Assessment of ecosystem health and driving forces in response to landscape pattern dynamics: the Shibing Karst world natural heritage site case study

Abstract

The Shibing Karst constitutes a pivotal component of the "South China Karst," and its ecosystem health integrity crucially influences the Outstanding Universal Value (OUV) of the corresponding Natural World Heritage (NWH). Consequently, robust ecosystem health assessment (EHA) is imperative for the judicious conservation and management of this heritage, as well as for the sustainable progression of the region. This research assessed the health of the Shibing Karst ecosystem from 2004 to 2020 by employing changes in landscape patterns through the Vigor-Organization-Resilience-Ecosystem Services (VORS) model. Spatial autocorrelation was employed to discern the spatial coherence and evolutionary patterns of ecosystem health, whereas a geo-detector ascertained the pivotal determinants impacting regional ecosystem vitality. The findings revealed that: (1) The landscape patterns distribution in the study area exhibited considerable constancy, primarily comprising forest land, with a rising trajectory in construction land and water, juxtaposed with a recession in shrubland, grassland, paddy land, and dryland expanses. (2) From 2004 to 2020, the ecosystem of the study area maintained its health and remained stable, with mean values of 0.8303, 0.7689, 0.6976, and 0.7824, respectively, showing an evolutionary trend of an initial downtrend trend followed by an upswing, with 2016 marking a pivotal juncture. (3) Spatial clustering analysis highlighted significant clustering characteristics of ecosystem health, with a nominal decrease in the Global Moran's I index from 0.666 to 0.665, which is indicative of a subtle decrease in clustering over time. High-high clustering areas were predominantly located within the World Heritage Site (WHS), while low-low clustering areas were mainly distributed in the southeastern part of buffer zone. (4) Land use and cover change (LUCC) and Ecosystem Services (ESs) were identified as the primary indexes of EHA, with Ecosystem Resilience (ER), Ecosystem Vigor (EV), and Ecosystem Organization (EO) exerting relatively mild influences. This study provides a scientific framework for policymakers in local governance to devise strategies for ecosystem conservation and management, enhances the analytical perspective on the integrity and conservation of Karst Natural World Heritage (KNWH).

Introduction

Ecosystem services, including resource provisioning and environmental services, are fundamental for human survival and development. Ecosystem health is defined as the ability of ecosystems to maintain their spatial structure and ecological processes, self-regulate, recover from external disturbances, and provide essential resources and services for human economic development [1,2,3]. Rapid global economic development and population growth have made human impacts on ecosystems increasingly apparent. The Millennium Ecosystem Assessment (MA) reports that nearly 60% of terrestrial ecosystems are degraded or unsustainable [4]. Consequently, maintaining ecosystem health and achieving harmony between human activities and the natural environment have emerged as critical academic concerns. NWH represents a unique, transboundary, and invaluable asset for human society, possessing OUV that transcends national boundaries and is irreplaceable [5]. NWHs are not only subject to natural decay but also vulnerable to interference from human social and economic activities [6]. Consequently, protecting their OUV and integrity presents significant challenges. As of 2021, 52 WHS have been included in the World Heritage List (WHL) as being in jeopardy. The United Nations Educational, Scientific, and Cultural Organization (UNESCO) mandates that the countries hosting these sites implement requisite measures to restore and conserve WHS [5], underscoring the urgency of protecting WHS. EHA is recognized as an intuitive reflection of ecosystem quality and a vital instrument for monitoring and conserving ecosystems [7]. Ecosystems serve as both the natural substrates and physical carriers of the NWH’s OUV. Monitoring their ecosystem health is crucial for effective conservation and management [8]. This is particularly urgent for KNWH in an ecologically fragile region where there is a pressing need to develop scientific and effective diagnostic approaches for ecosystem health [9, 10].

