Spatial distribution characteristics and influencing factors of Suzhou traditional villages from the perspective of “Millennium Village”

Abstract

Amid rapid urbanization, current traditional village (TV) conservation faces limitations: overly simplistic identification criteria heavily rely on official lists, and an overemphasis on tangible heritage, which fails to fully reveal their long-term adaptive characteristics. Drawing on Japan’s “Millennium Village” theory, this study takes Suzhou as the case, using multi-source historical data cross-validation, ArcGIS spatial analysis, and Geodetector models to identify 107 traditional villages with historical continuity and analyze their spatial patterns, formation mechanisms, and survival logic. Results show that (1) They present a “wide dispersion, small agglomeration, high core density, low marginal density” spatial pattern; (2) Natural factors underpin their formation, and socioeconomic factors drive their long-term survival through non-linear interactive enhancement with natural factors. (3) Less than 5% overlap with official lists and only 13.08% lie within urbanized areas, providing a basis for supplementary conservation objects and references for TV adaptive development amid rapid urbanization.

Introduction

Traditional villages contain rich material and intangible cultural heritage^1,2, and their protection and sustainable development have become a common issue urgently needing to be addressed by countries around the world^3,4,5,6,7,8. This constitutes the core problem to be focused on in this study: with the evolution of the natural environment and changes in the social environment⁹, traditional villages are facing a dual crisis both internally and externally. Internally, hollowing and population aging continue to erode their development vitality^9,10; externally, natural disasters directly threaten their material foundation and village pattern^11,12. These factors jointly restrict the development of traditional villages and even lead to their decline, decay, or even disappearance¹³.

Relevant data at home and abroad have confirmed the urgency of this problem. World Bank data shows that between 1960 and 2022, the proportion of the global rural population decreased by 23.30%¹⁴. In countries such as Hungary, the United Kingdom, and Japan, rural labor has moved away from mainstream agriculture, and the number of villages has decreased^15,16,17. In China, according to data from the China Urban and Rural Construction Statistical Yearbook, the number of natural villages decreased from 3.63 million in 2000 to 2.33 million by the end of 2022, showing an overall decreasing trend year by year. Protecting traditional village heritage has become a global consensus. However, current identification and evaluation standards—both domestic and international—still focus on the preservation of material cultural heritage and regional material form characteristics^18,19,20,21. Most central/local governments and international organizations conduct selection based on these standards^22,23,24, with Chinese scholars also frequently using the results as case data²⁴. This practice has excluded numerous villages that possess historical continuity and embody survival wisdom. Without targeted exploration and protection, such “nameless” but valuable villages are likely to gradually disappear amid rapid urbanization.

To solve the above problem, it is necessary to first sort out the existing research progress in the field. Traditional villages are living spaces that carry historical and cultural heritage as well as human survival wisdom. For their sustainable protection and development, a core prerequisite is understanding the survival logic behind their distribution patterns. This is directly related to the accuracy of protection strategies and development paths. Existing studies have carried out explorations from multi-scale, multi-disciplinary, and multi-method perspectives.

In terms of spatial scale, the research scope has covered the national, provincial, municipal, and county levels and typical village cases^{25,26,27,28,29}. Empirical research areas are concentrated in Hunan³⁰, Henan^31,32, Zhejiang^33,34, Southwest China³⁵, etc. These studies generally reveal the characteristics of spatial agglomeration and the uneven distribution of villages. In terms of research content, multiple disciplines such as architecture and human geography have focused on the impact of natural and socio-economic factors on distribution^36,37, confirming that the spatial heterogeneity of villages is the result of long-term human-environment interaction^24,38. In addition, with changes in the external environment, relevant studies have gradually extended from the influencing factors of distribution to the improvement of adaptive capacity^7,39. Regarding research methods, studies often combine qualitative methods (field surveys, oral history, etc.) with quantitative methods (buffer analysis, kernel density estimation (KDE), etc.)^40,41, which can effectively analyze the distribution pattern and driving mechanism^{42,43,44,45,46,47}.

Despite the above progress, existing research still has room for improvement, especially in sample selection, research focus, and method integration, making it challenging to fully explain the survival logic of “Millennium Village.” From the perspective of sample selection, the core samples of current traditional village research are mostly derived from protection lists established based on specific selection frameworks, but such lists have inherent limitations. Firstly, the selection criteria have undergone phased evolution—from the three first-level indicators (original appearance preservation degree, current scale, value characteristic certification) initially set in the Measures for the Selection and Evaluation of National Famous Historical and Cultural Towns (Villages) in 2003 to the five first-level indicators (historical and cultural accumulation, site selection, pattern and texture, traditional architecture, intangible cultural heritage, village vitality) in the 6th round of selection in 2022^18,48,49. Yet a single selection framework still struggles to accurately capture the differences in traditional village heritage worldwide. Traditional villages globally have a large quantity, wide distribution, and significant heterogeneity in history, space, and culture, which are shaped by ethnic diversity and varied geographical and climatic conditions^50,51. Secondly, most selections follow a fixed cycle with low update frequency. They fail to timely track the dynamic changes in village survival status or incorporate newly discovered villages and those requiring priority attention amid environmental changes. This leads to insufficient completeness and timeliness of the lists. Thirdly, both national and local-level selections are difficult to completely avoid the influence of subjective factors such as policy orientation and judges’ personal preferences, which may lead to the neglect of some villages with unremarkable material remains but strong historical continuity. These problems ultimately lead to insufficient representativeness of research samples and highlight the necessity of expanding sample sources.

From the perspective of research focus, existing studies only concentrate on the current spatial distribution characteristics⁵². They fail to deeply explore the internal resilience and dynamic adaptation process of villages, and lack the core evaluation dimension of sustainability, making it difficult to judge the survival capacity of villages. From the perspective of research methods, there is a lack of an integration path to convert non-standardized historical information into accurate spatial data. This deficiency makes it hard for research to break free from reliance on protection lists.

In view of the above research gaps, this study sets the following research objective: to clarify the spatial distribution characteristics, formation mechanisms, and survival logic of traditional villages with historical continuity in the Suzhou area, and to compensate for the deficiencies in existing research on sample selection, research focus, and method integration. To achieve this goal, this study will introduce the theory of “Millennium Village” into the research on traditional Chinese villages and conduct an in-depth analysis of 107 such villages in Suzhou through multi-source historical data cross-validation, GIS spatial analysis, and the Geodetector model. Ultimately, this study aims to provide theoretical support and practical reference for the protection and sustainable development of traditional villages.

