Multi-Scale Ecological Coupling Mechanisms of Environment, Pattern, and Architecture in Traditional Villages of Southern Shaanxi

Abstract

Traditional villages represent vital living heritage in China. We develop a multi-scale eco-coupling framework integrating GIS spatial analysis and 3D laser scanning to analyze the natural and social environment, spatial patterns, and architectural forms across macro–meso–micro levels in traditional villages of southern Shaanxi, and use partial least squares structural equation modeling (PLS-SEM) to test the hypothesized cross-scale pathways. The results show significant spatial clustering, mainly in the water-adjacent low-mountain valleys and under moderate gradients of GDP, population density, and road density. The morphology is classified as clustered, linear, or scatter shaped, while buildings are dominated by courtyard, patio, and single-row layouts with timber structures, rammed earth or stone walls, and double-pitched roofs. After reliability and validity checks, the PLS-SEM confirms significant macro–meso–micro pathways, with the meso scale as a key mediator. Overall, the study reveals that persistence depends on long-term coupling among multi-scale factors, providing theoretical and methodological support for conservation and sustainable development.

1. Introduction

Traditional villages are important conservers of China’s agrarian civilization [1,2]. They preserve historical spatial structures, local construction traditions, and cultural memory. However, rapid urbanization has intensified regional development imbalances and has placed increasing pressure on many traditional villages, threatening their survival [3,4]. Since the 1990s, resource depletion and sustained population outmigration have accelerated village decline, while successful cases of endogenous and sustainable revitalization remain limited [5]. In response, the Chinese government has established national conservation programs to coordinate heritage protection, rural development, and modernization [6,7].

With the parallel advancement of cultural heritage protection and rural revitalization strategies, traditional villages have attracted sustained attention across multiple academic disciplines. Recent studies in sociology [3,8,9,10], geography [11,12,13,14], anthropology [15,16,17], and ecology [18,19,20,21] have progressively expanded the research scope to encompass village spatial organization, architectural form characteristics, sociocultural system dynamics, and their interactions with their ecological environments. For example, Wang et al. [11] examined the spatial distribution mechanisms of and influencing factors on traditional villages in the Jiangnan region and proposed corresponding strategies for sustainable development. Huang et al. [14] used GIS to analyze 204 ancient settlement points in the Nanxi River Basin, revealing their spatiotemporal evolution and the important role of intangible cultural heritage. Zhou et al. [22] proposed an automated approach to 3D modeling of traditional village architectural heritage in Jiangnan using deep learning and remote sensing imagery, aiming to achieve efficient and low-cost digital heritage conservation. Wang et al. [19] investigated the evolution and driving mechanisms of ecological–production–living spatial carrying capacity in tourism-oriented traditional villages, offering practical implications for sustainable tourism development. Tang et al. [23] examined the cultural transmission capacity of 16 tourism-oriented traditional villages in Beijing by constructing an evaluation index system and identifying four pathways for enhancing cultural heritage transmission. Recent methodological developments have moved traditional village research beyond literature reviews [24,25,26] and field investigations [27,28] toward data-driven and multidimensional analyses [29,30,31]. At the same time, research perspectives have shifted from isolated macro- or micro-scale studies to integrated multi-scale frameworks covering macro, meso, and micro levels [32,33]. This transition has provided a stronger basis for understanding traditional village systems and for supporting heritage conservation and rural revitalization.

As critical spatial units embodying regional historical memory and cultural identity, traditional villages reflect the evolution of human–environment relationships and represent concentrated expressions of regional social structures and ecological wisdom [7,34]. The current research on traditional villages is dominated by Geographic Information System (GIS)-based spatial analysis, with techniques such as kernel density estimation, Moran’s I, and geographical detectors being widely applied to quantify spatial patterns and identify the driving mechanisms [1,27,35,36]. Moreover, advances in digital surveying technologies have expanded architectural heritage research through the application of three-dimensional laser scanning, oblique photogrammetry, and BIM modeling, enabling fine-grained documentation of architectural morphology and systematic identification of regional characteristics [9,10,37,38,39]. In addition, multi-criteria decision-making approaches, including analytic hierarchy processes and entropy weight methods, along with system dynamics models, have been employed to evaluate the coupling relationships among village conservation, resource utilization, and tourism development [28,40,41,42,43]. Despite these methodological advances, the existing studies remain largely confined to single dimensions and lack integrated multi-scale perspectives across macro, meso, and micro levels, limiting systematic examination of the coupled interactions among the spatial, architectural, ecological, and social dimensions. Moreover, quantitative modeling and qualitative interpretation are still insufficiently integrated. This limitation has prevented the formation of a coherent analytical framework for explaining how spatial, architectural, ecological, and social factors interact across scales. A more integrated framework is therefore needed to connect empirical analysis with conservation practice.

Recent studies have increasingly shifted from broad spatial analysis to more detailed investigations of village morphology and architectural heritage [44,45]. Architectural heritage is the material foundation of village continuity. It records the spatial form, construction knowledge, cultural texture, and social memory. Nevertheless, the relationship between village-scale spatial organization and building-scale heritage characteristics remains insufficiently examined. The existing studies have also tended to focus on specific administrative areas or on regions in southwestern, central, and eastern China [6,46,47], while traditional villages in northwest China have received less attention. Southern Shaanxi therefore provides a meaningful case for examining traditional village conservation and architectural heritage renewal in a transitional geographical region [48,49].

This study takes southern Shaanxi as the research area. The region is an important demonstration zone for rural revitalization in Shaanxi Province and has made continuous progress in the protection and development of traditional villages [50]. However, compared with the Guanzhong Plain and the Loess Plateau of northern Shaanxi, systematic research on traditional villages in southern Shaanxi remains limited, especially in relation to spatial organization, architectural typologies, and construction techniques [51]. At the same time, rapid urban development has intensified the decline of architectural heritage, and increased village hollowing and the loss of traditional construction skills. To address these issues, this study develops a multi-scale eco-coupling framework to examine spatial patterns, village morphology, and architectural forms across macro, meso, and micro levels. PLS-SEM is further used to test the cross-scale linkage pathways among environment, village pattern, and architectural heritage (Figure 1).

2. Materials and Methods

2.1. Research Materials

Southern Shaanxi is located in the southern part of Shaanxi Province. It includes Hanzhong, Ankang, and Shangluo, covering 28 districts and counties with a total area of approximately 69,500 km² [52]. The region is dominated by the Qinba mountainous terrain. The Qinling Mountains form its northern boundary and separate it from the Guanzhong Plain, while the Bashan Mountains connect it with the Chu–Shu region to the south [53]. Southern Shaanxi also forms an important watershed between the Yellow River Basin and the Yangtze River Basin. Climatically, it lies in the subtropical zone and is characterized by distinct seasons, a mild and humid climate, abundant precipitation, and favorable ecological conditions [54] (Figure 2).

2.2. Data Sources

The village dataset is based on the six batches of the National List of Traditional Villages released since 2012 by the Ministry of Housing and Urban–Rural Development and other national departments. The list includes 8155 traditional villages nationwide, of which 57 are located in southern Shaanxi. In addition, Shaanxi Province has released four batches of provincial-level traditional villages since 2015, with 484 villages in total. Among them, 128 are located in southern Shaanxi, ranking first in the province. The geographic coordinates of each traditional village were extracted from Google Earth and then vectorized in ArcGIS 10.8 for data preprocessing. The digital elevation model (DEM) data for traditional villages in southern Shaanxi were obtained from the SRTM database, while additional geographic data were sourced from the National Geomatics Center of China (https://www.tianditu.gov.cn/, accessed on 20 August 2024), the Geo-Information (GIS) and Remote Sensing (RS) platform (https://www.gisrs.cn/, accessed on 1 January 2012), and the Resource and Environment Data Center of the Chinese Academy of Sciences (https://www.resdc.cn/, accessed on 20 August 2024). The socioeconomic and demographic data were primarily obtained from the China Statistical Yearbook (https://www.stats.gov.cn/sj/ndsj/, accessed on 1 January 2026) and Shaanxi Statistical Yearbook (http://tjj.shaanxi.gov.cn/tjsj/ndsj/tjnj/, accessed on 1 January 2026).

