Spatial agglomeration patterns and associated factors of intangible cultural heritage in the Guangdong–Hong Kong–Macao greater bay area, China

Abstract

Introduction:

This study examines the overall, typological, temporal, and contextual geography of 188 national-level intangible cultural heritage (ICH) projects in the Guangdong–Hong Kong–Macao Greater Bay Area (GBA), China.

Methods:

Global Moran’s I, kernel density estimation, Getis–Ord Gi* analysis, standard deviational ellipse analysis, category-specific distribution profiles, spatial-coupling analysis, and directional historical-cultural and hydrological correspondence tests were applied.

Results:

The cumulative inventory shows a core-dominated but weak and scale-sensitive recognition pattern centred on Guangzhou and Foshan (Moran’s I = 0.189, p = 0.035), whereas no individual inscription batch reaches statistical significance. Category-specific analysis shows that this aggregate core is not reproduced uniformly across ICH types: 60.0% of traditional fine arts projects are located in Guangzhou and Foshan, whereas traditional dance and music display broader estuarine distributions. Spatial-coupling analysis identifies the closest city-level distributional similarity between traditional music and traditional dance (cosine similarity = 0.911). A reconstructed register of 438 nationally or provincially protected historical-cultural sites in the nine mainland GBA cities further reveals differentiated directional correspondences between ICH categories and historical material environments, although no category-level association survives false discovery rate adjustment. Two hydrological indicators, the 2018 water surface ratio and its 2000–2018 change, likewise show heterogeneous rather than uniform relationships with different ICH categories.

Discussion:

These findings are interpreted as directional spatial tendencies rather than causal effects. Conceptually, the study argues that the GBA’s aggregate ICH pattern is better understood as the superimposition of category–specific recognition geographies displaying differentiated relationships with historical, territorial, and hydrological contexts.

1 Introduction

Intangible cultural heritage (ICH) represents an important component of regional cultural identity and historical continuity. In recent years, increasing attention has been paid to the spatial distribution, conservation, and sustainable utilization of ICH resources under rapid urbanization and regional development (Naheed and Shooshtarian, 2022; Zhang et al., 2024). With the rapid development of geographic information systems (GIS) and spatial statistical techniques, studies have increasingly applied methods such as kernel density estimation (KDE), Moran’s I, hotspot analysis, standard deviational ellipse (SDE), and Geographic Detector models to investigate the spatial characteristics and influencing factors of ICH resources (Zhang et al., 2022; Pang and Wu, 2023; Zhang et al., 2024). Existing studies have identified significant clustering characteristics and spatial heterogeneity of ICH resources in regions such as the Yellow River Basin, the Beijing–Tianjin–Hebei region, Shandong Province, and Northwest China (Pang and Wu, 2023; Meng et al., 2022; Yue et al., 2025).

Despite these advances, research on the Guangdong–Hong Kong–Macao Greater Bay Area (GBA) remains relatively limited. As one of China’s most economically dynamic and highly urbanized regions, the GBA is also an important cultural region of Lingnan culture, with abundant national-level ICH resources distributed across Guangzhou, Foshan, Shenzhen, Hong Kong, Macao, and other metropolitan areas. Under the combined influence of historical-cultural accumulation, economic development, urbanization, and cross-border governance, the spatial organization of ICH resources in the GBA exhibits complex agglomeration characteristics (Ren et al., 2025; Mo et al., 2025). Recent GBA studies have documented overall spatial distribution patterns and associated factors, but they have generally treated the inventory as a static spatial snapshot rather than as the cumulative outcome of successive inscription rounds (Ren et al., 2025; Mo et al., 2025). More importantly, aggregate inventories often obscure typological heterogeneity. Craft traditions, ritual practices, music, dance, drama, medicine, and other forms of ICH differ substantially in their modes of production, transmission, community embedding, and institutional visibility; there is therefore little reason to assume that all categories follow the same spatial agglomeration mechanism. Existing GBA research has also paid limited attention to the spatial co-distribution or distributional coupling among different ICH types, or to whether the aggregate pattern is produced by overlapping category-specific geographies. In addition, heritage concentration is structured across three administrative systems—mainland Guangdong, Hong Kong, and Macao—whose nomination and recognition pathways are not fully equivalent. This cross-jurisdictional structure is analytically important because differences in national-level project counts may reflect nomination and recognition mechanisms as well as underlying cultural geography.

Previous studies suggest that the spatial distribution of ICH is associated with historical-cultural accumulation, socio-economic development, urbanization, and environmental conditions (Han et al., 2024). Yet these dimensions are themselves internally heterogeneous. Immovable historical-cultural resources may occur at the scale of settlements, districts, building complexes, or individual structures, and their functional contexts may be ritual, residential, defensive, maritime, commercial, educational, or otherwise specialized. Pooling these resources into a single historical-site count can therefore obscure more specific territorial correspondences with different ICH categories. Hydrological conditions also require explicit attention in the GBA. The Pearl River Delta is a riverine and estuarine landscape in which water systems have historically structured settlement connectivity, exchange, occupational specialization, and inter-community contact. Accordingly, the present study distinguishes historical-site types and incorporates water-surface indicators as contextual spatial variables. More broadly, recent GIS-based spatial planning research has demonstrated the value of integrating heterogeneous spatial indicators, environmental conditions, and spatial pattern evidence within transparent analytical frameworks that support geographically differentiated interpretation and planning decisions (Karadeniz et al., 2026a; 2026b). Although these studies address ecosystem services and ecological suitability rather than cultural heritage, their spatial-indicator logic provides a useful methodological parallel for the present integration of typological, historical-cultural, hydrological, and regional contextual evidence. Because the contextual indicators in this study are cross-sectional or temporally adjacent to a cumulative 2006–2021 recognition inventory, the resulting relationships are interpreted as directional spatial correspondences rather than causal determinants of heritage origin or inscription.

Against this background, the study integrates spatial pattern, typological, batch-wise, and contextual analyses of national-level ICH recognition in the GBA. Rather than treating the 188 projects as a homogeneous inventory, the study distinguishes the ten official ICH categories, compares their city-level concentration profiles, and evaluates spatial coupling among category-specific recognition geographies. The inventory is also decomposed across successive inscription batches to identify directional changes in cumulative recognition. Historical-cultural context is examined through a differentiated register of immovable heritage by official type, spatial scale, and functional context, while hydrological context is assessed using water surface ratio and long-term change in water surface ratio. Throughout, registered project locations are interpreted as recognition locations rather than direct representations of heritage origins or practice sites. Specifically, the study addresses four research questions. First, what are the overall spatial distribution and cumulative recognition characteristics of national-level ICH projects in the GBA? Second, how do concentration profiles and spatial coupling differ among major ICH types? Third, what directional changes are visible across successive inscription batches? Fourth, how do differentiated historical-cultural and hydrological contexts correspond directionally with the spatial distributions of the overall inventory and individual ICH categories?

