village conservation¹³. Liu proposes vitality assessment frameworks integrating both protection and development dimensions¹⁴. Based on Huizhou empirical research, Li identifies dual revitalization paths: enhancing internal competitiveness and strengthening external support through policies and infrastructure¹⁵. Many scholars explore tourism’s potential in driving village protection and development^16–18, while others investigate internal spatial characteristics including spatial form^19–21, public space quality²², and human settlement quality²³. These approaches provide valuable policy insights but often address symptoms rather than underlying vulnerability dynamics.

Geographic analyses have revealed important spatial patterns while exposing methodological limitations. Early studies examined isolated factors like topography or accessibility, struggling to capture complex village-environment interactions due to their focus on static conditions. Recent research has identified key determinants including natural and financial capital assets²⁴, demographic changes²⁵, institutional land regimes²⁶, location features²⁷. Studies reveal distinct distribution patterns across scales: national analysis shows concentrations in low-altitude, humid areas of southern and eastern China²⁸, while regional patterns vary from Jiangnan’s centralized clustering²⁹ to Henan’s “three cores and one belt” structure³⁰ and Fujian’s mountain-dominated distribution³¹. For Chinese traditional settlements, factors are often categorized as natural environment and social economy^32,33. However, these traditional geographic approaches fail to explain why villages with similar conditions exhibit divergent resilience trajectories, missing crucial sensitivity-resilience interactions that determine sustainability in fragile environments.

Notably, while international research has extensively examined rural settlements, significant gaps remain in studying traditional villages within ecologically fragile heritage areas. In regions like the Three Parallel Rivers, traditional villages face unique constraints: ecological fragility limits development options while heritage status prohibits transformative solutions like relocation. This dual constraint makes vulnerability assessment essential for identifying feasible adaptation pathways. Previous studies analyzing traditional village distribution rarely employed vulnerability perspectives, resulting in one-sided protection strategies. Villages in ecologically fragile areas face interconnected challenges—ecological degradation, economic hardship, cultural erosion, and inadequate infrastructure—that form vicious cycles threatening sustainability. Recent scholarship thus emphasizes the need for comprehensive vulnerability frameworks that capture complex human–environment interactions³⁴.

Vulnerability assessment offers a more holistic analytical framework than traditional geographic approaches. Vulnerability denotes the condition arising from internal or external disruptions induced by natural or anthropogenic factors, emerging from intricate interaction of ecological, social, and economic systems³⁵. Diverse ecosystems possess distinct features profoundly influenced by both natural processes and human activities^36,37, making standardized vulnerability indicators challenging³⁸. In socio-ecological systems, resilience represents “the capacity of a system to absorb disturbance and reorganize while undergoing change so as to retain essentially the same function, structure, and feedback”^34,39–42. Within vulnerability assessment, resilience functions as an inverse component where Vulnerability = f (exposure, sensitivity, inverse of resilience). Our S-R (Sensitivity-Resilience) framework operationalizes these concepts: sensitivity captures the system’s susceptibility to disturbances while resilience represents adaptive capacity. Villages with high sensitivity but strong resilience may maintain moderate vulnerability, whereas those with both high sensitivity and low resilience face critical threats.

Northwest Yunnan exemplifies these challenges as a typical ecologically fragile area with high traditional village concentration⁴³.The region hosts 248 traditional villages (31.9% of Yunnan’s total), facing dual crises: excessive development and commercialization leading to environmental degradation, and urbanization and the outflow of labor causing “hollowing out” and the decline of living environments⁴⁴. This study addresses three critical questions: (1) What multi-scale spatial patterns and heterogeneities characterize vulnerability distribution in traditional villages across ecologically fragile areas? (2) What are the key factors influencing spatial distribution from a vulnerability perspective, and what mechanisms drive these patterns? (3) How can resilience be enhanced to achieve sustainable development in ecologically fragile heritage areas? We employ ecological vulnerability assessment to establish spatial zoning systems, then analyze distribution characteristics through spatial analysis. Geographic detectors identify influencing factors and their interaction mechanisms, providing insights for protection and development constraints.

This research bridges the gap between environmental vulnerability research and traditional village studies, offering a new perspective for ecologically fragile heritage areas. By integrating sensitivity and resilience dimensions, our framework transforms static geographic analysis into dynamic vulnerability assessment. The findings provide theoretical support and practical verification for ecological protection and scientific development of traditional human settlements. Furthermore, this study establishes a reference framework for systematically studying traditional villages in ecologically vulnerable areas with heritage sites, contributing to both conservation theory and sustainable development practice.