Karst refers to the combination of soluble rocks and integrated systems above and below the ground that have evolved into distinct landforms. Globally, karst landscapes cover an area of 22 million km² [11], constituting 15% of the Earth's land area [12]. Karst extends over approximately one-third of the national territory in China, primarily across eight southwestern provinces, including Guizhou, Guangxi, Yunnan, Chongqing, Hunan, and three others [13]. First, due to their unique geological evolution and significant scientific and aesthetic value, karst regions have fostered the OUV of NWH, making them some of the best landscapes globally [14]. Second, karst landscapes are fragile ecosystems with deficient environmental carrying capacities and elasticity coefficients [15]. The problem of rocky desertification, aggravated by irrational human activities, has become a serious limitation on these areas' economic and social development [16, 17]. Consequently, the dichotomy between resource utilization and conservation is particularly apparent in karst regions, posing a significant threat to protecting their OUV and integrity [18]. Ecosystem health highlights an ecosystem's capacity to sustain and regenerate itself and its ability to meet the reasonable demands of human societal development. Positive ecosystem health exemplifies the effective synergy between conservation and development at KNWH. Since ecosystems vary in structure, function, and human activity intensities, the mechanisms for assessing ecosystem health and optimizing strategies also vary. As regions with significant conservation value are located within ecologically fragile areas, KNWH requires more accurate and objectively targeted approaches for adequate protection and management. Scientifically diagnosing the health status of ecosystems at such KNWH and developing effective conservation and management strategies have become critical scientific challenges that demand immediate attention.

Typically, research on EHA focuses on land use pattern changes [19], landscape pattern changes [7, 20], ESs [21], climate change [22], and management effectiveness [23] as entry points for analyzing the ecosystem health of various regions and landscapes, such as provinces [24], cities [25, 26], economic zones [22, 27], wetlands [28], lakes [28], watersheds [29], and plateaus [30]. The objective is to provide scientific references for ecological management, zoning, and ecosystem optimization. While the focus is predominantly on macro and mesoscales such as provinces and cities, emphasizing the impacts of urbanization and economic development, there is a notable scarcity of studies on the EHA of KNWH. The existing research frameworks and optimization paths are often unsuitable for KNWH due to their fragile ecology and OUV. Studies on KNWH typically employ the Remote Sensing Ecological Index (RSEI) [31], analyze land use pattern changes [32], and heritage tourism [33, 34] to assess ecosystem quality and influencing factors, however, these studies seldom provide a comprehensive measurement framework of ecosystem health. The RSEI model with its relatively limited evaluation indicators, lacks the explanatory power to fully capture ecosystems' structural and functional integrity. Moreover, studies focused on land use pattern changes often overlook the functions of ESs and the needs of human societal development, while research on heritage tourism primarily examines the negative impacts of tourism activities on KNWH.

As a geographical unit comprising various ecosystems, the landscape is crucial for maintaining ecological and environmental processes across different spatial scales. It offers a spatial perspective on changes in ecosystem health, thereby providing scientific references for protecting the integrity of WHS. The landscape stability of KNWH is notably unstable, with patterns susceptible to rapid changes. These changes can significantly disrupt ecosystem integrity, but their impact on ecosystem health remains poorly understood [35]. Integrity involves preserving the wholeness, unblemished state, and ancestral significance of WHS. To enhance the management of WHS, UNESCO underscores the necessity of regular monitoring, which involves the ongoing collection and analysis of data and timely adjustments to management policies [36, 37]. Ecosystem health is a critical indicator for assessing WHS's integrity, enabling identifying the specific regions and factors driving ecosystem health through landscape changes. This approach helps preserve the integrity and diversity of ecosystems, aligning with the requirements for safeguarding the integrity of KNWH. Thus, it provides a scientific basis for the effective management and dynamic monitoring of KNWH.

Current evaluation models of ecosystem health are primarily classified into three categories: (1) Vigor-Organizational-Resilience (VOR) and VORS model evaluation frameworks [38]. The former effectively explains ecosystems' structural characteristics and functional integrity but struggles to reflect the relationship between human development needs and ecosystem health. (2) Pressure-State-Response (PSR) [8] and Driving Force-Pressure-State-Impact-Response (DPSIR) evaluation frameworks [39]. These models emphasize the direct impacts of human activities on ecosystems but tend to overlook the integrity and complexity of ecosystems, rendering them inappropriate for assessing microregions with restricted human activities or limited socio-economic data availability. (3) Structure–Function-Process-Human Health-Development (SFPHD) evaluation framework [40]. Based on the theory of system ecology, SFPHD analyzes EHA across four dimensions: structure, function, process, and development. However, this approach does not adequately address the significance of changes in landscape patterns for EHA [41]. The VORS model, built on the VOR model, incorporates the concept of ESs to address the former’s lacking focus on human needs. This assessment framework has been applied across various ecosystems, including ecologically fragile zones [42, 43], forests [44], watersheds [45, 46], and urban areas [26, 35]. It is particularly effective in ecologically fragile zones and geographic systems with minimal human interference. Socio-economic activities have a more subtle impact on ecosystems in KNWH than in urban environments. Obtaining comprehensive, long-term socio-economic data is challenging because these sites are meso-micro in scale and not administrative regions. Consequently, models such as the PSR and SFPHD are less suitable for assessing ecosystem health. The VORS model assesses ecosystem health by examining natural ecosystem quality (landscape pattern changes) and ESs quality function. It effectively measures the current status of ecosystem conservation at the KNWH and aligns with the practical needs of human development. Matching the concept of efficiently managing WHS for regional sustainable development. Consequently, the VORS model was chosen to evaluate the ecosystem health of KNWH.

