Introduction

The accumulation of global ecological risks, the abnormal intensification of climate change, and resource shortages have led to the weakening of the ecosystem’s self-regulation function, seriously threatening the sustainable development of the natural-social-economic composite ecosystem1,2. Since the United Nations proposed the 2030 Sustainable Development Goals in 2015, evaluating ecosystem carrying capacity (ECC) has become a key topic in global environmental change research3. Initially, carrying capacity was applied to engineering geology4. With the acceleration of industrialization and urbanization, issues such as land loss and environmental pollution have continuously emerged. Consequently, the term carrying capacity has been used to describe the maximum capacity of ecosystems to withstand environmental changes5,6. The concept of carrying capacity was first introduced in 19217,8 and formally incorporated into ecology in 1922. With the ongoing development of the social economy, carrying capacity has become closely linked to environmental pollution, urban congestion, population density, and ecological deterioration. The concept of carrying capacity has evolved to accommodate different research directions9,10,11. In China, around the 1990s, scholars provided various interpretations of ECC, emphasizing the interactive system between humans and nature, the self-sustaining and self-regulating nature of ecosystems, the symbiosis and compatibility of resources and the environment, and the support for social and economic development12,13. Currently, the development of carrying capacity research is in a rapid growth stage, focusing primarily on the ECC of tourism, agriculture, forestry, fisheries, and river basins14,15,16,17. Numerous studies have examined ECC in karst areas; however, these studies are relatively simple, concentrating mainly on the ECC of water resources, mountain cities, and world natural heritage sites (HS) in karst mountainous regions, with less emphasis on composite ECC studies18,19. With ongoing improvements in population quality and economic development, ECC evaluations have expanded beyond single factors and current conditions to focus on multi-factor comprehensive influences and long-term scales20. Common research methods include the ecological footprint model21, comprehensive index method22, fuzzy comprehensive evaluation method, and technique for order preference by similarity to an ideal solution method23,24. These methods are mature, practical, and produce credible results. However, practical research has revealed significant discrepancies between research results and reality due to differences in research scale and the difficulty of obtaining crucial research parameters, leading to less accurate results25,26. Currently, ECC research is primarily conducted on a large scale, encompassing global countries, watersheds, urban agglomerations, and counties. Research on the ecological environment of key areas is relatively shallow, resulting in a superficial analysis of the research objects and an incomplete explanation of their ecological environment characteristics.

World heritage (WH) represents the pinnacle of human ingenuity and natural evolution27. It is an irreplaceable and valuable resource with outstanding universal value28,29. WH serves as an important testament to social development, geological evolution, and scientific progress, while also enhancing national pride. As of 2022, there are 167 WH contracting states, 1154 world HS, and 30 karst world heritage sites (KWHS) globally. In the South China KWHS, the core area is the KWHS and the buffer zone is its peripheral protection area30,31, which can protect the integrity and sustainability of the site from threats32. According to the WH Center in 2021, 30% of the world’s natural HS face threats from natural factors such as extreme weather, wildfires, and climate change, significantly impacting these sites. Human activities also pose substantial threats to WH33. Specifically, the tourism-driven economic development spurred by the WH brand effect34,35 and environmental pollution have severely threatened the aesthetic value, integrity, and biodiversity protection of these sites.

China’s 12 world natural HS have become major attractions for both domestic and international tourists due to their outstanding universal value36,37. In terms of development conditions, degree of development, and landscape aesthetics, the karst regions in southern China are exemplary. These regions belong to the tropical-subtropical monsoon type and are listed on the WH List under criteria (vii) and (viii)38,39. The unique geological formations, evolutionary processes, ecosystems, and biodiversity of the southern Chinese karst regions contribute significantly to their global value and importance40,41.

The fact that KWHS are nurtured in the ecologically fragile karst region and are threatened by the threats caused by the continuous development of heritage tourism has aroused scholars’ attention to the ecological conservation of KWHS. Fang et al.42 found that awareness of heritage values and positive tourism impacts are important for the formation of the concept of conservation responsibility among residents in the buffer zone of HS, and Zhang et al.43 revealed that the habitat quality of KWHS generally maintained a better level from 2000 to 2020, and that there was a positive correlation between the habitat quality and the altitude. Zhang et al.44 found that relatively stable land use types promoted ecosystem stability in KWHS. These studies provide relevant information about the ecosystem of the KWHS, but most of them are limited to the buffer zone only, and there are few studies on the evaluation of the ecological environment status that spatially integrates the natural, socio-economic, and tourism landscape elements.

In summary, this paper evaluates the spatial and temporal evolution trend of ECC using the comprehensive evaluation method, assesses the spatial aggregation characteristics of ECC using cold hotspot analysis, and finally explores the driving factors of ECC based on geo-detectors. The objectives of this study are to (1) assess the spatial and temporal evolution characteristics of the ECC of the KWHS from 2010 to 2020 (2) reveal the spatial agglomeration of the ECC of the KWHS (3) explore the key driving factors affecting the ECC. Investigating the ecological environment of KWHS and conducting large-scale, high-precision assessments of ECC impact factors will enhance our understanding of the current environmental status and challenges faced by these HS. This understanding is crucial for the rational use of heritage resources, promoting the harmonious development of heritage protection and economic growth, and providing valuable insights for ecological conservation in other regions.