Study area

The research region includes four prefecture-level units in northwestern Yunnan—Lijiang City, Diqing Tibetan Autonomous Prefecture, Nujiang Lisu Autonomous Prefecture, and Dali Bai Autonomous Prefecture, covering approximately 24.0°–29.7°N and 98.0°–101.8°E (Fig. 1). The physical landscape is structured by three north–south trunk valleys—the Nu (upper Salween), Lancang (upper Mekong), and Jinsha (upper Yangtze)—separated by elevated ridgelines. Elevation ranges from 700–800 m in the hottest sections of the Nu River valleys to 6,740 m at Kawagebo Peak (Meili Snow Mountains), producing steep local relief (often > 1,500 m within 5 km) and strong vertical environmental gradients. This valley–ridge configuration facilitates longitudinal movement along valley floors while constraining cross‑valley movement across watershed divides. Parts of Diqing, Nujiang, and Lijiang fall within the Three Parallel Rivers of Yunnan Protected Areas, inscribed on the UNESCO World Heritage List in 2003 and extended in 2007.

Fig. 1.

Study area.

Within the four prefectures, there are 248 traditional villages officially designated by the Ministry of Housing and Urban–Rural Development, representing about 31.9% of Yunnan’s provincial total. As Fig. 2 illustrates, many sites lie along valleys and basin margins; we formally evaluate their relationships to elevation, local relief, hydrologic proximity, and accessibility by operationalizing these variables on a 1 km grid within a S–R framework and testing contributions with the geographic detector. This design links the region’s physical constraints to socio-spatial patterns, enabling subsequent tests of trade-offs between cultural landscape conservation and accessibility (or ecological connectivity, if modeled) under a coupled human–natural system.

Fig. 2.

Traditional human settlements in Northwestern Yunnan, China.

Research framework

Drawing on socio-ecological systems theory and vulnerability science^45–47, this study develops a vulnerability-based analytical framework to examine traditional village distribution in Northwestern Yunnan (Fig. 3). The framework integrates vulnerability assessment, spatial analysis, and factor detection to reveal how villages navigate ecological constraints in this fragile mountain region.

Fig. 3.

Research framework.

The core of our framework is the S-R model, where vulnerability emerges from the imbalance between environmental stressors and adaptive capacity. Sensitivity indicators capture Northwestern Yunnan’s environmental constraints: topographic factors (elevation, slope, terrain roughness) reflect susceptibility to geological hazards and agricultural limitations in this plateau-edge region; climatic variables (temperature, precipitation) determine agricultural system stability across vertical climate zones. Resilience indicators encompass both ecological buffers—FVC for ecosystem health and water network density for resource availability—and socio-economic adaptive capacity through infrastructure accessibility (road density), human capital (population), and economic resources (GDP, nighttime lights). This selection recognizes that in traditional village contexts, human presence maintains rather than threatens cultural landscapes, making socio-economic factors essential components of system resilience rather than external pressures.

The framework operates across three analytical dimensions. First, vulnerability assessment generates a continuous surface across the study area using weighted integration of these 11 indicators, creating a spatial gradient from low to very high vulnerability zones. Second, overlay analysis examines how the 248 traditional villages distribute across these vulnerability zones, revealing preferential occupation patterns. Third, geographic detector analysis quantifies factor contributions and their interactions, identifying the mechanisms that drive observed spatial patterns.

Data and processing

The list of traditional villages was obtained from the Traditional Chinese Village Museum. As of 2023, six batches had been released, comprising 8156 officially recognized traditional villages. Village coordinates were obtained from Gaode Map, and spatial visualization was conducted in QGIS 3.40.1. To operationalize the vulnerability assessment framework, we assembled 11 indicators across sensitivity and resilience dimensions (Table 1).

Table 1.

Vulnerability assessment indicator system for Northwestern Yunnan.

Dimension	Indicator	Unit	Range	Weight	Attribute
Sensitivity	Altitude	m	748.7–6414.4	0.0499	+
	Slope	°	6.304–83.382	0.0767	+
	Terrain roughness(TR)	m	24–1681	0.0877	+
	Temperature(T)	°C	− 13.88 ~ 21.66	0.0203	–
	Precipitation(P)	mm	0 ~ 22,310	0.0657	–
Resilience	Fraction vegetation coverage (FVC)	–	0–1	0.0259	+
	Water network density (WND)	km/km2	0 ~ 6.12	0.0712	+
	Road network density (RND)	km/km2	0 ~ 3.933	0.2878	+
	Population density (PD)	people/km2	0 ~ 39,066	0.1415	–
	Gross Domestic Product (GDP)	10000yuan/km2	0 ~ 7734.5	0.0827	–
	Nighttime lighting	W·cm⁻2·sr⁻1·μm⁻1	0–63	0.0907	–

