Characteristics and Influencing Factors of Spatiotemporal Distribution of Rural Houses Construction Development in Mountainous Villages of China (1980–2019): A Case Study of Qingyuan Town

Abstract

Rural house is a fundamental component of rural settlements, and understanding its construction and development characteristics is crucial for rural land use and development planning. This paper focuses on the spatiotemporal characteristics and influencing factors of Rural Houses Construction Development (RHCD) from 1980 to 2019 with a case study of Qingyuan Town in China. Based on the literature review and filed research, a set of evaluation indicators for RHCD was established. The article calculates RHCD indicators from temporal and spatial dimensions, uses the location entropy method to demonstrate the spatial distribution of indicators, and classifies the RHCD type of 14 villages in Qingyuan Town using clustering algorithms. It also analyzes the influencing factors of spatiotemporal distribution. The results show that the RHCD in Qingyuan Town exhibits typical characteristics of mountainous areas and aligns with the development trends of rural society in China. Population growth, geographical location, and economic development are the primary driving factors for the quantity indicator (Qi), while economic growth, construction technology, industrial development, and policy adjustments are the key factors influencing the form indicator (Fi). In future policy-making, greater emphasis should be placed on optimizing development strategies, improving data and monitoring systems, and integrating administrative strength with actual development needs.

1. Introduction

House and land represent the primary assets of farmers, with rural house intricately linked to farmers’ livelihoods and production methods, serving as a significant manifestation of unique regional cultures [1]. Studies on rural house have predominantly concentrated on its spatial functionality evolution [2], transitions in materials and structures [3], and the embedded social and cultural meanings [4].

Since the Reform and Opening-up began in 1978, China’s rural area has witnessed significant socio-economic shifts, catalyzing a substantial increase in rural house construction. Contrasted with the centralized urban house construction model, rural house often shows characteristics such as family-based decision-making, self-building, and self-organization. This autonomy has led to a varied and complex evolution in the structural, material, spatial, and locational aspects of rural house [5].

Rural house construction is influenced by political, economic, social, and cultural factors [6]. The policies and mores of the national and local government are inevitably and effectively generating outcomes in the rural houses market [7], and economic growth and enhanced farmer incomes have primarily fueled the rural house construction surge [8]. Also, some studies attempt to understand and explain the complexity, individual differences, and dynamic effects of historical inheritance in rural changes from a purely theoretical perspective [9]. Simultaneously, industrialization and lifestyle shifts have encouraged the adoption of modern, easier-to-maintain house types, gradually supplanting traditional construction methods and living patterns [10,11]. Furthermore, urbanization has prompted significant rural depopulation and reductions in family sizes, resulting in increasing abandonment and vacancy of rural houses [12,13]. Although the reform era has improved rural living conditions, the over-construction has led to land resource wastage, ecological degradation, and a dilution of traditional cultural elements [14,15,16]. Thus, properly planned rural house policies and management strategies are essential for fostering orderly and healthy development in rural house, encompassing the establishment of construction and land use standards, and providing construction subsidies and financial support [17,18,19,20]. Moreover, the crucial roles of rural planning and architectural design in preserving traditional culture, enhancing house quality, and promoting sustainable development have been emphasized repeatedly in academic research [21,22,23]. In addition, in the context of globalization and urbanization, the existing problems in rural areas are complex and multidimensional. Issues such as environmental pollution and treatment [24], rural landscape protection [25], the change and renewal in the morphology of traditional villages [26], and the participation of new villagers and social reconstruction [27] have also attracted widespread concern and discussion of different disciplines. The contemporary rural construction needs more systematic and refined policies.

Rural House Construction Development (RHCD) is intrinsically linked to economy and social development, land resource management, housing construction, and settlement space planning (Figure 1). Although most studies predominantly adopt a macroscopic perspective on land use, there is a noticeable deficit in detailed analyses at the micro-level concerning individual architectural buildings.

This paper, therefore, concentrates on 14 villages within Qingyuan Town in Fujian Province of China, analyzing RHCD through the established set of evaluation indicators, which contains quantity indicators and architectural form indicators. Those indicators are calculated from the temporal aspect (from 1980 to 2019) and spatial aspect (14 villages). (Figure 2).

This approach provides a comprehensive display of the RHCD in Qingyuan Town. Ultimately, using cluster analysis to categorize villages and comparing the social, economic, and geographic transportation aspects of each clustered village, the key factors influencing RHCD are analyzed. This study reveals the general patterns of RHCD in China’s mountainous rural area after the Reform and Opening-up, offering important references for guiding the development of mountainous rural areas in the future.

2. Literature Review

2.1. Rural House Construction Development in China

Since China’s Reform and Opening-up commenced in 1978, its rural areas have witnessed profound transformations across economic, social, and cultural dimensions [28,29,30]. Initially, rural lifestyles and production methods evolved [31,32,33], unleashing a rapid release of pent-up demand for house improvements. This led to an unprecedented surge in rural house construction between 1978 and 19921. Starting in the early 2000s, the prevalent lax land management and unplanned house construction highlighted the issues of disorganization within rural settlements. Concurrently, the increasing disparity between urban and rural living standards significantly drove rural residents towards urban centers [34].