2 Materials and methods

2.1 Study area

The Guangdong–Hong Kong–Macao Greater Bay Area (GBA) is located in southern China and includes nine cities in Guangdong Province together with the Hong Kong and Macao Special Administrative Regions (Figure 1). Figure 1 maps the registered declaration or recognition locations of the national-level ICH projects used in the spatial analysis, rather than claiming that all points represent original sites of cultural practice. As one of China’s most economically developed and highly urbanized regions, the GBA contains abundant ICH resources associated with Lingnan culture. Differences in historical-cultural accumulation, economic development, transportation accessibility, and urbanization levels make the region a suitable case for examining whether and how nationally recognized ICH projects are spatially concentrated.

FIGURE 1

2.2 Data sources

This study collected 188 national-level ICH projects distributed across the GBA. The data were obtained from the official website of the Ministry of Culture and Tourism of the People’s Republic of China and supplemented by provincial and municipal databases. The dataset included project names, categories, declaration units, administrative locations, and geographic coordinates. National-level ICH projects were selected because they provide the most standardized publicly documented recognition inventory available for regional-scale comparison. However, nomination and recognition pathways are not institutionally equivalent across mainland Guangdong, Hong Kong, and Macao; cross-jurisdictional differences are therefore interpreted as part of the recognition geography rather than as directly comparable measures of underlying cultural abundance. Geographic coordinates were obtained through geocoding and manual verification based on the registered declaration locations of each project.

The geocoding procedure followed a three-step workflow. First, the registered declaration unit or officially listed administrative location of each project was used as the primary address. Second, coordinates were obtained through geocoding and checked against official government webpages, local ICH databases, gazetteer records, and online map records. Third, ambiguous cases were manually inspected and assigned to the most specific verifiable location available. Where only a city-level declaration unit was available, the point represents the registered recognition location rather than the original practice site. For jointly declared projects, the project was retained in the inventory but assigned to one representative location for point-based spatial analysis to avoid duplicate counting. This treatment may slightly affect fine-scale grid and hotspot results, but it does not alter the city-level inventory or the interpretation of the results as recognition geography rather than origin geography. A two-level typological framework was used to characterize heterogeneity within the 188-project inventory. At the first level, each project was assigned to one of the ten official ICH categories: folk literature, traditional music, traditional dance, traditional drama, quyi or folk performing arts, traditional sports, games and acrobatics, traditional fine arts, traditional handicraft skills, traditional medicine, and folk customs. These ten categories served as the primary spatial-comparison units. At the second level, the project-level inventory retained more detailed subcategory or cultural-form descriptors, which were used as an interpretive layer for identifying the specific cultural forms contributing to category-level concentrations. Because many subcategories contain only one or two projects, separate inferential spatial tests were not conducted at the subcategory level. The supplementary workbook provides the complete 188-project inventory (Supplementary Appendix A), the reconstructed mainland historical-site register, detailed historical-site correspondence results, and supporting analyses for MAUP, KDE bandwidth, Gi* FDR correction, Geographic Detector outputs, category distributions, spatial coupling, and hydrological correspondence (Supplementary Tables S1–S8).

To examine broader city-level contextual conditions associated with the spatial distribution of recognized ICH projects, socio-economic, demographic, industrial, urbanization, natural-environmental, and spatial-control indicators were assembled at the city level (Table 5). The twelve contextual screening indicators were drawn primarily from the 2021 statistical yearbooks of Guangdong, Hong Kong, and Macao and related governmental geographical databases. Historical-cultural context was examined separately using official national and Guangdong provincial protected cultural relic catalogues current to December 2023, while the hydrological indicators were derived from the Landsat-based water-surface series reported by Jiang et al. (2022). All spatial point data and regional variables were harmonized within the CGCS2000 spatial reference framework and processed in ArcGIS 10.8 where spatial GIS operations were required.

2.3 Spatial analysis methods

This study employed Global Moran’s I, kernel density estimation (KDE), Getis–Ord Gi* hotspot analysis, standard deviational ellipse (SDE), and Geographic Detector models to analyze the spatial agglomeration characteristics and associated factors of ICH resources in the GBA.

2.3.1 Global Moran’s I

Global Moran’s I was used to examine the spatial autocorrelation characteristics of ICH resources (Moran, 1950). Prior to analysis, ICH point data were aggregated into regular 5 km × 5 km grid cells, and the number of ICH projects within each grid cell was used as the attribute value for Moran’s I calculation. Global Moran’s I was calculated using Equation 1:where n represents the number of spatial units, xᵢ and xⱼ denote the number of ICH projects in spatial units i and j respectively, is the mean value, and wᵢⱼ represents the spatial weight matrix based on inverse Euclidean distance. Positive values indicate spatial clustering, whereas negative values indicate spatial dispersion. The threshold distance used in the spatial weight matrix was determined through Incremental Spatial Autocorrelation analysis and corresponded to the first statistically significant peak z-score distance (95.8 km), which was adopted as the main analytical distance band. Because grid-based spatial autocorrelation may be affected by the modifiable areal unit problem (MAUP), a supplementary sensitivity analysis was conducted using alternative grid sizes and distance bands. Specifically, the representative ICH point dataset was re-aggregated to 2.5 km, 5 km, and 10 km grids, and Moran’s I was recalculated across the occupied grid cells using distance bands of 75 km, 95.8 km, and 125 km. This sensitivity workflow differs from the main ArcGIS specification, which was based on the full analytical grid, including zero-count cells; it was used as a supplementary scale-sensitivity check rather than as an exact replication of the main Moran’s I estimate. The purpose was to assess whether the direction and substantive interpretation of cumulative spatial dependence remained stable under reasonable changes in spatial aggregation and distance specification.

2.3.2 Kernel density estimation

Kernel density estimation (KDE) was applied to transform discrete point data into a continuous surface to identify the spatial agglomeration intensity and high-density clustering areas of ICH resources, as expressed in Equation 2 (Silverman, 1986):where n represents the number of sample points, h denotes the bandwidth parameter, K is the kernel function, and (x − xᵢ) represents the spatial distance between the estimation location and sample point xᵢ. Kernel density estimation was implemented using the Kernel Density tool in ArcGIS 10.8 to generate a continuous density surface from discrete ICH point locations. A raster cell size of approximately 1 km × 1 km and a search radius of approximately 20 km were adopted for the main KDE maps. To assess bandwidth sensitivity, additional KDE surfaces were generated using 10 km, 20 km, and 30 km search radii. The comparison focused on whether the Guangzhou–Foshan core, the eastern Pearl River Estuary secondary belt, and the weak peripheral pattern remained identifiable across bandwidth choices. The 20 km radius was retained because it preserved local spatial variation while avoiding excessive fragmentation at 10 km and over-smoothing at 30 km. The bandwidth sensitivity results are provided in Supplementary Table S2. Areas with higher density values indicate stronger spatial concentration of ICH resources.