Environmental and socio-economic covariates were assembled from authoritative sources: annual China land cover data by Yang Jie and Huang Xin (Wuhan University)⁴⁸; monthly climate grids from the National Tibetan Plateau Data Center, aggregated to 1 km annual mean temperature and annual precipitation^48,49; 250m fraction vegetation coverage (FVC) from the same institution⁵⁰; OpenStreetMap (OSM) hydrography and road networks; Landsat-derived gridded population density⁵¹; 1 km GDP rasters from the Resource and Environmental Science Data Registration and Publication System⁵²; and a European Space Agency (ESA) DEM from which elevation, slope, and terrain roughness were derived⁵³. All datasets were projected to a common coordinate reference system, harmonized in extent, and resampled to a 1 × 1 km grid. This resolution is widely used in macro-scale assessments, balancing data availability, computational feasibility, and the ability to reveal regional spatial patterns^54,55. The grid size approximates the mean land area of Chinese rural settlements (17.11 hm²)⁵⁶, supporting adequate representation of village footprints within cells; and it enables coherent integration of environmental and socio-economic variables at the regional scale, which is essential for diagnosing broad spatial gradients and influencing factors of vulnerability.

The critic-entropy weighting method was applied to determine indicator weights based on each variable’s information content and statistical properties. The resulting weights (Table 1) show considerable variation, with road network density receiving the highest weight (0.2878), suggesting high spatial variability and low redundancy with other indicators. Temperature (0.0203) and FVC (0.0259) received lower weights, potentially indicating either limited spatial variation across the study area or higher correlation with other variables.

Methods

Vulnerability assessment approach

Indicator weighting

To objectively determine the relative importance of vulnerability indicators in Northwest Yunnan’s heterogeneous mountain landscape, we applied a combined CRITIC-Entropy method. The Entropy Method determines the weight of indicators by assessing the degree of data dispersion; the greater the dispersion across villages, the more information that indicator provides for distinguishing vulnerability levels. The steps are as follows:

Calculate the proportion of the ith indicator in the total indicators

1

Calculate the entropy value of the i-th indicator

2

Calculate the redundancy of information entropy

3

Calculate the weight of each indicator

4

The CRITIC method is an objective weighting method that determines the weight of indicators by comparing intensity and conflict through a tabulation process. The greater the comparison intensity, the higher the indicator weight; the conflict indicator represents the correlation between indicators, the higher the correlation, the lower the weight. The specific calculation steps are as follows:

Calculate the intensity of comparison of indicators:

5

Calculate the conflict of indicators:

6

Calculate the weight of indicators:

7

By combining the Entropy Method and CRITIC method, the weight of each indicator can be determined more scientifically and reasonably⁵⁷. The formula is as follows:

8

Vulnerability index calculation

To quantify the vulnerability of traditional villages within our Sensitivity-Resilience (S–R) framework, we calculated vulnerability as the weighted balance between environmental stressors and adaptive capacity. The vulnerability index calculation formula is as follows⁵⁸.

9

where VI is the vulnerability index,S_i is the sensitivity indicator,R_j is the resilience indicator, w_i,w_j are the weights of S_i and R_j, respectively.

Spatial correlation analysis

To identify spatial clustering patterns and dependencies in the distribution of traditional villages across vulnerability zones, we conducted spatial autocorrelation analysis. This reveals whether vulnerable villages concentrate in specific regions or scatter randomly across the landscape, providing insights for targeted conservation strategies.

First, we calculated the global Moran’s I statistic to test for overall spatial autocorrelation in the vulnerability values. The global Moran’s I ranges from − 1 to 1, where positive values indicate spatial clustering of similar values. Second, we conducted Local Indicators of Spatial Association (LISA) analysis to identify local spatial clusters and outliers. LISA classifies significant local spatial associations into four types: High-High (HH) clusters indicating high vulnerability areas surrounded by high values, Low-Low (LL) clusters indicating low vulnerability areas surrounded by low values, High-Low (HL) outliers, and Low–High (LH) outliers. This analysis helps identify not only clusters but also spatial anomalies that may require special attention in conservation planning.

Village distribution patterns

To identify the spatial organization of traditional villages and their concentration hotspots across Northwest Yunnan, we applied kernel density estimation (KDE) and nearest neighbor analysis (NNI).