In 2005, China introduced the “Socialist New Countryside2” initiative aimed at fostering balanced urban–rural development. This initiative saw the widespread construction of house areas, improvement of basic infrastructure, and large-scale renovations of unsafe rural house, alongside the progressive enhancement of the legislative framework governing rural house construction. Nevertheless, this period’s top-down approach somewhat overlooked the nurturing of rural social culture and the conservation of traditional rural aesthetics. In response, the Ministry of Agriculture initiated the “Beautiful Countryside” campaign in 2013, focusing on rural social culture and the preservation of rural landscapes.

The Rural Revitalization Strategy launched in 2017 prioritized cultural revitalization and significantly intensified the urban contribution to rural development. This shift in policy facilitated substantial investment in rural areas, emphasizing the preservation of traditional rural landscapes and architectures and sparking a boom in rural tourism [35]. This development particularly benefited mountainous settlements which, due to their geographical isolation, had remained largely unaltered by urbanization and industrialization. These areas, now tourism focal points, experienced rapid socio-economic advancement, and both rural construction and the modernization of rural house have been propelled forward.

2.2. Mountainous Rural Settlements and Land Use

Settlements refer to the concentrated living spaces within specific geographic areas where social and life organizations form, involving issues discussed by disciplines such as geography, anthropology, economics, sociology, planning, and architecture. The spatial configuration of mountainous rural settlements is heavily influenced by natural geographical factors like terrain, climate, and water resources, which collectively dictate their location, size, and form [36,37,38]. Moreover, the spatial distribution of these settlements often follows certain patterns, such as river alignments and mountain adaptations [39], with the architecture within these settlements seamlessly integrating with the terrain to exploit mountain slopes and create harmonious, livable spaces [40,41]. Given their geographical constraints, mountainous settlements often maintain well-preserved local cultural resources, including architectural styles, religious practices, and traditional crafts. Preserving and passing down these cultural heritages are pivotal for maintaining local identity and enhancing cultural diversity [42,43,44,45].

Since 1978, the accelerated social changes in China have led rural settlements through a dynamic, multi-scaled, and mixed restructuring process [36,46]. Urbanization has notably reshaped rural settlements [47,48,49,50], with rural hollowing-out becoming a focal point of research [51,52,53,54].

In rural areas, farmland and residential land are essential for rural development and land use transformation [55]. The coexistence of rural hollowing-out and expansion issues in new house construction has resulted in significant rural land wastage and severe damage to arable land resources, prompting scholars to analyze changes in settlement patterns and explore the immense potential for rural land consolidation [56,57,58]. Rational, intensive land use strategies in mountainous areas are crucial for advancing sustainable rural development [59,60,61].

2.3. Quantitative Methods in Rural Built Environment

The built environment in rural areas, which includes both houses and their surrounding areas, has been subject to many analytical approaches. The study of the built environment can be categorized into three scales: building scale, settlement scale, and regional scale.

At the scale of buildings, the focus is primarily on architectural features and forms within the discipline of architecture. Scholars constructed the morphological indicators of the building to evaluate the form evolution. For example, Wang designed the indicators system with the building area, rotation, and distance between buildings to describe the form of rural settlement characteristics [40].

At the settlement scale, the kernel density estimation, average nearest neighbor, and standard deviation ellipse are common algorithms or indicators used to describe certain variables of the settlement [62]. Indicators such as boundary shape, spatial structure, and architectural order are used to describe the morphological characteristics of the settlement. The space syntax model proposed by Bill in 1984 has been widely applied to analyze the internal spatial structure of settlements [63,64,65]. More complexly, when analyzing the relationships between different characteristics of the settlement, various correlation analysis methods are commonly used, such as Pearson Correlation Analysis and Spearman Correlation Analysis [66]. Beyond static analysis, quantitative methods also show a trend towards dynamic analysis. Theobald used cellular automata to predict the development of rural land use [67], which can also be seen as a shift in research methodology from inductive to deductive reasoning.

At the regional scale, the relationships between different areas within the study range become the focus of research. From a typological perspective, a series of clustering methods, such as K-means, are applied to classify settlement types within the region [68]. Spatial autocorrelation analysis is used to study the correlation and degree of correlation of variables within a spatial range [69]. For variables of different magnitudes, the location entropy algorithm is applied to express them on the same scale, thus more clearly demonstrating the differences between regions [70]. Moreover, Wang and others have proposed the Geodetector method, which seeks correlating factors based on the premise of spatial distribution similarity between independent variables and dependent variables [71].

3. Research Subjects and Methods

3.1. Scope of Research

Qingyuan Town in Shouning County, located in the Jiufeng Mountains of northeastern Fujian Province (Figure 3), is chosen as the subject of this study for the following primary reasons:

(1)

Qingyuan Town exemplifies a typical mountainous settlement, with approximately 90% of its area comprising mountainous landscapes characterized by complex and fragmented terrain and deep ravines.

(2)

The per capita land area in Qingyuan Town is merely about 55 m², significantly below the national average3, which means the contradiction of high population density with limited land is even more prominent. Therefore, the land use pattern needs urgent transformation.

(3)

In recent years, the Jiufeng mountain region has been under the focal point during the rural revitalization and the preservation of traditional villages, with considerable policy support and financial investment in these rural areas.

