Spatial Evolution and Driving Mechanisms of Rural Settlements in National New-Type Urbanization Pilot Areas: A Case Study of She County

Abstract

Using She County, a national new-type urbanization comprehensive pilot area, as a case study, this research develops a multi-layered “static–dynamic–driver” analytical framework based on rural settlement data. By integrating GIS spatial overlay, landscape pattern indices, average nearest neighbor analysis, kernel density estimation, and cold–hotspot analysis, the study systematically characterizes the spatiotemporal evolution and driving mechanisms of rural settlements from 1980 to 2020. The results reveal that: (1) settlement evolution exhibits distinct phase-specific patterns, encompassing four primary types of transformation: localized expansion and consolidation, individual disappearance, rapid expansion, and the emergence of new settlements with peripheral extension; (2) landscape pattern and aggregation analyses indicate continuous growth in both total area and number of settlements, accompanied by increasing irregularity and fragmentation of patches; settlement size aggregation shows a fluctuating decline followed by recovery, overall spatial clustering intensity trends upward, and high-density kernel areas shift from the central–western to the northwestern region; (3) under multi-factor interactions, settlement layouts transitioned from an early “survival–location dependent” pattern dominated by natural constraints and transportation accessibility, to a mid-stage rapid aggregation driven by economic development and public service provision, ultimately evolving into a composite pattern balancing economic drivers and ecological constraints. The findings underscore the nonlinear superimposed effects of natural environment, economic development, transportation accessibility, public service availability, and ecological carrying capacity, providing a robust scientific basis for optimizing rural settlement spatial arrangements and informing rural development policy under the context of national new-type urbanization.

1. Introduction

Rural areas continue to serve as the primary living environment for nearly half of the global population and represent essential spaces for supporting socio-economic activities, safeguarding food production, and maintaining ecological security [1]. As complex human–land systems shaped by the interactions between natural processes and anthropogenic activities, the spatial evolution of rural areas not only reflects intrinsic rural development dynamics but also embodies the transformation of urban–rural relationships, thereby offering critical insights into human–land interaction mechanisms [2,3]. Accordingly, the systematic identification of spatial distribution patterns, morphological evolution, and functional transformation pathways of rural settlements is fundamental for optimizing rural spatial organization, formulating differentiated rural development policies, and promoting coordinated urban–rural integration [4]. Furthermore, this research is closely associated with global food security, ecological sustainability, and the achievement of the Sustainable Development Goals, thus demonstrating considerable international research significance [5].

With continuous advancements in technological capabilities and data accessibility, the spatial evolution of rural settlements has gradually become a focal research topic within rural geography, land-use science, and urban–rural planning. Early investigations, constrained by limited image resolution and analytical methodologies, predominantly concentrated on village morphology, spatial texture, and basic distribution characteristics [6,7]. In recent years, however, the integration of geographic information systems (GIS), remote sensing technologies, spatial syntax analysis, and landscape pattern indices has facilitated multi-scale and multi-temporal identification of settlement evolution processes [8,9,10,11,12]. These methodological advances have contributed to the establishment of comprehensive quantitative analytical frameworks by examining spatial density, morphological structure, and agglomeration patterns, thereby providing strong technical support for exploring rural settlement transformation.

Substantial empirical evidence suggests that the formation and evolution of rural settlement patterns are influenced by a combination of natural geographical conditions, transportation accessibility, socio-economic development, policy interventions, and household behavioral decisions [13,14,15]. At broader spatial scales, topographic and geomorphological characteristics determine settlement suitability and spatial carrying capacity, while transportation infrastructure affects production efficiency and village development vitality. Meanwhile, policy and economic factors frequently induce rapid and significant adjustments in settlement configurations over relatively short periods [16,17,18]. Nevertheless, due to regional differences in resource endowment, urbanization trajectories, demographic change, and rural development strategies, rural settlement evolution exhibits pronounced spatial heterogeneity across regions [19,20]. Therefore, systematically identifying region-specific driving mechanisms within distinct geographical contexts is essential for comprehensively understanding rural spatial restructuring processes [21,22].

Notably, under the national strategy of new-type urbanization, the spatial pattern of rural settlements is influenced by the superimposed effects of multiple factors such as population flow, industrial transformation, land consolidation, and rural construction, presenting complex characteristics of scale change, morphological restructuring, and functional reshaping [16,23,24]. Although existing studies have made certain progress in spatial pattern identification and driving factor analysis, several limitations still exist: First, most studies focus on a single scale or method, lacking a comprehensive analytical framework for “pattern evolution–morphological characteristics–driving mechanism” of rural settlements. Second, the identification of driving factors mostly remains at the level of correlation analysis, without systematic characterization of the interactive effects of multiple factors and their spatial differentiation. Third, empirical studies targeting typical pilot areas under the new-type urbanization policy are still relatively insufficient, which limits the policy interpretability and applicability of the research conclusions.

To bridge the above gaps, this paper takes She County, a national comprehensive pilot zone for new-type urbanization, as the study area, and constructs a multi-level comprehensive analysis framework of “static pattern–dynamic evolution–driving factors”. Compared with existing studies, the main contributions of this paper are threefold: (1) From the research perspective, this study integrates spatial pattern, morphological characteristics and evolutionary process to systematically reveal the multi-dimensional changes in rural settlements from “distribution–agglomeration–evolution”; (2) In terms of methodology, this study integrates multiple approaches including GIS spatial overlay analysis, landscape pattern indices, average nearest neighbor index, kernel density analysis and hot spot analysis to achieve multi-scale and multi-dimensional spatial pattern identification; combined with the GeoDetector model, it quantitatively characterizes the multi-factor driving mechanism and their interactive effects, effectively alleviating the potential multicollinearity problem in traditional methods; (3) In terms of research object, this study focuses on the new-type urbanization pilot area, exploring the spatial restructuring process of rural settlements under policy intervention, providing a typical case support for understanding the transformation of urban-rural relations.