Methods

Localized adaptation and operational criterion definition of the “Millennium Village” theory

The core of the “Millennium Village” theory, put forward by Japanese scholars, lies in identifying settlements that have sustained human activities for a millennium⁵³. It holds that such villages embody survival logic characterized by adaptation to environmental changes, disaster resistance, and resource stability maintenance. This study applies the theory to the research on traditional villages in Suzhou, China, and the feasibility of this application is rooted in the compatibility between China and Japan in terms of agricultural civilization foundation, historical time scale, and evolutionary logic.

On the one hand, from the perspective of civilization foundation, the origin of China’s agricultural civilization can be traced back to the Neolithic Age. Since the Holocene epoch, the gradual formation of the alluvial plain in the Yangtze River Delta has endowed Suzhou with gentle terrain, stable water systems, and fertile soil. This is highly consistent with the core proposition of Japan’s “Millennium Village” theory that “the long-term survival of a region depends on natural foundations such as terrain, water systems and arable land.” On the other hand, from the perspective of the historical cycle, “a millennium,” as a medium-to-long time scale, not only conforms to the chronological recording feature based on dynastic changes in Chinese historical documents, but also effectively transcends the interference of short-term social unrest and regime alternations, thus focusing on the fundamental mechanism of long-term regional survival. Furthermore, in terms of evolutionary logic, the core logic of Japan’s “Millennium Village” lies in the iteration of administrative systems while preserving the essence of settlements—from ancient “Gō” (townships) to modern “Ōaza” (large administrative districts), although administrative names and levels have undergone adjustments, the core residential space, toponym inheritance, and survival functions of villages have remained continuous. This is highly consistent with the millennium-long evolution law of Chinese villages: from ancient “Li—Cun—Du” to modern “administrative villages—natural villages,” while upper-level administrative units have dynamically adjusted with dynastic changes, the geographical space, toponym identification, and agricultural survival functions of grass-roots settlements have always remained continuous. This forms a cross-cultural echo with the Japanese theory, further consolidating the rationality of theoretical adaptation.

Japanese academic circles have developed a mature empirical research paradigm for the application of this theory, which provides a key methodological reference for this study. For example, architectural historian Norihito Nakatani and his team successfully identified “Millennium Villages” by systematically comparing the locations of “go” (villages) recorded in the Wamyō Ruiju Shō of the Heian period with those of modern “ōaza” (large administrative districts). They found that these villages are generally distributed in areas with diverse terrain, proximity to water sources, availability of arable land, and stable foundations⁵³, which directly reveals the survival wisdom behind their site selection. Subsequent studies have further expanded the application dimensions of the theory: for instance, Daiki Takahashi and Tsuyoshi Kinoshita extended it to the field of disaster risk management, demonstrating the effectiveness of retaining traditional agricultural land within “ōaza” units in reducing flood risks⁵⁴, and deepening the understanding of village resilience mechanisms. Separately, studies on heritage sites (e.g., Shirakawa-mura) have highlighted the core role of community participation in sustainable conservation⁵⁵. Collectively, these studies have constructed a methodology covering theory, spatial identification, mechanism analysis, and conservation practice, providing important references for global traditional village research.

To ensure the rigor and verifiability of the theory in application, this study defines the operational definition of “millennium-long continuous habitation/activities” in accordance with the characteristics of China’s historical document system and formulates specific procedures and operational criteria. This definition does not aim to prove the absolute uninterrupted continuity of villages, but rather refers to the following criterion: clear evidence confirms that the name of a settlement and its corresponding geographical entity repeatedly appear across multiple major historical nodes spanning hundreds or even thousands of years. Notably, there is no conclusive evidence of permanent abandonment or long-distance overall relocation of the settlement. Meanwhile, we fully recognize that between discrete historical recording points, there may theoretically be a temporary weakening of village activities or a selective omission in documents. Therefore, the core objective of this study is to identify settlements that exhibit significant stability and resilience from a long-term historical perspective.

Based on this definition, this study adopts a prudent and pragmatic principle in handling the “interruption” of villages. Firstly, we explicitly acknowledge that the absence of records in documents does not equate to the disappearance of settlements—the compilation of local chronicles in China is periodic and selective, with inherent gaps in records. Meanwhile, we have set clear exclusion criteria: if explicit documentary evidence confirms that a place was completely abandoned or relocated elsewhere due to disasters or wars, with no evidence of subsequent reconstruction near the original site, it is regarded as a break in continuity. Additionally, cases where the place name has been completely changed—with no traceable linguistic, geographical, or genealogical connections, and essentially referring to an entirely new settlement—are also excluded. In addition, in practical operation, a reasonable margin of error is allowed for the central point of a settlement. Specifically, if historical records indicate relocation or reconstruction, and the straight-line distance between the new and original settlement centers does not exceed 3 km while the relative orientation to key natural landmarks remains essentially unchanged, continuity is still recognized; otherwise, it is regarded as an interruption.

In terms of identification and verification procedures, this study formulates screening and identification criteria based on multi-source data. These criteria follow the principles of continuous historical records, detailed information collection, and cross-validation of evidence (see Fig. 1 for the Screening and Decision-Making Flowchart; see Supplementary Table I for typical village examples). To ensure the consistency and objectivity of the screening and verification process, this study adopts a “two-person independent review and cross-recheck” mechanism: two researchers conduct independent verification in accordance with the above criteria. If the consistency rate of their judgments reaches ≥90%, the consensus is directly adopted; if the consistency rate is < 90%, all controversial cases are submitted for collective review and comprehensive adjudication. Firstly, village information (name, duration of existence, location, etc.) is extracted from Xianchun Piling Gazetteer (1268) and Zhengde Gusu Gazetteer (1506), which are included in The Jiangsu Complete Compendium of Local Gazetteers. Secondly, the village information is cross-validated with subsequent materials (such as Yutu Map of Jiangsu Province (1895), Atlas of Jiangsu Province (1978), etc.). Subsequently, the villages that have passed the historical evidence chain verification are accurately located and their current state of existence is checked using modern GIS data, place name directories and remote sensing images, to confirm that they still exist as identifiable geographical entities or communities to this day.

Fig. 1: Screening and decision-making flowchart for “Millennium Village” research samples.