2.3.3 Getis–Ord Gi* hotspot analysis

The Getis–Ord Gi* statistic was employed to identify statistically significant local hotspot and coldspot regions of ICH resources (Getis and Ord, 1992):where and S denote the mean and standard deviation of the attribute values across all spatial units, respectively. High positive Z-scores indicate significant hotspots, while low negative Z-scores indicate coldspots. Because Gi* analysis involves multiple local significance tests, Benjamini–Hochberg false discovery rate (FDR) correction was applied to the local Gi* p-values as a robustness check. The main Gi* maps report the nominal confidence levels generated by the local statistic, but the revised interpretation distinguishes nominal local signals from FDR-adjusted significance. Where no local unit survives FDR correction, hotspot and coldspot patterns are interpreted as exploratory local indications rather than strong inferential evidence.

2.3.4 Standard deviational ellipse analysis

Standard deviational ellipse (SDE) analysis was used to examine the directional distribution, dispersion trends, and orientation characteristics of ICH resources (Yuill, 1971):where (xᵢ′, yᵢ′) represent the coordinates of individual ICH spatial points, while (′, ȳ′) denote the mean centre coordinates. The SDE analysis was performed using projected coordinates (CGCS 2000 3-degree Gauss-Kruger, central meridian 114°E) within ArcGIS 10.8, so that the standard distances along the major and minor axes are expressed in kilometres. For ease of interpretation, the mean-centre coordinates in Table 4 are reported in geographic coordinates (longitude and latitude), while the ellipse axes and rotation angles are derived from the projected coordinates.

2.3.5 Typological spatial pattern and coupling analysis

Typological analysis was conducted using the ten official ICH categories as the primary comparison units. For each category, project counts and proportional distributions were calculated across the eleven GBA jurisdictions. Category-specific concentration profiles were then compared across the Guangzhou–Foshan core, the eastern Pearl River Estuary corridor represented by Dongguan, Shenzhen, and Hong Kong, and the remaining jurisdictions. Detailed project subcategories were retained as an interpretive layer for examining the cultural forms contributing to category-level concentrations. Because category sizes ranged from only 3 to 47 projects, separate inferential point-pattern tests were not conducted for individual categories; category-specific agglomeration was therefore assessed descriptively through jurisdictional concentration profiles to avoid unstable small-sample spatial inference.

Spatial coupling among ICH categories was assessed from their city-level distribution profiles. For each category, an eleven-jurisdiction count vector was constructed, and pairwise cosine similarity was calculated between category vectors. Values approaching one indicate highly similar relative distributions across jurisdictions, whereas lower values indicate more differentiated recognition geographies. The coupling measure describes spatial co-distribution within the national recognition inventory and does not imply cultural dependence, diffusion, or causal interaction between ICH types.

2.3.6 Geographic Detector and correlation analysis

The relationship between the city-level distribution of ICH projects and regional development conditions was examined using a Spearman rank correlation together with the Geographic Detector model (Wang et al., 2016). Because the explanatory variables were available only at the city level, the analysis was conducted at the city scale (N = 11) rather than at the grid level, in order to avoid the pseudo-replication that arises when city-level values are duplicated across many grid cells. The number of national-level ICH projects in each city was used as the dependent variable. Most explanatory indicators refer to 2021, whereas the dependent variable is the cumulative 2006–2021 inventory of nationally recognized ICH projects. Accordingly, these analyses are interpreted as cross-sectional associations with the contemporary spatial pattern of recognition and not as explanations of earlier inscription decisions or batch-specific recognition processes. Twelve socio-economic, demographic, industrial, urbanization, natural-environment, and spatial-control indicators were retained for the original N = 11 city-level screening analysis (Table 5). The previous aggregate ancient-cultural-site indicator was removed from the principal correlation framework because it pooled heterogeneous immovable heritage resources across different spatial scales and functional contexts.

Historical-cultural context was therefore examined separately through a reconstructed typological register assembled from the official national key cultural relic protection catalogue and the Guangdong provincial cultural relic protection catalogue, both current to December 2023. The analysis was restricted to the nine mainland GBA cities because these jurisdictions share the same national–provincial designation framework; Hong Kong and Macao use institutionally different statutory heritage-register systems and were not pooled into the matching analysis. The resulting register contained 438 protected sites, including 71 national-level and 367 provincial-level records.

Each historical site retained its official heritage category and was additionally coded by spatial scale and broad functional context. Spatial-scale classes distinguished archaeological or site-based resources, individual sites, site complexes, individual buildings or structures, building complexes, and settlement- or district-scale heritage. Functional coding distinguished ritual and communal, residential and settlement, defensive and military, maritime and transport, commercial and craft, performing and public-cultural, administrative and educational, funerary and memorial, and residual archaeological, architectural, or modern-civic contexts. Coding used official designation categories together with explicit functional or spatial information available in official catalogue records and designation names; no functional attribution was inferred solely from presumed cultural association, and ambiguous records were retained in residual categories.

For each of the ten official ICH categories, city-level project counts across the nine mainland GBA cities were compared with counts of historical-site categories and functional types using Spearman rank correlations. These analyses were treated as directional and exploratory correspondence tests. Benjamini–Hochberg false discovery rate adjustment was applied separately to the official-category and functional-type comparison families.

The separate N = 11 city-level contextual screening proceeded in two steps. First, a non-parametric Spearman rank correlation was computed between the total recognized-project count and each of the twelve retained contextual indicators. Because twelve contextual indicators were screened, a Bonferroni-adjusted threshold of α = 0.0042 (0.05/12) was used as a conservative multiple-comparison benchmark. Nominal p-values are retained for descriptive transparency, but coefficients that do not meet the adjusted threshold are interpreted only directionally. Second, the Geographic Detector was applied as an exploratory complement: each indicator was discretized into three classes, and single-factor q-values and pairwise interaction q-values were computed. The q-statistic is defined as:where h = 1, … ,L represents the stratification of factor X; N_h and N denote the number of units in stratum h and the entire study area; and σ² represent the variance of Y within stratum h and the entire study area. The range of q is [0,1], where larger values indicate stronger explanatory power. Given that the analysis rests on only eleven city-level units discretized into three classes, the Geographic Detector is used solely as an exploratory descriptive complement to the Spearman correlation. In particular, pairwise interaction q-values are highly sensitive to sparse strata because overlaying two factors can create classes containing only one or two cities. Therefore, the q-values are reported for transparency but are not used as inferential evidence of causal mechanisms or substantive factor interactions.

3 Results

3.1 Overall composition and categorical structure of ICH resources

The 188 national-level ICH projects span all ten official categories, but the inventory is typologically uneven (Figure 2a). Traditional handicraft skills form the largest category (47 projects, 25.0%), followed by folk customs (40, 21.3%), traditional fine arts (20, 10.6%), and traditional music (20, 10.6%). Together, these four categories account for 67.6% of the inventory. Traditional dance (17, 9.0%), traditional medicine (14, 7.4%), and traditional drama (13, 6.9%) form an intermediate group, whereas folk literature (8, 4.3%), traditional sports, games and acrobatics (6, 3.2%), and quyi or folk performing arts (3, 1.6%) are comparatively rare.