It should be clarified that the sample used in this study represents officially recognized traditional villages rather than all rural settlements in southern Shaanxi. The use of national- and provincial-level lists ensures data comparability and policy relevance, but it may also introduce a list-based selection bias. Villages that have not yet been included in official lists, especially localized and peripheral settlements with fragmentary architectural remains, local construction practices, or living heritage value, are not fully covered by the present dataset. Therefore, the spatial patterns identified in this study should be interpreted as the distribution pattern of recognized traditional village heritage resources, rather than as a complete representation of all heritage-related rural settlements in southern Shaanxi. In addition, the socioeconomic variables used in this study are mainly employed to characterize the current regional context of officially recognized traditional villages. Although statistical yearbooks provide the possibility of constructing multi-year datasets, the present study does not establish a full longitudinal time-series model. Therefore, the socioeconomic indicators should be interpreted as cross-sectional explanatory variables for identifying regional associations, rather than as evidence of long-term dynamic causality.

2.3. Research Framework

This study focuses on traditional villages in southern Shaanxi and develops a multi-scale eco-coupling analytical framework based on a macro–meso–micro hierarchy (Figure 3). The framework follows a progressive logic of spatial identification, morphological interpretation, architectural evidence extraction, and mechanism validation. At the macro scale, the natural–socioeconomic environment is examined using GIS-based spatial analysis. Village coordinates extracted from Google Earth are processed in ArcGIS to identify spatial distribution patterns, clustering characteristics, and environmental associations. At the meso scale, village spatial patterns and morphologies are analyzed through typological classification and field-based interpretation, with attention to the relationship between settlement form, road–river corridors, terrain constraints, and functional organization. At the micro scale, architectural heritage is documented through field surveys and 3D laser scanning, focusing on courtyard layout, building structure, material use, and construction details.

These three levels correspond to the three latent systems in the proposed E–P–A model: environment (E), village spatial pattern and morphology (P), and architectural heritage (A). On this basis, PLS-SEM is introduced not as an additional descriptive method, but as a validation tool for testing whether the macro-scale environment influences the architectural heritage directly and indirectly through the meso-scale village morphology. This framework therefore links qualitative architectural interpretation with quantitative path testing and provides a coherent analytical chain for explaining the ecological coupling mechanism of traditional villages.

From an ecological coupling perspective, the sustainable development of traditional villages is primarily influenced by three key factors: the natural–socioeconomic environment (E), village spatial pattern and morphology (P), and architectural heritage (A). These correspond to the three latent variables in the structural model. Since these latent variables cannot be directly observed, this study references the existing “Evaluation Index System for China’s Historical and Cultural Towns (Villages),” the “Evaluation and Recognition Index System for Traditional Villages (Trial),” and indicator systems developed in prior research for traditional village conservation and development. Considering the practical constraints of data collection from field surveys, the measurement indicators for this study are defined (

Table 1. Model observed variables.

Latent Variables	Observed Variables	Measurement Method
Natural–Socioeconomic Environment	Elevation	Represented by the elevation values of the DEM grid at the village point or village center, the elevation-based classification can be assignedis defined as follows: below 100 m = 1; 100–200 m = 2; 200–300 m = 3; 300–500 m = 4; above 500 m = 5.
	Slope	Calculate the slope rasters from the DEM data, with the slope values representing the gradient at each settlement point. Assign values based on the slope magnitude: <5° = 1; 5–10° = 2; 10–15° = 3; 15–25° = 4; >25° = 5.
	GDP	Represented by the county (or township) GDP per unit area (RMB 10,000/km2) where the village is located, assign values based on the regional statistical quantiles or natural breaks: low = 1; below average = 2; medium = 3; above average = 4; high = 5.
	Population density	Represented by the population density (people/km2) of the county (or township) where the village is located, it is similarly divided into five levels using the quantile or natural breakpoint method: extremely low = 1; low = 2; medium = 3; high = 4; extremely high = 5.
	Urbanization rate	Represented by the percentage of the county (or township) where the village is located that is classified as urban population, using data from statistical yearbooks. Assign values in tiers based on thresholds, such as 20%, 40%, 60%, and 80%, to reflect the urbanization gradient.
	Road density	Expressed as the total road length per unit area (km/km2) within a village or grid, calculated using road vector data; assign values based on the density levels, with moderate density corresponding to medium-high grades.
	Distance to water source	The Euclidean distance (km) from the village center to the nearest river is grouped and scored based on distance thresholds: <0.5 km = 5; 0.5–1 km = 4; 1–2 km = 3; 2–4 km = 2; >4 km = 1.
Village Spatial Pattern and Morphology	Village spatial density	Perform a kernel density estimation on traditional village points using software such as ArcGIS. Select the appropriate search radius and grid size to extract the kernel density values for the grid cells containing village points.
	Spatial distribution pattern	Using the local Moran’s I (local spatial autocorrelation), with villages as the spatial units, a spatial weight matrix is constructed to calculate the local I value or Z value for each village. This serves to characterize the degree of clustering or dispersion at its location.
	Plan morphology	Based on the village’s boundary contours and building distribution patterns, settlements are classified into cluster-type, linear-type, and scatter-shaped-type categories. Within the model, these classifications are represented through dummy variable coding (e.g., cluster-type = 1, others = 0) or assigned in an ordinal form according to their dominant spatial morphological characteristics.
Architectural Heritage	Courtyard spatial layout	Through field surveys and mapping, traditional courtyard houses or multi-courtyard layouts within a village are evaluated on a scale of 1 to 5. The average score is calculated as the “Courtyard Spatial Layout Index.”
	Individual building structure	Based on 3–4 structural elements—including the structural system, number of bays and depth, roof and eaves construction, and joint detailing—typical traditional dwellings within a village are rated on a scale of 0–1 or 1–5. After weighted summation and normalization, the “individual building construction index” is derived; at the village level, the average value of sampled buildings is calculated.

). To validate the aforementioned E-P-A coupling mechanism at the quantitative level, this study constructs a structural equation model based on the preceding analysis and indicator systems, and proposes the following research hypotheses:

H1: 

The natural–social environment system (E) exerts a significant effect on the village spatial pattern and morphology system (P).

H2: 

The village spatial pattern and morphology system (P) has a significant effect on the architectural heritage system (A).

H3: 

After controlling for the village spatial pattern and morphology system (P), the natural–social environment system (E) still has a significant direct effect on the architectural heritage system (A).

H4: 

The village spatial pattern and morphology system (P) plays a partial mediating role in the relationship between the natural–social environment system (E) and the architectural heritage system (A).

2.4. Research Method

To avoid treating the methods as separate procedures, this study organizes them according to the macro–meso–micro analytical framework. GIS-based methods, including kernel density estimation, geographic concentration index, spatial autocorrelation, OLS, and GWR, are used collectively to identify the macro-scale spatial patterns and environmental–socioeconomic associations. Field investigation, typological classification, and 3D laser scanning are used to interpret the meso-scale settlement morphology and micro-scale architectural heritage characteristics. PLS-SEM is then used to integrate these observed variables into a path model and test the hypothesized E–P–A coupling mechanism. Therefore, the methodological design follows a sequential logic from spatial description to driver interpretation and finally to mechanism validation.

2.4.1. Kernel Density Analysis

Kernel density estimation (KDE) is employed to characterize the spatial distribution density of traditional villages in southern Shaanxi. By treating each village location as a probabilistic contribution and aggregating these contributions across space, KDE enables the quantification of spatial clustering through the construction of a continuous probability density surface [55]. In this approach, the estimation location x and the bandwidth (h) represent the key parameters governing the smoothing process and directly influence the resulting density pattern. The kernel density function is expressed as follows:

$$\mathrm{f}\left(\mathrm{x}\right)={\displaystyle \frac{1}{\mathrm{nh}}}{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}\mathrm{k}\left({\displaystyle \frac{{\mathrm{d}}_{\mathrm{i}}}{\mathrm{n}}}\right)$$

(1)

$${\mathrm{d}}_{\mathrm{i}}=\mathrm{x}-{\mathrm{x}}_{\mathrm{i}}$$

(2)

where $\mathrm{f}\left(\mathrm{x}\right)$ denotes the kernel density estimate at a given point within the study area; n represents the total number of observations; h is the search radius (bandwidth) and $\mathrm{h} > 0$; and ${\mathrm{d}}_{\mathrm{i}}$ is the absolute distance between the estimation point and the sample point. Through iterative computation, an appropriate bandwidth is determined, and the kernel density distribution map of traditional villages in southern Shaanxi is subsequently generated.