KDE is a crucial method for measuring the spatial distribution of objects of interest within a region and is widely used in the study of the spatial distribution of geographical entities such as villages⁵⁹. The formula is as follows.

10

In the formula, epresents the density value at point x, with a higher value indicating a greater concentration of traditional villages in the region,  is the amount of traditional villages .h is the bandwidth, which controls the smoothness of the kernel function. K is the kernel function. x_i is the value of each data point.

NNI is an important tool for evaluating the spatial distribution pattern of traditional villages⁶⁰. It measures this by calculating the ratio of the average observed distance between a village and its nearest neighboring village to the expected average distance. If the NNI equals 1, it indicates a random distribution of villages; less than 1 suggests a clustered distribution; and greater than 1 indicates a dispersed distribution. The formula is as follows.

11

where R is the NNI, is the average observed distance between all traditional villages and their nearest neighboring village.   is the average expected distance between all traditional villages and their nearest neighboring village under the assumption of random distribution.

Criteria for Interpretation:

If R = 1, the traditional villages are spatially distributed at random.

If R < 1, the traditional villages are spatially clustered.

If R > 1, the traditional villages are spatially dispersed.

Driving factor analysis

To identify which factors most strongly influence the spatial distribution of traditional villages and understand their complex interactions in this ecologically fragile heritage area, we employed the Geographic Detector model. It is a tool that excels in revealing the heterogeneity of spatial distributions and identifying the primary forces driving these patterns^61,62. It particularly shows its strengths in exploring complex interactions and nonlinear effects among various influencing factors. In this study, firstly, “Factor Detector” (formula 12) is used to determine the extent to which each independent variable affects the vulnerability of traditional villages.

12

where Q represents the explanatory power of a factor, Nh is the number of samples in stratum h,  is the variance of traditional village vulnerability in stratum h, N is the total number of samples, and  is the global variance of traditional village vulnerability.

Then the Interaction Detector is used to assess the interaction between two factors affecting the vulnerability of traditional villages. This interaction is determined by comparing the Q-values of single-factor interactions with the sum of the Q-values of two-factor interactions. The types of two-factor interactions are shown in Table 2.

Table 2.

The types of factor interactions.

Description	Interaction
q(X1 ∩ X2) < Min(q(X1), q(X2))	Weaken, nonlinear
Min(q(X1),q(X2)) < q(X1 ∩ X2) < Max(q(X1)), q(X2))	Weaken, uni-
q(X1 ∩ X2) > Max(q(X1), q(X2))	Enhance, bi-
q(X1 ∩ X2) = q(X1) + q(X2)	Independent
q(X1 ∩ X2) > q(X1) + q(X2)	Enhance, nonlinear

Results

Distribution pattern of traditional villages in Yunnan Northwest under vulnerability assessment

Vulnerability assessment results

Based on the spatial auto-correlation analysis, the Moran’s I index (Fig. 4) indicates a spatial characteristic of clustering within areas with different levels of vulnerability In northwestern Yunnan, there are almost no areas with low vulnerability.

Fig. 4.

The Moran’s I index of vulnerability In Northwestern Yunnan.

After evaluating the spatial vulnerability of northwestern Yunnan, we found that the overall vulnerability of the region is at a relatively high level, as shown in Fig. 5. Areas with lower vulnerability account for about 0.63% of the study area, mainly distributed in the west and south of Erhai Lake, as well as in the urban area of Lijiang City. Moderate vulnerability areas account for about 16.7% of the study area, primarily found in the areas around lakes, along river valleys, and in mountain valleys. Areas with higher and high vulnerability account for about 44.9 and 37.6% of the study area respectively, widely distributed across the region, mainly in the Hengduan Mountains and areas bordering the Qinghai-Tibet Plateau. The overall spatial pattern of vulnerability is characterized by higher levels in the northwest and lower levels in the southeast. From an administrative perspective, these areas are primarily concentrated in Diqing County, the northern part of Nujiang, and the northern part of Lijiang City. Areas with lower vulnerability are mainly distributed in Dali county and the central part of Lijiang.

Fig. 5.

(a) Spatial pattern of vulnerability in northwestern Yunnan, (b) LISA cluster map of vulnerability. The maps were generated by QGIS 3.40.1 (https://qgis.org/project/overview/).