On this basis, this paper establishes a technical route of “pattern identification–evolution analysis–driving mechanism interpretation”: firstly, identify the spatial distribution pattern and evolution characteristics of rural settlements in She County from 1980 to 2020 based on multi-period remote sensing data; secondly, depict the morphological structure and agglomeration characteristics using landscape indices and spatial statistical methods; finally, combine the GeoDetector model to analyze the driving effects and interaction mechanisms of natural, location and socio-economic factors on the spatial evolution of settlements.

Based on the above analytical framework, this paper aims to: (1) Systematically identify the spatiotemporal evolution characteristics of rural settlements in She County from 1980 to 2020; (2) Quantitatively resolve the main driving factors and action mechanisms of its spatial evolution; (3) Propose spatial optimization strategies for rural revitalization and urban–rural integrated development. The findings can provide a scientific basis for rural spatial governance and resource allocation in mountainous counties under the background of new-type urbanization.

2. Study Area and Data Sources

2.1. Study Area

She County (36°17′ N–36°55′ N, 113°26′ E–114°00′ E) is located in Handan City, in the southwestern part of Hebei Province. Situated at the eastern foot of the Taihang Mountains and at the junction of Shanxi, Hebei, and Henan provinces, it serves as a pivotal transition zone connecting the Beijing–Tianjin–Hebei region with the Central Plains (Figure 1). Designated as one of the third batch of National New-type Urbanization Comprehensive Pilot Areas in 2016, the county holds significant policy implications for county-level urban–rural spatial restructuring, land system innovation, and the equalization of public services. The landscape is dominated by deep mountainous terrain, with the spurs of the Taihang Mountains traversing the entire territory. The elevation ranges from 203 to 1562.9 m, sloping from northwest to southeast, and is characterized by fragmented topography, dense valleys, and scattered basins. She County experiences a warm temperate semi-humid continental monsoon climate, with an average annual precipitation of approximately 571.7 mm. As of 2021, the county had a total population of 433,500 and an urbanization rate of 64.03%. The regional Gross Domestic Product (GDP) reached 20.39 billion CNY, while the per capita disposable incomes for urban and rural residents were 28,840 CNY and 17,416 CNY, respectively [16].

She County was selected as the study area primarily based on four considerations. First, as a designated National New-type Urbanization Comprehensive Pilot Area, the county has undertaken extensive exploratory practices in household registration system reform, land factor allocation, and urban–rural integrated development, thereby offering a distinct institutional context and policy advantage. Second, the region’s typical mountainous and hilly topography imposes rigid constraints on village layout and settlement evolution, making it an ideal case for identifying general patterns of settlement dynamics in mountainous counties. Third, driven by the coordinated development of the Beijing–Tianjin–Hebei region and regional industrial transfer, the county has experienced accelerated urbanization. This process has triggered significant shifts in the scale, structure, and spatial organization of rural settlements, providing an exemplary sample for examining settlement restructuring in rapidly transforming regions. Finally, characterized by multifunctionality—including agricultural production, ecological conservation, mining development, and cultural tourism—the spatial morphology of rural settlements in She County exhibits high complexity and representativeness, holding substantial value for both theoretical and practical research.

2.2. Data Sources

The data used in this study are categorized into physical geographical data and socio-economic data (

Table 1. Data sources and descriptions.

Category	Data Type and Source	Resolution	Time Period
Physical geographical data	Land use data (https://www.resdc.cn/)	30 m	1980–2020
	DEM data (https://www.resdc.cn/)	30 m	-
	Water body data (https://www.resdc.cn/)	-	-
	Annual precipitation data (https://www.resdc.cn/)	1000 m	1980–2020
	Mean annual temperature data (https://www.resdc.cn/)	1000 m	1980–2020
	NDVI (https://www.resdc.cn/)	1000 m	1980–2020
Socio-economic data	Population densit (https://www.resdc.cn/)	1000 m	1995–2020
	GDP (https://www.resdc.cn/)	1000 m	1995–2020
	Railways (https://www.resdc.cn/)	-	1995–2020
	Roads (https://www.resdc.cn/)	-	1995–2020
	POI (Scenic spots, Commercial facilities, Schools, Clinics) (https://www.resdc.cn/)	-	2005–2020

). The physical geographical data include land use, Digital Elevation Model (DEM), water bodies, annual precipitation, mean annual temperature, and Normalized Difference Vegetation Index (NDVI), which were used to characterize the regional terrain, climate, and ecological conditions. The socio-economic data comprise population density, GDP, railways, roads, and Points of Interest (POI), utilized to reflect population distribution, economic development, transport accessibility, and living service facilities. Regarding data processing, patches of cultivated land, woodland, grassland, water bodies, urban land, rural settlements, and other construction land were extracted by reclassifying the current land use data. Slope data were derived from the DEM, while the POI data encompassed scenic spots, commercial facilities, schools, and clinics. Additionally, the spatial distances from rural settlements to cultivated land, water bodies, scenic spots, commercial facilities, clinics, schools, city centers, and roads (highways and railways) were calculated using the Proximity tools in ArcGIS. These calculations provided the foundational data support for the subsequent spatial analysis and the identification of driving mechanisms.