2.4.2. Geographic Concentration Index

The geographic concentration index is employed to assess the degree of spatial concentration of traditional villages across counties [56]. The calculation formula is expressed as follows:

$$\mathrm{G}=100\times \sqrt{{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}{\left({\displaystyle \frac{{\mathrm{X}}_{\mathrm{i}}}{\mathrm{T}}}\right)}^2}$$

(3)

where $\mathrm{G}$ represents the geographic concentration index of traditional villages; n denotes the total number of counties; ${\mathrm{X}}_{\mathrm{i}}$ is the number of traditional villages in county i; and T denotes the total number of traditional villages. The value of $\mathrm{G}$ ranges from 0 to 100, with higher values indicating a more concentrated spatial distribution and lower values indicating a more scatter-shaped pattern.

2.4.3. Spatial Autocorrelation Analysis

The degree of spatial association of traditional villages within the study area is assessed using a spatial autocorrelation analysis. A global spatial autocorrelation reflects whether the spatial distribution of a given phenomenon exhibits significant clustering or dispersion across an entire region [11]. In this study, the global Moran’s I index is employed to evaluate the overall spatial autocorrelation of traditional villages in southern Shaanxi and measure the degree of similarity among spatially adjacent units. When I > 0, the spatial distribution is characterized by clustering; when I≈0, the distribution approximates randomness; and when I < 0, the distribution is scatter-shaped. The global Moran’s I is calculated as follows:

$$\mathrm{I}={\displaystyle \frac{\mathrm{N}}{{\displaystyle {\sum }_{\mathrm{i}=1}^{\mathrm{N}}}{\displaystyle {\sum }_{\mathrm{j}=1}^{\mathrm{N}}}{\mathrm{w}}_{\mathrm{ij}}}}·{\displaystyle \frac{{\displaystyle {\sum }_{\mathrm{i}=1}^{\mathrm{N}}}{\displaystyle {\sum }_{\mathrm{j}=1}^{\mathrm{N}}}{\mathrm{w}}_{\mathrm{ij}}\left({\mathrm{x}}_{\mathrm{i}}-\bar{\mathrm{x}}\right)\left({\mathrm{x}}_{\mathrm{j}}-\bar{\mathrm{x}}\right)}{{\displaystyle {\sum }_{\mathrm{i}=1}^{\mathrm{N}}}{\left({\mathrm{x}}_{\mathrm{i}}-\bar{\mathrm{x}}\right)}^2}}$$

(4)

To further identify the areas exhibiting high clustering, low clustering, and spatial outliers, a local Moran’s I analysis is conducted based on the results of global Moran’s I. The local spatial patterns are classified into four cluster types: high–high (H–H), high–low (H–L), low–high (L–H), and low–low (L–L). The local Moran’s I is expressed as follows:

$${\mathrm{I}}_{\mathrm{i}}={\displaystyle \frac{\left({\mathrm{x}}_{\mathrm{i}}-\bar{\mathrm{x}}\right)}{{\mathrm{s}}^2}}\sum _{\mathrm{j}\ne \mathrm{i}}\mathrm{N}{\mathrm{w}}_{\mathrm{ij}}\left({\mathrm{x}}_{\mathrm{j}}-\bar{\mathrm{x}}\right)$$

(5)

$${\mathrm{s}}^2={\displaystyle \frac{{\displaystyle {\sum }_{\mathrm{j}=1}^{\mathrm{n}}}{\left({\mathrm{x}}_{\mathrm{j}}-\bar{\mathrm{x}}\right)}^2}{\mathrm{n}-1}}$$

(6)

where $\mathrm{N}$ denotes the total number of counties within the study area; ${\mathrm{w}}_{\mathrm{ij}}$ represents the spatial adjacency relationship between counties i and j (where ${\mathrm{w}}_{\mathrm{ij}}$ = 1 if the two counties are adjacent, and ${\mathrm{w}}_{\mathrm{ij}}$ = 0 otherwise); ${\mathrm{x}}_{\mathrm{i}}$ and ${\mathrm{x}}_{\mathrm{j}}$ denote the number of traditional villages in counties i and j; and $\mathrm{x}$ is the mean number of traditional villages within the study area.

2.4.4. OLS and GWR

An ordinary least squares (OLS) regression is applied to examine the global relationships between the spatial distribution of traditional villages and their influencing factors. Subsequently, a Geographically Weighted Regression (GWR) is employed to extend the OLS framework by capturing the spatial variations in these relationships across different locations. By performing locally weighted regressions at each geographic location, the GWR generates a set of local regression coefficients (LRCs) that can be spatially visualized to reveal the underlying spatial heterogeneity [44]. The combined application of OLS and GWR enables a multilevel analytical perspective that bridges the overarching global trends and fine-scale local characteristics. The OLS model is expressed as follows:

$${\mathrm{y}}_{\mathrm{i}}={\mathsf{\beta }}_0+{\displaystyle \sum _{\mathrm{k}=1}^{\mathrm{p}}}{\mathsf{\beta }}_{\mathrm{k}}{\mathrm{x}}_{\mathrm{ik}}+{\mathsf{\epsilon }}_{\mathrm{i}}$$

(7)

where ${\mathrm{y}}_{\mathrm{i}}$ denotes the density of traditional villages (dependent variable); and ${\mathrm{x}}_{\mathrm{ik}}$ represents the k (k = 1,2,…,n) influencing factor of sample i, such as elevation, slope, GDP, or population density (independent variables); ${\mathsf{\beta }}_0$ is the intercept term (constant); ${\mathsf{\beta }}_{\mathrm{k}}$ is the regression coefficient corresponding to the k independent variable (constant in the global OLS model); and ${\mathsf{\epsilon }}_{\mathrm{i}}$ is the random error term.

$${\mathrm{y}}_{\mathrm{i}}={\mathsf{\beta }}_0\left({\mathrm{u}}_{\mathrm{i}},{\mathrm{v}}_{\mathrm{i}}\right)+{\displaystyle \sum _{\mathrm{k}=1}^{\mathrm{p}}}{\mathsf{\beta }}_{\mathrm{k}}\left({\mathrm{u}}_{\mathrm{i}},{\mathrm{v}}_{\mathrm{i}}\right){\mathrm{x}}_{\mathrm{ik}}+{\mathsf{\epsilon }}_{\mathrm{i}}$$

(8)

where (${\mathrm{u}}_{\mathrm{i}},{\mathrm{v}}_{\mathrm{i}}$) represents the geographic coordinates of sample point $\mathrm{i}$; and ${\mathsf{\beta }}_{\mathrm{k}}$ denotes the local regression coefficient at location (${\mathrm{u}}_{\mathrm{i}},{\mathrm{v}}_{\mathrm{i}}$), which varies spatially. All remaining symbols retain the same meaning as defined in the OLS model.