Spatial distribution of traditional villages

Through the nearest neighbor index analysis, we found that the spatial distribution of traditional villages in Yunnan Northwest exhibits characteristics of clustering distribution, with a z-score of − 9.257. Additionally, through kernel density analysis, we obtained the spatial distribution pattern of traditional villages in Yunnan Northwest (Fig. 6). The overall distribution of traditional villages in Yunnan Northwest is mainly concentrated in the central and southeastern parts of the study area, forming two core areas and sub-core areas (Fig. 6a). From an administrative perspective, they are mainly concentrated in east of Jianchuan County, and around the urban area of Gucheng and southeastern of Yulong, with the rest exhibiting linear distribution characteristics along rivers and lakes.

Fig. 6.

Traditional village kernel density (a), average nearest neighbour index (b). The maps were generated by QGIS 3.40.1 (https://qgis.org/project/overview/).

Traditional village distribution under vulnerability zoning

By overlaying the traditional villages with the vulnerability assessment map, we obtained the vulnerability levels of the traditional villages within the study area (Fig. 6). Statistical analysis reveals that there are 27 villages with lower vulnerability (10.9%), 174 with moderate vulnerability (70.2%), 44 with higher vulnerability (17.7%), and 3 with high vulnerability (1.2%) (Fig. 7). The distribution shows a pronounced concentration in the moderate vulnerability category, forming a bell-shaped pattern. Villages with lower and moderate vulnerability account for 81.04% of the total number of traditional villages, while areas with lower and moderate vulnerability only occupy 17.33% of the study area. This indicates a strong spatial concentration of traditional villages in areas with lower to moderate vulnerability levels.

Fig. 7.

Spatial distribution of traditional villages based on vulnerability assessment.

This “middle-heavy” distribution pattern reflects multiple factors. Low vulnerability areas (flatlands with good accessibility) experienced extensive urbanization since the Reform and Opening-up, with many potential traditional villages absorbed or modernized before China initiated its Traditional Village List in 2012. High vulnerability areas with extreme environmental conditions could rarely support settlements substantial enough to develop rich cultural heritage. The 70.2% concentration in moderate vulnerability zones thus represents an optimal balance: sufficiently peripheral to avoid development pressure while hospitable enough for sustained habitation. Our S–R analysis further reveals this isn’t merely favorable site selection but dynamic equilibrium—villages developed resilience to offset environmental sensitivities. This quantitative finding transforms intuitive observations into precise evidence, showing these villages exist in delicate balance requiring targeted conservation strategies.

Through examining the distribution patterns of traditional villages with different vulnerability levels (Fig. 8), distinct spatial characteristics emerge. Villages in low vulnerability areas are concentrated in the central study area and around large lakes and towns. In contrast, both moderate and high vulnerability villages display linear patterns along river corridors, though with different topographic characteristics.

Fig. 8.

Spatial distribution of traditional villages with different degrees of vulnerability.

Our field investigations conducted during the rainy season (June 2024) in the Jinsha River basin of northwestern Yunnan provide crucial insights into this riverine distribution pattern. While both moderate and high vulnerability villages follow river systems, moderate-vulnerability villages typically occupy elevated positions on valley slopes rather than flood-prone valley floors, reflecting traditional site selection wisdom. This topographic positioning—providing protection from seasonal flooding while maintaining water access—is captured through our slope and terrain relief indicators. The vertical separation between villages and river channels (as shown in Fig. 2b) helps explain why these riverside settlements register as moderate rather than high vulnerability despite their proximity to water courses. High vulnerability villages, conversely, tend to occupy more precarious positions with steeper slopes and greater exposure to geological hazards, contributing to their elevated risk classification.

Key factors affecting the spatial distribution of traditional villages within vulnerability zoning

The spatial distribution of traditional villages in northwestern Yunnan exhibited a characteristic agglomeration, with significant regional disparities. Spatial differentiation arises from the combined influence of multiple factors encompassing both natural and social settings. The use of geographic detectors in single factor detection and interactive factor detection helps explain the key factors affecting the spatial distribution of traditional villages. By analyzing 11 key factors, including slope, elevation, relief amplitude, vegetation coverage, average temperature, precipitation, GDP, river network density, road network density, population density, and night light brightness, we can identify the unique determinants that drive village clustering and their vulnerability.

Single factor detection results

As shown in Fig. 9a, the single factor detection results provide a quantitative measure of the explanatory power of these 11 factors in shaping the spatial distribution of traditional villages. By calculating the q values, we identified the relative importance of each factor and their role in explaining the agglomeration patterns of villages.

Fig. 9.

Factor detection (a), Interaction factors detection (b).