FIGURE 2

By inscription batch (Figure 2b), the inventory was built up mainly through the 2011 round of national recognition, which alone contributed 43.1% of all projects (81 projects). The 2021 (39 projects, 20.7%) and 2008 (36 projects, 19.1%) batches form a second tier, while the 2006 (23 projects, 12.2%) and 2014 (9 projects, 4.8%) batches are comparatively small. The 2006, 2008, and 2011 batches together account for 74.5% of the inventory, indicating that the core of the system was established before 2014, while the sizeable 2021 batch shows continued region-wide supplementation in recent years.

At the city scale, the resources are strongly concentrated in the regional core (Table 1). Guangzhou (42 projects, 22.3%) and Foshan (28 projects, 14.9%) jointly account for 37.2% of the total, followed by Shenzhen and Zhaoqing (16 each, 8.5%) and Hong Kong (15, 8.0%), while Zhuhai and Macao hold the fewest. This distribution indicates a marked concentration of nationally recognized ICH projects in the historic Guangzhou–Foshan core relative to other GBA cities.

3.2 Category-specific spatial patterns and spatial coupling

The category-specific distributions show that the aggregate Guangzhou–Foshan core is not reproduced uniformly across all ICH types. Traditional fine arts display the clearest core concentration: 12 of 20 projects (60.0%) are located in Guangzhou and Foshan. Traditional drama and traditional medicine also show comparatively strong Guangzhou–Foshan concentration, with 46.2% and 42.9% of their projects, respectively, located in the two-city core. Traditional handicraft skills are more numerous but less exclusively concentrated; 17 of 47 projects (36.2%) occur in Guangzhou and Foshan.

By contrast, traditional dance, traditional music, and folk customs show broader spatial profiles. Only four of 17 traditional dance projects (23.5%) are located in Guangzhou and Foshan, whereas six projects (35.3%) occur across Dongguan, Shenzhen, and Hong Kong. Traditional music is distributed equally between the Guangzhou–Foshan core and the eastern Pearl River Estuary corridor, with six projects (30.0%) in each zone. Folk customs likewise show a distributed configuration, with 13 of 40 projects (32.5%) in Guangzhou–Foshan and 12 (30.0%) in Dongguan, Shenzhen, and Hong Kong. The three numerically smaller categories show less stable concentration profiles. Folk literature is distributed across several jurisdictions rather than forming a dominant two-city concentration, while traditional sports, games and acrobatics remain sparse across the regional inventory. Quyi or folk performing arts comprises only three projects, making any apparent concentration highly sensitive to individual project locations. Overall, the aggregate Guangzhou–Foshan core is disproportionately reinforced by selected ICH categories rather than being reproduced uniformly across all types. The complete jurisdictional distributions of all ten categories are reported in Supplementary Table S6.

The spatial-coupling analysis further indicates that some ICH categories share closely aligned recognition geographies, whereas others have more differentiated city-level profiles. The highest distributional similarity occurs between traditional music and traditional dance (cosine similarity = 0.911), followed by traditional handicraft skills and traditional medicine (0.908), traditional fine arts and traditional medicine (0.907), and traditional drama and folk customs (0.893). These coupling values describe similarity in city-level recognition distributions and do not imply functional dependence, cultural transmission, or causal interaction between ICH types. Full category-by-jurisdiction distributions and the pairwise coupling matrix are reported in Supplementary Tables S6 and S7.

3.3 Global spatial autocorrelation

Under the main specification, the cumulative inventory showed weak positive spatial autocorrelation (Table 2). For all 188 projects, Moran’s I reached 0.189, with a z-score of 2.11 and a p-value of 0.035, indicating a statistically detectable but substantively modest clustering tendency rather than a strong region-wide agglomeration pattern. The analysis used an inverse-distance conceptualization with Euclidean distances and a threshold distance of approximately 95.8 km.

When the analysis is repeated for individual inscription batches, none of the single-batch distributions reaches statistical significance (Table 2). The 2006 (I = 0.128) and 2008 (I = 0.154) batches show weak, non-significant positive autocorrelation, the 2011 batch is essentially random (I = 0.037), and the 2014 (I = −0.137) and 2021 (I = −0.292) batches display a non-significant tendency toward dispersion. These results suggest that the weak positive autocorrelation observed in the full inventory under the main specification may reflect the cumulative aggregation of successive inscription batches rather than any single round of inscription. However, the batch-specific point patterns are based on markedly different and, in some cases, sparse project counts (9–81 projects), which may also affect the stability and detectability of batch-level spatial autocorrelation.

The supplementary MAUP sensitivity analysis further indicates that the cumulative spatial pattern is scale-sensitive. This analysis re-aggregated the representative ICH point dataset and calculated Moran’s I across occupied grid cells; it therefore serves as a scale-sensitivity check rather than an exact replication of the full-grid main specification reported in Table 2. Moran’s I remained positive across all tested grid sizes and distance bands, but its statistical strength declined as the grid became coarser. At the 2.5 km grid scale, Moran’s I ranged from 0.288 to 0.295 across the 75 km, 95.8 km, and 125 km distance bands, with permutation p-values of 0.011. At the 5 km grid scale, Moran’s I ranged from 0.159 to 0.169, with permutation p-values between 0.066 and 0.070, indicating only marginal support. At the 10 km grid scale, Moran’s I declined to 0.059–0.069 and was not statistically significant. Taken together, the sensitivity results preserve the positive direction of spatial dependence but show that its magnitude and statistical support are sensitive to spatial aggregation. The cumulative pattern is therefore interpreted as limited and scale-sensitive rather than as robust region-wide clustering. Full results are provided in Supplementary Table S1.

3.4 Kernel density estimation

The kernel density surface shows a core-dominated pattern with several lower-intensity secondary concentrations (Figure 3). For the full inventory (panel a, peak density 1,214.21), the highest density values are concentrated in the Guangzhou–Foshan contiguous zone, forming the primary recognition core. A secondary, lower-intensity belt extends along the eastern bank of the Pearl River Estuary through Dongguan, Shenzhen, and Hong Kong, with smaller concentrations around Zhongshan and the western-bank cities. These patterns describe concentrations of registered recognition locations rather than the original spatial diffusion of cultural practices.