2.4.5. Structural Equation Modeling

Traditional villages constitute complex systems shaped by multiple latent variables that are not directly observable and therefore must be represented by directly measurable manifest indicators. To investigate the causal relationships among influencing factors, this study employs partial least squares structural equation modeling (PLS-SEM), within which both the measurement model and the structural model are specified in a unified framework to identify the latent constructs and their effect pathways. This approach helps elucidate the underlying mechanisms governing traditional village development and provides methodological support for promoting sustainability in traditional villages. The specific equations are as follows:

$$\mathrm{Y}={\Lambda }_{\mathrm{y}}\mathsf{\eta }+\mathsf{\epsilon }$$

(9)

$$\mathrm{X}={\Lambda }_{\mathrm{x}}\mathsf{\xi }+\mathsf{\delta }$$

(10)

where Equations (9) and (10) correspond to the exogenous and endogenous measurement models, respectively. Here, ${\Lambda }_{\mathrm{x} }$ denotes the factor loading matrix linking the observed exogenous indicators to the exogenous latent construct $\mathsf{\xi }$, and ${\Lambda }_{\mathrm{y} }$ denotes the factor loading matrix linking the observed endogenous indicators to the endogenous latent construct $\mathsf{\eta }$. The terms $\mathsf{\delta }$ and $\mathsf{\epsilon }$ represent the measurement error vectors of the respective measurement models. The structural model specifies the causal relationships between the endogenous and exogenous latent constructs. The specific equations are as follows:

$${\mathsf{\eta }}_{\mathrm{j}}=\sum _{\mathrm{k}}{\mathsf{\beta }}_{\mathrm{j}\mathrm{k}}{\mathsf{\eta }}_{\mathrm{k}}+\sum _{\mathrm{m}}{\mathsf{\gamma }}_{\mathrm{j}\mathrm{m}}{\mathsf{\xi }}_{\mathrm{m}}+{\mathsf{\zeta }}_{\mathrm{j}}$$

(11)

where ${\mathsf{\eta }}_{\mathrm{j}}$ denotes the endogenous latent variable, ${\mathsf{\eta }}_{\mathrm{k}}$ denotes other endogenous latent variables included as predictors, and ${\mathsf{\beta }}_{\mathrm{j}\mathrm{k}}$ is the structural path coefficient between the endogenous constructs. ${\mathsf{\xi }}_{\mathrm{m}}$ denotes the exogenous latent variable, ${\mathsf{\gamma }}_{\mathrm{j}\mathrm{m}}$ is the path coefficient, and ${\mathsf{\zeta }}_{\mathrm{j}}$ is the residual term.

3. Results

3.1. Spatial Distribution Patterns of Traditional Villages in Southern Shaanxi

3.1.1. Spatial Distribution Characteristics and Patterns

ArcGIS was used to visualize and analyze the spatial distribution of traditional villages in southern Shaanxi. The kernel density analysis showed that the villages are mainly concentrated in Ankang City, forming distinct high-density clusters, with the density decreasing eastward and westward to produce a pattern of “central concentration and east–west dispersion”. The primary high-density area occurs in central Ankang (Hanbin District, Shiquan County, and Hanyin County), with kernel density values of 0.010–0.014, while secondary clusters appear in southern Shangluo (Shanyang and Shangnan counties), with values of 0.0073–0.010. In contrast, villages are relatively sparse in Hanzhong, the northern and southern parts of Ankang, and northern Shangluo (0.0006–0.0051) (Figure 4). The ANN value is <1, indicating an overall clustered distribution with significant variations existing in spatial density.

3.1.2. Degree of Spatial Distribution Equity

The degree of spatial distribution equity in the traditional villages was quantified using the imbalance index. The calculated imbalance index for southern Shaanxi was S = 0.723, which approached 1. This indicated a markedly uneven spatial distribution. The geographic concentration index further reflected the degree of clustering in the spatial distribution of traditional villages. The computed value for southern Shaanxi was G = 32.064, which was substantially higher than the regional average value of G₀ = 7.14. This demonstrates a relatively high level of spatial concentration. The most densely clustered traditional villages are located in Hanbin District, followed by Shiquan County and Hanyin County (Figure 5).

3.1.3. Significant Spatial Autocorrelation

The spatial autocorrelation at both the global and local levels was examined using Moran’s I for traditional villages in southern Shaanxi. The results indicate a significant positive global spatial autocorrelation (p < 0.01; z = 3.98 > 2.58), demonstrating an overall clustered spatial distribution pattern, with Moran’s I = 0.70 (>0). The local spatial autocorrelation analysis reveals that among the 28 counties, four exhibited strong spatial clustering. Ankang City and Hanyin County were classified as H–H clusters, whereas Shanyang County and Shangnan County were identified as H–L clusters. In addition, eight counties displayed L–L clustering, including Lueyang, Liuba, Chenggu, and Zhenba in Hanzhong; Zhenping in Ankang; and Zhen’an, Zhashui, and Luonan in Shangluo (Figure 6).

However, low density in the listed village dataset should not be equated with the absence of rural heritage value. Some peripheral areas may contain unlisted villages with localized architectural features, dispersed heritage remains, or living construction traditions that require further field verification.

3.2. Factors Influencing the Spatial Distribution of Traditional Villages in Southern Shaanxi

3.2.1. Natural Environmental Factors

Natural environmental factors constitute the fundamental conditions shaping the formation and persistence of traditional village spatial patterns and exert a strong influence on village economic structures, architectural characteristics, and cultural landscapes [4]. Their effects are reflected in the following aspects (Figure 7).

(1)

Geographical factors

The elevation, slope, and aspect jointly regulate village site selection at different spatial scales. Traditional villages in southern Shaanxi have a clear preference for low-elevation environments, being predominantly distributed in valleys, basins, and hilly areas below 1000 m. The highest concentration occurs between 200 and 800 m, accounting for approximately 71.3% of all villages. In terms of slope conditions, the average slope of village sites is 11°, and approximately 95.2% of villages are located on slopes within the range of 0.5–35°. Pronounced clustering is observed within the gentle slope interval of 5–20°.

(2)

Climatic and hydrological factors

Traditional villages in southern Shaanxi are mainly distributed in areas characterized by distinct seasons, with mild and humid climates, moderate precipitation, and abundant water resources [57]. Approximately 81.6% of villages are located in zones with an annual mean temperature of 15–17 °C and annual precipitation of 700–800 mm. Spatially, villages are predominantly situated within river buffer zones. Specifically, 78.3% of traditional villages lay within 0–1 km of a water source, with the highest proportion (51.6%) concentrated within 0–0.5 km, followed by 26.7% within 0.5–1 km. Only 7.6% of villages are located more than 2.5 km from a water source. Overall, traditional villages exhibit an evident water-oriented settlement pattern, reflecting not only the requirements of domestic water use and agricultural irrigation but also the historical role of rivers as transportation corridors.

3.2.2. Socioeconomic Factors

Socioeconomic factors play a bridging role in shaping the spatial distribution patterns of traditional villages. In southern Shaanxi, the GDP level, population density, transportation conditions, and urbanization intensity are significantly associated with village distribution (Figure 8).

(1)

GDP: Traditional villages are predominantly distributed in areas with GDP values of 5–20 billion CNY, accounting for 79.2% of all villages. Within this interval, the village density shows a negative relationship with the GDP (a higher GDP corresponds to lower village density). In contrast, the village density increases sharply in areas with GDP of 30–50 billion CNY, mainly concentrated in Ankang City.

(2)

Population density: Traditional villages are mainly located in areas with population densities <500 persons/km², with 96.2% falling within the range of 50–250 persons/km². Village numbers are relatively low in both very high-density zones (>2000 persons/km²) and very low-density zones (<50 persons/km²), forming an inverted U-shaped distribution pattern.

(3)

Road density: The road density is generally higher in central areas and lower toward the periphery. Village distribution exhibits a pronounced nonlinear relationship with road density, with most villages located within 0.56–2.11 km/km² of roads.

(4)

Urbanization: Areas with low urbanization rates contain relatively large numbers of villages; the highest concentration of villages occurs within the moderate urbanization range of 0.8–1.8%. Village numbers decline sharply at higher urbanization levels (>2.0%).

3.3. Morphological Classification of Traditional Villages in Southern Shaanxi

The traditional villages in southern Shaanxi exhibit pronounced morphological diversity and strong regional adaptability. The settlement forms can be classified into three primary types—clustered, linear, and scatter shaped. Despite these differences, their spatial structures share several fundamental components: road–alley systems that organize circulation, public spaces for social interaction and collective activities, production spaces linked to agricultural practices, religious and ritual spaces that reinforce clan cohesion and local belief systems, and dwelling spaces centered on courtyard-based residential units (

Table 2. Morphological classification of traditional villages.