In summary, this study chose Shibing Karst, a typical representation of KNWH, as the study area comprising the WHS and buffer zone. It measures KNWH's ecosystem health and analyzes the driving forces from the perspective of landscape patterns. The study contains the following components: (1) Construct an EHA framework for the KNWH using the VORS model and measure the ecosystem health levels in the study area, (2). Analyze changes in the landscape pattern and the spatial and temporal evolution of ecosystem health from 2004 to 2020, identifying non-healthy areas to determine key future conservation and management areas (3). Explore the spatial clustering characteristics and near-neighbor effects of ecosystem health using spatial autocorrelation and identify key influencing factors with the geo-detectors model. The results provide a scientific basis for the zoning management and effective conservation of the KNWH, offering practical insights into the integrity of OUV conservation and environmental management of the WHS within the same region.

Materials and methodology

Study area

The Shibing Karst, a significant component of the "South China Karst" second phase, was successfully included in the WHL in 2014, meeting the World Heritage (WH) criteria (vii) and (viii). This area is situated in the northern region of Shibing County within Guizhou Province. It features a gradual decrease in topography from northwest to southeast, with an average elevation of 912 m. The total area of the Shibing Karst is 28,295 hectares, including a heritage site covering 10,280 hectares and a buffer zone spanning 18,015 hectares (Fig. 1). The center of the latitude and longitude coordinates is located at 108°05′40″E, 27°10′16″N. The Shibing Karst experiences simultaneous heat and precipitation, and soluble rock widely distributed. The ecological environment boasts lush vegetation, and over millions of years, the processes of dissolution and erosion have shaped distinctive karst landforms. The primary geomorphological landscapes include karst cliffs, valleys, and tower-like formations atop the peaks. This region represents the most typical dolomite karst in tropical and subtropical zones worldwide [34].

Fig. 1

Location map of the Shibing Karst. a Location of the "South China Karst" in China; b Location of the Guizhou Province in "South China Karst"; c Location of the Shibing Karst in Guizhou Province; d Study area of the Shibing karst

The ecological environment of the Shibing Karst is extremely fragile, serving as a critical intersection of abundant heritage treasures and a lagging regional economy. Resident incomes in the Shibing Karst are significantly below the Chinese average. The local government's urgent need for economic development, along with growing tourism spurred by the heritage site's brand, has escalated human activity in the region. This upsurge exacerbates the tension between resource conservation and local development, with potential changes in regional landscape patterns posing a significant risk to both ecosystem health and the integrity of the Shibing Karst. Accordingly, the geographical characteristics, ecological environment, and socio-economic development level of Shibing Karst highlight the common challenges in conserving ecosystem health and integrity at KNWH. Utilizing the Shibing Karst as a study area to assess its ecosystem health, the results of this study can serve as a scientific reference for enhancing the synergy between conservation and development at similar KNWH. Additionally, these findings provide valuable information for local governments when formulating environmental management and governance policies.

Data sources and processing

Data sources

The data used for this study primarily consisted of remote-sensing image data. Remote sensing images of the study area were downloaded from the Geospatial Data Cloud (http://www.gscloud.cn/). Considering the scale of the study area, the ultimate choice encompassed Landsat 5 and Landsat 8 OLI image data acquired in 2004, 2010, 2016, and 2020 (Table 1). Landsat 5TM has a spatial resolution of 30 m. Landsat 8 has two main instruments: the OLI (Operational Land Imager) and TIRS (Thermal Infrared Sensors). The OLI instrument features nine bands; bands 1 to 7 each have a spatial resolution of 30 m, and the panchromatic band (band 8) has a spatial resolution of 15 m.