Below is detailed information on Qingyuan Town:

Qingyuan Town is approximately 8 km from the county city and is considered a suburban town. It spans 10 km from east to west and 10.8 km from north to south, with a total area of roughly 78.4 km². It administers 16 administrative villages and 81 natural villages, encompassing over 4560 households and a population exceeding 19,000 (Figure 4).

3.2. Indicators for RHCD

This study defines the RHCD by two main aspects: the quantity changes and the form changes. Accordingly, two primary sets of indicators are set: quantitative indicators and form indicators (

Table 1. Indicator system for Rural House Construction Development.

Primary Indicator	Secondary Indicator	Definition	Unit
Quantity Indicator (QI)	Number of Houses by Decade (Ni)	Total number of houses existed in decade i in current conditions	-
	Growth Rate of Houses by Decade (Gi)	Growth rate of houses in decade i	%
Form Indicator (FI)	Number of Floors (NF)	Average Number of Floors of houses	-
	Total Floor Area (TFA)	Average Area per Floor of houses	m2
	Average Area per Floor (AAF)	Average Area per Floor of houses, which reflects the area of the residential land	m2
	Proportion of Concrete Structures (PCS)	Proportion of brick-concrete structure houses, which reflects the impact of industrialization on house construction mode	%

).

3.3. Data Collection and Preprocessing

In mountainous rural areas of China, it is common that historical data of houses’ construction are lacking. Therefore, only current data can be accessed, which poses challenges to understanding the trends in RHCD. Unlike urban areas, houses built in rural areas since the 1980s are seldom demolished; thus, it is assumed that the number of existing houses from each decade after the 1980s represents the net increase in houses for that period. For example, if there are 100 houses from the 1990s in the current houses stock, it is assumed that the increase for that decade is 100 houses. Based on this reality, we can estimate the growth rates of houses in different decades for the Qingyuan area through surveys and calculations.

The current data primarily come from the Shouning County Housing and Construction Bureau and field surveys conducted by our research team, covering 100% of the rural residences in the Qingyuan area. The data collected for each house include the following: year of construction, structure type, area, and number of floors. During data processing, it was found that the data for Houyang Village and Cunwei Village among the 14 villages had anomalies, so they were not included in the study. Also, since Qingyuan Village and Waiwei Village, as well as Yangwei Village and Daiyang Village, were planned as one sector in government planning, the data for these two sets of villages were merged for analysis.

The growth rates of houses construction by (Gi) for different decades are inferred from the current status of residences and through sampling statistics. The specific methodology is as follows:

To compute the growth rate of houses in the 1980s for a given spatial area, the formula is the following:

G₁₉₈₀ = N₁₉₈₀/A₁₉₇₉

Here, G₁₉₈₀ represents the growth rate of houses during the 1980s, N₁₉₈₀ denotes the number of houses constructed during the 1980s, and A₁₉₇₉ is the total houses number in 1979. Similarly, the growth rates for subsequent decades, the 1990s, 2000s, and 2010s, are calculated as follows:

G₁₉₉₀ = N₁₉₉₀/A₁₉₈₉

G₂₀₀₀ = N₂₀₀₀/A₁₉₉₉

G₂₀₁₀ = N₂₀₁₀/A₂₀₀₉

The data for N₁₉₈₀, N₁₉₉₀, N₂₀₀₀, and N₂₀₁₀ are based on current data, whereas data for A₁₉₇₉, A₁₉₈₉, A₁₉₉₉, and A₂₀₀₉ must be estimated due to the unknown number of demolished houses in each village over time.

To develop an effective estimation method, the research team conducted a sampling survey in three strategically selected villages within Qingyuan Village, Sanwangyang Village, and Yushangang Village. These villages represent three types of towns based on China’s rural village system4, respectively, providing a representative sample. The team gathered data on the number of houses built before 1979 (N <₁₉₈₀), and the count of such houses demolished each decade thereafter (d₁₉₈₀, d₁₉₉₀, d₂₀₀₀, d₂₀₁₀) through interviews and on-site observations (

Table 2. Data from sampled villages.

Village	Sampled Area	N <1980	d1980	d1990	d2000	d2010
Qingyuan		109	2	4	5	15
Sanwangyang		68	1	3	3	9
Yushangang		62	0	2	4	7

).

Additionally, the dataset includes the number of houses constructed before the 1980s, labeled as N₁₉₈₀. Assuming that the demolition ratio of older buildings across each village aligns with those observed in our sample, we can estimate the house stock for previous years as follows:

A₁₉₇₉ = N₁₉₈₀ × r₁₉₇₉

A₁₉₈₉ = N₁₉₈₀ × r₁₉₈₉

A₁₉₉₉ = N₁₉₈₀ × r₁₉₉₉

A₂₀₀₉ = N₁₉₈₀ × r₂₀₀₉

Here, ‘r’ represents the restoration coefficient, calculated as the following:

r₁₉₇₉ = [N₁₉₈₀ + (d₁₉₈₀ + d₁₉₉₀ + d₂₀₀₀ + d₂₀₁₀)]/N₁₉₈₀

r₁₉₈₉ = [N₁₉₈₀ + (d₁₉₉₀ + d₂₀₀₀ + d₂₀₁₀)]/N₁₉₈₀

r₁₉₉₉ = [N₁₉₈₀ + (d₂₀₀₀ + d₂₀₁₀)]/N₁₉₈₀

r₂₀₀₉ = [N₁₉₈₀ + d₂₀₁₀]/N₁₉₈₀

Using averaged data from the three sampled areas, we adopt the following values for ‘r’ (

Table 3. Value of ‘r’.

r1979	r1989	r1999	r2009
1.228	1.217	1.179	1.128

).