Analysis results

In terms of natural factors, slope (q = 0.7213) and elevation (q = 0.6912) are the dominant natural factors, reflecting the significant constraints imposed by the rugged terrain of northwestern Yunnan. Villages are concentrated in low-slope, mid-altitude areas where agricultural potential and accessibility are optimized, while high-slope areas are generally unsuitable for habitation. Relief amplitude (q = 0.6035) and river network density (q = 0.5821) emphasize the importance of topographical variability and water resources in shaping settlement patterns. Climatic conditions, including average temperature (q = 0.5718) and precipitation (q = 0.5549), play a moderate role, while vegetation coverage (q = 0.4437) shows a comparatively lower impact, indicating that ecological factors are secondary to topographical constraints.

Socio-economic factors dominate the results, with GDP (q = 0.8868) and night light brightness (q = 0.7159) having the highest explanatory power. These factors reflect the influence of economic activity and human development on village clustering. Road network density (q = 0.6619) and population density (q = 0.6442) are also significant, underscoring the importance of accessibility and human settlement dynamics.

The results reveal a clear distinction between the roles of natural and socio-economic factors. While natural factors such as slope and elevation dominate spatial constraints, socio-economic factors like GDP and road network density are critical for driving village clustering in economically active areas.

Key factors and their regional context

The rugged terrain of northwestern Yunnan imposes significant constraints on village distribution. Vulnerability analysis highlights that areas with steep slopes and high relief amplitude are prone to ecological risks like landslides, erosion, and limited agricultural viability. Consequently, villages cluster in low-slope areas, such as valley floors, where accessibility and arable land are better. This pattern is confirmed by single factor results, with slope (q = 0.7213) and relief amplitude (q = 0.6035) as key determinants. Elevation also plays a crucial role. Villages are concentrated between 2000 and 3000 meters, where climatic conditions support agriculture and habitation. Vulnerability analysis shows that extreme altitudes (too high or low) increase exposure to climate stresses, such as frost or limited growing seasons. Mid-altitude areas provide a balance between ecological stability and agricultural productivity. Water accessibility, indicated by river network density (q = 0.5821), is another critical factor. Vulnerability analysis emphasizes that proximity to water reduces ecological risks and supports sustainability. Villages near rivers benefit from essential resources for irrigation and transportation, while areas far from water often face compounded vulnerabilities due to limited agricultural potential.

Socio-economic factors significantly shape village distribution and vulnerability. Vulnerability analysis indicates that economically marginalized areas with low GDP and poor infrastructure face higher risks due to restricted access to resources, services, and markets. This is supported by single factor results, with GDP (q = 0.8868) and road network density (q = 0.6619) emerging as the most influential socio-economic factors. Villages in economically vibrant regions, such as Lijiang and Dali, benefit from better infrastructure, tourism-driven growth, and enhanced accessibility, reducing their vulnerability. Economic activity in these regions provides resources for disaster mitigation, cultural preservation, and development. Conversely, remote villages, such as those in Diqing, face isolation due to underdeveloped roads and limited opportunities. While this isolation can preserve cultural authenticity, it also exacerbates risks related to poverty and lack of access to healthcare, education, and markets. Accessibility, reflected in road network density and population density (q = 0.6442), is key to reducing vulnerability. Villages along major valley roads benefit from connectivity, enabling economic integration and reducing isolation. Vulnerability analysis confirms that well-connected villages are better equipped to respond to ecological and socio-economic challenges, while remote mountainous settlements remain highly vulnerable due to poor infrastructure and limited adaptability.

Interactive factor detection results

Interactive factor detection provides a deeper understanding of how the combination of different factors influences the vulnerability of traditional villages. The results show that all factor interactions exhibit either enhanced or non-linearly enhanced explanatory power, meaning that the combined effect of two factors is always stronger than the impact of any single factor alone (Fig. 9b). This highlights the complexity of the factors shaping traditional village vulnerability, as their combined influence often reflects the interplay between natural, socio-economic, and infrastructural conditions.

Notably, the interaction between GDP and other factors significantly improves explanatory power. For example, when GDP interacts with road network density, the combined effect highlights how economic development and accessibility together determine the spatial distribution and vulnerability of villages. Similarly, the interaction between GDP and night light brightness reflects how moderate economic activity supports village resilience, while excessive urbanization risks eroding traditional village characteristics. These findings underscore the critical role of economic development in shaping traditional village layouts, as settlements are deeply influenced by human interaction with their environment, infrastructure, and socio-economic conditions.

Overall, the results emphasize that traditional villages are not shaped by isolated factors but by the dynamic interplay of economic, natural, and social forces. The enhanced explanatory power of factor interactions underscores the importance of considering combined influences when analyzing spatial patterns and vulnerabilities, particularly in regions like northwestern Yunnan, where economic activity, physical geography, and cultural preservation are intricately linked.