3. Methodology

To systematically reveal the evolutionary characteristics of rural settlement spatial patterns in She County, this study constructed a “Static-Dynamic-Driving” multi-layer analysis framework. First, the static analysis utilizes the Average Nearest Neighbor (ANN) Index to characterize the overall distribution and aggregation features of settlements. It employs Kernel Density Estimation (KDE) and Cold/Hot Spot Analysis to identify spatial agglomeration centers and density gradients, while integrating landscape pattern indices to assess the fragmentation and diversity of the spatial structure across different periods. Second, the dynamic analysis, based on the superposition of multi-temporal data, identifies types of settlement changes—including expansion, newly built, disappearance, and merger. This component reveals evolutionary modes and their spatiotemporal heterogeneity, reflecting the dynamic development process of rural settlements. Third, the driving force analysis applies the Geodetector method to quantitatively assess key factors and their interactions influencing settlement distribution, covering physical geography, transport accessibility, public services, and socio-economic conditions. This step aims to elucidate the primary mechanisms underlying spatiotemporal evolution. Progressing from static characterization to dynamic analysis and finally to the revelation of driving mechanisms, this framework establishes a systematic and comprehensive methodological system for studying rural settlement spatial patterns.

3.1. Analysis of Static Spatial Pattern Characteristics

3.1.1. Identification of Distribution Morphology and Aggregation Characteristics of Rural Settlements Based on Average Nearest Neighbor (ANN)

The Average Nearest Neighbor (ANN) is a classical method for identifying the spatial distribution patterns of point features, which is suitable for characterizing the spatial agglomeration characteristics of rural settlement patches [25]. Its basic principle is to compare the observed average nearest neighbor distance with the expected average nearest neighbor distance under random distribution, so as to determine the type of spatial pattern.

Specifically, first, point features are extracted based on the centroids of rural settlement patches. The distance from each patch centroid to its nearest neighboring patch centroid is calculated, and the average nearest neighbor distance is obtained. Second, assuming random distribution of point features, the expected nearest neighbor distance is calculated according to the area of the study area (represented by the minimum bounding rectangle) and the number of patches. On this basis, the ANN index is computed as the ratio of the observed distance to the expected distance, with the formula as follows:

$$\mathrm{A}\mathrm{N}\mathrm{N}={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{d}}_{\mathrm{i}}/\mathrm{m}}{\frac{\sqrt{\mathrm{n}/\mathrm{R}}}{2}}}$$

(1)

where ${\mathrm{d}}_{\mathrm{i}}$ is the distance from the centroid of the i-th rural settlement patch to its nearest neighboring patch centroid (unit: meter), n is the total number of patches, and R is the area of the minimum bounding rectangle occupied by all rural settlements in the study area (unit: square meter). When ANN = 1, the spatial distribution is random; when ANN < 1, rural settlements show a clustered distribution; when ANN > 1, they exhibit a dispersed distribution.

Furthermore, to further determine the significance of the spatial distribution type, a standardized Z-score is introduced for significance testing [25]. The Z-score is calculated by comparing the difference between the observed nearest neighbor distance and the expected value, and then it is compared with the standard normal distribution. The formula is as follows:

$$Z={\displaystyle \frac{\overline{d_l}-E\left(d\right)}{\sqrt{\mathrm{var}\left(\overline{d_l}-E\left(d\right)\right)}}}$$

(2)

When the Z-score is less than −1.96, rural settlements are significantly clustered; when the Z-score is greater than 1.96, they are significantly dispersed; when the Z-score ranges between −1.96 and 1.96, there is no significant difference between their spatial distribution and random distribution.

The above analysis was implemented using the ArcGIS 10.8 software platform.

3.1.2. Characterization of Spatial Aggregation Centers and Gradient Patterns Based on Kernel Density and Cold/Hot Spot Analysis

(1)

Kernel Density Estimation (KDE)

Kernel Density Estimation (KDE) is an effective spatial analysis method designed to calculate the unit density of point or line features within a specific neighborhood [26]. As the kernel density value increases, the distribution density of these features rises correspondingly. Consequently, this method demonstrates significant advantages in detecting spatial hotspots or identifying spatial clustering of rural settlements. By employing KDE, it is possible to precisely identify the spatial clustering locations of high-value or low-value features, thereby intuitively revealing the spatial heterogeneity in the density of rural settlements. The calculation formula for this method is as follows:

$$\mathrm{f}\left(\mathrm{x},\mathrm{y}\right)={\displaystyle \frac{1}{{\mathrm{n}\mathrm{h}}^2}}{\sum }_{\mathrm{i}=1}^{\mathrm{n}}\mathrm{k}\left({\displaystyle \frac{{\mathrm{d}}_{\mathrm{i}}}{\mathrm{n}}}\right)$$

(3)

This formula characterizes the estimated kernel density value at point (x,y), where $\mathrm{h}$ represents the smoothing parameter or bandwidth, K denotes the kernel function, and di indicates the distance between point (x,y) and the i-th observed point. The bandwidth parameter h has a significant impact on the kernel density results. This study determined the optimal bandwidth through multi-scheme comparison by combining the scale of the study area and the characteristics of sample distribution, ensuring the expression of the overall trend while taking into account the local agglomeration characteristics.