Materials and methods

Research area

Shi Bing KWHS

Shibing KWHS is located in the central and eastern part of Guizhou Province and the northern part of Shibing County, Qiannan Prefecture. The total area is 282.98 km², with the HS covering 102.80 km² and the buffer zone spanning 180.5 km². It is situated at coordinates 27°10′16′′N, 108°05′40′′E, with an average annual precipitation of 1220 mm and an average annual temperature of 16 °C. The elevation ranges from 600 to 1250 m, with an average altitude of 912 m (Fig. 1). Shibing Karst was successfully designated as a World HS in 2014. It is an exemplary representative of the world’s tropical and subtropical dolomite karst, characterized by its unique geomorphological features, long evolutionary history, and diverse landforms, which significantly contribute to the integrity of the South China Karst series.

Fig. 1: The location of Study area.
figure 1

a represents Shibing heritage site; b represents Libo-Huanjiang heritage site.

Full size image

Libo - Huanjiang KWHS

The Libo-Huanjiang KWHS is situated at the junction of Libo County, Guizhou Province, and Maonan Autonomous County, Guangxi. The central coordinates are 25°12′08′′N, 107°58′40′′E. The HS receives an average annual precipitation of 1752 mm, with an average annual temperature of 15 °C and an average altitude of 758 m. The total area of the HS is 730.16 km², with the core area encompassing 295.18 km² (Fig.1). Libo-Huanjiang Karst was successfully designated as a HS in 2007. It showcases the complex geological background necessary for the development of conical karst, including lithology and structure, explains the intricate history of climate evolution in this region, and preserves a variety of surface and subterranean landscapes along with unique primitive forest ecosystems. It is a quintessential representative of conical karst.

Data sources

Based on Landsat 5 TM and Landsat 8 OLI remote sensing satellite images with a band resolution of 30 m, Land use cover change data was obtained by combining supervised classification and visual interpretation (Table 1). Land use data were obtained every 5 years from 2010 to 2020. NDVI data, with a resolution of 30 meters, was sourced from the National Ecological Science Data Center(http://www.geodata.cn). Net Primary Production (NPP) data was derived from the MOD17A3HGF dataset with a spatial resolution of 500 m. Landscape fragmentation and landscape disturbance indices were calculated using Fragstats. Terrain data, primarily from NASA (https://doi.org/), with a spatial resolution of 30 m, was used to process topographic relief and elevation data.

Table 1 Data source and description
Full size table

The habitat quality index was simulated using the Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) model. Soil erosion data was calculated using the Revised Universal Soil Loss Equation (RUSLE). Population density data, with a resolution of 100 m, were sourced from the World Population Database (https://hub.worldpop.org/). GDP data, with a resolution of 250 m, was sourced from the Geographic Remote Sensing Ecological Network (www.gisrs.cn). Ecological sensitivity and human disturbance indices were calculated based on land use data. Meteorological data, with a resolution of 1 km, was sourced from the Geographic Resource Data Cloud (www.gscloud.cn).

To ensure spatial consistency of all data, it was processed by clipping, projection, and resampling to a spatial resolution of 30 m.

Methods

Description and extraction of ECC assessment factors

Amount of soil erosion

Soil erosion is a fundamental indicator of ecological quality and illustrates the ecological impact of human interactions45. The RUSLE model is a basic tool for assessing soil erosion46. Its formula is as follows:

$${\rm{A}}={\rm{R}}\times {\rm{K}}\times {\rm{L}}\times {\rm{S}}\times {\rm{C}}\times {\rm{P}}$$
(1)

Among them, \({\rm{A}}\) is the actual annual average soil erosion modulus; \({\rm{R}}\) is rainfall erosivity factor; \({\rm{K}}\) is soil erodibility factor; \({\rm{L}}\) and \({\rm{S}}\) are slope length and slope factor, dimensionless; \({\rm{C}}\) is the surface vegetation coverage and management factor, dimensionless; \({\rm{P}}\) is the factor of soil and water conservation measures, dimensionless.

Habitat quality index

Habitat quality index represents the self-regulation and self-sustaining ability of regional ecosystems. In this paper, the InVEST model is used to evaluate the habitat quality index47. The calculation method is shown in the formula:

$${{\rm{Q}}}_{{\rm{xj}}}={{\rm{H}}}_{{\rm{j}}}\left\{1-\left(\frac{{{{\rm{D}}}^{{\rm{z}}}}_{{\rm{xj}}}}{{{{\rm{D}}}^{{\rm{z}}}}_{{\rm{xj}}}+{{\rm{K}}}^{{\rm{z}}}}\right)\right\}$$
(2)

In the formula, \({{\rm{Q}}}_{{\rm{xj}}}\) is the habitat quality of patches in land use type \({\rm{j}}\) ; \({{\rm{H}}}_{{\rm{j}}}\) is the habitat suitability of patches in land use type \({\rm{j}}\) ; \({{\rm{D}}}_{{\rm{xj}}}\) is the threatened level of patches in land use type \({\rm{j}}\), and \({\rm{z}}\) is a constant.

Landscape fragmentation and landscape disturbance index

Different landscape types play different roles in maintaining biodiversity, improving landscape structure, and promoting the overall function of the landscape. At the same time, they have different resistance to external disturbances. Landscape disturbance and landscape fragmentation represent the pressure of environmental damage. In this paper, the landscape disturbance index is constructed with reference to48,49, and the formula is:

$${\rm{LD}}{{\rm{I}}}_{{\rm{i}}}={\rm{a}}{{\rm{C}}}_{{\rm{i}}}+{\rm{b}}{{\rm{H}}}_{{\rm{i}}}+{\rm{c}}{{\rm{F}}}_{{\rm{i}}}$$
(3)

Among them, \({\rm{a}}\), \({\rm{b}}\) and \({\rm{c}}\) are the weights of landscape fragmentation, diversity index and fractal dimension, respectively, reflecting the influence of each index on the ecological environment represented by the landscape to varying degrees, in which \({\rm{a}}+{\rm{b}}+{\rm{c}}=1\) ; the weights of research results \({\rm{a}}\), \({\rm{b}}\) and \({\rm{c}}\) in related fields are 0.5, 0.3 and 0.2 ; \({\rm{LD}}{{\rm{I}}}_{{\rm{i}}}\) is landscape disturbance index, \({{\rm{C}}}_{{\rm{i}}}\) is landscape fragmentation, landscape fragmentation, \({\rm{H}}\) is diversity index, \({{\rm{F}}}_{{\rm{i}}}\) is fractal dimension.