Table 1 Remote sensing image data

Data processing

This study employed several software tools, including ENVI 5.3, ArcGIS 10.2, Fragstats 4.2, and GeoDa, to conduct spatial analysis and quantify the study data. ENVI 5.3 software was utilized for data preprocessing tasks such as radiometric calibration, atmospheric correction using Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes (FLAASH), and image cropping. The land use classification of the research region was established using the China Land Use Classification Standard (GB/T 21010–2007), and adopting the support vector machine method in remote sensing visual interpretation and supervised classification, the land use types within the study region were categorized under seven separate categories: forest land, shrubland, dry land, paddy land, grassland, construction land, and water. The accuracy of the classification data was evaluated using the Kappa coefficient, mapping accuracy, and user accuracy. The Kappa coefficients for the land use classification results exceeded 0.89, meeting the research requirements for EHA. Fragstats 4.2 was employed to calculate various landscape pattern indexes, while ArcGIS 10.2 facilitated the overlay analysis of EV, EO, ER, and ESs to examine the pattern and distribution of EHA within the designated study area. Finally, GeoDa was used to analyze the spatial clustering characteristics of the ecosystem health status between 2004 and 2020.

Methodology

The ecosystem serves as the carrier of the WHS’s OUV. Dynamic monitoring of the KWHS's ecosystem health offers crucial scientific data for maintaining its integrity and ensuring effective management. It also offers decision-making references for managing natural zones and developing local governance strategies. However, existing research has not sufficiently addressed this area. Integrating the research framework's suitability with the study area and fully considering the method's advantages alongside the ecological and socio-economic context and constraints, this study utilizes remote sensing technology to gather analytical data. Ecosystem health was comprehensively assessed based on landscape pattern changes using the VORS model. Spatial autocorrelation analysis and the geo-detector model are employed to reveal spatial clustering characteristics, nearest-neighbor effects, driving factors, and their interactions, enhancing the explanatory power of the research framework. The VORS model offers a more comprehensive approach than single-factor change analysis. Unlike integrated socioeconomic indicator systems, the VORS model explicitly analyzes the integrity of ecosystem structure, function, and process, emphasizing the provision of ESs [47]. It effectively addresses the challenges posed by the Shibing Karst's natural environment, workforce limitations, financial resources, technology, and difficulties collecting socio-economic data in localized areas. This makes it an efficient and precise method for dynamically evaluating ecosystem health. In conclusion, this research framework efficiently and accurately identified the ecosystem health status and its driving factors at KNWH. It effectively captures the nuances of ecosystem changes and benchmarks integrity protection by assessing the ecosystem health status of KNWH. Furthermore, it provides essential technical support for effective WHS management and regular monitoring.

The methodology encompasses three key aspects: (1) Collecting and processing vector files and remote sensing image data. (2) Developing Vigor, Organization, Resilience, and ESs indices for ecosystem health based on remote sensing indicators and constructing an EHA framework tailored to the Shibing Karst. (3) Analyzing the spatial and temporal evolution trends and spatial aggregation characteristics of ecosystem health values in the study area and identifying future focus areas for protection and ecological optimization. Details of the specific index system and calculation process are provided in Fig. 2 and Table 2.

Fig. 2

Ecosystem health assessment framework for the Shibing Karst

Table 2 Shibing Karst ecosystem health assessment indexes system

Ecosystem health assessment framework

Healthy ecosystems retain their spatial structure, preserve ecosystem processes, and exhibit self-recovery capabilities when faced with external stresses and disturbances. It can ensure a sustainable supply of ESs to meet the rational requirements of human social development [1, 48]. Drawing from the conceptual significance of ecosystem health, the research objectives, and the particular circumstances of the study site, we established a framework for evaluating the ecosystem health of the Shibing Karst using the VORS model (Fig. 2). EHA was calculated as follows:

$$ {\text{EHA}} = \sqrt[4]{{\text{EV*EO*ER*ES}}} $$

(1)

Given the importance of the four indexes in assessing ecosystem health, equal weights are assigned to each index. However, due to data interpretation and scale differences among the various indexes, direct interpretation of ecosystem health proved challenging. The research data were standardized using Eqs. (2) and (3) to account for the varying roles of different indexes in the ecosystem [21, 49]. EHA, EV, EO, ER, and ES were standardized within the range [0, 1]. Based on the research findings and classification criteria from WHS and national parks [50, 51], the standardized ecosystem health level was categorized into five grades, from low to high, by applying the equal-interval method [52]: degraded (0–0.2), unhealthy (0.2–0.4), average healthy (0.4–0.6), suboptimal healthy (0.6–0.8), and highest healthy (0.8–1.0).