(2)

Cold/Hot Spot Analysis

This study employs a spatial association measurement model to analyze the spatial aggregation characteristics of rural settlement scale in She County. At the global level, the Getis-Ord General G statistic is utilized to examine the overall spatial association features. The calculation formula for the global clustering statistic, $\mathrm{G}\left(\mathrm{d}\right)$, is as follows:

$$\mathrm{G}(\mathrm{d})={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i}\mathrm{j}}(\mathrm{d}){\mathrm{x}}_{\mathrm{i}}{\mathrm{x}}_{\mathrm{j}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{x}}_{\mathrm{i}}{\mathrm{x}}_{\mathrm{j}}}}$$

(4)

The standardized statistic is expressed as:

$$\mathrm{Z}\left(\mathrm{G}\right)={\displaystyle \frac{\mathrm{G}\left(\mathrm{d}\right)-\mathrm{E}\left[\mathrm{G}\left(\mathrm{d}\right)\right]}{\sqrt{\mathrm{Var}\left[\mathrm{G}\left(\mathrm{d}\right)\right]}}}$$

(5)

where ${\mathrm{x}}_{\mathrm{i}}$ and ${\mathrm{x}}_{\mathrm{j}}$ represent the attribute values of rural settlement scale for spatial units i and j (obtained by converting settlement raster data to polygons and assigning values); ${\mathrm{w}}_{\mathrm{i}\mathrm{j}}$(d) denotes the spatial weight constructed based on a given distance threshold d, typically employing binary adjacency or a distance decay function; E[G(d)] and Var[G(d)] denote the expectation and variance, respectively, under the random null hypothesis (assuming attributes are randomly distributed in space); and Z(G) is the standardized statistic of spatial association used for significance testing.

Significance Determination: When Z(G) > 1.96 (with p < 0.05), it indicates a significant clustering of high values at the global scale; when Z(G) < −1.96 (with p < 0.05), it indicates a significant clustering of low values; if |ZG| ≤ 1.96, the spatial association is considered statistically insignificant. Furthermore, empirical p-values obtained via Monte Carlo permutation tests in ArcGIS are utilized to enhance the robustness of the significance testing.

Since the global spatial association index characterizes the overall distribution pattern of the study object using a single value, it fails to detect association patterns within specific local areas. To address this limitation, the local spatial association index is capable of effectively identifying “hotspots” as well as spatial heterogeneity. Therefore, this study further employs the local spatial association index (Getis-Ord Gi*) to identify hotspots and cold spots of rural settlement scale. The calculation formula is as follows:

$${\mathrm{G}}_{\mathrm{i}}^{\ast }\left(\mathrm{d}\right)={\displaystyle \frac{\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}\left(\mathrm{d}\right) {\mathrm{x}}_{\mathrm{j}}-\overline{\mathrm{x}}{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}\left(\mathrm{d}\right)}{\sqrt{\frac{{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}^2\left(\mathrm{d}\right)-{\left({\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}\left(\mathrm{d}\right)\right)}^2}{\mathrm{n}-1}}}}$$

(6)

where ${\mathrm{x}}_{\mathrm{j}}$ is the scale attribute value of spatial unit j; $\overline{\mathrm{x}}$ represents the global mean of the scale attribute values; s denotes the standard deviation of the scale attribute values; and other symbols retain their previously defined meanings.

Decision Rule: When Gi* is a significantly positive value (e.g., Z > 1.96 and p < 0.05), spatial unit i and its neighborhood form a high-value cluster (Hot Spot). Conversely, when Gi* is a significantly negative value (e.g., Z < −1.96 and p < 0.05), they constitute a low-value cluster (Cold Spot). If |Z| ≤ 1.96, the area exhibits no significant spatial aggregation characteristics.

3.1.3. Assessment of Spatial Structure Fragmentation and Diversity Based on Landscape Pattern Indices

Landscape pattern indices play a crucial role in studying the evolution of land use patterns, as they can deeply reveal the structural characteristics and spatial configuration of the landscape [27,28]. To comprehensively and specifically reflect the spatial layout and scale evolution of rural settlements, this study selected indices covering three aspects: density, area, and shape, to characterize the evolutionary features of rural settlement spatial patterns. All indices selected in this paper were calculated using the Fragstats 4.2 software package, based on land use data of rural settlements. The definitions and calculation formulas for each index are presented in

Table 2. Landscape pattern indices and their descriptions.

Primary Index	Secondary Index	Description	Formula	Note
Density and Differentiation Indices	Number of Patches (NP)	The total number of patches of a specific landscape type.	$\mathrm{N}\mathrm{P}={\mathrm{n}}_{\mathrm{i}}$	${\mathrm{n}}_{\mathrm{i}}$ represents the number of patches of a specific type in the landscape (unit: count).
	Mean Patch Size (MPS)	Represents the average condition and indicates the degree of landscape fragmentation. A smaller MPS value implies a more dispersed patch type and higher fragmentation.	$\mathrm{M}\mathrm{P}\mathrm{S}={\displaystyle \frac{\mathrm{C}\mathrm{A}}{\mathrm{N}\mathrm{P}}}$	CA is the area in hm2;NP is the total number of patches.
Area Indices	Class Area (CA)	Reflects the scale of a specific patch type in the landscape; serves as the basis for calculating other indices.	$\mathrm{C}\mathrm{A}={\displaystyle \sum _{\mathrm{j}=1}^{\mathrm{n}}}{\mathrm{a}}_{\mathrm{i}\mathrm{j}}\times {\displaystyle \frac{1}{1000}}$	${\mathrm{a}}_{\mathrm{i}\mathrm{j}}$ is the area of patch ij (unit: m2), with CA ≥ 0. The converted unit for CA is hm2.
	Largest Patch Index (LPI)	Identifies the dominant patch type in the landscape; indirectly reflects the direction and magnitude of human disturbance.	$\mathrm{L}\mathrm{P}\mathrm{I}={\displaystyle \frac{\mathrm{a}}{\mathrm{C}\mathrm{A}}}$	a is the area of the largest patch of a specific type (unit: hm2);CA is the total area of that patch type (unit: hm2).
Shape Index	Landscape Shape Index (LSI)	Reflects the irregularity or complexity of patches. A larger LSI value indicates more irregular and elongated patch shapes.	$\mathrm{L}\mathrm{S}\mathrm{I}={\displaystyle \frac{0.25{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{c}}_{\mathrm{i}}}{\sqrt{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{a}}_{\mathrm{i}}}}}$	${\mathrm{c}}_{\mathrm{i}}$ is the perimeter of the i-th patch (unit: m); ${\mathrm{a}}_{\mathrm{i}}$ is the area of the i-th patch (unit: m2).