Other factors

Ecological sensitivity is an important indicator of regional ecosystem stability. According to related research, the ecological sensitivity evaluation results were calculated based on five factors: vegetation cover, elevation, slope, slope direction, and land use type50. Land use intensity and development interference index reflect the degree of human exploitation of the ecological environment, and both factors are calculated based on land use data51,52. Climate indicators are fundamental components of ecosystems, driving material cycles and energy flows within the system53. Among them, temperature, precipitation, and evapotranspiration directly reflect regional climatic conditions, which have significant impacts on regional ecosystems, residents’ lives and social production. Vegetation condition directly affects the structure and function of ecosystems54,55. NDVI and NPP directly reflect the growth condition of regional vegetation and the quality of ecological environment56. Topographic relief and elevation data directly or indirectly affect the regional ecological environment through microclimate differentiation. In addition, human activities also affect the ecological environment, such as population density and economic development index reflect the degree of exploitation of nature by human activities to a certain extent.

ECC assessment

Analytic Hierarchy Process

The Analytic Hierarchy Process (AHP) is a decision-making method that decomposes a problem into multiple levels of objectives, criteria, and alternatives, allowing for both qualitative and quantitative analysis57. The specific steps of the AHP are as follows:

  1. (1)

    Use Yaahp10.5 software to build a hierarchical structure model, including the target layer, influencing factor layer, and index layer.,

    Table 2 Proportion scale table
    Full size table
    Table 3 Evaluation index system of ECC
    Full size table
  2. (2)

    Set the impact indicators. The \({\rm{n}}\) indexes that affect the ECC are set as \({{\rm{y}}}_{1},{{\rm{y}}}_{2},{{\rm{y}}}_{3},\cdots {,{\rm{y}}}_{{\rm{n}}}\).

  3. (3)

    Construct a comparison matrix. Starting from the second layer, use the comparison matrix and the 1–9 scale method to measure the importance of each pair of indicators. Experts in the field score the relative importance of each index, and the average values are calculated. The Delphi method is then used to obtain the comparison scale value for each index, forming the judgment matrix. Based on the judgment matrix, the relative importance of indicators at this level is compared in pairs. Matrix analysis allows for a clear understanding of the main influencing factors at this and the previous levels. The 1–9 scale method is used to evaluate the judgment matrix, as shown in Table 2.

  4. (4)

    Construct the judgment matrix. The importance of different indicators is scored according to Tables 2, 3, and the judgment matrix is constructed:

    $${\rm{X}}=\left[\begin{array}{ll}{{\rm{x}}}_{1,} \qquad{\scriptstyle{1}^{\displaystyle{\cdots}{{\rm{x}}}}}_{1,\quad{\rm{n}}}\\ \quad\vdots \qquad\ddots \quad\vdots \\ {{\rm{x}}}_{{\rm{n}},}\qquad{\scriptstyle{1}^{\displaystyle{\cdots} {{\rm{x}}}}}_{{\rm{n}},\quad{\rm{n}}}\end{array}\right]$$
    (4)
  5. (5)

    Calculate the maximum eigenvalue and its corresponding eigenvector of matrix \({\rm{X}}\):

    $${{\rm{\lambda }}}_{\max }=\frac{1}{{\rm{n}}}\times \mathop{\sum }\limits_{{\rm{i}}=1}^{{\rm{n}}}\mathop{\sum }\limits_{{\rm{i}}=1}^{{\rm{n}}}{{\rm{x}}}_{{\rm{i}}\,,{\rm{j}}}\times \frac{{{\rm{\omega }}}_{{\rm{j}}}}{{{\rm{\omega }}}_{{\rm{i}}}}$$
    (5)
  6. (6)

    Standardize the eigenvector58,59. The \({{\rm{\lambda }}}_{\max }\) eigenvector, is calculated and normalized, ω\(=\left({{\rm{\omega }}}_{1},{{\rm{\omega }}}_{2},{{\rm{\omega }}}_{3},\ldots ,{{\rm{\omega }}}_{{\rm{n}}}\right)\), and satisfies \({{\rm{\omega }}}_{{\rm{n}}}=\mathop{\sum }\nolimits_{{\rm{j}}=1}^{{\rm{n}}}{{\rm{x}}}_{{\rm{i}},{\rm{j}}}/\mathop{\sum }\nolimits_{{\rm{j}}=1}^{{\rm{n}}}\mathop{\sum }\nolimits_{{\rm{j}}=1}^{{\rm{n}}}{{\rm{x}}}_{{\rm{i}},{\rm{j}}}\). Using the power method, the maximum eigenvalue of the matrix \({\rm{X}}\) and its corresponding eigenvector \({\rm{\omega }}\) are calculated and standardized.