Morphological Type	Layout Plan	Existing Villages	Basic Characteristics
Clump shaped			• Clustered villages represent a typical concentrated settlement form, generally developing under relatively favorable natural conditions. • Their spatial structure is characterized by high compactness and strong centralization. • They enable efficient land use while reinforcing social cohesion and defensive capacity.
Line shaped			• Linear villages exhibit a transportation-oriented and watercourse-dependent settlement pattern, with pronounced directionality and structural continuity. • Their spatial organization is typically structured along roads or rivers as the primary axes, with residential units extending linearly to form a unified and orderly spatial arrangement. • This form reflects an adaptive development model shaped by the combined influence of natural geographical constraints and transportation needs.
Scatter shaped			• Scattered settlements represent a terrain-constrained scatter-shaped settlement pattern, primarily shaped by mountainous topography and natural environmental limitations. • These villages exhibit a loose spatial organization with indistinct boundaries, characterized by scatter-shaped residential clusters lacking clear centrality and social cohesion. • Significant deficiencies are evident in transportation accessibility, economic development, and the provision of public services.

).

The differentiation among these settlement types arises primarily from variations in the configuration and relative dominance of these fundamental spatial units [58]. Clustered villages feature concentrated residential and religious spaces, forming compact patterns under relatively favorable natural conditions. Linear villages typically develop along major transport routes or river corridors, unfolding longitudinally with strong directionality and continuity. Scatter-shaped villages comprise relatively independent spatial units distributed across the landscape, reflecting constraints imposed by mountainous topography and environmental conditions. Overall, these forms and spatial structures reflect adaptive strategies and diversified development patterns in response to complex terrain, as well as human–environment interactions and the spatial expression of regional cultural traditions.

By overlaying the OLS and GWR driver regimes with the mapped village morphology types, we observe a clear spatial correspondence between the local driver effects and morphology differentiation. Clump-shaped villages are more frequently distributed in areas characterized by weaker topographic constraints and stronger positive local coefficients for road density, suggesting that compact settlement forms tend to co-occur with relatively favorable buildability and higher accessibility contexts. Line-shaped villages are concentrated in zones where the hydrological factor exhibits a strong and significant positive local coefficient, indicating that a linear morphology is closely aligned with hydro-geomorphic corridors. Scatter-shaped villages show a pronounced correspondence with areas where the slope and elevation coefficients are locally significant, consistent with their predominance under more restrictive terrain conditions.

3.4. Residential Courtyard Typologies and Architectural Forms of Traditional Villages in Southern Shaanxi

The previous sections examined the spatial distribution and morphological types of traditional villages at the macro and meso scales. These spatial patterns are ultimately expressed through residential courtyards and architectural forms. Therefore, the analysis now turns to the micro scale, focusing on courtyard organization, building structure, and construction details. This step links village-scale conservation with building-scale heritage documentation and restoration [59,60,61].

It should be noted that traditional residential typologies are not merely morphological categories; they are the result of the interactions among ways of living, productive activities, family structure, and environmental adaptation. The heritage value of different housing types lies not only in their courtyard layout, structural systems, and material craftsmanship, but also in their continuing capacity to support daily life, agricultural production, ritual activities, climatic regulation, and spatial order. Therefore, a discussion of architectural heritage sustainability requires linking the typological characteristics with the original functions, current use status, and contemporary adaptability.

3.4.1. Courtyard Layout Characteristics

The courtyard layouts of traditional dwellings in southern Shaanxi show strong regional variation. They include both quadrangle courtyards, which are common in traditional Chinese architecture, and locally adapted skywell courtyards. Although these forms are generally enclosed, they differ from northern quadrangle courtyards and from the more refined skywell houses of southern China. In some areas, front rooms facing the street gradually developed into shopfront spaces, creating a distinctive mixed commercial–residential form. Constrained by local topographic conditions, the single-row courtyard layout emerged as the most prevalent form in the region, reflecting its strong adaptability to locally available materials. In stone-rich areas, dwellings were commonly constructed as slate or stone houses, whereas in bamboo-abundant regions, bamboo–timber composite structures were more frequently adopted (

Table 3. Courtyard layout types and characteristics.

Courtyard Type	Courtyard Name	Layout Plan	Current Situation	Courtyard Features
Courtyard-style residences	The Guo Family Courtyard in Zhongshan Village			• The Guo Family Compound is located in Zhongshan Village, Xunyang County, Shaanxi Province. Constructed during the mid-Qing Dynasty, it evolved from a private school into a three-courtyard compound. • The architecture follows the natural hillside contours with a north–south orientation, forming an orderly layout that integrates harmoniously with the surrounding landscape. • The stone carvings and gable wall craftsmanship exhibit high construction skill and reflect the regional characteristics and artistic values of southern Shaanxi vernacular architecture.
	The Hu Family Courtyard in Yunzhen Village			• The Hu Family Courtyard is situated in Yunzhen Village, Yungai Temple Town, Zhen’an County, Shangluo City, and was built in the late Qing Dynasty. It represents a typical three-sided courtyard residential form. • The buildings align along the street and are divided into front and rear courtyards, creating a compact and well-organized spatial layout. • The overall structure adopts a beam-and-post framework with gray tile roofing, gabled fireproof walls, and a bluestone foundation, reflecting the regional architectural style of southern Shaanxi dwellings.
	The Sun Family Courtyard in Gaoshan Village			• Gaoshan Village is located in the Qinling Mountains of Ankang and is characterized by complex mountainous terrain, with settlements predominantly distributed on hillsides and mountaintops. • The Sun Family Compound is constructed against the mountainside with a north–south orientation and a clearly defined hierarchical layout shaped by elevation differences. • The architecture employs combined post-and-lintel and beam-and-column systems, with stone extensively used for the foundations and steps, reflecting the ritual order and regional characteristics of mountain dwellings in southern Shaanxi.
Skywell (patio) residences	The Skywell in the Old Street of Manchuanguan			• Manchuanguan Ancient Town is located in Shangluo, southern Shaanxi, along the historical boundary between the Qin and Chu regions. Its ancient streets predominantly follow a front-shop, rear-residence courtyard housing pattern with central courtyards. • The residential layouts are compact and adopt hybrid brick-and-wood structural systems combining beam-and-post and post-and-lintel techniques, typically topped with gable roofs. The courtyard forms vary widely, integrating commercial and residential functions while reflecting a hierarchical spatial order and the regional architectural characteristics.
	The Liu Family Courtyard in Yunzhen Village			• Yunzhen Village is located at a transportation crossroads linking the Qin, Jin, and Shu regions, historically attracting merchant activity. Its ancient streets feature courtyard houses with shop-front spaces facing the street and residences arranged at the rear. • The Liu Family Courtyard, a Qing Dynasty relic, exhibits a symmetrical layout organized around a central skywell with four-sided drainage converging inward. • Constructed using brick-and-wood framing and rammed-earth walls, the buildings feature ornate ridge decorations that reflect the ritual order and the aesthetic characteristics of southern Shaanxi vernacular architecture.
	No. 39, Old Street, Chengguan Village			• Chengguan Village, located in Liuba County, Hanzhong, functions as a cultural node integrating Sichuan and Shaanxi traditions. Its historic streets are characterized by skywell courtyards, with shops fronting the street and residential spaces arranged behind. • Courtyard No. 39 features a narrow layout centered on a skywell connecting the front and rear sections and is primarily constructed using beam-and-post structural systems. • The gabled walls and extended eaves reflect an architectural fusion of Guanzhong and Bashu styles and demonstrate adaptation to the local environment.
Single-row residences	The Stone-Paved House of the Wang Family in Wangzhuang Village			• Wangzhuang Village is situated in the Qinba mountainous and hilly region of Hanbin District, Ankang. The dwellings are scattered along roadsides and hillsides, forming a scatter-shaped settlement pattern without distinct boundaries. • The architectural forms are dominated by stone-slab houses with simple structures and practical functions, employing post-and-lintel frameworks constructed from locally sourced stone. • Stone walls, stone tile roofs, and stone-heated beds exemplify the ingenuity of mountain dwellings in southern Shaanxi, emphasizing adaptation to local conditions and economic durability.

).

In terms of utility, courtyard-style residences were originally not single-purpose living spaces but multifunctional units integrating domestic life, production, and ritual activities. Courtyards supported daily household activities, crop drying, agricultural tool storage, temporary work, festive rituals, and neighborhood interaction, while also providing daylighting, ventilation, drainage, and a spatial order that reflected family hierarchy. Skywell residences were more common in compact street-based settlements or areas with active commercial exchange. A skywell improved daylighting and ventilation, organized rainwater drainage, and supported the front-shop–rear-residence pattern of mixed commercial and domestic use. Single-row residences were mainly responses to mountainous terrain and limited building land; their spatial organization emphasized contour adaptation, reduced construction cost, local material use, and basic residential and production-related functions.