.

3.2. Analysis of Dynamic Spatial Pattern Evolution

To systematically reveal the spatiotemporal evolutionary characteristics of rural settlements in She County, this study introduced a spatial overlay identification method. Based on the vector data of rural settlements from 1980, 1990, 2000, 2010, and 2020, the Spatial Overlay tool in ArcGIS was utilized to perform period-by-period superposition comparisons of settlement patches between adjacent years. This process finely characterized the changes in the spatial location of settlements over the time series. Furthermore, by integrating characteristics such as patch area variation, land use conversion in surrounding areas, and the distance between adjacent patches, the evolution types of settlements were identified and categorized into four groups: Spatial Expansion Type, Spatial Consolidation Type, Spatial Decline Type, and Spatial Emergence Type. This method intuitively reflects dynamic processes including expansion, merger, degradation, and new growth, providing key technical support for revealing the phasic evolutionary characteristics of the spatiotemporal patterns of rural settlements in She County.

3.3. Identification of Driving Factors for Rural Settlement Spatial Patterns

3.3.1. Selection of Driving Factors

In the identification of driving factors, to ensure scientific rigor and logical consistency, an indicator system was constructed based on the dominant mechanisms governing the formation and evolution of rural settlement patterns. The driving factors were categorized into three groups: Location & Accessibility, Socio-economic Conditions, and Natural Environment Conditions (

Table 3. Evaluation indicator system for driving factors of rural settlement spatial distribution in She County.

Category	Indicator	Function/Mechanism
Location & Accessibility	X1: Distance to cultivated land	Influences settlement expansion and location choice; optimizes production efficiency.
	X2: Distance to water bodies	Constraints related to agricultural production and convenience of daily life.
	X3: Distance to scenic spots	Proximity to scenic spots provides more economic opportunities (e.g., tourism).
	X4: Distance to commercial facilities	Enhances living convenience and agglomeration of economic activities.
	X9: Distance to town centers	Proximity to urban centers attracts more settlements.
	X10: Distance to roads	Improves transport connectivity and logistics; promotes agglomeration.
	X11: Distance to railways	Enhances long-distance transport accessibility; influences layout.
Socio-economic Conditions	X5: Population size	Population concentration drives the agglomeration of economic and social resources.
	X6: GDP	Regions with a strong economic foundation attract settlements, affecting density and scale.
	X7: Distance to clinics	Living convenience and accessibility to medical services influence layout.
	X8: Distance to schools	Educational resources attract family settlement; influences spatial clustering.
Natural Environment Conditions	X12: Slope	Steep slopes are unsuitable for construction and agriculture, limiting spatial expansion.
	X13: Elevation	Climate and production conditions in high-altitude areas constrain settlement layout.
	X14: Annual precipitation	Influences agricultural production and water supply; constrains distribution.
	X15: Mean annual temperature	Agricultural and residential suitability conditions influence spatial layout.
	X16: NDVI	Vegetation cover and ecological quality affect production and living conditions.

). Specifically, Location & Accessibility factors characterize the spatial relationships between settlements and key locational elements—such as cultivated land, water bodies, urban centers, and transport facilities—reflecting the guiding and constraining effects of locational conditions on settlement agglomeration and expansion [16]. Socio-economic Conditions factors, centering on population size, economic development level, and accessibility to public services, reveal the driving effects of production activity agglomeration and improved living convenience on the adjustment of settlement layouts [29]. Natural Environment Conditions factors represent the foundational constraints imposed by ecological carrying capacity and development suitability on the spatial distribution of settlements, focusing on aspects such as terrain, elevation, and climate-vegetation conditions [21].

3.3.2. Geodetector Analysis

GeoDetector is a statistical method for identifying spatial heterogeneity and its driving forces, which can quantitatively evaluate the explanatory power of each factor on spatial heterogeneity and its interactive effects. Based on the Geographical Detector package in R 4.3.1 software, this study adopted the factor detector and interaction detector to analyze the driving mechanism of spatial differentiation of rural settlements [30]. The graded kernel density results of rural settlements were taken as the dependent variable (Y), and various driving factors were taken as independent variables (X). The q-statistic was used to measure the explanatory power of a single factor, and a larger q-value indicates a stronger explanatory power for the spatial distribution. On this basis, the interaction detector was applied to analyze the change in explanatory power after the superposition of multiple factors, and identify the synergistic enhancement or independent action relationship between factors.

Compared with traditional regression models, the GeoDetector does not need to meet the assumptions of normal distribution and linearity. It can effectively reduce the impact of multicollinearity on results when multiple factors are highly correlated, showing good explanatory robustness, which is suitable for analyzing the driving mechanism of complex human-earth system.

The calculation formula of q-statistic is as follows:

$$q=\left(N{\sigma }^2-{\displaystyle \sum _{h=1}^L{N}_h{\sigma }_h^2}\right)/N{\sigma }^2$$

(7)

where q (0 ≤ q ≤ 1) represents the explanatory power of the factor for spatial differentiation of rural settlements; $\mathrm{L}$ denotes the number of strata; $N_h$ and $N$ refer to the number of samples in the $\mathrm{h}$-th stratum and the whole region, respectively; ${\sigma }_h^2$ and ${\sigma }_h$ represent the variance of the $h$-th stratum and the whole region respectively. The closer the $q$-value is to 1, the stronger the explanatory power.