  7. (7)

    Calculate the consistency index. Consistency indicators:

    $${\rm{CI}}={{\rm{\lambda }}}_{\max }-{\rm{n}}/{\rm{n}}-1$$
    (6)
  8. (8)

    Consistency test. By looking up the table to analyze the \({\rm{n}}\) value, the \({\rm{CR}}\) value of the random index is obtained, and the \({\rm{CI}}/{\rm{CR}}\) value is calculated. If the \({\rm{CI}}/{\rm{CR}}\) value is less than or equal to 0.1, it indicates that the inconsistency of the matrix \({\rm{X}}\) is acceptable, that is, the hierarchical total ranking meets the requirements of the consistency test.

CRITIC weight method

The CRITIC weighting method, which determines weight through the correlation between indicators, is an objective weighting method. Its core principle is to comprehensively consider the indicators of contrast strength and conflict60. The specific steps are as follows:

  1. (1)

    Data standardization:

    Differences in the dimensions and attributes of the evaluation indicators need to be standardized. Based on their connotations, indicators are defined as positive or negative. Positive indicators have a positive impact on the corresponding aspects of ECC, while negative indicators have a negative impact.

    Positive indicators:

    $${{\rm{Y}}}_{{\rm{ij}}}={{\rm{x}}}_{{\rm{j}}}-{{\rm{x}}}_{\min }/{{\rm{x}}}_{\max }-{{\rm{x}}}_{\min }$$
    (7)

    Negative indicators:

    $${{\rm{Y}}}_{{\rm{ij}}}={{\rm{x}}}_{\max }-{{\rm{x}}}_{{\rm{j}}}/{{\rm{x}}}_{\max }-{{\rm{x}}}_{\min }$$
    (8)

    In the formula: \({{\rm{Y}}}_{{\rm{ij}}}\) represents the standard value of the \({\rm{i}}\)-year and \({\rm{j}}\) index, while \({{\rm{X}}}_{{\rm{ij}}}\) is the original value of the \({\rm{I}}\) year and \({\rm{j}}\) index, and \({{\rm{X}}}_{\max }\) and \({{\rm{X}}}_{\min }\) represent the maximum and minimum values of the j-index, respectively.

  2. (2)

    Calculate the variability \({{\rm{S}}}_{{\rm{j}}}\) within each index:

    The index variability is expressed in the form of standard deviation. \({{\rm{S}}}_{{\rm{j}}}\) represents the standard deviation of the jth index, where \({\bar{{\rm{X}}}}_{{\rm{j}}}\) is the average value of the jth index.

    $${\bar{{\rm{X}}}}_{{\rm{j}}}=\frac{1}{{\rm{m}}}\mathop{\sum }\limits_{{\rm{i}}=1}^{{\rm{m}}}{{\rm{X}}}_{{\rm{ij}}}$$
    (9)
    $${{\rm{S}}}_{{\rm{j}}}=\sqrt{\frac{1}{{\rm{m}}-1}}\mathop{\sum }\limits_{{\rm{i}}=1}^{{\rm{m}}}{\left({{\rm{X}}}_{{\rm{ij}}}-{\bar{{\rm{X}}}}_{{\rm{j}}}\right)}^{2}$$
    (10)

    The CRITIC method uses the standard deviation to quantify the degree of variation in the values of each index. A larger standard deviation indicates a wider range of numerical variation for the index. Therefore, an index with a larger standard deviation has a stronger evaluation ability and should be assigned a higher weight.

  3. (3)

    Calculate the conflict \({{\rm{R}}}_{{\rm{j}}}\) between indicators:

    $${{\rm{R}}}_{{\rm{j}}}=\mathop{\sum }\limits_{{\rm{j}}=1}^{{\rm{n}}}\left(1-{{\rm{r}}}_{{\rm{ij}}}\right)$$
    (11)

    The correlation coefficient measures the correlation between different indicators. If an index has a high correlation with other indicators, it indicates overlap in evaluation content, so the weight of such an index should be reduced.

  4. (4)

    Calculate the amount of information \({{\rm{C}}}_{{\rm{j}}}\) for each indicator:

    $${{\rm{C}}}_{{\rm{j}}}={{\rm{S}}}_{{\rm{j}}}\mathop{\sum }\limits_{{\rm{j}}=1}^{{\rm{n}}}\left(1-{{\rm{r}}}_{{\rm{ij}}}\right)={{\rm{S}}}_{{\rm{j}}}{{\rm{R}}}_{{\rm{j}}}$$
    (12)

    In the formula, the larger the \({{\rm{C}}}_{{\rm{j}}}\), the greater the role of the jth index in the entire evaluation index system, and more weight should be assigned to it.

  5. (5)

    Calculate the weight of each index \({{\rm{W}}}_{{\rm{j}}}\):

$${{\rm{W}}}_{{\rm{j}}}=\frac{{{{\rm{C}}}_{{\rm{j}}}}_{{\rm{j}}}}{\mathop{\sum }\limits_{{\rm{j}}=1}^{{\rm{n}}}{{\rm{C}}}_{{\rm{j}}}}$$
(13)

Comprehensive evaluation method and classification

ECC is a critical index for characterizing the current state of the regional ecological environment. A higher ECC indicates a better ecological environment. The ECC results are obtained by applying the weight of each parameter to the evaluation parameters and then summing the results61. To facilitate the comparison of multi-period ECC index, standardization is necessary.

$${\rm{ECC}}=\mathop{\sum }\limits_{{\rm{i}}=1}^{{\rm{n}}}{{\rm{X}}}_{{\rm{i}}}\times {{\rm{W}}}_{{\rm{i}}}$$
(14)
$${{\rm{I}}}_{{\rm{ECC}},{\rm{i}}}=\frac{{{\rm{ECC}}}_{{\rm{i}}}-{{\rm{ECC}}}_{{\rm{i}},\min }}{{{\rm{ECC}}}_{{\rm{i}},\max }-{{\rm{ECC}}}_{{\rm{i}},\min }}$$
(15)