3.4.2. Architectural Structure of Individual Buildings

(1)

Traditional dwellings in southern Shaanxi generally include entrance steps and a platform base. Three-tiered steps are most common, built of rammed earth or crushed stone and paved with gray stone slabs. Larger or higher-status residences often have seven-tiered steps for symbolic significance. Their platform bases are mainly stone: on flat terrain, low-embedded platforms with three-step entrances prevail, while in mountainous areas elevated stone bases using stepped or dry-laid masonry are widely applied to reduce moisture intrusion and enhance stability and ceremonial expression (

Table 4. Architectural structure of individual buildings and detailed construction.

Building Construction	Type Composition			Characteristics
Building fundamentals				• Residential buildings in southern Shaanxi predominantly feature square stone platforms, typically constructed using stone masonry with stepped or dry-laid techniques to enhance moisture resistance and structural stability. • These platforms commonly incorporate three-step or seven-step bluestone staircases, which function both as circulation elements and as symbolic representations of household status and auspicious meaning.
Building framework	(a) Single-projecting eaves; (b) Single-projecting eaves with corbels; (c) Double-projecting eaves.	Post-and-lintel construction	Hybrid timber construction	• The primary structural system is a pier-and-beam framework composed of interconnected columns and beams with load-bearing purlins, forming a lightweight and flexible configuration well suited to small-scale buildings and efficient material use. • Hybrid structural systems are also widespread, combining through-beam and post-and-lintel construction methods and incorporating rammed-earth walls and gable walls as needed to balance structural stability, moisture control, and fire resistance. • In some ceremonial buildings, inserted-beam structures are adopted, in which beam ends are embedded into columns; these structures are characterized by layered compositions and gently sloping roofs and are predominantly used in halls and ancestral temples.
Building walls	Exterior wall (a) Earthen wall (b) Brick wall (c) Stone wall	Gable wall (firewall) (a) Three-mountain style; (b) Five-mountain style; (c) Arch Style.		• Exterior walls are mainly constructed of rammed earth, brick, or stone masonry. Through the combined use of these three materials, distinctive regional wall forms emerge, including locally recognized features, such as “tiger-skin” walls. • Gabled firewalls are widely applied in front-shop–back-dwelling residences and commonly appear in tiered “three-peak” or “five-peak” configurations, serving both fireproofing and decorative functions and reflecting the architectural artistry of southern Shaanxi dwellings.
Building roof	Roof ridge (a) Plain ridge tile; (b) Decorative ridge tile; (c) Curved ridge tile.	Roof tile (a) Cold-laid tiles (b) Slate tiles		• Roofs in southern Shaanxi predominantly adopt gable or hip-and-gable double-sloped forms, with exposed ridge tiles being a common feature. The degree of roof ornamentation often corresponds to the social standing of the household. • Roofing materials are mainly flat tiles, small blue-gray tiles, or slate tiles, while barrel tiles are additionally employed in mountainous areas and in larger buildings.
Doors and windows	Building doors (a) Framed panel door (b) Street-facing panel door	Building windows (a) Fixed window; (b) Casement window; (c) Sash window; (d) High window.		• Framed panel doors are widely used. Elaborately decorated main entrances indicate household status, whereas removable panel doors are commonly applied in shops and residential entrances. Interior partition doors contribute to improved natural lighting and visual quality. • Window forms are diverse, including vertical-slat windows, casement windows, lattice windows, and high windows. Together, these designs achieve a balance between daylighting, ventilation, and decorative expression.

).

(2)

The wall systems comprise the exterior walls and firebreak gable walls. The exterior walls are typically rammed earth, often reinforced with brick plinths to form earth–brick composite walls; multi-courtyard residences may use hollow-brick walls or exposed brickwork. In the stone-rich mountains, rough stone or schist masonry were used to form mottled “tiger-skin walls”. Firebreak gable walls, introduced in Huizhou architecture, became common in skywell dwellings, providing fire separation and stepped street-facing elevations, often “three-peak” or “five-peak”, with painted embellishments in ancestral halls and ritual buildings.

(3)

The structural frameworks follow the timber frame tradition but vary in terms of their materials and cultural influences. Ankang and Shangluo commonly employed hybrid systems combining post-and-lintel and through-beam construction, whereas Hanzhong is dominated by the through-beam system; mixed earth–timber and brick–timber forms are widespread.

(4)

The roofs are mainly double-pitched, including flush gable and overhanging gable types, with the latter often used for skywell and courtyard residences. The ridges include plain ridges, ornamental brick ridges, and upturned ridges (plain ridges most prevalent). The roofing is dominated by flat tiles laid with cold overlay techniques using small grey tiles; slate tiles are frequent in mountainous areas.

(5)

The doors include framed plank doors, paneled doors in skywell courtyards, and interior plank or lattice-panel doors. The windows include vertical lattice, pivot lift, lattice panel, and high-set windows, with mullion design as a key ornamental element.

By overlaying the OLS- and GWR-identified driver regimes with the coded architectural form features, we observe systematic differences in the architectural configurations across spatial contexts. In zones where the hydrological coefficient is significant, the buildings more frequently present a consistent configuration of raised foundations, reinforced wall bases, and a more explicit drainage organization. In zones characterized by weaker topographic constraints and stronger road density coefficients, the buildings exhibit greater diversity in their courtyard depths and openness, with a stronger emphasis on courtyard enclosures and the organization of public spaces, and they predominantly adopt post-and-lintel construction. In zones where the slope and elevation coefficients are significant, courtyard spaces tend to be more uniform, buildings are typically constructed with through-beam construction, the use of locally available materials is more evident, and the construction detailing more frequently emphasizes structural stability and moisture protection.

3.5. Validation of Ecological Coupling Mechanisms Using PLS-SEM

PLS-SEM validation and path analysis of cross-scale evolutionary pathways: Firstly, a reliability and validity assessment of the model is conducted. In PLS-SEM, the model may include both formative and reflective indicators; their reliability is evaluated via outer weights and outer loadings, respectively. As a general rule, bootstrap-based inference is used, and the formative indicators are considered meaningful when the outer weights exceed 0.20 and are statistically significant. For the reflective indicators, LGBARIA et al. suggested a minimum acceptable loading of 0.40 as evidence of significance, whereas loadings above 0.70 typically indicate that the indicator sufficiently reflects the latent construct. In addition, multicollinearity should be assessed because high correlations among indicators can bias estimation; this is commonly evaluated using the variance inflation factor (VIF). Hair et al. proposed that a VIF < 10 is acceptable for collinearity diagnostics, while Diamantopoulos et al. argued for a stricter threshold of VIF < 3.3.

As shown in Table 2, all the formative indicators have outer weights greater than 0.20 and pass the bootstrap significance test. For the reflective indicators, all outer loadings exceed 0.70, except one loading of 0.558 (>0.40). The VIF values range from 1.000 to 3.011. Collectively, these results indicate that the measurement model demonstrates satisfactory reliability (

Table 5. Results of validity assessment and collinearity (VIF) diagnostics.

Latent Construct	Indicator (Observed Variable)	Outer Weight	Outer Loading	Outer VIF Values	p-Value
Natural–Socioeconomic Environment	Elevation	0.673	-	1.013	0.000
	Slope	0.377	-	1.013	0.004
	GDP	-	0.714	2.072	0.000
	Population density	-	0.835	1.001	0.062
	Urbanization rate	-	0.611	1.079	0.000
	Road density	-	0.874	1.079	0.037
	Distance to water source	-	0.912	1.243	0.000
Village Spatial Pattern and Morphology	Village spatial density	-	0.783	2.051	0.000
	Spatial distribution pattern	-	0.606	1.982	0.016
	Plan morphology	-	0.935	1.771	0.000
Architectural Heritage	Courtyard spatial layout	-	0.813	3.001	0.000
	Individual building structure	-	0.772	2.675	0.072

).