This figure presents a five-step framework for screening and validating “Millennium Village” research samples, anchored by the core criterion of “Millennium-Long Continuity of Habitation/Activity.” Step 1 establishes operational definitions, including continuity standards, interruption duration limits, relocation rules, and toponym evolution criteria. Step 2 generates an initial candidate list from The Jiangsu Complete Compendium of Local Gazetteers (518–1949), focusing on village age and historical continuity records. Step 3 constructs an evidence chain using the 1868 Yutu Map of Jiangsu Province and the 1978 Atlas of Jiangsu Province to cross-validate place names and spatial positions. Step 4 verifies current status via modern Tianditu maps, classifying locations as villages, towns, or residential areas, and assigning coordinates. Step 5 involves a dual independent review to finalize stable, resilient village samples for the study.

In terms of administrative unit correspondence, the natural village is adhered to as the basic research unit, and the merger of modern administrative villages does not affect the independent identification of the historical natural villages under their jurisdiction. In handling place name evolution, changes conforming to the rules of Wu dialect phonetic shifts, abbreviations, refinement, or taboo avoidance are accepted. The judgment needs to meet one of the following conditions: clear annotations in local chronicles, distinct phonetic change rules with supporting examples, or complete overlap or immediate adjacency in geographical location. If the place name is completely changed, it must meet at least two of the following: a clear record of renaming in local chronicles, full geographical overlap with the original settlement, or unbroken resident inheritance proven by genealogy or inscriptions. In accordance with the principle of prudence, this study incorporates a small number of cases with uncertain or questionable matching results into the final sample as boundary cases, with explanations of boundary cases (including specific identification basis and considerations) provided in Supplementary Table II.

In summary, a village’s evidence is deemed reliable only if its information is fully and continuously recorded in three sources: early local chronicles, at least one subsequent historical material, and modern geographical data. Meanwhile, evidence strength is quantified by the number of verification instances (distinguished by solid and hollow dots in Supplementary Table I). This process yields 107 stable traditional villages with continuous, reliable evidence across historical documents (see Supplementary Table III).

Overview of the study area

Suzhou City is located in East China, the middle of the Yangtze River Delta, and the southeast of Jiangsu Province, with geographical coordinates ranging from 119°55′E to 121°20′E and 30°47′N to 32°02′N. The city administers five municipal districts and has jurisdiction over four county-level cities. It borders Shanghai to the east, Zhejiang Province to the south, Taihu Lake to the west, and the Yangtze River to the north, covering a total area of approximately 8657.32 km² (Fig. 2). The terrain of Suzhou is flat, with a dense river network and numerous lakes. It has a subtropical monsoon marine climate, featuring mildness, humidity, and abundant rainfall—all of which provide a natural foundation for the long-term survival of traditional villages.

Fig. 2: Distribution of historically formed villages in Suzhou City, categorized by their current status.

This figure shows the spatial distribution of historically formed villages in Suzhou City, classified into three categories according to their current status. The map on the right shows the study area within the jurisdiction of Suzhou City. Blue dots represent (1) “Currently existing as residential areas”: the original sites of historical villages have been transformed into urban residential districts. Red dots represent (2) “Currently existing as villages”: the historically formed village pattern has been maintained to the present day, and village-level administrative units still exist. Yellow dots represent (3) “Currently existing as towns”: historical villages have developed into town government seats and regional centers. The inset map at the top left locates the study area within the national context of China. The bar chart at the bottom left quantifies the number of villages in the ten county-level districts of Suzhou. The coordinate system is CGCS2000; base map data are from Amap (Map Approval No.: GS (2024) 1158); village coordinates are generated by verifying literature in the National Digital Library of China.

As the birthplace of the Wu culture in the Jiangnan region, Suzhou has long been the administrative, economic, and cultural center of Jiangnan in history. Human activities here can be traced back more than 4000 years⁵⁶, and Neolithic cultural sites, together with historical documents, jointly confirm the continuity of its civilization. The unique location, natural environment, and historical heritage have nurtured a large number of traditional villages. These villages have interacted deeply with the environment during their evolution and have not been interrupted by disasters such as earthquakes and floods, which fully reflects the site selection wisdom of ancient people. After verification, a total of 107 traditional villages conforming to the characteristics of millennium-long existence were identified in this study. The spatial distribution, quantity characteristics, and current status of these villages (Fig. 2) further confirm the representativeness of Suzhou as a research area for “millennium village.”

The selection of Suzhou as the research area stems from its dual values. On the one hand, as the core area of Jiangnan culture, its villages are highly representative of the farming traditions of the Wu region and the patriarchal clan social structure. This aligns with the common background of similar settlements in the Yangtze River Delta and even East Asia, enabling the study to provide targeted plans for village heritage protection in similar regions. On the other hand, Suzhou has a large number of villages with millennium-long survival characteristics, which embody the wisdom of site selection for risk avoidance. This not only provides a path for the protection of traditional villages in water network areas of southern China but also offers experience for the sustainable development of waterfront heritage sites worldwide.

Data sources

To avoid the limitations of a single data source and ensure data reliability and continuity, this study adopts multi-source data to support the analysis (see Table 1 for specific data types and sources). Among them, the historical and geographical information of traditional villages is mainly derived from the Chinese Digital Resources Collection of the National Library of China (http://read.nlc.cn/user/), supplemented and verified by existing academic achievements; the basic geographical data of modern villages is obtained from the National Geomatics Center of China (https://www.tianditu.gov.cn/).

Table 1 Main materials used in the research

The precise coordinates of traditional villages are acquired by extracting the longitude and latitude of the central point via Baidu Maps and uniformly converting them to the CGCS2000 National Geodetic Coordinate System. For villages that have undergone urbanization transformation, the coordinates of the original seat of the township government are used as a substitute.

After identifying the 107 sample villages, to explore the driving mechanism of their spatial distribution, this study selected 9 key influencing factors (including natural environment and socio-economic factors; see Table 2 for details) based on existing research consensus^27,57,58 and the availability of regional data in Suzhou. Explanations for data processing are as follows: The 1960 climate data adopts long-term average values to reflect stable natural characteristics. The 2019 road data represents the latest modern transportation pattern. The 1995 population and GDP data are the earliest complete statistical data of the study area, which is consistent with the historical survival stage of the villages. All data were standardized to the 107 village locations using ArcGIS 10.8 to ensure a unified analysis scale.