\({\rm{ECC}}\) is the ECC, \({{\rm{X}}}_{{\rm{i}}}\) is the standardized value of the parameter, \({{\rm{W}}}_{{\rm{i}}}\) is the weight value of the parameter, \({{\rm{I}}}_{{\rm{ECC}},{\rm{i}}}\) is the standardized value of the ecosystem in the \({\rm{i}}\) year, and \({{\rm{ECC}}}_{{\rm{i}},\max }\) is the maximum value of the ECC index in the \({\rm{i}}\) year. \({{\rm{ECC}}}_{{\rm{i}},\min }\) is the minimum value of the ECC index in the \({\rm{i}}\) year.

Based on the natural geographical characteristics of Shibing and Libo-Huanjiang KWHS, and referencing the ECC and study area characteristics, the ECC is divided into five levels: weakest (0–0.2), weaker (0.2–0.4), medium (0.4–0.6), stronger (0.6–0.8), and strongest (0.8–1).

Analysis of spatial aggregation of ECC

In this study, the Getis-Ord Gi * index is used to analyze the high/low spatial aggregation of ECC. The spatial distribution of cold and hot spots is calculated as follows62:

$${{\rm{Gi}}}^{* }=\frac{\mathop{\sum }\limits_{{\rm{j}}=1}^{{\rm{n}}}{{\rm{w}}}_{{\rm{i}},{\rm{j}}}{{\rm{x}}}_{{\rm{j}}}-\bar{{\rm{X}}}\mathop{\sum }\limits_{{\rm{j}}=1}^{{\rm{n}}}{{\rm{w}}}_{{\rm{i}},{\rm{j}}}}{\root{{s}}\of{\left[{\rm{n}}\mathop{\sum }\limits_{{\rm{j}}=1}^{{\rm{n}}}{{{\rm{w}}}^{2}}_{{\rm{i}},{\rm{j}}}-{\left(\mathop{\sum }\limits_{{\rm{j}}=1}^{{\rm{n}}}{{\rm{w}}}_{{\rm{i}},{\rm{j}}}\right)}^{2}\right]/\left({\rm{n}}-1\right)}}$$
(16)
$$\bar{{\rm{X}}}=\frac{1}{{\rm{N}}}{\sum }_{{\rm{j}}=1}^{{\rm{n}}}{{\rm{x}}}_{{\rm{j}}}\sqrt{\left(\frac{1}{{\rm{n}}}\mathop{\sum }\limits_{{\rm{j}}=1}^{{\rm{n}}}{{{\rm{x}}}_{{\rm{j}}}}^{2}-{\bar{{\rm{X}}}}^{2}\right)}$$
(17)

In the formula: \({\rm{n}}\) represents the total number of patches in the region, \({{\rm{Gi}}}^{* }\) is the output statistical \({\rm{Z}}\) score, \({{\rm{x}}}_{{\rm{j}}}\) is the attribute value of the spatial unit j, and \({{\rm{w}}}_{{\rm{i}},{\rm{j}}}\) is the spatial weight between the adjacent spatial units and.

Analysis of ECC drivers

Geodetector are powerful tools for detecting spatial heterogeneity and identifying the main drivers of spatial heterogeneity, including factor detection, interaction detection, ecological detection and risk area detection63. This paper uses a factor test to explore the specific effects of the 16 drivers on ECC (i.e., explanatory power q-values), and an interaction test to confirm the extent to which any two drivers interact to contribute to ECC. The specific steps are as follows:

  1. (1)

    Factor detector is mainly used to measure the spatial differentiation of ECC \({\rm{Y}}\) and the degree of interpretation of different influencing factors \({\rm{X}}.\) The calculation formula is:

    $${\rm{q}}=1-\frac{{\sum }_{{\rm{h}}=1}^{{\rm{L}}}{{\rm{n}}}_{{\rm{h}}}{{{\rm{\delta }}}_{{\rm{h}}}}^{2}}{{{\rm{n}}{\rm{\delta }}}^{2}}=1-\frac{{\rm{SSW}}}{{\rm{SST}}}$$
    (18)
    $${\rm{SSW}}=\mathop{\sum }\limits_{{\rm{h}}=1}^{{\rm{L}}}{{\rm{n}}}_{{\rm{h}}{{\rm{\delta }}}^{2}}\,{\rm{SST}}=\mathop{\sum }\limits_{{\rm{h}}-1}^{{\rm{L}}}{{\rm{n}}{\rm{\delta }}}^{2}$$
    (19)

    In the formula, \({\rm{q}}\) is the factor explanatory power, \({\rm{h}}=1,2,\cdots ,{\rm{L}}\) is the number of classifications; \({{\rm{n}}}_{{\rm{h}}}\) and \({\rm{n}}\) are the number of sample units in layer \({\rm{h}}\) and the whole region, respectively. \({{{\rm{\delta }}}_{{\rm{h}}}}^{2}\) and \({{\rm{\delta }}}^{2}\) are the variance of layer h and the whole region, respectively. \({\rm{SSW}}\) is the sum of the intra-layer variance, and \({\rm{SST}}\) is the sum of the total variance of the whole region. The value range of \({\rm{q}}\) is [0,1]. The larger the \({\rm{q}}\) value is, the stronger the explanatory power of the impact factor to the spatial differentiation of ECC is.