Discussion

Theoretical implications for vulnerability and resilience

Our findings reveal a counterintuitive pattern: 81.04% of traditional villages concentrate in low-to-moderate vulnerability zones despite only 17.33% of the study area offering such conditions. This concentration pattern reflects adaptive settlement strategies, where villages deliberately position themselves in zones of moderate constraint that balance resource access with manageable environmental challenges. This suggests villages accept moderate vulnerability as a trade-off for other benefits such as defensive positions, resource diversity, and cultural continuity—a pattern distinct from the risk minimization assumed in conventional settlement theories. Such adaptive behavior aligns with socio-ecological resilience theory, where systems persist not through stability but through maintaining flexibility within dynamic equilibrium zones.

By adopting a vulnerability-based perspective, our study provides new insights into the complex mechanisms shaping the spatial distribution of traditional villages. In previous studies on traditional villages, scholars have often argued that natural factors play a dominant role^63,64. Our vulnerability-based analysis reveals that the dominant factors driving vulnerability distribution vary significantly across regions. Studies in other regions have documented different factor dominance patterns^,32,58, while our northwestern Yunnan case reveals a unique pattern where natural and socio-economic factors achieve near-parity (with socio-economic factors slightly dominant). This reflects the region’s transitional character between Tibet’s extreme natural constraints and Yunnan’s developing economy. This finding indicates these villages exist in what rural development theory terms a “transitional phase”—no longer purely determined by environmental constraints yet not fully integrated into modern economic systems, creating dual vulnerability through isolation from development opportunities and exposure to market forces that commodify cultural resources.

The geographic detector’s revelation that all factor interactions produce enhanced effects (15–30% increases in q-values) extends vulnerability science beyond linear causation, demonstrating that vulnerability emerges from systemic relationships rather than additive stressors.

Policy and management implications

Our vulnerability assessment framework reveals differentiated policy needs across Northwest Yunnan’s traditional village landscape. The geographic detector results demonstrate that slope (q = 0.7213) and elevation (q = 0.5960) are the dominant natural factors, while road network density (q = 0.6619) and GDP (q = 0.8943) represent the most critical socio-economic factors. This spatial differentiation demands tailored strategies.

For high-slope mountainous villages, where topographical constraints dominate, policies must work with rather than against natural limitations. Infrastructure investments should prioritize environmentally sensitive connectivity solutions^65,66. The persistence of villages in these high-vulnerability zones represents a deliberate trade-off—accepting environmental hardship in exchange for cultural preservation and community autonomy. Policy interventions should respect these trade-offs by supporting in-situ adaptation rather than forced relocation.

For villages in river valley corridors, the moderate influence of river network density (q = 0.4636) combined with relatively better accessibility suggests these areas can serve as development nodes. Policy should focus on strengthening these villages as service centers that support more remote settlements, creating a hierarchical support system that reduces pressure on highly vulnerable locations. The concentration of villages in these zones likely reflects generations of adaptive learning about optimal ecological niches, where traditional water management practices transform seasonal flooding into productive agricultural cycles.

Near urban centers, while our analysis shows lower environmental vulnerability, villages face different pressures from rapid tourism development and cultural commodification. Villages that historically adapted to physical isolation now must adapt to tourism pressure and market integration. Management strategies should emphasize development controls, visitor capacity limits, and mechanisms to ensure local communities capture tourism revenue.

The significant interaction effects between natural and socio-economic factors (particularly slope-GDP interactions with q = 0.92) indicate that single-sector policies will be insufficient. Economic isolation in mountainous areas creates compounded vulnerability that neither topographical nor economic factors alone explain. This demands integrated landscape management approaches that coordinate environmental protection, infrastructure development, and cultural preservation across administrative boundaries. Such integration requires establishing inter-agency coordination mechanisms and developing vulnerability-based zoning regulations that recognize the coupled human–environment dynamics our study reveals.

Limitations and future research

This study employed 1 × 1 km resolution data, which is standard for regional vulnerability assessments and provides an optimal balance between spatial coverage and data availability. However, this resolution may not fully capture micro-scale interactions between villages and their immediate environments. Future research could incorporate higher-resolution data and multi-scale analysis to refine village-level assessments while maintaining regional comparability. The geographic detector method, though powerful for identifying factor importance, cannot capture the adaptation mechanisms, social networks, or traditional knowledge systems that enable villages to manage vulnerability trade-offs. Mixed methods combining quantitative assessment with ethnographic interviews and participatory mapping would help understand how communities navigate trade-offs between environmental security, economic opportunity, and cultural continuity.