4. Research Results

4.1. Spatial and Scale Agglomeration Characteristics

From the perspective of point distribution pattern, rural settlements in Shexian County always showed significant spatial agglomeration characteristics from 1980 to 2020. The Average Nearest Neighbor Index (ANN) was less than 1 in all periods, and the Z-values were all less than −1.96. The ANN changed from 0.78 in 1980 to 0.80 in 1990 and 0.81 in 2000, then decreased to 0.73 in 2010 and 0.70 in 2020, showing a trend of “first increasing and then decreasing”, indicating that the degree of spatial agglomeration increased significantly after 2010 (Figure 2).

Kernel density analysis shows that the overall spatial distribution intensity of rural settlements increased, and the scope of high-value areas continued to expand. From 1980 to 2000, high-value areas were mainly concentrated along the Qingzhang River and around towns; after 2010, high-value areas obviously migrated to the northwest; by 2020, high-value areas were further concentrated, and the spatial agglomeration pattern was significantly strengthened (Figure 3).

The results of hot-spot analysis show that the scale pattern of settlements presented a phased change of “agglomeration-dispersion-re-agglomeration”. From 1980 to 2000, the Z-value of the global Getis-Ord G statistic rose from −0.66 to −0.12 (p < 0.05), showing significant agglomeration; from 2000 to 2010, the Z-value rose to −0.01 (p = 0.11), with insignificant spatial agglomeration and a dispersed pattern; from 2010 to 2020, the Z-value dropped to −0.35 (p = 0.03), and the agglomeration effect strengthened again, forming an axial hot-spot structure along “Shecheng Town-Jingdian Town-Gengle Town-Guxin Town”, while the scope of cold-spot areas in mountainous regions expanded (Figure 4).

Overall, the spatial pattern of rural settlements in Shexian County is characterized by coexistence of phased fluctuations in agglomeration intensity and overall enhancement, reflecting the superimposed effect of physical geographical constraints and urbanization process.

4.2. Evolution of Landscape Patterns

From 1980 to 2020, both the total area and patch number of rural settlements in Shexian County increased continuously, and the degree of landscape fragmentation intensified. Over the 40 years, the total area increased by 1448.19 hectares, and the number of patches increased by 89. In stages, the growth was relatively slow from 1980 to 2000, with the area increasing by 190.62 hectares and 64.89 hectares, respectively, and the average annual growth rate was lower than 1%; it entered a rapid expansion stage from 2000 to 2010, with the area increasing by 870.93 hectares, the average annual growth rate reaching 2.66%, and the number of patches increasing by 5.70% annually; the growth slowed down from 2010 to 2020, with the area increasing by 321.75 hectares and the average annual growth rate dropping to 0.78% (

Table 4. Landscape Pattern Indices of Rural Settlements in She County, 1980–2020.

Index		Year
Primary Index	Secondary Index	1980	1990	2000	2010	2020
Density and Differentiation Indices	Number of Patches (NP)	152.00	151.00	149.00	234.00	241.00
	Mean annual change rate		−0.07%	−0.13%	5.70%	0.30%
	Mean Patch Size (MPS) (ha)	19.89	21.29	22.01	17.74	18.56
Area Indices	Class Area (CA) (ha)	3023.73	3214.35	3279.24	4150.17	4471.92
	Mean annual change rate		0.63%	0.20%	2.66%	0.78%
	Largest Patch Index (LPI) (%)	7.30	6.88	7.41	3.54	3.68
Shape Index	Landscape Shape Index (LSI)	15.22	15.27	15.08	24.61	22.65

).

Landscape morphology and structure indicators further indicate that rural settlements show a significant trend of irregularity and fragmentation. The Landscape Shape Index (LSI) rose from 15.22 in 1980 to 22.65 in 2020, indicating that patch morphology tended to be complex; the Mean Patch Size (MPS) decreased from 19.89 to 18.56, and the Largest Patch Index (LPI) decreased from 7.30 to 3.68, reflecting that while scale expanded, patch segmentation enhanced and the overall structure tended to be dispersed.

Overall, the evolution of rural settlement landscape pattern in Shexian County presents a synergistic change characteristic of “accelerated expansion (2000–2010)—morphological complexity—structural fragmentation”.

4.3. Evolutionary Patterns of Rural Settlements

From 1980 to 2020, rural settlements in Shexian County mainly experienced four types of evolution: expansion, merger, disappearance and new addition, with obvious phased characteristics (Figure 5). From 1980 to 1990, settlements were dominated by local individual expansion and small-scale integration, with a steady increase in total quantity; from 1990 to 2000, local expansion and disappearance coexisted, and the spatial pattern showed a trend of decentralized adjustment. From 2000 to 2010, large-scale individual expansion and local disappearance coexisted, with a rapid growth in quantity, reflecting the significant promoting effect of urbanization promotion and land consolidation policies on spatial layout. From 2010 to 2020, expansion tended to be localized, while new settlements increased significantly, and the overall spatial distribution became more balanced. Overall, the evolution of rural settlements in Shexian County presents a dynamic process of “local expansion—merger regulation—local disappearance—new supplementation”, showing the continuous impact of population agglomeration, land use and infrastructure development on the spatial pattern of settlements, and reflecting the regularity of phased adjustment of spatial structure over time.