  2. (2)

    Interaction detection is used to identify the interaction between different influencing factors, that is, to assess whether the combined effects of \({{\rm{X}}}_{1}\) and \({{\rm{X}}}_{2}\) will increase or decrease the explanatory power of ECC or the effects of these factors on the spatial distribution of ECC are independent of each other. The interaction mainly identifies the influence of the interaction between the two influencing factors on the change of ECC by calculating the sum of the \({\rm{q}}\) values of the two factors and the \({\rm{q}}\) value of the combination of the two influencing factors. Where \({\rm{q}}\)(\({\rm{x}}\) ∩ \({\rm{y}}\)) denotes the interaction between \({\rm{x}}\) and y,\({\rm{Min}}\)(\({\rm{q}}\)(\({\rm{x}}\)),\({\rm{q}}\)(\({\rm{y}}\))) denotes the minimum value between \({\rm{q}}\)(\({\rm{x}}\)) and \({\rm{q}}\)(\({\rm{y}}\)),\({\rm{Max}}\)(\({\rm{q}}\)(\({\rm{x}}\)),\({\rm{q}}\)(\({\rm{y}}\))) denotes the maximum value between \({\rm{q}}\)(\({\rm{x}}\)) and \({\rm{q}}\)(\({\rm{y}}\)), and \({\rm{q}}\)(\({\rm{x}}\))+ \({\rm{q}}\)(\({\rm{y}}\)) denotes the sum of \({\rm{q}}\)(\({\rm{x}}\)) and \({\rm{q}}\)(\({\rm{y}}\)).

Results

Construction and weighting of ECC indicator system

The ECC of KWHS is affected by a combination of factors. And considering the representativeness and availability of the influencing factors, this paper screened 16 factors from three aspects, namely, ecological resilience, resource and environmental supply capacity, and socio-economic coordination capacity, and obtained their weights according to the weight calculation method. The influence factors are represented by X, i.e., NDVI (X1), NPP (X2), elevation (X3), temperature (X4), precipitation (X5), evaporation (X6), topographic relief (X7), soil erosion (X8), habitat quality index (X9), ecological sensitivity (X10), landscape fragmentation (X11), landscape disturbance index (X12), land use intensity (X13), population density index (X14), economic development index (X15) and development disturbance index (X16) (Fig. 2 and Table 3).

Fig. 2: Influence factor weight ranking.
figure 2

a represents Shibing heritage site; b represents Libo-Huanjiang heritage site.

Full size image

Spatial variation characteristics of ECC

The spatial differentiation of ECC in the KWHS is pronounced, presenting an overall spatial layout of high core area and low buffer zones (Fig. 3), which is closely related to the natural and socio-economic environment of the KWHS.

Fig. 3: Spatial distribution of ECC from 2010 to 2020.
figure 3

a represents Shibing heritage site; b represents Libo-Huanjiang heritage site.

Full size image

From 2010 to 2020, the weakest and weaker ECC zones of Shibing HS have changed little, with a slow growth trend over the decade, and are mainly located in the buffer zone. The zones with medium, stronger and strongest ECC are mainly distributed in the core zone and the buffer zone with less human activities, in which the medium ECC zone shows a ‘V’-shaped trend of change, with a slow overall change, while the stronger and strongest ECC zone show significant changes. It is worth noting that from 2015 to 2020, the core area of Shibing HS is mainly at the medium ECC level.

From 2010 to 2020, the zones with weakest and weaker ECC in the Libo-Huanjiang HS change less, showing an inverted‘V’trend, and are mainly distributed in the buffer zone, with only sporadic distribution in the core zone. On the other hand, the area of medium, stronger and strongest ECC zones changes significantly, of which medium ECC zones are mainly distributed in the buffer zone and the transition area between the buffer zone and the core zone, stronger ECC zones have larger areas and are mainly distributed in the buffer zone and part of the core zone, and the strongest ECC zones are more centrally distributed, mainly in the southeastern part of the site and near the watershed of the buffer zone. It is worth noting that the ECC of the core area of the Libo-Huanjiang HS has slightly decreased from 2015 to 2020.

Temporal variation characteristics of ECC

To further clarify the spatial distribution of ECC changes in the KWHS, this study utilized ArcGIS 10.6 software to analyze ECC results and obtained dynamic monitoring maps for the periods 2010–2015, 2015–2020, and 2010–2020 (Figs. 4–5).

Fig. 4: The proportion of different ECC area.
figure 4

a represents Shibing heritage site; b represents Libo-Huanjiang heritage site.

Full size image
Fig. 5: Time evolution characteristics of ECC.
figure 5

a represents Shibing heritage site; b represents Libo-Huanjiang heritage site.

Full size image

In 2010–2015, the ECC of Shibing HS is relatively stable, the area of medium, stronger and strongest ECC area accounts for 96.32% of the total area of the HS, the area of weakest ECC area is slightly larger than the area of weaker ECC area, mainly scattered in the buffer zone and the middle of the core area. 2015–2020, the area of medium, stronger and strongest ECC area shows a trend of shrinkage, only accounting for 70.19% of the total area of the of Shibing HS, while the area of the weaker ECC area increases significantly. 2010–2020 the area of the strongest ECC area of the Shibing HS is drastically reduced, and the area of the medium, stronger and strongest ECC area is only 168.46 km2 accounting for 59.66% of the total area of the HS. Overall, the ecological and environmental conditions of Shibing HS regress from 2010 to 2020.