Table 2 Data sources of natural geographic and socio-economic indicators for traditional villages in Suzhou

Nearest Neighbor Index (NNI)

The Nearest Neighbor Index (NNI) is used to quantify the spatial distribution pattern of point features (clustered, uniform, or random). Its core principle is to determine the degree of aggregation by comparing the average distance of actual nearest neighbors with the average distance under a theoretical random distribution⁵⁹. The statistical significance of the NNI is tested by the Z-score: |Z| > 1.96 (P < 0.05) indicates that the distribution pattern is significant at the 95% confidence level. |Z| > 2.58 (P < 0.01) indicates that the distribution pattern is significant at the 99% confidence level (see Supplementary Table IV for the definitions of symbols, variables, and units in each method formula).

$$R=\frac{{r}_{1}}{{r}_{e}}$$

(1)

$${r}_{e}=\frac{1}{2}\sqrt{\frac{A}{n}}$$

(2)

In the formula, $R$ represents the NNI; ${r}_{1}$ represents the observed average distance of nearest neighbors; ${r}_{e}$ represents the expected average distance of nearest neighbors under random distribution; $A$ represents the area of the study region; $n$ represents the number of point features. When $R > 1$, the distribution is uniform; when $R=1$, the distribution is random; when $R < 1$, the distribution shows a significant clustered pattern.

Kernel density estimation (KDE)

KDE is a non-parametric spatial statistical method. Its core function is to estimate the spatial distribution density of point features, which can smooth discrete point data into a continuous density surface. This allows for the visualization of high-density areas (hotspots) and low-density areas (coldspots)⁶⁰. The calculation formula is as follows:

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

(3)

Among them, $f\left(x\right)$ represents the estimated density value at position $x$; $n$ denotes the total number of point features; $h$ stands for the bandwidth, which is used to control the smoothness of the density surface; $k$ is the kernel function; $x-{x}_{i}$ refers to the distance between the target coordinate $x$ (where density is to be estimated) and the position coordinate ${x}_{i}$.

Spatial autocorrelation analysis (Moran’s I)

Global Moran’s I is a statistical tool for quantifying the global spatial autocorrelation of spatial data. It is mainly used to test whether the entire study area exhibits a significant clustered or dispersed pattern⁶¹. The calculation formula is as follows:

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

(4)

Among them, $n$ represents the number of spatial units; ${x}_{i}$ and ${x}_{j}$ denote the attribute values of the $i$th and $j$th spatial units, respectively; $\bar{x}$ stands for the average value of attribute values across all units; ${w}_{{ij}}$ is the spatial weight matrix. The value range of Moran’s I is usually$\,[-1,\,1]$, and its interpretation is as follows: When $I > 0$, the spatial distribution shows positive autocorrelation; the larger the value, the more significant the correlation. When $I=0$, the spatial distribution presents a random pattern. When $I < 0$, the spatial distribution exhibits negative autocorrelation; the smaller the value, the more significant the correlation.

Density-Weighted Gini Coefficient

The Density-Weighted Gini Coefficient is a spatial extension of the traditional Gini Coefficient. It can measure the distribution balance of resources after accounting for regional area differences and intuitively reflect the degree of inequality per unit area. Compared with the traditional Gini Coefficient, which only focuses on quantity distribution and ignores regional area differences, this coefficient analyzes the spatial imbalance of traditional villages in Suzhou through density per unit area, making the results more in line with geographical reality. Derived from the area relationship between the Lorenz curve (with cumulative area proportion on the X-axis and cumulative density weight on the Y-axis) and the line of absolute equality, its calculation formula is as follows:

$$G=1-2{\sum }_{i=1}^{n}\left(\frac{{D}_{i-1}+{D}_{i}}{2}\right)\times \left({A}_{i}-{A}_{i-1}\right)$$

(5)

Among them, ${D}_{i}$ is the cumulative density weight of the $i$th region; ${A}_{i}$ is the cumulative area proportion of the $i$th region; $n$ is the total number of regions; ${A}_{i}-{A}_{i-1}$ is the difference in cumulative area proportion between adjacent regions; ${D}_{i-1}+{D}_{i}$ is the sum of cumulative density weights of adjacent regions. The Density-Weighted Gini Coefficient ranges from 0 to 1. A value of 0 indicates an absolutely even distribution of traditional villages, while a value of 1 indicates an absolutely uneven distribution. Its evaluation criteria are as follows: Coefficient < 0.2: Highly balanced; 0.2 ≤ Coefficient < 0.3: Relatively balanced; 0.3 ≤ Coefficient < 0.5: Moderately unbalanced; 0.5 ≤ Coefficient < 0.7: Highly unbalanced; Coefficient > 0.7: Extremely unbalanced.

Geodetector

Geodetector is a spatial statistical method used to identify the spatial differentiation law of geographical factors and quantify the explanatory power of different factors on geographical phenomena. Its core principle is: if a geographical factor has a significant impact on the distribution of the target phenomenon, the spatial patterns of the two should show a significant correlation⁶². The calculation formula is as follows:

$$q=1-\frac{{\sum }_{h=1}^{L}{N}_{h}{\sigma }_{h}^{2}}{N{\sigma }^{2}}$$

(6)

Among them, $L$ represents the stratification of variable $Y$ (dependent variable) or factor $X$ (independent variable); ${N}_{h}$ and $N$ denote the number of units in stratum $h$ and the entire region, respectively; ${\sigma }_{h}^{2}$ and ${\sigma }^{2}$ stand for the variance of $Y$ values in stratum $h$ and the entire region, respectively; $q$ indicates the explanatory power of each influencing factor X on the dependent variable Y, with a value range of [0, 1]. The interpretation of the $q$value is as follows: A larger $q$value means the spatial differentiation of the dependent variable is more significant, and the explanatory power of the independent variable on the dependent variable is stronger (i.e., the factor has a greater impact on the distribution of traditional villages); a smaller $q$ value indicates the opposite. In extreme cases, $q=1$ means the independent variable completely controls the spatial distribution of the dependent variable, while $q=0$ indicates no correlation between the two.