4.4. Driving Factors Analysis

Based on Geodetector analysis, the driving mechanism of spatial pattern evolution of rural settlements in Shexian County from 1980 to 2020 shows significant phased characteristics, with various factors playing a leading role alternately in different stages (Figure 6). Spatial location and accessibility factors played a leading role in the early stage. From 1980 to 1990, settlement layout was highly dependent on traffic accessibility, with the average distance to railways having the highest explanatory power (q-value 0.302). Meanwhile, climatic conditions remained an important constraint, with the q-value of annual average precipitation being 0.281. From 1990 to 2000, the impact of traffic location decreased slightly (q-value 0.242), while the role of economic development and public service factors enhanced, and the explanatory power of GDP and distance to schools rose simultaneously, indicating that settlements tended to gather in economically active areas and educational resource clusters. In the 2010 stage, climatic constraints regained dominance, with the q-value of annual average temperature being 0.144, while the role of traffic and commercial accessibility factors weakened significantly. By 2020, socio-economic factors strengthened again, with the explanatory power of GDP rising to 0.222, and the explanatory power of cultivated land and commercial service-related indicators rebounded simultaneously, reflecting that production convenience and economic development became important driving forces for settlement agglomeration and expansion, while the constraints of natural conditions tended to be stable.

Multi-factor interaction detection further reveals that the evolution of settlement spatial pattern is not dominated by a single factor, but the result of the synergistic effect of multiple factors, and the interaction structure evolves dynamically over time (Figure 7). In 1980, the interaction between educational facility accessibility and climatic conditions was the most significant; in 1990, highway accessibility and slope synergistically enhanced; in 2000, the driving force of various factors tended to be balanced, the cross-system interaction of economic, traffic and topographic factors significantly enhanced, and the driving mechanism changed from single dominance to multi-factor compound. In 2010, the focus of interaction shifted to urban location factors, reflecting the development trend of settlements gathering to central towns and attaching to traffic corridors; in 2020, the synergistic effect of traffic location and topographic conditions enhanced again, while the interaction between economic factors and multiple factors remained at a high level, forming an overall comprehensive driving pattern of multi-factor coupling and mutual reinforcement.

Overall, the driving mechanism of settlement evolution presents an alternating dominant mode of “natural conditions—traffic location—socio-economy—natural conditions—socio-economy”, with the interaction of single and multiple factors working together, reflecting the dynamic regulation of location, ecological and socio-economic factors.

5. Discussion

5.1. Driving Mechanism Affecting the Spatial Distribution of Rural Settlements

The formation and evolution of the spatial pattern of rural settlements in Shexian County are the result of the synergistic effect of three factors: spatial location and accessibility, socio-economic conditions and natural environment constraints. There are significant differences in the dominant factors at different stages, and the overall evolution process is from the dominance of natural and location constraints to the enhancement of socio-economic driving forces.

In the early stage of evolution (1980–1990), the settlement layout was mainly restricted by the joint constraints of spatial location and natural environment conditions. In terms of traffic accessibility, the explanatory power of the average distance to railways and the average distance to town centers was 0.302 and 0.256, respectively; in terms of natural environment, the q-values of annual average precipitation in 1980 and 1990 were 0.281 and 0.283, respectively, indicating that hydrothermal conditions had a sustained restrictive effect on settlement site selection. In this stage, settlements were distributed along traffic corridors and water sources, and the spatial pattern was characterized by the coexistence of local expansion and integration, mainly concentrated in Ping’an Subdistrict, Henandian Town, Shentou Township and Xida Town (Figure 5a), reflecting the evolutionary feature of “synergistic dominance of location dependence and ecological constraints”.

The period from 1990 to 2000 was a stage where socio-economic factors gradually intervened. With the development of township enterprises and the increase in residents’ income, the level of economic development and the accessibility of public services were significantly enhanced, and the explanatory power of GDP and distance to schools continued to rise, promoting settlements to gather in areas with good economic foundation and perfect public services [16]. At the same time, spatial location factors still played a basic role, while natural environment constraints were relatively weakened. The spatial pattern was characterized by the coexistence of local expansion and disappearance. Individual expansion was mainly concentrated in the central part of Suobao Town, the western part of Jingdian Town, the western part of Gengle Town and the central part of Guxin Town, while local disappearance occurred in the central part of Ping’an Subdistrict and Shecheng Town (Figure 5b), reflecting the transformation of settlement layout from “natural constraint dominance” to “economic and service-driven”.

From 2000 to 2010, settlements entered a stage of rapid expansion, with the parallel effect of socio-economic driving forces and natural environment constraints. On the one hand, economic growth and increased housing demand promoted the expansion of settlements in the whole region; on the other hand, the role of ecological constraints was enhanced, and the regulatory effect of natural factors such as temperature, precipitation and terrain on the spatial pattern was improved (the q-value of relevant factors rose to 0.144). The spatial pattern was characterized by the coexistence of expansion and local disappearance, mainly distributed in Shecheng Town, Jingdian Town, Gengle Town, Henandian Town and Shentou Township (Figure 5c), reflecting the composite mechanism of “economic driving-ecological constraints-spatial optimization”.

From 2010 to 2020, the evolution of settlements entered a stage of comprehensive regulation and functional restructuring. Socio-economic factors became the dominant driving force, the explanatory power of GDP increased to 0.222, and the accessibility of commercial services and public facilities was simultaneously enhanced, promoting settlements to gather in areas with obvious location advantages. At the same time, the constraints of the natural environment tended to be stable, the ecological threshold formed a basic constraint on spatial expansion, and the traffic location continued to play a guiding role. The spatial pattern was characterized by the coexistence of local expansion and new addition. Expansion mainly occurred in the central part of Xirong Town and the central part of Suobao Town, and new settlements were concentrated in Henandian Town, Shentou Township and Jingdian Town (Figure 5d), reflecting the transformation of settlements from a single residential function to a compound pattern of residence and industry.