In 2010–2015, the ECC of Libo-Huanjiang HS is comparatively good, with 92.49% of the total area of the site in the medium, stronger, and strongest ECC zones, while the area of the weakest ECC zone increases slightly, mainly in the vicinity of populated settlements in the buffer zone and in the core zone where anthropogenic modifications are obvious. In 2015–2020, the area of medium ECC area increases significantly, reaching 141.95 km2, while the area of strongest ECC area is opposite, and the area of stronger ECC area remains stable, which is mainly manifested in the local degradation of the core area with high ECC to medium ECC area. 2010–2020, the area of low ECC area gradually expands, while the area of high ECC area locally degrades to medium ECC area. In general, the overall degradation of the ECC in 2010–2020 is obvious.

Spatial aggregation characteristics of ECC

The ECC of the KWHS for the years 2010, 2015, and 2020 were calculated, and the average values were derived using a 300 × 300 grid. Subsequently, a spatial cold and hot spot cluster analysis was performed, followed by a significance test. The results are presented in Fig. 6. Hot spot clusters represent areas with high ECC values, while cold spot clusters represent areas with low ECC values.

Fig. 6: Spatio-temporal pattern characteristics of spatial cold and hot spots of ECC.
figure 6

a represents Shibing heritage site; b represents Libo-Huanjiang heritage site.

Full size image

The hotspot gathering area of Shibing HS is mainly distributed in the core area, the western transition zone between the core area and the buffer zone, and the northern part of the area where human interference is weak (99% confidence level), which is dominated by the forest ecosystem, with a more complete system structure, less anthropogenic interference, and a high ECC of the ecosystem. The cold spot aggregation area is mainly distributed in the anthropogenic activity area of the buffer zone (confidence level 95–99%), where the intensity of anthropogenic interference is also relatively high, the system structure is simple, and the vegetation cover is poorer than that of the core area, so the ECC is lower. The ECC in 2015 was the best within the decade.

The hotspot aggregation area of Libo-Huanjiang HS is mainly distributed in the primitive forest area in the southeast of the site, and there are also sporadic distributions in the central part of the buffer zone (confidence level 99%). These areas are free of interference from human activities, have a warm climate, abundant precipitation, and have an intact ecosystem structure, rich biodiversity, and a high level of ecological resilience, and thus have a high ECC of the ecosystem. Cold spot aggregation areas are mainly distributed in anthropogenic activity areas in the buffer zone, the transition zone between the core zone and the buffer zone, and local areas in the core zone (confidence level 95–99%), which have higher exposure to human activities, have lower vegetation cover than the primary forest area, and are vulnerable to global climate change, and thus have a lower ECC. The ECC in 2010 was the best in the decade.

Obstacle factor diagnosis of ECC

Single factor detection of ECC

The main influencing factors of the ECC of the KWHS were analyses from the perspective of natural and social factors, and the results are shown in Fig. 7. The factors with the strongest influence on the spatial differentiation of the ECC of the Shibing HS in 2010 were the land-use intensity and the economic development index, respectively, and the development interference index and the land-use intensity in 2015, and the development interference index and the habitat quality index in 2020. The explanatory power of the above factors on ECC is greater than 0.61, while the explanatory power of natural factors such as NPP, temperature, and soil erosion is smaller, indicating that the ecological environment in this area is more strongly affected by human interference.

Fig. 7: Detection of influencing factors of ECC.
figure 7

a represents Shibing heritage site; b represents Libo-Huanjiang heritage site.

Full size image

The dominant factors causing spatial differentiation in ECC of Libo-Huanjiang HS from 2010 to 2020 are ecological sensitivity and habitat quality index, both of which have an explanatory power of more than 0.38 for ECC, followed by land-use intensity and landscape disturbance index, and natural factors such as elevation, evaporation, and amount of soil erosion have less explanatory power, indicating that the ecosystems of the Libo-Huanjiang HS are is more complete and more strongly driven by natural factors.

Interactive detection of ECC factors

Factor interaction test can further analyses the effect of the interaction between different factors on the ECC, the results are shown in Fig. 8. 2010–2015, the stronger interaction of Shibing HS on the ECC is the land use intensity∩NDVI, the economic development index∩NDVI, and their explanatory power is more than 0.71. Secondly, the land use intensity, the development interference index has a strong interaction with other factors. The interactions with stronger explanatory power in 2020 were development disturbance index∩NPP, land use intensity∩development disturbance index and, landscape disturbance index∩development disturbance index, with explanatory powers of 0.8042, 0.8039, and 0.8029, respectively, and both habitat quality index and development disturbance index had strong interactions with other factors, indicating that anthropogenic activities have always been an important factor influencing the ecosystem’s development, indicating that anthropogenic activities have always been an important factor affecting the stability of the ecosystem.

Fig. 8: Interactive detection of ECC factors.
figure 8

a represents Shibing heritage site; b represents Libo-Huanjiang heritage site.

Full size image

In 2010, the interactions with higher explanatory power for ECC at Libo-Huanjiang HS were habitat quality index∩ecological sensitivity, land use intensity∩ecological sensitivity, with explanatory powers of 0.6297 and 0.5574, respectively. 2015 was the same as that of 2010, followed by stronger interactions between the habitat quality index, ecological sensitivity and other factors. The interactions with higher explanatory power for ECC in 2020 were habitat quality index∩NDVI, ecological sensitivity∩habitat quality index, and the interactions between other factors of land use intensity were significantly larger than those in 2010 and 2015. Overall, the interactions among multiple factors significantly enhanced the spatial variability of ECC.