FIGURE 3

The batch-specific surfaces reveal how this recognition pattern accumulated over time. In 2006 (panel b, peak density 186.58) and 2008 (panel c, peak density 274.03), the highest-density areas remained centred on the Guangzhou–Foshan core. The 2011 surface (panel d, peak density 232.03) retained this core while additional density areas appeared further south. The 2014 surface (panel e, peak density 11.34) is visually diffuse and should be interpreted cautiously because this batch contains only nine projects. The 2021 surface (panel f, peak density 64.02) shows a more dispersed and relatively polycentric pattern compared with the earlier batches. Overall, the KDE surfaces suggest a persistent Guangzhou–Foshan recognition core and a weaker peripheral pattern, with some broadening of the recognition footprint in the most recent batch. The bandwidth sensitivity analysis preserved the Guangzhou–Foshan concentration across the 10 km, 20 km, and 30 km search radii, although local fragmentation declined as bandwidth increased. The 10 km surface produced multiple local peaks, whereas the 20 km and 30 km specifications resolved these into broader concentrations. The main KDE result is therefore interpreted as a stable indication of a Guangzhou–Foshan recognition core rather than evidence of precise local boundaries. Full sensitivity results are provided in Supplementary Table S2.

3.5 Exploratory Getis–Ord Gi* local pattern analysis

The Getis–Ord Gi* analysis identified only spatially limited nominal local signals (Figure 4). For the full inventory (panel a), the unadjusted output shows a 90% nominal hotspot around the north-central Guangzhou–Foshan core and nominal coldspot signals in the south-eastern maritime area around Hong Kong and the estuary, including one small 95% coldspot. The Guangzhou–Foshan signal is spatially consistent with the broader KDE pattern, whereas the south-eastern coldspot signals describe the national recognition inventory rather than cultural scarcity. However, no local unit survived FDR correction, so these patterns are treated only as exploratory indications of localized tendencies. The batch-specific panels show similarly limited nominal signals.

FIGURE 4

Across individual inscription batches, nominal local signals were similarly limited, and no cell reached the 99% confidence level (Table 3). The minimum adjusted p/q value was 0.197 for the total inventory and 0.280, 0.365, 0.110, and 0.576 for the 2006, 2008, 2011, and 2021 batches, respectively; the 2014 batch was too sparse for stable FDR interpretation. Full correction results are provided in Supplementary Table S3.

3.6 Standard deviational ellipse analysis

As shown in Table 4 and Figure 5, the mean centre of ICH project locations remained relatively stable across batches, varying only between 113.390°E and 113.537°E in longitude and between 22.738°N and 22.910°N in latitude. This indicates that the centre of the recognition inventory remained within the Pearl River Delta core. The major-axis standard distance increased from 68.01 km in 2006 to 97.05 km in 2011, suggesting a broader batch-specific spread of recognized project locations during the early inscription period. After 2011, the major axis decreased to approximately 86 km in 2014 and 2021, while the minor axis fluctuated within a narrower range.

FIGURE 5

The major-axis orientation ranged from approximately 97°–121° across batches, so the major axis consistently follows a WNW–ESE alignment. Although moderate fluctuations occurred over time, the ellipses maintained a broadly consistent directional structure. Overall, the SDE results provide a descriptive indication of broader spatial dispersion up to 2011 followed by relative stabilization, while the mean centres remained concentrated within the Pearl River Delta core region. However, changes in ellipse size should be interpreted cautiously because the number of projects differs substantially across inscription batches. In particular, the large 2011 batch and the small 2014 batch may affect the estimated spread of the ellipse independently of any substantive change in the spatial organization of ICH recognition.

3.7 Correlation and Geographic Detector analysis

To examine broader city-level contextual conditions, a six-dimension screening system comprising twelve socio-economic, demographic, industrial, urbanization, natural-environmental, and spatial-control indicators was retained (Table 5). The demographic dimension includes resident population, urban population, and population density; the economic dimension includes GDP and GDP density; the industrial dimension includes industrial output and the industrialization index; the urbanization dimension includes urbanization rate and road density; the natural-environmental dimension includes average elevation and annual precipitation; and administrative land area is retained as a spatial control. Most indicators refer to 2021 and are matched to the cumulative 2006–2021 recognition inventory. They are therefore used for directional city-level screening rather than as explanatory determinants of individual inscription decisions. Historical-cultural context is examined separately through the differentiated typological correspondence analysis described above.

3.8 Hydrological directional correspondence analysis

Hydrological conditions were incorporated through two city-level indicators derived from the Landsat-based remote-sensing and GIS analysis of water-surface change in the GBA conducted by Jiang et al. (2022): the 2018 water surface ratio (WSR 2018) and the change in water surface ratio between 2000 and 2018 (ΔWSR 2000–2018). WSR represents water-surface area as a percentage of total spatial area and captures the relative prominence of riverine and estuarine water space, whereas ΔWSR describes directional change in that hydrological context over the study period.

The vector chart reported by Jiang et al. (2022) presents WSR trajectories for nine GBA jurisdictions—Macao, Dongguan, Foshan, Guangzhou, Huizhou, Jiangmen, Shenzhen, Hong Kong, and Zhaoqing—but does not include Zhuhai or Zhongshan. Values were digitized from the published vector chart and calibrated against the Foshan and Jiangmen 2000; 2018 values explicitly reported by Jiang et al. (2022). The hydrological correspondence analysis was therefore restricted to these nine jurisdictions.

Spearman rank correlations were calculated between WSR2018 or ΔWSR 2000–2018 and the total number of recognized ICH projects and, separately, the count of each official ICH category. Benjamini–Hochberg false discovery rate adjustment was applied separately to the two ten-category correlation families. Because two pairs of digitized hydrological values were close enough for minor extraction error to alter their rank order, a rank-swap sensitivity analysis was also conducted. These tests are interpreted as directional spatial correspondences rather than causal hydrological effects.

It should be noted that Spearman correlation and Geographic Detector analyses evaluate different analytical concepts. Spearman correlation measures the strength of monotonic associations between variables, whereas the Geographic Detector assesses the extent to which spatial stratification corresponds to spatial heterogeneity. Consequently, variables showing weak correlations may still exhibit relatively high q-values if their spatial stratification aligns with the observed distribution of ICH resources. As an exploratory complement, the Geographic Detector single-factor and interaction results are reported only for transparency in Supplementary Table S4.

3.9 City-level contextual screening

As summarized in Table 6, After removal of the heterogeneous aggregate ancient-site indicator, none of the twelve retained city-level socio-economic, demographic, industrial, urbanization, environmental, and spatial-control indicators met the Bonferroni-adjusted threshold of α = 0.0042. Resident population showed the largest positive directional coefficient (ρ = 0.66, nominal p = 0.028), followed by urban population (ρ = 0.55), average elevation (ρ = 0.54), administrative land area (ρ = 0.51), and annual precipitation in the negative direction (ρ = −0.51). The remaining indicators showed weaker associations. Given the small number of jurisdictions and the temporal mismatch between mainly 2021 contextual indicators and the cumulative 2006–2021 inventory, these coefficients are treated as directional screening signals rather than evidence of independent determinants. Geographic Detector outputs are retained only as exploratory descriptive checks.