Results

Statistical overlap between Suzhou’s traditional villages and the designated traditional villages list

As mentioned earlier, this study initially screened 907 village names based on the previously organized materials. After verifying each village in accordance with the screening criteria established earlier, 623 villages were excluded because they only met the requirements of one type of documentary record or failed to meet the coordinate accuracy standard, and 177 villages were excluded for meeting the requirements of only two types of documentary records. Eventually, only 107 villages passed the verification of three types of evidence chains and became the “Millennium Village” samples. To clarify the connection between this sample and the existing protection system, the samples were compared with the recognized national (Batches 1–6, up to 2023), provincial (Batches 1–8, up to 2025), and municipal-level (Batches 1–3, up to 2025) traditional village lists of Suzhou. This spatial matching was conducted at the natural village level through a two-step process. First, an initial automated match was performed based on administrative divisions and a fuzzy name matching algorithm (with a tolerance threshold of two character differences) to identify potential correspondences. The complete cross-reference table generated by this process is available in Supplementary Table V (or in the accompanying data repository: https://raw.githubusercontent.com/millennium-village/Cross-reference-Table-of-the-lists/refs/heads/main/compare_villages.py). Subsequently, all potential matches were subjected to a rigorous manual verification based on multi-source historical evidence cross-validation, incorporating judgments on the consistency of administrative divisions and the homology of toponym evolution, to precisely align the sample villages with the identified “Millennium Villages.” The results showed that the overlap rate between the two was less than 5%, with only 6 villages overlapping—namely Shiqiao Village, Kaixiangong Village, Yangwan Village, Ducheng Village, Lüshe Village, and Jiangze Village. This difference is not due to data deviation, but rather because the current official lists take “current protection value” as the core selection indicator, while the “Millennium Village” in this study take “survival capacity” as the core indicator. This indicates that the official lists have gaps in the dimension of historical depth, and this study can provide new evaluation standards and perspectives for the heritage protection of traditional villages.

Lorenz curve analysis

To quantify the spatial distribution imbalance of Suzhou’s 107 “Millennium Villages,” this study analyzed the “area proportion” and “village density weight” (share of Millennium Villages in a district to the city’s total) of each county-level administrative district. Relevant data are in Table 3, and the Lorenz curve is plotted in Fig. 3. Results show the villages are unevenly distributed, mostly in Wuzhong and Wujiang Districts (52.34% combined). The general rule is: smaller districts have higher village density, while larger ones have lower density.

Fig. 3: Lorenz curve of the spatial distribution of traditional villages across districts in Suzhou.

This figure presents a Lorenz curve illustrating the spatial distribution of traditional villages across districts in Suzhou. The x-axis represents the cumulative area proportion of each district, while the y-axis represents the cumulative density weight of traditional villages. The solid blue line (Cumulative Density Weight) shows the actual distribution of village density, and the dashed red line (Uniform distribution line) represents an idealized scenario of perfectly even density distribution. The deviation of the blue line from the red diagonal indicates the degree of unevenness in the spatial density distribution of traditional villages across districts in Suzhou.

Table 3 Statistical table of traditional villages by county-level jurisdictions in Suzhou City

Based on the curve shape and district features, the Lorenz curve is divided into four segments. Initial segment (far below the absolute equality line): Taicang City (area 9.67%, density weight 0.0210)—low density, little cumulative contribution. Rising segment (slope increasing): Huqiu District (area 3.97%, density weight 0.1795)—extremely high density, major contributor to imbalance. Middle segment (low slope): Changshu City (area 15.23%, density weight 0.0801) and Wuzhong District (area 26.63%, density weight 0.1298)—village distribution matches area. Final segment (sharp rise): Wujiang District (area 14.77%, density weight 0.1515) and Xiangcheng District (area 5.85%, density weight 0.1565)—high density; Xiangcheng is one of the main causes of imbalance.

Density-Weighted Gini Coefficient analysis

To quantify the overall spatial distribution imbalance of Suzhou’s 107 “Millennium Villages,” this study calculated the trapezoidal area under the Lorenz curve using “area proportion” and “village density weight” data of each county-level administrative district (Table 3). The Density-Weighted Gini Coefficient was further derived. A coefficient value <0.2 indicates that, at the city-wide scale, the distribution of “Millennium Villages” matches regional areas well, showing a highly balanced overall pattern. This result does not contradict the local imbalance revealed by the Lorenz curve: the former quantifies the overall balance across the study area, while the latter highlights extreme differences in local regions. In summary, the spatial distribution of “Millennium Village” features overall balance and local differentiation.

Spatial pattern analysis of traditional villages in Suzhou City

To determine the spatial distribution pattern of Suzhou’s 107 “Millennium Villages,” this study used the average NNI method for analysis, with results shown in Fig. 4. The “Millennium Villages” exhibit a clustered spatial distribution: the NNI R = 0.836023 (<1), and the Z-score is −3.245. Additionally, the p value is 0.001175 (far less than 0.01), indicating the probability of this clustering being caused by random factors is less than 1%, which is statistically significant at the 99% confidence level.

Fig. 4: Average Nearest Neighbor test report for the spatial pattern of traditional villages in Suzhou.

This figure shows the Average Nearest Neighbor test results for traditional villages in Suzhou. The nearest neighbor ratio is 0.836023, with a z-score of −3.244933 and a p value of 0.001175. The central yellow area of the normal distribution curve represents random patterns, while the shaded tails indicate significant clustering (left) or dispersion (right). The legend lists significance levels and critical z-scores. The z-score of −3.244933 indicates a less than 1% probability of the clustered pattern occurring by chance, confirming a statistically significant clustered spatial distribution.

To further identify the spatial clustering hotspots of Suzhou’s 107 “Millennium Villages,” this study used KDE for village locations and generated a kernel density distribution map (Fig. 5). Results show the overall distribution follows a pattern of “wide dispersion, small-scale clustering, more cores, fewer edges,” with density decreasing gradually from core to edge areas: High-density areas (kernel density: 351.5–618.2): Concentrated in northeastern and southeastern Wuzhong District, and the border area between northern Xiangcheng District and Changshu City. Secondary density areas (kernel density: 179.4–351.5): Covering southern Wujiang District, northern Xiangcheng District, central Kunshan City, and southern Zhangjiagang City. Sparse areas (kernel density: <179.4): Concentrated in Taicang City, western Wuzhong District, and northern Zhangjiagang City.

Fig. 5: Kernel density distribution pattern of traditional villages in Suzhou.

This figure illustrates the kernel density distribution pattern of traditional villages in Suzhou. The left panel is a spatial density map, where different color gradients represent varying kernel density levels (traditional villages per square kilometer), ranging from very low (0–41.2 villages/km²) to very high (460.6–618.2 villages/km²). Black dots mark the locations of traditional villages. The right panel is a bar chart showing the number of villages falling into each kernel density interval. The coordinate system is CGCS2000; base map data are from Amap (Map Approval No.: GS (2024) 1158); village coordinates are generated by verifying literature in the National Digital Library of China.

To verify the relevance of spatial clustering of Suzhou’s 107 “Millennium Villages” at the global scale, this study calculated the Global Moran’s I using spatial autocorrelation analysis in ArcGIS (Fig. 6). Results show: Global Moran’s I = 0.679, z = 11.235, p < 0.001. This indicates a significant global positive correlation in the spatial distribution of “Millennium Villages”—high-density village areas tend to adjoin other high-density areas, and low-density areas adjoin other low-density areas. The probability of this clustering pattern arising randomly is <0.1%, with statistical significance.