These coefficients facilitate the calculation of the houses’ growth rates for each decade, as detailed below:

G₁₉₈₀ = N₁₉₈₀/(N₁₉₈₀ × r₁₉₇₉)

G₁₉₉₀ = N₁₉₉₀/(N₁₉₈₀ × r₁₉₈₉)

G₂₀₀₀ = N₂₀₀₀/(N₁₉₈₀ × r₁₉₉₉)

G₂₀₁₀ = N₂₀₁₀/(N₁₉₈₀ × r₂₀₀₉)

3.4. Methods of Data Analysis

The research primarily consists of three parts for analyzing RHCD data, including temporal distribution analysis (Section 4.1), spatial distribution analysis (Section 4.2), and cluster analysis (Section 4.3). Each part corresponds to different data analysis methods, as detailed below:

3.4.1. Method for Temporal Distribution Analysis

To analyze the temporal distribution of RHCD, the primary focus is on qualitative analysis based on data. The QI and FI data for Qingyuan Town are averaged over decades, and line graphs are plotted to demonstrate the trend of changes.

3.4.2. Method for Spatial Distribution Analysis

To analyze the spatial distribution of RHCD, the average values of the QI and FI indicators for each village are calculated for the period from 1980 to 2019 and exhibited on maps. For each indicator, different colors are used to denote the numerical ranges of villages on the map, facilitating a more intuitive display of spatial distribution differences.

Due to the diverse dimensions of the indicator, the study employed the method of location entropy to standardize the absolute values of each indicator, thereby bringing them to the same scale. The specific methodology is as follows:

Location entropy (LE) is an analytical metric in geographic and regional studies that measures the concentration of a specific feature or activity at a particular location relative to a larger area. It is employed to evaluate whether the spatial distribution of a variable is uniform or tends to be concentrated in certain regions.

The formula for calculating location entropy is as follows:

$$L{E}_i=\frac{p_{ij}}{p_j}$$

Here, p_ij denotes the proportion of a feature or activity j in area i, and p_j is the total proportion of feature j across the entire study area. The values of location entropy indicate the following:

LE_i = 1: The distribution in the area mirrors that of the whole study area, indicating the average concentration.

LE_i > 1: Higher than average concentration in area i, suggesting clustering.

LE_i < 1: Lower than average concentration in area i, indicating sparsity.

3.4.3. Method for Cluster Analysis

K-Means, a prevalent unsupervised learning technique in data analysis, was initially proposed by Stuart Lloyd in 1957 for signal processing and later applied to data classification by James MacQueen in 1967. Its primary aim is to organize data points into K clusters where intra-cluster similarity is high and inter-cluster similarity is low, revealing underlying patterns within the data. Cluster quality is typically assessed by calculating the mean squared distance from each data point to its cluster center, expressed by the following:

$$\mathrm{D}=\frac{1}{N}{\sum }_{i=1}^N{\parallel x_i-{\mu }_k\parallel }^2$$

where x_i represents a data point, μ_k the centroid of its cluster, and N the total number of points. In architecture, clustering helps identify and analyze groups of buildings sharing similar features, facilitating insights into architectural characteristics and trends across different regions or types. This research leverages cluster analysis on rural house construction indices to synthesize development patterns and examine their distinctive and common features.

4. Results and Analysis

4.1. Temporal Distribution of RHCD in Qingyuan Town

4.1.1. Number of Houses by Decade (Ni)

Most of the existing residences in Qingyuan Town were constructed after China’s Reform and Opening-up policy was implemented (i.e., in the 1980s and later). Residences built in the 1970s and earlier account for only 22.9% of the total. Subsequently, the proportion of residences built in each decade increased from 11.5% in the 1980s to 25.2% in the 2010s (

Table 4. Distribution of construction year of existing houses.

Construction Year	Quantity	Proportion
before 1949	169	4.2%
1950s	253	6.2%
1960s	190	4.7%
1970s	318	7.8%
1980s	467	11.5%
1990s	709	17.5%
2000s	924	22.8%
2010s	1023	25.2%
Total	4053	100.0%

).

4.1.2. Temporal Distribution of Gi

Utilizing the methodology outlined in Section 3.3, this study calculated the houses’ growth rates for Qingyuan Town across different decades (refer to Table 4). Since the 1980s, the peak of house construction growth occurred in the 1990s, after which it has been in decline, dropping from 44.3% in the 1990s to 32.5% in the 2010s. The growth rates during the 1980s and the 2000s were relatively consistent, at 40.9% and 40.7%, respectively (Figure 5,

Table 5. GI by decades of Qingyuan Town.

G1980	G1990	G2000	G2010
40.9%	44.3%	40.7%	32.5%

).

4.1.3. Temporal Distribution of FI

This section explores the evolution of house form indicators (Fi) over the decades from 1980 to 2019, focusing on secondary Fi indicators: Number of Floors (NF), Average Area per Floor (AAF), Total Floor Area (TFA), and Proportion of Concrete Structures (PCS) (

Table 6. FI by decades in Qingyuan Town.