The structural model reveals significant cross-scale relationships among the three latent constructs. The natural–socioeconomic environment (E) has the strongest positive effect on village spatial pattern and morphology (P), with a standardized path coefficient of β = 0.472. This indicates that environmental and socioeconomic conditions play a fundamental role in shaping village distribution and morphological differentiation. The village spatial pattern and morphology (P) also has a positive effect on the architectural heritage (A), with β = 0.258, suggesting that more stable and organized settlement patterns are associated with better preservation of courtyard layouts and building structures. In addition, the natural–socioeconomic environment (E) retains a direct effect on the architectural heritage (A), with β = 0.176. These results indicate that P functions as a partial mediating layer between E and A (

Table 6. Path coefficients of the structural model.

Structural Relationship Between Latent Constructs	Path Coefficient
Natural–Socioeconomic Environment → Village Spatial Pattern and Morphology	0.472
Village Spatial Pattern and Morphology → Architectural Heritage	0.258
Natural–Socioeconomic Environment → Architectural Heritage	0.176

).

At the measurement level of the model, the latent construct natural–socioeconomic environment is mainly driven by indicators such as road density, urbanization rate, and population density; the elevation and the water system buffer provide moderate loading support, whereas the loading of slope is comparatively lower. The latent construct village spatial pattern and morphology shows the highest loading for plan morphology, followed by spatial density and distribution pattern. The latent construct architectural heritage is jointly reflected by the courtyard spatial layout and building structure, with loadings both above 0.70. Overall, the model demonstrates good theoretical consistency and statistical significance in both the structural paths and the measurement layer, providing robust empirical evidence for an eco-coupling mechanism (Figure 9).

From the perspective of E–P–A eco-coupling, the sustainable development of traditional villages is not driven by a single factor, but emerges from the long-term interaction among the macro-scale environmental base (E), meso-scale spatial pattern (P), and micro-scale architectural heritage (A) within the same regional system. The E system—through topographic and hydrological conditions, resource endowments, and gradients of accessibility and development—provides the fundamental constraints and support for site selection, habitat safety, and development boundaries, functioning as an external “ecological field” for village formation and persistence. Building on this foundation, the P system translates the environmental conditions into a concrete spatial order and morphological structure; through clustering intensity, distribution patterns, and plan form it regulates human–land tensions and land-use efficiency and serves as a key mediating link for structural stability and functional flexibility. The A system, expressed in courtyard organization, construction systems, and detailing, embodies the accumulated long-term adaptation to environmental constraints and spatial patterns, and constitutes the core carrier through which cultural continuity and local identity are maintained. Accordingly, the E–P–A coupling mechanism implies that sustainable development under contemporary transformation pressures can be achieved only by respecting the natural and socioeconomic contexts, rationally guiding and optimizing settlement spatial patterns, and simultaneously conserving and activating architectural heritage and construction systems with strong local adaptability, thereby integrating environmental safety, spatial order, and cultural continuity.

4. Discussion

The spatial patterns and architectural heritage of traditional villages are co-formed through the interaction of natural conditions, socioeconomic dynamics, and historical–cultural processes [62,63]. Southern Shaanxi was selected as the study area because it lies within a north–south transitional zone with distinctive geographical characteristics, providing an appropriate human–environment context for this research. At the same time, traditional villages in southern Shaanxi are currently facing multiple pressures, including spatial restructuring and the gradual decline of architectural heritage. In response, this study adopts a multi-scale and interdisciplinary analytical framework. From integrated macro-, meso-, and micro-level perspectives, it systematically examines the spatial distribution patterns, settlement morphology types, and evolutionary mechanisms of architectural forms. Guided by explicit research hypotheses, a PLS-SEM approach is further employed to verify the cross-scale coupling relationships and establish a coherent logical closed loop. This framework not only helps reveal the intrinsic coupling logic of traditional village systems at the theoretical level, but also provides strategic support for heritage conservation and rural revitalization in practice.

The advancement of multi-scale and interdisciplinary research has provided new perspectives and methodological approaches for the academic study of traditional villages. On this basis, the comprehensive eco-coupling analytical framework developed in this study mainly includes the following components: (1) Traditional villages in southern Shaanxi exhibit an overall clustered distribution, with high-density concentrations in Hanbin, Shiquan, and Hanyin in Ankang, while the peripheral areas remain relatively scattered. This pattern indicates that natural factors—such as terrain, elevation, slope, and proximity to water systems—provide the fundamental constraints on settlement formation, while socioeconomic variables, including population density, transport accessibility, and moderate urbanization, also significantly influence their spatial persistence and evolution. (2) The morphology of traditional villages in southern Shaanxi is diverse and can be classified into clustered, linear, and scatter-shaped types. These differences reflect both the region’s transitional north–south geographic position and its long history of multi-ethnic and multicultural interaction, representing the integration of ecological adaptation, spatial order, and cultural continuity. (3) Traditional villages in southern Shaanxi express environmental adaptation and cultural symbolism through courtyard organization and architectural detailing. From an architectural heritage perspective, this study reveals the dynamic coupling among environmental adaptation, craftsmanship transmission, and cultural expression, and demonstrates that scientific documentation, material restoration, and craft reproduction support the maintenance of authenticity and continuity in architectural heritage. (4) Building on the multi-scale ecological coupling analysis, the PLS-SEM validates the cross-scale coupling pathways and confirms the research hypotheses, demonstrating that the sustainable development of traditional villages is not driven by a single factor but results from the long-term interaction among macro-level environmental conditions, meso-level spatial patterns, and micro-level architectural heritage within the same regional system.

Compared with existing studies, the contribution of this study does not lie in replacing established GIS-based spatial analytical methods, but in reorganizing them within a cross-scale eco-coupling framework and further extending the analytical chain to the architectural heritage level. In recent years, a large body of research has employed methods such as ArcGIS [64,65], Geodetector [6,56,66], and GWR [67,68] to identify the regional spatial distribution patterns of and influencing factors on traditional villages, substantially advancing the understanding of clustering characteristics, environmental constraints, and socioeconomic associations across different provincial and river basin scales. However, most of these studies have remained focused primarily on the macro-scale distribution layer, with relatively limited efforts to systematically integrate settlement morphology and architectural heritage evidence into a unified explanatory framework.

At the meso scale, existing morphological and typological studies have provided an important foundation for settlement classification and the identification of spatial order; however, morphology has often been treated as a descriptive endpoint rather than an explanatory mediating layer. The existing research has predominantly emphasized the diversity and spatial logic of village morphologies. For example, Lin et al. [12] analyzed Jiuguan Village using facility clustering, road network analysis, and spatial form indices, demonstrating the multifaceted influence of transportation systems on historical development processes and adaptive spatial patterns. Jiang et al. [69] employed the CityEngine platform in combination with digital and 3D visualization techniques to explore the self-organizing rules governing the spatial morphology of Xiaoxi Village, thereby contributing to landscape resource integration, feature assessment, optimization strategies, and spatial management. In contrast, this study defines village morphology as a mediating structural system that receives and translates the effects of macro-scale natural–socioeconomic gradients onto the architectural level. This treatment shifts morphology from a mere typological label to a mechanism-bearing component within a cross-scale analytical logic.

At the micro scale, the existing architectural heritage studies have often focused on typological documentation, material and thermal performance evaluation, or the analysis of individual building components. For example, Wang et al. [70] combined subjective surveys with objective measurements to evaluate indoor thermal comfort, proposing improvements to enclose the thermal performance and an evaluation framework for residents in cold regions. Similarly, Xie et al. [71] employed numerical simulations to assess the effects of insulation materials, glazing types, wall thickness, insulation layer configuration, and solar utilization on the energy performance of traditional buildings in Beijing. Building on this literature, the distinct contribution of this study lies in integrating courtyard organization, structural systems, wall foot detailing, roof–eaves configurations, and drainage characteristics into a comparative architectural-response framework that can be examined across different spatial regimes and interpreted under varying driver contexts. In other words, the architectural layer is no longer treated merely as a cultural inventory, but is instead understood as a locally adaptive response system with heritage science relevance, including aspects such as moisture control, drainage performance, terrain adaptation, and maintainability.