Fig. 6: Global spatial autocorrelation test results for traditional villages in Suzhou.

This figure shows the results of a global spatial autocorrelation test (Moran’s I) for traditional villages in Suzhou. The Moran’s Index is 0.679209, with a z-score of 11.234514 and a p value of 0.000000. The central yellow area of the normal distribution curve represents random patterns, while the shaded tails indicate significant clustering (right) or dispersion (left). The legend lists significance levels and critical z-scores. The z-score of 11.234514 indicates a less than 1% probability of the clustered pattern occurring by chance, confirming a statistically significant positive spatial autocorrelation.

The relationship between the spatial distribution pattern of traditional villages and influencing factors in Suzhou City

In terms of elevation (Fig. 7a), the number of villages first increases then decreases with rising elevation, generally favoring low-elevation flat terrain, distributed between 0–30 m (average: 11 m, concentrated in alluvial plains). Among them, 85.0% (91 villages) lie within [5, 15) m, showing an overall right-skewed distribution—reflecting a preference for low-elevation sites. Such areas offer easy access to water, lower disaster risks, and better conditions for transportation, farming, and daily life, laying a solid foundation for the villages’ millennium-long survival. Only a few villages (e.g., 4 in Wuzhong District above 20 m, including 2 on Dongshan Island and 1 in Linhu Town) are in relatively high elevations, due to combined factors like terrain, climate, flood defense, and resource culture.

Fig. 7: Relationship between traditional village distribution and variations in elevation, slope, and aspect.

This figure illustrates the relationship between the spatial distribution of traditional villages and topographic factors in the Suzhou region. The top row comprises three spatial maps: a elevation, b slope, and c aspect, with black dots marking the locations of traditional villages. The bottom row presents corresponding bar charts that quantify the number of villages within each interval. Map (a) employs a color gradient to represent elevation, ranging from 0 m (dark green) to 348 m (dark red). Map (b) uses a color gradient to represent slope, ranging from 0° (dark green) to 38° (dark red). Map (c) utilizes a color gradient to represent aspect, ranging from 0° (light blue) to 359.7° (dark red). The coordinate system is CGCS2000; base map data are sourced from Amap (Map Approval No.: GS (2024) 1158); the data set is provided by the Geospatial Data Cloud site, Computer Network Information Center, Chinese Academy of Sciences. (http://www.gscloud.cn).

Regarding slope (Fig. 7b), the number of villages also first increases, then decreases with steeper slopes, favoring low-slope terrain, ranging from 0.34° to 8.09° (average: 2.68°). Further statistics show ~92.6% (99 villages) are distributed within 0°–5°, where terrain benefits residential stability and disaster resistance: 1.9% (2 villages) in 0°–0.5° (flat land), 37.4% (40) in 0.5°–2° (gentle slope), 53.3% (57) in 2°–5° (moderate slope), and 7.5% (8) in 5°–15° (steep slope). Notably, village numbers decline beyond 5°, but 6 villages (5.6%) still exist in 5°–8°, reflecting adaptive use of steeper terrain in local areas; no villages are found in slopes >15°, consistent with traditional societies’ avoidance of extreme terrain.

For aspect selection (Fig. 7c), unlike mountain villages that strongly prefer sunny slopes, Suzhou’s “Millennium Villages” show no strict constraints. Data reveals a wide aspect range (3.69°–333.68°) covering all directions, slightly favoring sunny slopes (90°–270°) with 57 villages (53.27%). This distribution is not random but influenced by Suzhou’s low-elevation terrain—with an average elevation of only 11 m, most areas have weak aspect variations, making the aspect’s impact on light intensity and temperature far smaller than in mountainous regions.

The climatic commonality of Suzhou’s “Millennium Villages” is not generalized mildness and humidity, but a focus on moderately suitable ranges of precipitation and temperature to avoid extreme values, reducing flood and drought risks. In terms of climatic characteristics (Fig. 8a, b): Precipitation: The annual average precipitation of villages mainly ranges from 1231.1–1513.4 mm (mean 1379.8 mm), with 76.6% (82 villages) concentrated in 1308.2–1470.1 mm—balancing irrigation needs and disaster avoidance. Temperature: The distribution range is extremely narrow (15.5–16.3 °C, mean 15.9 °C), which not only meets crop growth needs but also avoids risks of high-temperature drought and low-temperature waterlogging.

Fig. 8: The relationship between the distribution of traditional villages and climatic and hydrological factors.

This figure illustrates the relationship between traditional village distribution and climatic and hydrological factors in the study area. The top row shows three spatial maps: a annual precipitation, b annual average temperature, and c distance to rivers, with black dots marking village locations. The bottom row presents corresponding bar charts of village counts per interval. Map (a) uses a color gradient for annual precipitation (1148.5–1554.5 mm). Map (b) uses a color gradient for annual average temperature (14.9–16.3 °C). Map (c) uses a color gradient for distance to rivers (0–3 km). The coordinate system is CGCS2000; base map data are from Amap (Map Approval No.: GS (2024) 1158); precipitation and temperature data are from the Resource and Environmental Science Data Center of CAS; distance-to-river data are from the National Geomatics Center of China.

The single-peak distribution of climatic factors (temperature peak: 15.8–16.0 °C; precipitation peak: 1380–1410 mm) is the result of active village site selection. Beyond the peak ranges, village density decreases significantly, indicating that extreme climates damage the agricultural foundation for survival. For a few villages deviating from the commonality (13.1% with precipitation <1308.2 mm, 11.2% with >1470.1 mm), low-precipitation areas compensate for water shortages through water storage and diversion projects, while high-precipitation areas resolve flood risks with dikes and polders—converting disadvantages into agricultural advantages. This adaptation mechanism not only reflects the fundamental constraints of the natural geographical environment but also human initiative in transforming the environment, jointly shaping the distribution wisdom of pursuing benefits and avoiding harm.

In terms of water source distance selection, “Millennium Villages” cluster within a suitable range by controlling distance to rivers, meeting water needs for agricultural irrigation, daily life, and historical transportation. Buffer zone analysis of key rivers (Fig. 8c) shows: The average distance from villages to water sources is 644.0 m (maximum 3509.5 m). 58.8% (63 villages) are distributed within 500 m for immediate water access. 86.9% (93 villages) lie within 1000 m to cover water transportation needs (village density decreases significantly beyond this distance). 97.2% (104 villages) concentrate within 2500 m to avoid high water-use costs. It is evident that the commonality in village water source selection centers on low-cost and highly stable utilization, with core clustering within the critical 1000-m distance around rivers.