Time Period	NF	TFA (m2)	AAF (m2)	PCS (%)
1970s and before	1.97	103.72	199.13	15.6
1980s	2.14	83.58	175.03	37.0
1990s	2.64	72.40	181.19	77.3
2000s	2.83	75.23	208.86	89.2
2010s	2.92	76.38	205.83	88.9

NF: Number of Floors; TFA: Total Floor Area; AAF: Average Area per Floor; PCS: Proportion of Concrete Structures.

, Figure 6).

The NF has progressively increased, rising from 1.97 in the 1970s or earlier to 2.92 in the 2010s, with a deceleration in the rate of increase starting in the 1980s.

Furthermore, the AAF per residence initially experienced a sharp decline, followed by a modest recovery. It decreased significantly from 103.72 m² prior to the 1980s to 72.40 m² in the 1990s, followed by a slight rebound to 76.38 m² in the 2010s.

The TFA has shown variability, with a notable decrease in the 1980s, a gradual increase through the 2000s, and a slight decline in the 2010s.

Regarding structural forms, PCS has markedly risen, moving from a lesser-used form to the predominant method of construction, remaining at around 90% since the 2000s.

4.2. Spatial Distribution of RHCD in Qingyuan Town

4.2.1. Spatial Distribution of Ni

Based on the analysis of data from 12 sampled administrative villages in Qingyuan Town, these villages are divided into two groups based on the quantity of houses. The first group includes villages with more houses (exceeding the average of 337.75): Qingyuan Village + Waiwei Village, Yangwei Village + Daiyang Village, Sanwangyang Village, and Tongyang Village.

The second group consists of villages with fewer houses (below the average of 337.75 units): Yeyangpu Village, Guiyang Village, Zhuping Village, Jiaolin Village, Yushangang Village, Fujiadang Village, Pingyan Village, and Shaotuo Village (Figure 7).

4.2.2. Spatial Distribution of Gi

This section presents the calculated houses’ growth rates for Qingyuan Town’s villages during the 1980s, 1990s, 2000s, and 2010s, alongside their spatial distribution (

Table 7. Decadal houses’ growth rates in Qingyuan Town Villages.

Village	1980s	1990s	2000s	2010s
Qingyuan Village + Waiwei Village	61.8%	49.8%	57.4%	55.7%
Yangwei Village + Daiyang Village	48.4%	50.0%	43.9%	51.2%
Sanwangyang Village	21.8%	40.7%	34.2%	16.1%
Tongyang Village	36.6%	33.6%	55.4%	31.1%
Yeyangpu Village	32.6%	74.2%	32.6%	2.6%
Guiyang Village	21.4%	58.1%	46.5%	14.8%
Zhuping Village	99.8%	41.0%	14.7%	13.0%
Jiaolin Village	102.9%	44.6%	59.3%	17.9%
Fujiadang Village	16.3%	12.7%	9.0%	44.9%
Yushangang Village	105.4%	39.8%	23.7%	26.3%
Shaotuo Village	10.5%	9.6%	9.0%	66.0%
Pingyan Village	30.2%	35.0%	24.6%	7.2%
Overall Qingyuan Town	40.9%	44.3%	40.7%	32.5%

, Figure 8).

The houses’ growth rates in Qingyuan Town’s villages generally showed an initial increase followed by a subsequent decrease (Figure 5).

During the 1980s, the average growth rate across the villages is 40.9%, with the highest increases observed in Yushangang Village (105.4%), Jiaolin Village (102.9%), and Zhuping Village (99.8%), which are positioned in the southwestern and northeastern sectors of the town, respectively (Figure 8a).

By the 1990s, the average growth rate across the villages rose to 44.3%, with Yeyangpu Village leading (74.2%). However, Shaotuo Village and Fujiadang Village exhibited the lowest growth rates at 9.6% and 12.7%, respectively. Most other villages during this decade experienced growth rates within the 30% to 60% range, showing no notable differences in distribution (Figure 8b).

In the 2000s, the average growth rate slightly decreased to 40.7%. The majority of the villages saw a reduction in house construction growth compared to the 1990s, except for Jiaolin Village (59.3%), Qingyuan Village + Waiwei Village (57.4%), and Tongyang Village (55.4%), which maintained higher rates. Conversely, Shaotuo Village, Fujiadang Village, and Zhuping Village had the lowest rates, at 9.0%, 9.0%, and 14.7%, respectively. The rest ranged from 20% to 50% (Figure 8c).

By the 2010s, the average growth rate further declined to 32.5%. Shaotuo Village recorded the highest rate at 66.0%. Notably, Qingyuan Village + Waiwei Village, Yangwei Village + Daiyang Village, and Fujiadang Village also had significant growth rates exceeding 40%, whereas other villages remained below this threshold (Figure 8d).

4.2.3. Spatial Distribution of FI

This section focuses on the spatial distribution of Fi in Qingyuan Town. It collects all the villages’ Fi indicators (Number of Floors (NF), Average Area per Floor (AAF), Total Floor Area (TFA), and Proportion of Concrete Structures (PCS)) and then calculates the location entropy for these indicators (

Table 8. Numerical values and location entropy of house forms in villages of Qingyuan Town.