Multi-factor interaction analysis further shows that the evolution of settlement spatial pattern is not dominated by a single factor, but the result of the synergistic effect of the three types of factors. In the early stage, it was dominated by the synergy of traffic accessibility and climatic conditions; in the middle stage, it was driven by the superposition of economic development, public services and traffic conditions; in the later stage, it reflected the synergistic regulation of socio-economic factors and ecological constraints. Overall, under the long-term constraints of the natural environment, the spatial evolution of rural settlements in Shexian County is supported by spatial location, driven by socio-economic development and population flow, and the coupling of multiple factors promotes the gradual evolution of settlements from scattered layout to relative agglomeration and from single function to compound function.

5.2. Limitations and Applicability of the Study

Within the established research framework, this paper constructs a relatively complete analysis chain from spatial pattern identification and evolution process characterization to driving mechanism analysis, and realizes the comprehensive application of multi-source data and multi-methods at the county scale. The research conclusions have good logical consistency and explanatory rationality on the whole. However, there are still some limitations due to the constraints of research scale, data accuracy and methodological conditions.

Firstly, in terms of research scale, this paper takes the county as the basic unit, which is difficult to characterize the spatial differences within villages and at the micro scale, and the applicability of the conclusions at a finer scale needs to be verified. Secondly, in terms of data, the accuracy of some early data is limited, resulting in slight uncertainty in the characterization of local spatial characteristics. Thirdly, at the methodological level, although the Geodetector can identify multi-factor interactions, it has limitations in explaining causal relationships, and the conclusions need to be further verified by combining other models.

In addition, this paper only conducts analysis based on a single county, and the promotion of conclusions is constrained by differences in regional resource endowments and urbanization processes, which are only applicable to areas with similar natural and socio-economic conditions. Future research can further test and expand the conclusions on the basis of multi-region comparison and multi-scale analysis.

5.3. Policy Implications

Research shows that the spatial evolution of rural settlements in Shexian County is a comprehensive result driven by socio-economic development, improved traffic accessibility and differential allocation of public service facilities under the long-term constraints of physical and geographical conditions. Therefore, facing the development practice of new-type urbanization comprehensive pilot areas, we should focus on optimizing urban–rural spatial organization and resource element allocation, and comprehensively promote the layout optimization of settlements and rural function restructuring [31].

In terms of cultivated land resource protection and production space guarantee, the red line of cultivated land protection should be strictly implemented, the systems of occupation-compensation balance and use control of new construction land should be strengthened to prevent the encroachment of high-quality cultivated land by the disorderly expansion of settlements [32,33]. Meanwhile, combined with the spatial endowment differences of cultivated land, water resources and tourism resources, characteristic agriculture and rural tourism should be developed in suitable areas to promote industrial diversification, enhance the employment absorption capacity of rural areas, and thus improve the coordination between population and industrial space [34,35].

In terms of public service and infrastructure allocation, the hierarchical layout of medical, educational and other public service facilities should be optimized according to the agglomeration pattern of settlements and population change trends, so as to improve service accessibility, focus on improving areas with insufficient service supply, and narrow urban–rural and regional differences [36]. Meanwhile, by improving the trunk highway and village-road network system, traffic accessibility should be enhanced, the radiation driving role of central towns should be strengthened, and settlements should be guided to gather orderly along traffic corridors and around central towns, so as to promote the coordinated layout of population, industry and public services [37].

In terms of ecological security control, focusing on ecologically sensitive areas and important ecological function areas, the disorderly expansion of settlements should be strictly controlled to promote the coordination between spatial development and ecological protection. For settlements located in areas with strong ecological constraints, the coordinated promotion of ecological restoration and improvement of living environment can be realized through reasonable guidance of relocation and centralized resettlement [38,39,40].

Overall, on the basis of respecting the constraints of the natural environment, guided by the optimization of traffic location and centered on socio-economic development and public service improvement, we should promote the transformation of rural settlements from scattered layout to moderate agglomeration, and facilitate the optimization and coordinated functional development of the urban–rural spatial structure in the county.

6. Conclusions

Through a systematic analysis of the 40-year evolution of rural settlements in She County, this study elucidates the overarching patterns of spatial transformation and the underlying driving mechanisms. First, rural settlements exhibit pronounced phase-specific evolutionary characteristics. During the early stage, characterized by a limited economic base, settlement layouts were predominantly constrained by basic transportation accessibility and natural environmental conditions. With the advancement of socio-economic development and the enhancement of public service facilities, settlements progressively aggregated in areas offering higher economic activity and greater convenience for daily life. In recent years, the interplay between economic growth and ecological constraints has guided settlement distribution toward regions that optimize locational advantages and service accessibility while maintaining ecological carrying capacity. Second, landscape pattern analysis reveals that the sustained increase in settlement number and total area has been accompanied by heightened irregularity and fragmentation of settlement forms, indicating ongoing spatial restructuring amid rapid expansion. Aggregation analysis further demonstrates that settlement size and density evolved in a fluctuating manner, with centers of spatial clustering shifting in response to changing economic activities and transportation conditions. Finally, multi-factor interaction analysis indicates that the spatial organization of rural settlements is driven by the nonlinear superposition of production, livelihood, and ecological factors. No single factor can fully account for the observed patterns, emphasizing the integrated influence of socio-economic development, infrastructure accessibility, and ecological constraints in directing spatial evolution.

Overall, the evolution of rural settlements in She County reflects a progressive “natural constraint–economic driving–ecological regulation” pattern, illustrating the dynamic balance among development forces, livelihood needs, and environmental capacity. These findings provide an empirical foundation for informed regional planning, rural development strategies, and land management practices.