In terms of road density selection (Fig. 9a), the average road network density in the study area is 1.21 km/km² (ranging 0.36–2.24 km/km²). The maximum road density in areas with “Millennium Villages” is 2.24 km/km², with a critical value of 1.7 km/km². 87.9% (94 villages) are distributed in areas with road density ≤1.7 km/km²; Beyond this value, the number drops sharply. This choice keeps villages in a low-disturbance environment, laying a stable foundation for their millennium-long survival.

Fig. 9: The relationship between the distribution of traditional villages and socioeconomic factors.

This figure illustrates the relationship between traditional village distribution and socioeconomic factors in the study area. The top row shows four spatial maps: a road density, b population density, c GDP density, and d urbanization rate, with black dots marking village locations. The bottom row presents corresponding bar charts of village counts per interval. Map (a) uses a color gradient for road density (0–2.57 km/km²). Map (b) uses a color gradient for population density (248.3–5657 persons/km²). Map (c) uses a color gradient for GDP density (248–7397 ×10,000 yuan/km²). Map (d) uses a color gradient for the urbanization index (0–49979.1). The coordinate system is CGCS2000; base map data are from Amap (Map Approval No.: GS (2024) 1158); road density data are from the National Geomatics Center of China; population and GDP density data are from the Resource and Environmental Science Data Center of CAS.

Suzhou’s “Millennium Villages” concentrate in low-population-density areas. Specifically, their population density ranges 251–5577 people/km², with an average of 809 people/km² (standard deviation: 770, clustering in low-density areas despite local variations). Approximately 86.0% of villages lie in areas with population density <1000 people/km²; The number decreases significantly as density rises (Fig. 9b). This preference for low population density reduces over-development, protects agricultural resources from occupation, and maintains stable traditional order.

In terms of GDP density, “Millennium Villages” are distributed across 355–7293 thousand yuan/km² (average: 1239 thousand yuan/km², standard deviation: 1014), with numbers decreasing as GDP density rises: 78.5% are in low-development-pressure areas with GDP density <1396 thousand yuan/km², and only 21.5% in higher-density areas (Fig. 9c). This reflects active avoidance of high-intensity development zones to preserve traditional production structures.

Regarding distance to administrative units, “Millennium Villages” tend to locate within 20 km for moderate radiation: 74.8% (80 villages) are in this range, and only 5.6% (6 villages) are over 30 km away (Fig. 9d). This choice ensures sustained access to administrative support and public service resource allocation while avoiding excessive administrative intervention, providing institutional guarantees for village survival.

Discussion

In the context of rapid urbanization, traditional villages in high-intensity urbanization plain areas represented by Suzhou are facing risks of disappearance and homogeneous destruction. Existing studies on traditional villages mostly focus on mountainous areas, with few focusing on plain areas, and lack quantitative analysis of spatial distribution driving forces and factor interaction mechanisms. Against this background, this study introduces Japan’s “Millennium Village” theory, identifies 107 traditional villages with historical continuity in Suzhou through collating historical documents, and uses tools such as GIS and Geodetector to comprehensively consider natural and socioeconomic influencing factors, constructing an overall research framework for the spatial distribution and influencing factors of Suzhou’s “Millennium Villages,” so as to reveal their spatial distribution patterns and survival logic.

To quantify the impact intensity and interaction effects of factors (e.g., natural geography, socioeconomic conditions) on the spatial distribution of Millennium Villages, this study used a factor detector to evaluate the independent explanatory power of each factor (Fig. 10a) and an interaction detector to identify inter-factor interactions (Fig. 10b). Results show that among the 10 influencing factors, only road accessibility, population density, and GDP density passed the significance test (p < 0.001), exerting a significant impact on the distribution of traditional villages. The remaining factors did not reach a significant level (p > 0.05). Further Geodetector analysis shows Suzhou’s Millennium Villages’ spatial distribution is mainly driven by socioeconomic factors. Population density, economic development level, and road accessibility are core drivers, while natural factors have relatively weak independent impacts at this regional scale. This aligns with the previously summarized “preference for low-disturbance environments,” confirming village site selection is not random but an active response to core drivers. Specifically, population density has the strongest explanatory power (q = 0.278), verifying the rationality of choosing areas with density ≤1391 people/km². Historically, this population density range ensures livability and productivity while avoiding hollowing caused by population loss. Currently, it also mitigates the impacts of urbanization. GDP density (q = 0.279) and road accessibility (q = 0.322) further support this preference for low-disturbance environments. For GDP density, village survival does not depend on high economic levels but rather on a balance between basic vitality and avoiding over-development. Road accessibility: Only low accessibility is needed for basic material and information exchange, avoiding development pressure and homogeneous invasion from high accessibility.

Fig. 10: Results of geodetector analysis on the distribution of traditional villages in Suzhou.

This figure presents the results of a Geodetector analysis on traditional village distribution in Suzhou. Subplot (a) is a bar chart showing the q-statistic values for single-factor detection. Subplot (b) is a heatmap displaying interaction-factor detection results, with a color gradient representing interaction strength (red for stronger, green for weaker). The coordinate system is CGCS2000; base map data are from Amap (Map Approval No.: GS (2024) 1158); factor data are from the Resource and Environmental Science Data Center of CAS and the National Geomatics Center of China.

The spatial pattern of Suzhou’s Millennium villages is shaped by the interaction of socioeconomic and physical geographical factors. Results from the geographical detector interaction analysis (Fig. 10b) further confirm that all two-factor combinations exhibit significantly higher explanatory power for village distribution compared to single factors, with all showing non-linear enhanced effects. This synergy is key to the villages’ long-term survival. Among interaction results, elevation and population density exhibited the highest explanatory power (q = 0.619). The mechanism is as follows: “Millennium villages” are concentrated in low-elevation areas with agricultural potential, and low population density also strongly influences their distribution. This dual characteristic ensures the long-term preservation of farming capacity. The interactions between temperature and road accessibility (q = 0.586) and temperature and GDP density (q = 0.569) reveal that suitable temperature ranges for agriculture form the basic condition for village distribution, within which settlements further align with development intensity. Similarly, interactions between precipitation and population density (q = 0.580), and slope aspect and population density (q = 0.575), show that natural factors first delineate agriculturally viable areas, while population density acts as a key balancing factor for village survival, ultimately shaping the observed distribution patterns. In summary, the persistence of millennium village does not rely on the absolute dominance of a single factor, but rather on the synergy between factors, which naturally formed an adaptive distribution range over long-term development.