Discussions

Spatial and temporal characteristics of ECC in KWHS

2010–2020, Shibing HS ECC overall degradation evolution trend, of which in 2010 the stronger ECC area dominated, 2015 strongest ECC area is the largest, 2020 degradation to the stronger ECC area is dominated. The main reason is that in 2010–2015 Shibing HS vegetation coverage, complex landscape structure and ecological functions, and slow economic development, tourism development is immature, the exploitation rate of nature and lower. Due to the’WH’brand effect, from 2015 to 2020 Shibing HS tourism development is becoming more mature, the level of ecological environment development and utilization increased64,65,66.

From 2010 to 2020, the ECC of Libo-Huanjiang HS showed an overall degradation trend, in which the ECC status was the worst in 2020, which showed that the strongest ECC area was degraded to a stronger ECC area, and the medium ECC area within the core area was expanded. The main reason is that the tourism development of Libo-Huanjiang HS is earlier, the tourism infrastructure is mature, the interference of human activities on the natural environment is enhanced, and its ecological environment is increasingly fragile43,67. Overall, the ECC status of Libo-Huanjiang HS is better than that of Shibing HS from 2010 to 2020, and the degradation rate of ECC of Shibing HS is more obvious than that of Libo-Huanjiang HS.

ECC drivers for KWHS

Natural attributes and anthropogenic factors affect the spatial and temporal changes of regional ECC68. This study found that the ECC of the KWHS was correlated with the habitat quality index, NDVI and ecological sensitivity factors. It reveals that natural factors are potential elements affecting regional ecological stability. Among the driving mechanisms of ecological risk in Bayinbuluk World HS in Xinjiang, landscape factors and natural environment factors are stronger drivers of ecological risk69. In addition, land use intensity, development interference index and population density have strong correlation with ECC. The impacts caused by anthropogenic activities on regional ecological stability are rapid and strong. Natural HS can be ecologically restored to a certain extent under slow economic development70. While agricultural production and tourism economic activities have negative ecological effects on KWHS, population density, tourism factors and woodland conservation have significant correlation with ecological vulnerability71.

In summary, the direction of sustainable development of the KWHS is indeed crucial in the context of increasing environmental problems. As the site is located in the southwest inland plateau with fragile ecological fundamentals, how to find a balance between community economic development and heritage ecological protection has become a central issue in guiding the development of the KWHS.

Suggestions on ecological protection and management of KWHS

The South China KWHS is an exemplary karst landform and natural landscape, possessing significant aesthetic and tourism value72,73,74. It is significantly affected by the vulnerability of the karst environment and human activities, resulting in considerable ecological damage and numerous environmental problems. The uniqueness, vulnerability, and complexity of the KWHS ecological environment highlight the need for75,76. To achieve stable and sustainable development of KWHS, it is crucial to address the close relationship between ecological environment protection and socio-economic development, particularly tourism economic development77,78.

  1. (1)

    Insist on following the principles of sustainable development, community participation in publicity and education, and local residents as key factors in WH conservation and sustainable development79.

  2. (2)

    Adhere to the development of ecotourism without damaging the health of the ecosystem, and encourage community participation in ecotourism activities80. Increase pollution control in the upper reaches of the study area and adjacent areas, reduce the direct entry of tourism pollutants into natural areas, and take protective measures such as wastewater interception and treatment, river cleaning, and ongoing maintenance and monitoring81.

  3. (3)

    Enhance the environmental awareness of the main bodies related to tourism activities, multi-level units of government joint prevention, control and supervision to improve the ecological environment and maintain ecological stability. The sustainable development of WH tourism is the key to coordinating the relationship between heritage protection and utilization82.

To sum up, enhancing ecological security from multiple perspectives, such as increasing people’s participation, encouraging the development of eco-tourism, as well as enhancing the environmental protection awareness of relevant subjects by the joint government, is an important way to enhance the ECC, and an important ecological foundation for the promotion of local economic development.

Limitations and future directions

This study is based on the evaluation of ECC by a combination of RS/GIS technology. However, on the one hand, due to the special positioning of the KWHS, limited socio-economic data about human activities can be collected. Therefore, the indicator system is constructed more mainly based on the natural environment elements. On the other hand, in the identification of impact factors, no clear indicator system has been found at home and abroad for reference, which may still be subjective according to the specific reality of the research object and the operability of each method, and may not completely and accurately reflect the ECC of the study area. To sum up, it is a difficult problem to construct the ECC indicator system in a scientific and reasonable way83. Therefore, it is necessary to further explore the future development trend of the ECC of the KWHS, with a view to providing reference for its future construction and development.

Conclusions

In this study, the spatial and temporal distribution of ECC of Shibing and Libo- Huanjiang HS among the KWHS in South China was analyzed, and the spatial aggregation characteristics and the driving factors of ECC were found during the period of 2010–2020. The conclusions are as follows:

  1. (1)

    Spatially, the ECC of the core areas of the two HS is significantly greater than that of the buffer zones, and the ECC of the buffer zones with no human disturbance is greater than that of the buffer zones with strong human activities.

  2. (2)

    The changes of ECC of the two HS are similar in time, and both of them show a degradation trend. Among them, the degradation rate of ECC in Shibing HS is significantly greater than that in Libo - Huanjiang HS.

  3. (3)

    The ECC of the two HS has strong spatial agglomeration and obvious spatial differentiation. The core area is the hotspot aggregation area, and the buffer area is the cold spot aggregation area.

  4. (4)

    The dynamic change of the ECC of the KWHS is the result of a combination of internal and external factors. The main driving factor of the ECC of Shibing HS is the natural environment factor, while the main driving factor of the ECC of Libo - Huanjiang HS is the socio-economic factors.