Village	TFA	LE	AAF	LE	NF	LE	PCS	LE
Fujiadang Village	107.54	0.54	43.75	0.57	2.59	0.94	0.79	0.99
Guiyang Village	230.92	1.16	81.46	1.06	2.93	1.07	0.88	1.09
Jiaolin Village	204.96	1.03	87.26	1.13	2.61	0.95	0.73	0.91
Pingyan Village	224.77	1.13	111.86	1.45	1.98	0.72	0.64	0.80
Qingyuan Village + Waiwei Village	236.94	1.19	90.85	1.18	2.80	1.02	0.83	1.04
Yeyangpu Village	79.78	0.40	35.50	0.46	2.59	0.95	0.91	1.13
Sanwangyang Village	206.46	1.03	79.35	1.03	2.81	1.03	0.84	1.05
Tongyang Village	267.58	1.34	92.59	1.20	2.94	1.07	0.82	1.03
Shaotuo Village	296.58	1.49	151.57	1.97	2.33	0.85	0.86	1.08
Yangwei Village + Daiyang Village	157.46	0.79	61.34	0.80	2.68	0.98	0.74	0.93
Yushangang Village	94.10	0.47	40.58	0.53	2.54	0.93	0.72	0.90
Zhuping Village	112.57	0.56	56.59	0.74	1.98	0.72	0.29	0.37

NF: Number of Floors; TFA: Total Floor Area; AAF: Average Area per Floor; PCS: Proportion of Concrete Structures; LE: Location Entropy.

, Figure 9).

The variation in NF across villages is relatively small, with the location entropy slightly lower in Zhuping Village (0.72), Pingyan Village (0.72), and Shaotuo Village (0.85), while the rest of the villages range between 0.93 and 1.07 floors.

There is a greater disparity in AAF among the villages. Shaotuo Village has the highest location entropy at 1.97, with six villages ranging between 1.03 and 1.45, and the remaining villages between 0.46 and 0.80.

The distribution of TFA is similar to AAF, with location entropy ranging from 0.40 to 1.49. Villages with location entropy greater than 1 are mostly located in the northern and central parts of Qingyuan Town.

Most villages have a relatively high PCS; thus, the overall differences are not significant. Only Zhuping Village (0.37) has significantly lower location entropy, with the rest of the villages ranging between 0.8 and 1.13.

4.3. Cluster Analysis of RHCD in Qingyuan Town

This section conducts a cluster analysis based on the Gi and Fi in each village from 1980 to 2019.

4.3.1. Cluster Analysis of Gi among Villages

Based on the Gi data from 1980 to 2019, the villages are classified into three categories (Figure 10,

Table 9. 1980–2019 houses’ Gi cluster information.

Category	Cluster 1	Cluster 2	Cluster 3
Village Names	Guiyang Village, Sanwangyang Village, Yeyangpu Village, Tongyang Village, Waiwei Village + Qingyuan Village, Pingyan Village, Yangwei Village + Daiyang Village	Yushangang Village, Zhuping Village, Jiaolin Village	Fujiadang Village, Shaotuo Village
Cluster Typical Values (Centroids)	G1980 = 0.361, G1990 = 0.488, G2000 = 0.421, G2010 = 0.255	G1980 = 1.027, G1990 = 0.418, G2000 = 0.326, G2010 = 0.191	G1980 = 0.134, G1990 = 0.112, G2000 = 0.090, G2010 = 0.555
Type Definition	Developing Villages	Declining Villages	Abnormal Villages

):

• Cluster 1 shows a relatively balanced growth, especially high Gi in the 1990s and 2000s, and a slowdown in the 2010s. Compared to other groups, recent RHCD remains relatively active.
• Cluster 2 exhibits high Gi in the 1980s, which gradually decreased thereafter, declining Gi after the 1980s, and a clear trend of RHCD decline.
• Cluster 3 has uneven Gi, with a significant high in the 2010s, showing a marked difference from other periods, indicating a recent surge in RHCD in these villages.

4.3.2. Cluster Analysis of FI among Villages

The clustering based on FI from 1980 to 2019 also groups the villages into three clusters (

Table 10. 1980–2019 houses’ Fi cluster information.

Cluster	Cluster 1	Cluster 2	Cluster 3
Village Names	Guiyang Village, Jiaolin Village, Pingyan Village, Qingyuan Village + Waiwei Village, Sanwangyang Village, Tongyang Village	Fujiadang Village, Yeyangpu Village, Yangwei Village + Daiyang Village, Yushangang Village, Zhuping Village	Shaotuo Village
Cluster Typical Values (Centroids)	Total Floor Area (TFA) = 1.15, Average Area per Floor (AAF) = 1.18, Number of Floors (NF) = 0.98, Proportion of Concrete Structures (PCS) = 0.99	Total Floor Area (TFA) = 0.55, Average Area per Floor (AAF) = 0.62, Number of Floors (NF) = 0.90, Proportion of Concrete Structures (PCS) = 0.87	Total Floor Area (TFA) = 1.49, Average Area per Floor (AAF) = 1.97, Number of Floors (NF) = 0.85, Proportion of Concrete Structures (PCS) = 1.08
Type Definition	Balanced Villages	Cramped Villages	Spacious Villages

, Figure 11):

• Cluster 1: Features TFA and AAF near the mean values, showing the highest.
• Cluster 2: Displays the lowest TFA, AAF, and PCS but maintains moderately high NF.
• Cluster 3: Stands out with high TFA and AAF, relatively high PCS, and low NF.