Our vulnerability framework captures current conditions based on 2000–2020 data, which may not fully reflect recent intensification of extreme weather events. While our field observations in June 2024 validated current vulnerability patterns—confirming that riverside villages maintain strategic positions above flood-prone areas—climate change projections suggest increasing frequency and intensity of extreme precipitation that could exceed historical patterns. The moderate vulnerability classification of riverside villages represents current equilibrium between exposure and adaptive capacity, which may shift as climate impacts intensify.

Future research should incorporate scenario-based modeling to examine how vulnerability patterns may evolve under different climate and development trajectories. Integration of climate projection models would be particularly valuable for developing dynamic vulnerability assessments that can guide long-term conservation planning. Understanding how traditional adaptation strategies might inform contemporary climate resilience represents a critical research frontier where vulnerability science must engage with local knowledge. Such forward-looking approaches would help policymakers design targeted strategies that anticipate and mitigate emerging risks to traditional villages.

Conclusion

This study advances vulnerability theory by revealing how traditional villages in Northwestern Yunnan exhibit a counterintuitive “moderate vulnerability preference”—with 81.04% occupying low-to-moderate vulnerability zones despite only 17.33% of the region offering such conditions. This pattern challenges conventional vulnerability theory that assumes settlements minimize exposure, instead demonstrating place-based resilience where communities strategically balance environmental constraints against resource accessibility. Such positioning reflects rural sustainability strategies evolved over centuries, where moderate vulnerability becomes optimal for maintaining both cultural continuity and economic viability in ecologically fragile areas.

Our geographic detector analysis contributes to understanding coupled human–environment systems by quantifying how socio-economic factors slightly dominate natural factors in shaping vulnerability patterns—a finding unique to this transitional region between Tibet’s extreme natural constraints and inland China’s economic development. The prevalence of interaction effects (all showing enhanced or nonlinear enhancement) reveals that vulnerability emerges from systemic relationships rather than additive stressors, extending vulnerability science beyond linear causation models. This systemic perspective reframes traditional villages not as passive victims of environmental hazards but as active agents navigating complex vulnerability landscapes.

The vulnerability framework exposes fundamental tensions in traditional village conservation that resist technical solutions. Tourism development epitomizes these contradictions: essential for economic sustainability yet accelerating cultural erosion through commodification. Infrastructure improvements present similar paradoxes—road construction reduces isolation-induced vulnerability but facilitates urbanization pressures that threaten village integrity. These conflicts reflect deeper contradictions between heritage preservation’s static ideals and rural sustainability’s dynamic requirements. Current policies attempting to balance these tensions through zoning or visitor limits fail to address underlying structural issues of unequal benefit distribution and external capital domination.

Critical implementation barriers transcend technical challenges. Fragmented governance across ministries creates policy conflicts where villages simultaneously receive contradictory directives to preserve, develop, and modernize. The absence of integrated vulnerability assessments means interventions address symptoms rather than systemic vulnerabilities. Most critically, current approaches lack mechanisms for maintaining place-based resilience—the adaptive capacity that enabled villages to persist through centuries of environmental and social change. Without recognizing this resilience as central to sustainability, conservation efforts risk creating “specimen villages” that preserve form while losing function.

This vulnerability-based approach ultimately redefines conservation objectives for traditional villages. Rather than eliminating vulnerability or freezing villages as heritage sites, the goal becomes maintaining adaptive capacity within acceptable vulnerability thresholds. Understanding how villages have historically navigated rather than avoided vulnerability provides crucial insights for climate adaptation strategies. As environmental changes intensify, the place-based resilience embodied in traditional villages offers lessons for rural sustainability beyond heritage preservation—demonstrating how human communities can persist within, rather than despite, ecological constraints.

Author contributions

Ruiqu MA: Conceptualization, Writing, Funding acquisition. Linyue LIU: Investigation, Data analysis, Writing. Jian XU: Resources, Conceptualization. Milun YANG: Investigation, Methodology. Gengyu XU: Validation, Editing. Huirong YANG: Editing. Ming MA: Validation. All authors reviewed and approved the final manuscript.

Funding

This work was supported by the Applied Basic Research Foundation of Yunnan Province (grant number 202301AT070177) and the Yunnan Office of Philosophy and Social Science (grant number QN202309).

Data availability

Data are available in a public, open access repository and can be freely downloaded from the official websites, as detailed in the “Data and processing” section. Any additional raw data are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.