3.10 Historical-site typological correspondence

The reconstructed mainland GBA historical-cultural register contained 438 nationally or provincially protected sites. Ancient architecture was the largest official category (n = 190), followed by modern important historic sites and representative buildings (n = 176), ancient sites (n = 41), ancient tombs (n = 19), other designated resources (n = 10), and cave temples, stone carvings, and inscriptions (n = 2). The spatial-scale classification was dominated by individual buildings or structures (n = 248) and building complexes (n = 96), but also included archaeological or site-based resources (n = 41) and settlement- or district-scale heritage (n = 22). Functionally, ritual and communal heritage formed a substantial group (n = 77), alongside residential and settlement heritage (n = 49), defensive and military heritage (n = 29), maritime and transport heritage (n = 21), and commercial and craft heritage (n = 14). These distributions show that the former aggregate ancient-site indicator pooled materially and functionally distinct historical environments. The complete typological distribution by official heritage category, spatial scale, and functional context is reported in Supplementary Table S5.

The typological correspondence analysis revealed differentiated directional patterns rather than a single historical-site relationship shared by all ICH types. Traditional fine arts showed a comparatively strong positive correspondence with ancient architecture (Spearman’s ρ = 0.744, nominal p = 0.022), while traditional handicraft skills also showed a positive correspondence with ancient architecture (ρ = 0.684, nominal p = 0.042). At the functional level, traditional handicraft skills were positively aligned with commercial and craft heritage (ρ = 0.736, nominal p = 0.024), and traditional fine arts showed a positive directional correspondence with ritual and communal heritage (ρ = 0.667, nominal p = 0.050).

However, none of these associations survived Benjamini–Hochberg FDR adjustment within their respective comparison families. They are therefore interpreted as exploratory directional correspondences that identify candidate territorial relationships for further investigation rather than as statistically confirmed mechanisms.

3.11 Hydrological directional correspondence

The total national-level ICH inventory showed no strong directional relationship with either the 2018 water surface ratio (Spearman’s ρ = −0.227, p = 0.557) or the 2000–2018 change in water surface ratio (ρ = 0.303, p = 0.429) across the nine jurisdictions represented in the published hydrological series. The category-specific results, however, were heterogeneous in direction and magnitude.

The largest positive WSR2018 coefficient was observed for quyi or folk performing arts (ρ = 0.725, nominal p = 0.027), but this association did not survive FDR adjustment and the category comprised only three projects. Other category-level coefficients varied in direction, with negative coefficients for folk literature, traditional medicine, and traditional music and a positive coefficient for folk customs. For ΔWSR 2000–2018, the largest positive coefficients were observed for traditional fine arts, folk literature, and traditional medicine; none survived FDR adjustment. Full category-level coefficients are reported in Supplementary Table S8.

A rank-swap sensitivity check was conducted because the digitized 2018 WSR values for Hong Kong and Jiangmen and the ΔWSR values for Huizhou and Zhaoqing were closely spaced. The maximum absolute change in category-level Spearman’s ρ was 0.094 for WSR2018 and 0.069 for ΔWSR. No substantive interpretation changed. The complete rank-swap sensitivity results are provided in the supplementary sheet “Hydrology Rank Sensitivity.” Overall, the hydrological results do not support a single water-related spatial relationship shared by the entire ICH inventory; instead, they indicate category-dependent directional variation. Full hydrological correspondence results are reported in Supplementary Table S8.

4 Discussion

The findings identify a Guangzhou–Foshan-centred recognition core, but they do not support a strong or uniformly robust pattern of region-wide agglomeration. Positive spatial autocorrelation is limited to the cumulative inventory under the main specification, weakens with coarser spatial aggregation, and is not accompanied by FDR-robust local Gi* clusters. The central interpretive issue is therefore not simply whether nationally recognized ICH projects are spatially clustered, but how a limited aggregate core emerges from heterogeneous category-specific recognition geographies.

4.1 Typological differentiation within the core–periphery recognition pattern

The category-specific results show that the Guangzhou–Foshan core is disproportionately reinforced by selected ICH types, particularly traditional fine arts, rather than constituting a uniform regional pattern. The relative concentration of traditional fine arts, handicraft skills, and several related categories is consistent with the possibility that heritage forms tied more closely to localized production knowledge, built environments, and historically accumulated cultural infrastructure are more likely to reinforce the regional core.

By contrast, the broader distributions of music and dance are compatible with an interpretation in which performing traditions are less dependent on a single localized production environment. Their recognition geographies may be associated with circulation through performers, associations, schools, ritual communities, and cross-jurisdictional cultural networks. These processes were not measured directly and should therefore be treated as theoretically plausible explanations rather than empirically confirmed mechanisms. Folk customs occupy an intermediate spatial position. One plausible interpretation is that this broad official category combines place-embedded ritual traditions with festivals, procession practices, dragon and lion traditions, and other recurring community activities represented across multiple settlements.

The spatial-coupling results further suggest that some ICH types share similar recognition environments. The close distributional alignment of music and dance, and of drama and folk customs, does not demonstrate functional dependence but may reflect shared territorial settings, community networks, or recognition pathways. The GBA’s aggregate pattern is therefore better understood as the superimposition of category-specific recognition geographies rather than as the product of a single regional mechanism. Other category profiles require more cautious interpretation. The relatively strong Guangzhou–Foshan concentration of traditional medicine may be consistent with the accumulated institutional and urban infrastructure required for recognized medical traditions, while the spatial similarity between traditional drama and folk customs may reflect their shared embedding in recurring community events, temple-based activities, and local performance networks. By contrast, the sparse inventories of folk literature, traditional sports, games and acrobatics, and quyi do not support strong mechanism-specific conclusions. For these smaller categories, the observed distributions are better treated as provisional recognition profiles whose generative processes require project-level historical and transmission evidence.

The differentiated historical-site analysis further changes the interpretation of historical-cultural context. The former aggregate ancient-site variable pooled settlements, buildings, archaeological sites, ritual environments, defensive heritage, commercial and craft settings, and other materially distinct resources. After disaggregation, traditional fine arts and handicraft skills show comparatively strong directional alignment with ancient architecture, while handicraft skills also align directionally with commercial and craft heritage. None of these relationships survives FDR adjustment, so they cannot be treated as statistically confirmed mechanisms. Their value is more specific: they identify plausible territorial contexts that would have been obscured by a single aggregate historical-site count.

4.2 Directional trends in cumulative recognition and spatial broadening

Whereas previous GBA studies (Ren et al., 2025; Mo et al., 2025) analysed the inventory primarily as a static spatial snapshot, the batch-wise decomposition identifies a directional pattern of cumulative recognition. No individual batch shows significant global spatial autocorrelation, whereas the cumulative inventory exhibits weak positive autocorrelation under the main specification. The SDE results similarly indicate broader spatial spread up to 2011 followed by relative stabilization, with mean centres remaining in the Pearl River Delta core and a broadly stable WNW–ESE orientation.

Taken together, these results are consistent with a two-phase directional tendency: the 2006–2011 batches consolidated and extended recognition around the Guangzhou–Foshan core, while later batches show broader representation across peripheral and cross-jurisdictional areas. The negative Moran’s I values for 2014 and 2021 are not statistically significant, so this later pattern is interpreted as spatial broadening rather than a confirmed transition to dispersion. Because batch sizes differ and contextual indicators are unavailable longitudinally for each inscription round, the analysis identifies the direction of spatial change rather than its historical causes.