Suzhou’s millennium villages exhibit characteristics of clustering near water (86.8% within 1 km of rivers) and favoring low-lying, gentle terrain (92.6% on slopes below 5°). This essentially reflects their natural adaptation to agricultural needs, laying the foundation for their formation and continuity. However, the high explanatory power and strong interaction of population density (q = 0.322) and GDP density indicate that natural conditions alone cannot sustain their millennium-long survival. Historical development shows these villages are generally located in areas with low population, low GDP, and low road density—a pattern that maintains basic vitality and economic foundations while avoiding excessive urbanization impacts, achieving a balance between vitality and preservation. Compared to vanished or endangered villages, the core advantage of millennium village lies in their regions’ long-term livability, productive potential, and disaster resilience, which convert natural conditions into socioeconomic toughness. This adaptive mechanism manifests in the fact that these villages are not static historical relics but dynamic entities. Through residents’ active modifications, they continuously adjusted production and social structures during material-energy exchanges with the environment, ultimately achieving long-term survival.

The current traditional village selection system oriented by value identification has achieved remarkable results in the emergency protection of cultural heritage, but it has shortcomings such as the lag of static lists and selection bias. This study provides a new framework from the perspective of historical continuity, enriching the overall understanding of village heritage value. For “Millennium Villages” not included in the lists, their millennium-long survival itself proves that they contain historical information, adaptive wisdom, and cultural memory, and carry the key logic for the sustainable development of traditional villages. Notably, highly urbanized areas such as industrial parks still retain a considerable proportion of traditional villages (accounting for 3.21% of the area and 13.08% of the total villages), which indicates that urbanization and the survival of traditional villages are not absolutely opposed. While the regional economy brings development pressure, its advantages in capital, technology, and management can feed back to heritage protection, and create more job opportunities to inject vitality into villages, helping the modern transformation of traditional villages and ultimately realizing sustainable inheritance.

Compared to previous studies, the innovative academic contributions of this research are as follows: (1) Current research on traditional villages in China mostly focuses on mountainous and underdeveloped rural areas, with relatively few studies on plain areas with high urbanization levels, such as Suzhou. This study, taking Suzhou’s plain area as the research unit, explores the spatial distribution patterns and survival logic of “Millennium Village” in high-intensity urbanization plain areas, which further enriches and improves the regional characteristics and connotations of traditional village research in China’s different geographical units. (2) Existing studies on the spatial distribution of traditional villages often lack quantitative analysis of driving forces and interaction mechanisms between factors. This study combines GIS spatial analysis and Geodetector to quantitatively identify the core driving forces of “Millennium Village” distribution and the non-linear enhancement effect between natural and socioeconomic factors, breaking the limitations of qualitative description in previous related studies and improving the scientificity and rigor of research in this field. (3) Current research on traditional village protection mostly adopts a value-oriented static identification framework, while this study introduces Japan’s “Millennium Village” theory into the Chinese context, constructs a research paradigm integrating “historical continuity identification-spatial pattern analysis-survival logic exploration,” and supplements a new perspective for traditional village heritage protection, which is conducive to improving the comprehensiveness of traditional village value recognition.

The practical significance of this study is reflected in the following aspects: (1) For policy formulation, this study reveals the core conditions for the formation and continuation of “Millennium Village” (natural foundations such as slope, annual average temperature, and annual average precipitation; socioeconomic keys such as population density, GDP density, and road accessibility), which provides a scientific basis for the spatial layout optimization of traditional village protection in the Yangtze River Delta region, and supports the delineation of key protection areas and low-disturbance development boundaries of traditional villages, as well as the dynamic update of the protection list. (2) For protection practice, the study clarifies that the survival of “Millennium Village” relies on the balance between natural adaptation and low-intensity human activities, which guides the balance between traditional village heritage protection and rural revitalization. It is suggested that in the protection process, we should adhere to the principles of preserving the natural base, maintaining vitality, and avoiding excessive interference, and develop low-disturbance industries with regional characteristics to avoid homogeneous competition. (3) For regional governance, this study takes Suzhou as an example to explore the adaptive mechanism of traditional villages in high-intensity urbanization plain areas, which provides a reference for the protection and sustainable inheritance of similar historical villages in plain areas with high economic development intensity, and is conducive to promoting the coordinated development of traditional village cultural inheritance and man-nature harmony in a wider range.

This study also has certain limitations. Due to constraints on time and data availability, some unrecorded or poorly documented villages still await further exploration. From the Song-Yuan dynasties to the present, through multiple dynasties and social changes, some villages may have disappeared and then been rebuilt; verification at a single time point may not fully reflect their true historical status. Additionally, the accurate restoration of ancient village locations using modern GIS technology is still limited by the detail and accuracy of the original data. The 107 villages selected in this study are only potential “Millennium Village,” i.e., a candidate list. In the future, it is necessary to conduct more in-depth on-site research combined with field investigations, carrying out micro-level analysis and research on the villages from various aspects such as their history, natural environment, social organization and operation, physical spatial composition, and transportation. Ultimately, special discussions should be held to determine whether they meet the criteria of “Millennium Village.” Based on the above research conclusions and limitations, this study puts forward suggestions on the resilient development and sustainable inheritance of traditional villages: First, build a digital database and conduct systematic digital filing for traditional villages, relying on geographic information systems and digital technologies to comprehensively collect material and intangible heritage data and dynamic monitoring data, focusing on natural foundation data and core driving factors to provide data support for forward-looking protection. Second, improve the systematic evaluation system for traditional village viability, integrate indicators of historical continuity and village viability on the basis of existing static list identification, and formulate evaluation standards highlighting Suzhou’s regional characteristics. Third, develop low-disturbance industries with Suzhou’s regional characteristics, guide villages to leverage their own resources to develop characteristic experience economies, and encourage villages near cities and towns to integrate into regional economic development, converting urban economic momentum into sustainable resources for village development.

In conclusion, Suzhou’s “Millennium Villages” present a spatial distribution pattern of “overall balance and local agglomeration,” and their survival depends on the dual adaptation mechanism of natural conditions and low-intensity human activities. Socioeconomic factors are the core driving forces of their spatial distribution, and the synergistic effect between natural and socioeconomic factors is the key to their long-term survival. This study not only enriches the theoretical system of traditional village research but also provides practical guidance for their protection and sustainable inheritance, which is of great significance for promoting the coordinated development of cultural heritage protection and rural revitalization in high-intensity urbanization areas.