At the methodological level, many existing studies are able to identify “influencing factors” or describe multi-scale phenomena, but relatively few explicitly verify the causal relationships among these factors. By introducing PLS-SEM to test the cross-scale coupling logic, this study advances multi-scale analysis into a testable cross-scale path model, enabling the E–P–A relationships to be evaluated in terms of direct effects, indirect effects, and mediating effects.

Building on the above comparison with existing studies, the key advance of this study lies not only in integrating evidence across the macro, meso, and micro scales, but also in transforming this integration into a mechanistic interpretation of settlement morphology types and architectural heritage. Specifically, the influencing factors identified by OLS and GWR are fed back into the analyses of morphology types and architectural forms, and the typological elements are further operationalized into performance-relevant indicators. This shifts the study from a descriptive multi-scale juxtaposition toward an evidence-bounded cross-scale explanatory chain. This transition is critical for discussing the “eco-coupling mechanism,” because it moves coupling beyond a conceptual juxtaposition and reframes it as an interpretable set of linkages through which the natural–socioeconomic environment (E) is translated into architectural heritage (A) responses via the village spatial pattern and morphology (P).

The correspondences among the E–P–A systems reveal a stepwise cross-scale coupling process from the macro-scale driving mechanisms, to the meso-scale spatial organization, and further to the micro-scale architectural forms. On the one hand, areas with weaker topographic constraints and significant road density effects are more likely to form clustered villages, whereas areas with significant hydrological characteristics tend to generate linear villages organized along water-related geomorphic corridors; by contrast, steeper-slope or higher-elevation areas are more closely associated with scatter-shaped villages. On the other hand, from the perspective of heritage science and environmental adaptation, the traditional architectural system of southern Shaanxi is not a passive response to natural conditions; rather, it is the result of a regionally specific and integrated coping mechanism centered on the humid climate, steep terrain, and flood risk. Under the humid and high-rainfall conditions, the buildings reduce erosive exposure through construction strategies, such as raised foundations, reinforced wall bases, extended eaves, and explicit drainage organization. Under the topographic constraints, the settlement and building layouts follow contour lines, reflecting a spatial logic in which terrain adaptation is prioritized. In river valleys with higher flood risk, foundation elevation and clearly organized runoff and drainage paths are used to reduce risks of scouring and water accumulation. Taken together, these two major aspects not only demonstrate environmental adaptability, but also, in heritage science terms, embody risk reduction, improved maintainability, and more robust sustainability.

It should be noted that the sustainability of traditional housing typologies does not mean that their original functions can be fully preserved. In some traditional villages, courtyard-style, skywell, and single-row residences are being replaced to varying degrees by modern brick–concrete houses, newly built detached dwellings, and street-facing commercialized buildings. This transformation is not simply a matter of changing aesthetic preference; rather, it is driven by multiple practical factors. First, timber, earth, and stone structures often require relatively high maintenance costs and are vulnerable to material aging, wall dampness, and structural damage in humid and mountainous environments. Second, traditional houses often have limitations in terms of sanitation, heating, thermal insulation, parking, privacy, and the conveniences required by modern family life. Third, population outmigration and household miniaturization have weakened the social basis that once supported multi-generational cohabitation, agricultural production, and clan rituals in traditional courtyard houses. Finally, the spread of modern materials, standardized construction, and changing local building preferences has further accelerated the replacement of traditional housing typologies.

5. Conclusions

This study constructs an E–P–A eco-coupling framework to examine how the natural–socioeconomic environment, village spatial pattern, and architectural heritage interact across scales in the traditional villages of southern Shaanxi. Rather than treating the spatial distribution, settlement morphology, and architectural form as separate descriptive layers, the study links them into a cross-scale explanatory chain. The revised conclusions are as follows.

First, the spatial distribution of traditional villages in southern Shaanxi is highly uneven and shows a pattern of central concentration and peripheral sparsity. The core clusters are located mainly in Hanbin, Shiquan, and Hanyin in Ankang, while many peripheral counties contain scattered and low-density village groups. This finding suggests that conservation planning should not adopt a uniform regional policy. Instead, central Ankang should be treated as a priority conservation area where continuous village clusters, historical settlement corridors, and architectural heritage resources can be protected as an integrated cultural landscape. In the peripheral low-density areas, conservation should focus on network-based protection, selected key villages, and the maintenance of cultural routes rather than large-scale uniform intervention.

Second, the persistence of traditional villages is closely associated with specific environmental and socioeconomic thresholds. These villages are mainly distributed below 1000 m in elevation, on slopes of 5–20°, and within 1 km of water sources. They are also concentrated in areas with medium GDP levels, relatively low population density, moderate road density, and low-to-moderate urbanization intensity. These findings provide direct implications for planning control. Water-adjacent villages should prioritize river corridor protection, drainage improvement, and flood risk management; slope-based villages should strengthen foundation stability, retaining structures, and terrain-sensitive construction control; villages close to roads should balance accessibility improvement with the prevention of excessive reconstruction; and villages under higher urbanization pressure require stricter controls over new construction, façade alteration, and the replacement of traditional dwellings.

Third, three main village morphology types—clustered, linear, and scattered—correspond to different conservation priorities. Clustered villages, usually associated with weaker topographic constraints and stronger road effects, should be protected through the maintenance of street–alley networks, public spaces, courtyard continuity, and overall settlement texture. Linear villages, often distributed along rivers or transportation corridors, should emphasize the continuity of waterfront or roadside spatial sequences, the protection of front-shop–rear-residence patterns, and the control of linear expansion. Scattered villages, which are more closely related to slope and elevation constraints, require a different strategy: protection should focus on terrain adaptation, the stabilization of individual dwellings, the conservation of local materials, and the maintenance of dispersed cultural landscape units.

Fourth, architectural heritage in southern Shaanxi should be conserved not only as a set of visual forms but also as a system of environmental adaptation and daily use. Courtyard-style residences, skywell/patio residences, and single-row dwellings have different functional logics. Courtyards originally supported family organization, drying, storage, household rituals, ventilation, and daily interactions; skywells provided daylighting, ventilation, drainage, and support for compact commercial–residential use; single-row dwellings responded to mountainous terrain, limited land, and local material availability. Therefore, adaptive reuse should retain the useful spatial functions of courtyards and skywells while improving sanitation, heating, drainage, and structural safety. Repair work should prioritize stone foundations, raised platforms, reinforced wall bases, timber structural systems, roof drainage, eaves, and locally specific wall materials, because these elements directly support moisture control, terrain adaptation, and maintainability.

Fifth, the PLS-SEM results confirm that the natural–socioeconomic environment has the strongest effect on the village spatial pattern and morphology, while the village morphology further mediates the relationship between the environmental conditions and architectural heritage. This result indicates that architectural conservation cannot be separated from the wider settlement and environmental context. In practical terms, heritage protection should proceed through an integrated sequence: identifying environmental constraints, classifying village morphology, evaluating architectural heritage conditions, and then formulating differentiated conservation measures. This sequence provides a more operational pathway than treating each village or building as an isolated heritage object.

The conclusions of this study should be understood as regional-scale findings. They are useful for identifying conservation zones, selecting representative villages, classifying housing typologies, and formulating differentiated planning strategies in southern Shaanxi. However, they do not replace smaller-scale studies of individual villages, courtyards, or buildings. Future research should incorporate household interviews, current-use surveys, building pathology records, structural safety assessments, and detailed restoration evaluations to further improve the applicability of the framework to site-specific conservation practice.

Despite these findings, the study has limitations. (1) This study focuses on 57 national-level and 128 provincial-level traditional villages in southern Shaanxi; however, many locally designated villages that are not included in the national or provincial lists also retain substantial research value. Therefore, the sample scope of this study is concentrated on representative villages rather than achieving full coverage, and some limitations remain in terms of spatial representativeness. (2) The macro-level natural and socioeconomic drivers (e.g., climate, population, road networks, and urbanization) are inherently dynamic. As a result, analyses based on cross-sectional data from a single period are subject to temporal contextual dependency, and the estimated driver effects and threshold intervals may change as data are updated. (3) The methods employed in this study rely primarily on spatial data analysis and field investigations, and thus place greater emphasis on analyzing and identifying the current conditions of traditional villages, while lacking long-term tracking of historical evolution and social dynamics. Future research will further address these limitations in order to better support the conservation and development of traditional villages.