This apparent two-phase trajectory is more consistent with the timing of the national inscription programme than with evidence of intrinsic change in the heritage itself. The first three batches (2006, 2008, and 2011) followed the establishment and early consolidation of the national representative-item system (State Council of the People’s Republic of China, 2005) and account for 74.5% of the GBA inventory. The more geographically dispersed 2021 batch may be consistent with broader policy attention to balanced regional heritage development and capacity-building after the consolidation of the national ICH system (Ministry of Culture and Tourism of the People’s Republic of China, 2021). However, because policy implementation was not directly measured, this interpretation remains contextual rather than causal. Similar tendencies toward cumulative heritage concentration and subsequent spatial broadening have also been reported in other Chinese heritage regions (Han et al., 2024; Zhang et al., 2024).

4.3 Recognition geography of intangible cultural heritage

The spatial distribution of nationally recognized ICH projects should be understood as a geography of heritage recognition rather than a direct map of cultural origins. Because the dependent variable consists of officially inscribed projects, the observed pattern reflects cumulative nomination, documentation, evaluation, and inscription processes operating across jurisdictions and historical periods. The Guangzhou–Foshan concentration may therefore reflect a combination of long-term cultural continuity and cumulative recognition processes. Differences in the capacity to identify, document, and nominate heritage resources represent one plausible institutional explanation, but this capacity was not measured directly in the present study.

This distinction is interpretive rather than a direct empirical test of nomination capacity, for which the retained socio-economic indicators provide no strong proxy. Nevertheless, treating official inventories as recognition outcomes clarifies why lower project counts cannot automatically be interpreted as cultural scarcity and provides a basis for examining spatial inequalities in heritage governance.

This recognition-geography perspective also helps situate the GBA among other Chinese macro-regions. Heritage concentration has likewise been reported in the Yellow River Basin and the Yangtze River Basin, where historical-cultural, socio-economic, accessibility, and environmental conditions have been examined as associated spatial contexts (Chang et al., 2023; Wang et al., 2023). The GBA presents a different regional setting, however, because its recognition core is located within a historically commercial and maritime urban system rather than a region centred primarily on historical political capitals. This distinction provides a contextual basis for examining whether different regional histories are associated with different configurations of heritage recognition.

4.4 Hydrological context and category-sensitive spatial correspondence

The direct hydrological analysis qualifies any simple claim that ICH recognition is concentrated where water surfaces are more extensive. The aggregate inventory is only weakly related to both 2018 water-surface prominence and long-term change in water surface ratio. More importantly, the direction and magnitude of association vary across ICH categories. This heterogeneity suggests that riverine and estuarine environments operate as a territorial medium rather than as a universal determinant of ICH concentration.

Given the Pearl River Delta’s riverine and estuarine setting, water systems provide a plausible territorial context for mobility, exchange, settlement connectivity, and inter-community contact. Their relevance may therefore differ across cultural forms that rely on different combinations of community recurrence, circulation, localized production, and built environments. The present correlations do not identify these pathways directly, but they indicate where category-specific hydrological relationships may warrant closer historical and sub-city investigation.

The nominal quyi–WSR2018 coefficient should not be overinterpreted because quyi comprises only three projects and the association does not survive FDR adjustment. Overall, the heterogeneous coefficients support treating water systems as a category-sensitive geographical context rather than a uniform or independently causal influence.

The revised contextual analyses do not identify a single city-level factor that uniformly accounts for the geography of nationally recognized ICH. The retained socio-economic and demographic indicators show weak or statistically uncertain directional associations, while the disaggregated historical-site and hydrological analyses reveal substantial category-specific heterogeneity. Consistent with the recognition-geography framing, these patterns are interpreted as contextual spatial correspondences with inscription outcomes rather than as determinants of heritage origin. This interpretation is consistent with research indicating that financial resources, administrative institutions, stakeholder networks, museums, and documentation systems are important components of heritage preservation and cultural governance (Moreno-Mendoza et al., 2019; Hess et al., 2018; Seila et al., 2025).

The Geographic Detector interaction results are not used to support substantive claims about combined factor effects. With only eleven city-level units, the overlay of discretized factors can produce sparse strata and unstable interaction q-values. The interaction matrix is therefore retained only as a transparency check, whereas substantive interpretation focuses on the directional Spearman screening and the separately specified historical-site and hydrological correspondence analyses.

4.5 Implications for heritage conservation

The findings carry several implications for regional cultural policy. First, the primacy of the Guangzhou–Foshan core suggests that safeguarding resources and recognition capacity may be concentrated in historically central areas. However, lower national-level project counts in western, inland, or cross-jurisdictional areas should not be read as evidence of cultural poverty; they may instead reflect under-recognition, uneven nomination capacity, different listing pathways, or institutional mismatch.

Second, the directional broadening of recent recognition suggests a differentiated regional strategy: established core areas may require stronger transmission and continuity support, whereas under-recognized jurisdictions may benefit more from systematic surveys, documentation assistance, and nomination capacity-building.

Finally, the inclusion of Hong Kong and Macao within the same regional analytical frame highlights the need for greater comparability and communication across distinct heritage-governance systems. Rather than assuming that national-level project counts are directly equivalent across jurisdictions, regional cooperation could focus on compatible documentation practices, joint research and survey initiatives, and cross-jurisdictional transmission programmes for heritage forms shared across the GBA.

4.6 Limitations and future research

Several boundaries remain important when interpreting the findings. The study examines nationally recognized ICH project locations and therefore describes recognition geography rather than the complete geography of cultural practice. Point-based results remain sensitive to registration locations and spatial scale, as indicated by the MAUP and Gi* sensitivity analyses. The historical-site and hydrological correspondence tests are based on nine comparable or source-covered jurisdictions and involve multiple exploratory category-level comparisons; although FDR adjustment and rank-sensitivity checks were applied, these results should be interpreted as directional evidence rather than confirmatory mechanisms. In addition, mainly contemporary contextual indicators are compared with a cumulative 2006–2021 recognition inventory, so early inscription decisions cannot be attributed to present-day conditions. Future work can extend the present framework through longitudinal hydrological and institutional indicators, sub-city heritage data, and direct evidence on practice and transmission locations.

5 Conclusion

Using an integrated GIS framework, this study shows that national-level ICH recognition in the Guangdong–Hong Kong–Macao Greater Bay Area exhibits a weak, scale-sensitive, and core-dominated spatial pattern centred on Guangzhou and Foshan, but that this aggregate pattern is not shared uniformly across ICH types. Traditional fine arts show the clearest Guangzhou–Foshan concentration, whereas music, dance, and folk customs display broader estuarine distributions, and several categories show closely aligned city-level recognition profiles. Disaggregated historical-site and hydrological analyses likewise reveal category-sensitive directional correspondences, although no category-level relationship survives FDR adjustment. These findings support a recognition-geography perspective in which the observed regional pattern is understood as the superimposition of category-specific inscription geographies shaped by differentiated territorial contexts and recognition processes rather than as a direct map of cultural origins. For heritage governance, this implies that lower project counts in particular jurisdictions or categories should not automatically be interpreted as cultural scarcity and that documentation, nomination, and safeguarding strategies should be adapted to the spatial characteristics of different ICH types.