5. Discussion

Through data comparison5, on-site research, and interviews6, this chapter analyzes the main influencing factors affecting the temporal and spatial distribution differences of RHCD in Qingyuan Town (

Table 11. Main influencing factors of RHCD in Qingyuan Town.

Category	Explanatory Variable	Main Factor	Specific Comparison Content
Temporal Dimension	Qingyuan Town’s average QI by decade	Population	Trends in rural population development
		Economy	Per capita disposable income in rural areas
	Qingyuan Town’s average FI by decade	Technology	Transition timings of rural construction patterns
		Regulations and Policies	Key legal regulations and policy milestones
Spatial Dimension	Clustered by Growth Rate of Houses by Decade (Gi)	Transportation	Village distances from town center, county center, and county roads
	Clustered by Form Indicator (FI)	Economy	Per capita disposable income in rural areas, village industrial development

). It delves deeply into the mechanisms of these key influences and provides detailed explanations.

5.1. Temporal Factors Influencing RHCD

5.1.1. Influencing Factors of Quantity Indicator (QI) by Decade

Changes in population and economy are foundational elements affecting the temporal evolution of RHCD. Following the economic Reform and Opening-up policy, as the rural economy improved overall, the disposable income of rural residents continued to grow (Figure 12), enabling villagers to invest in housing construction. Consequently, there was a macro-level increase in the volume of residential construction. After 1990, the area of residences also grew, reaching a peak average size of 208.86 square meters for single-family homes in the 2000s.

Since the start of the Reform and Opening-up, particularly before 1995, the permanent rural population continued to rise, reaching a peak of 859 million. However, starting in 1996, there was a consistent decline, with the largest drop occurring in 2011, amounting to 21.24 million. From 1996 to 2021, the annual average decline in China’s permanent rural population was 13.89 million (Figure 5). The growth rate of residential construction in Qingyuan Town from 1980 to 2019 generally showed a positive correlation with the macro trends in rural population changes. Residential construction in Qingyuan Town began to increase significantly after the 1980s, reaching a peak growth rate of 44.3% in the 1990s before declining. In the 2000s, the growth rate of residential construction fell by approximately 4%, and by the 2010s, it had sharply decreased to 32.5%, consistent with the national trend of the rural permanent population. Thus, population growth is a primary factor influencing the amount of housing construction by decade (Ni).

5.1.2. Influencing Factors of Form Indicator (FI) by Decade

The transition in building technologies is a key element influencing the changes in RHCD in the Qingyuan region. Between 1980 and 2019, three major shifts in construction techniques occurred: from wooden structures, to brick structures, and finally to concrete structures (Figure 13). Initially, traditional wooden structures lacked the technological capacity to construct higher buildings. Consequently, after the 1990s, PCS progressively increased. NF in Qingyuan Town has consistently risen, from 1.97 floors before 1980 to 2.92 floors in the 2010s. This trend is further reflected in the average building area of the town, which initially decreased and then increased. The initial decrease was due to a reduction in residential land area; without changes in construction technology, a smaller plot size meant smaller housing areas. However, after the transition in construction technology, the increase in the number of floors compensated for the reduced land area, resulting in an increased building area.

Changes in national policies have been a restrictive factor affecting the temporal evolution of RHCD. Earlier national policies were loosely enforced at the grassroots level and regulations were not comprehensive; thus, RHCD developed spontaneously. With the continuous refinement of laws such as the Land Administration Law in 2004, the Urban and Rural Planning Law in 2008, and the issuance of the Central Government’s opinions in 2014 on pilot reforms of rural land acquisition, collective business construction land market entry, and residential land system reforms, the management of residential land became increasingly stringent7. This directly influenced the morphological indices in RHCD; for instance, the growth rate of NF slowed after 2000, and TFA that peaked in the 2000s began to decline to 205.83 square meters in the 2010s. Both the reduced growth in the number of floors and the decline in AAF reflect the restrictive policies on the number of floors and area that were implemented in the 2010s.

5.2. Spatial Factors Influencing RHCD

Based on the analysis presented earlier, this study categorizes the villages of Qingyuan Town into two clusters according to the Growth Rate of Houses by Decade (Gi) and form indicator (Fi)8, as illustrated in Figure 14. This chapter conducts a detailed analysis based on these classifications to explore the underlying mechanisms of their development.

5.2.1. Influencing Factors of QI-Based Clustering

Based on the analysis in the previous text, the villages are divided into three clusters according to the QI within their RHCD. The study finds that this classification is closely related to the accessibility of transportation and the annual per capita income of the residents.