References

• 1

  ChangB.DingX.XiJ.ZhangR.LvX. (2023). Spatial-temporal distribution pattern and tourism utilization potential of intangible cultural heritage resources in the yellow river basin. Sustainability15 (3), 2611. 10.3390/su15032611

  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  GetisA.OrdJ. K. (1992). The analysis of spatial association by use of distance statistics. Geogr. Anal.24 (3), 189–206. 10.1111/j.1538-4632.1992.tb00261.x

  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  HanQ.TaoF.HongZ.QinG.WeiY.ChenY.et al (2024). Research on the spatiotemporal distribution and factors influencing intangible cultural heritage in Fujian province from a multiscale perspective. Herit. Sci.12, 239. 10.1186/s40494-024-01359-2

  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  HessM.ColsonA.HindmarchJ. (2018). Capacity building and knowledge exchange of digital technologies in cultural heritage institutions. Mus. Int.70 (1–2), 48–61. 10.1111/muse.12192

  • CrossRef
  • Google Scholar
  • View reference in article

• 5

  JiangX.LiaoX.QuanX.XueJ.ZhongY.YuB. (2022). Analysis of water surface ratio evolution in guangdong-hong kong-macao greater Bay area based on landsat. Renmin Zhujiang43 (6), 21–27. 10.3969/j.issn.1001-9235.2022.06.004

  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  KaradenizE.ŞekeroğluA.KayaA. Y.ZhangM.AdigüzelF. (2026a). A spatial planning approach for the sustainability of provisioning ecosystem services with economic value: the case of Türkiye. J. Environ. Manag.399, 128652. 10.1016/j.jenvman.2026.128652

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  KaradenizE.ErS.AydoğduM.SünbülF. (2026b). Bridging expert knowledge and machine intelligence: a hybrid spatial indicator framework for ecotourism suitability. Ecol. Indic.183, 114678. 10.1016/j.ecolind.2026.114678

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  MengL.ZhuC.PuJ.WenB.SiW.-F. (2022). Study on the influence mechanism of intangible cultural heritage distribution from man–land relationship perspective: a case study in Shandong Province. Land11 (8), 1225. 10.3390/land11081225

  • CrossRef
  • Google Scholar
  • View reference in article

• 9

  Ministry of Culture and Tourism of the People's Republic of China (2021). The 14th Five-Year Plan for the Protection of Intangible Cultural Heritage. Available online at: https://zwgk.mct.gov.cn/zfxxgkml/fwzwhyc/202106/t20210609_925092.html (Accessed April 8, 2026).

  • Google Scholar

• 10

  MoB.ZhangX.ChenW.ZhouR.ZhangJ. (2025). Research on the spatial distribution characteristics and influencing factors of sports intangible cultural heritage in the Guangdong–Hong Kong–Macao Greater Bay area. Sci. Rep.15, 27147. 10.1038/s41598-025-10734-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11

  MoranP. A. P. (1950). Notes on continuous stochastic phenomena. Biometrika37 (1–2), 17–23. 10.1093/biomet/37.1-2.17

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12

  Moreno-MendozaH.Santana-TalaveraA.LeónC. J. (2019). Stakeholders of cultural heritage as responsible institutional tourism product management agents. Sustainability11 (19), 5192. 10.3390/su11195192

  • CrossRef
  • Google Scholar
  • View reference in article

• 13

  NaheedS.ShooshtarianS. (2022). The role of cultural heritage in promoting urban sustainability: a brief review. Land11 (9), 1508. 10.3390/land11091508

  • CrossRef
  • Google Scholar
  • View reference in article

• 14

  PangL.WuL. (2023). Distribution characteristics and influencing factors of intangible cultural heritage in beijing–tianjin–hebei. Herit. Sci.11, 19. 10.1186/s40494-023-00862-2

  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  RenK.LiC.ZhanJ.LamJ. F. (2025). Spatiotemporal distribution and influencing factors of intangible cultural heritage in the guangdong–hong kong–macao region. Herit. Sci.13, 407. 10.1038/s40494-025-01978-3

  • CrossRef
  • Google Scholar
  • View reference in article

• 16

  SeilaF.SelimG.NewisarM. (2025). A systematic review of factors contributing to ineffective cultural heritage management. Sustainability17 (1), 366. 10.3390/su17010366

  • CrossRef
  • Google Scholar
  • View reference in article

• 17

  SilvermanB. W. (1986). Density Estimation for Statistics and Data Analysis. London: Chapman & Hall. 10.1007/978-1-4899-3324-9

  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  State Council of the People's Republic of China (2005). Opinions of the General Office of the State Council on Strengthening the Protection of China's Intangible Cultural Heritage (Guobanfa [2005] No. 18). Beijing: State Council of the People's Republic of China.

  • Google Scholar
  • View reference in article

• 19

  WangJ. F.ZhangT. L.FuB. J. (2016). A measure of spatial stratified heterogeneity. Ecol. Indic.67, 250–256. 10.1016/j.ecolind.2016.02.052

  • CrossRef
  • Google Scholar
  • View reference in article

• 20

  WangJ.ChenM.ZhangH.YeF. (2023). Intangible cultural heritage in the Yangtze river basin: its spatial distribution characteristics and influencing factors. Sustainability15 (10), 7960. 10.3390/su15107960

  • CrossRef
  • Google Scholar

• 21

  YueZ.RenG.ZhangY.ZhangH.ChuF.QiX.et al (2025). Spatiotemporal distribution pattern and evolution characteristics of intangible cultural heritage in northwest China. Geocarto Int.40, 2567951. 10.1080/10106049.2025.2567951

  • CrossRef
  • Google Scholar
  • View reference in article

• 22

  YuillR. S. (1971). The standard deviational ellipse; an updated tool for spatial description. Geogr. Ann. Ser. B, Hum. Geogr.53 (1), 28–39. 10.1080/04353684.1971.11879353

  • CrossRef
  • Google Scholar
  • View reference in article

• 23

  ZhangZ.LiQ.HuS. (2022). Intangible cultural heritage in the yellow river basin: its spatial–temporal distribution characteristics and differentiation causes. Sustainability14 (17), 11073. 10.3390/su141711073

  • CrossRef
  • Google Scholar
  • View reference in article

• 24

  ZhangZ.CuiZ.FanT.RuanS.WuJ. (2024). Spatial distribution of intangible cultural heritage resources in China and its influencing factors. Sci. Rep.14, 4960. 10.1038/s41598-024-55454-2

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article