Cluster 1 consists of developmental villages, which have recently seen a gradual stabilization in the growth rate of residential construction. These villages are predominantly located along county roads, close to town centers and county seats (see Figure 5), providing a geographical advantage that has shaped the core area of Qingyuan Town. These areas continue to attract residents from surrounding villages, creating densely populated and relatively prosperous centers. Due to economic effects, the annual income of residents in these areas is also comparatively high (as shown in Figure 15). Notably, Qingyuan Village and Waiwei Village have merged to form the location of the Qingyuan Town People’s Government. Additionally, Yangwei Village and Daiyang Village have connected, with Yangwei Village being the site of the town government before the 1980s. Tongyang Village, located near the county seat, has been and remains an important transport hub both historically and presently. For mountainous villages, the accessibility of transportation decisively influences village development, with historically larger villages and market towns often located along major roads to attract a higher population and promote economic growth.

Clusters 2 and 3 represent declining and sudden growth-type villages, respectively, most of which are located in remote areas and have experienced slow economic development over a long period. Specifically, Cluster 3 villages have generally transitioned from the state of Cluster 2 due to industrial transformation. Taking Shaotuo Village as an example, a significant increase in the number of houses has recently occurred due to the influx of external capital and the development of tourism. Despite this, the per capita annual income in Shaotuo Village remains low. In-depth research indicates that despite tourism development driven by corporations, it has not significantly created employment opportunities for the villagers or increased land values, thus failing to effectively boost local income or enhance regional vitality.

5.2.2. Influencing Factors of FI-Based Clustering

Villages are divided into three clusters based on the FI characteristics within the RHCD. Upon comparing the basic information of villages in these three clusters, significant differences were observed among them in terms of per capita income (Figure 16), industrial development, and transportation, detailed as follows:

• Cluster 1 includes villages with convenient transportation and distinctive economic industries. Qingyuan Village and Waiwei Village not only serve as the commercial centers of the town but also, due to their advantageous geographic location and targeted governmental support, have become the driving forces of the regional economy, clustering the main commercial activities and institutions of the town. Tong Village leverages its historical role as an agricultural production hub to establish a rural integrated food industry (including shiitake and shimeji mushrooms) as the economic backbone of the village, enhancing the net income of its farmers. Guiyang Village, primarily focused on vegetable and fruit cultivation, has been recognized as a model village for comprehensive county-level development. Its agricultural production not only enriches the lives of the farmers but also stabilizes the village’s economic growth.
• Cluster 2 villages are typically located in more remote areas with sparse populations, relying primarily on traditional tea and grain industries. The lagging economic activities in these areas limit the villagers’ ability to construct larger residences and delay the adoption of new building models. Zhuping Village is a unique example within this cluster, focusing on the kiwifruit industry and employing a “village collective + company + cooperative + farmer” model to stimulate the local economy. However, despite some economic revitalization, recent changes in building morphology and villagers’ incomes remain at lower levels. This situation indicates that traditional agriculture struggles to fundamentally alter the economic income and stem population loss in remote mountain villages.
• In Cluster 3, Shaotuo Village maintains its architectural traditional spatial characteristics through a lake tourism development project. This project has led the village to maintain lower building heights and adopt concrete structures as the primary construction model. These measures cater to the needs of tourism development, ensuring that Shaotuo Village’s architecture retains its traditional charm while accommodating modern functional requirements.

6. Conclusions

This study employs a comprehensive cluster analysis of residential construction growth rates and forms, demonstrating the close connections between these indicators and population, economy, policy, and technology. The findings reveal that the house construction in Qingyuan Town not only reflects the unique developmental patterns of mountainous areas but is also closely linked to the overall evolution of rural society in China. Despite Qingyuan Town’s limited geographical area, there are notable differences in residential construction among its internal villages. The study identifies population growth, geographic location, and economic development as the main drivers of the house construction quantity indicator (QI), while economic growth, advancements in building technology, industrial development, and policy adjustments are key factors influencing the residential form indicator (FI).

• Optimization of Development Strategies: Rural depopulation is the main issue faced by mountain settlements. Through our classification research and analysis of the current housing situation, we can effectively determine which villages have the potential to continue as central living areas, and which are suitable for transformation into tourism or other industrial development. For example, developing villages, as key development areas, should strengthen their public services and infrastructure construction to cope with the population growth triggered by attracting residents from surrounding villages. For declining villages, merely changing basic crops cannot completely alter their developmental trends; they should consider emulating the developmental models of villages experiencing sudden changes, possibly focusing on traditional facade preservation through policy support for cultural heritage protection and considering sustainable development avenues like ecotourism. Abnormal villages9 should promptly conduct detailed development trend analyses to ensure local employment needs, economic sustainability, and ecological balance are met.
• Improvement of Data and Monitoring Systems: It is recommended to strengthen the systematic data collection and monitoring of rural residential construction, particularly in terms of detailed data on individual buildings. This will aid in more accurately identifying and responding to changes and needs within rural settlements.
• Alignment of Governance Village Classification with Development Reality: There is currently a discrepancy that the classification of villages under the government governance system does not match the actual development situation of the villages10. Policy-making should consider geographical constraints and existing advantages, timely adjusting the development positioning and policy support of villages and towns to match the actual situation.

The limitations of this study are primarily in the following aspects: firstly, the research scope is confined to observations within a single town. Future research should expand to more towns and villages to enhance the universality and depth of the study. This would help to validate and deepen the findings of this research and identify the unique characteristics and differences in residential construction among different regions or types of rural settlements, providing a more comprehensive understanding of the temporal and spatial distribution characteristics of residential construction in China’s mountainous rural settlements and their influencing factors.