Indexes with positive correlations are standardized with the formula:

$$ X_{i} = {{\left[ {X_{i} - Min_{{\left( {Xi} \right)}} } \right]} \mathord{\left/ {\vphantom {{\left[ {X_{i} - Min_{{\left( {Xi} \right)}} } \right]} {\left[ {Max_{{\left( {Xi} \right)}} - Min_{{\left( {Xi} \right)}} } \right]}}} \right. \kern-0pt} {\left[ {Max_{{\left( {Xi} \right)}} - Min_{{\left( {Xi} \right)}} } \right]}} $$

(2)

Indexes with negative correlations are standardized with the formula:

$$ X_{i} = {{\left[ {Max_{{\left( {Xi} \right)}} - X_{i} } \right]} \mathord{\left/ {\vphantom {{\left[ {Max_{{\left( {Xi} \right)}} - X_{i} } \right]} {\left[ {Max_{{\left( {Xi} \right)}} - Min_{{\left( {Xi} \right)}} } \right]}}} \right. \kern-0pt} {\left[ {Max_{{\left( {Xi} \right)}} - Min_{{\left( {Xi} \right)}} } \right]}} $$

(3)

Indexes for ecosystem health assessment

(1) Ecosystem vigor

EV pertains to an ecosystem's energy input and nutrient cycling capacity and characterizes its metabolic or primary productivity [53]. The Normalized Difference Vegetation Index (NDVI) reflects vegetation distribution and productivity along with spatial and temporal dynamics [54]. Previous research has demonstrated NDVI as an effective index for evaluating EV [2, 48, 55, 56]. Consequently, NDVI was chosen as the EV index. The calculation formula is as follows:

$$ {\text{NDVI}} = {{\left( {{\text{NIR}} - {\text{RED}}} \right)} \mathord{\left/ {\vphantom {{\left( {{\text{NIR}} - {\text{RED}}} \right)} {\left( {{\text{NIR}} + {\text{RED}}} \right)}}} \right. \kern-0pt} {\left( {{\text{NIR}} + {\text{RED}}} \right)}} $$

(4)

where NIR denotes the near-infrared band, and RED denotes the infrared band.

(2) Ecosystem organization

EO refers to the overall stability of ecosystem structure, encompassing landscape pattern, spatial heterogeneity, and landscape connectivity [35]. It is quantitatively evaluated based on landscape heterogeneity, landscape connectivity, and the connectivity of ecologically essential patches, incorporating landscape diversity, fragmentation, cohesion, and other indexes. Considering the two indexes represent distinct facets of ecosystem structure, the weight was set to 0.35 [2, 7]. Moreover, the ecological functions patch connectivity weight should not exceed the overall patch connectivity weight [35]. It was set to 0.30. Landscape diversity signifies the richness and area uniformity of landscape patch types and can be considered the core index of landscape heterogeneity. Given the significant function of forest land in the Shibing Karst ecosystem, the Index of Connectivity (IC) was quantified by assessing the fragmentation degree of forest land and the patch cohesion index. Fragmentation serves as the core index for determining connectivity. The weight for forest land fragmentation was higher than the cohesion index at 0.20 and 0.10, respectively [2]. The calculation formula for EO follows:

$$ \begin{gathered} {\text{EO}} = 0.35{\text{*LH}} + 0.35{\text{*LC}} + 0.3{\text{*IC}} \hfill \\ = 0.35{\text{*SHDI}} + 0.35{\text{*FN}} + 0.2{\text{*FN}}_{1} + 0.1{\text{*COHESION}} \hfill \\ \end{gathered} $$

(5)

where LH refers to Landscape Heterogeneity, LC refers to Landscape Connectivity, IC refers to the Connectivity of Patches with Critical Ecological Functions, SHDI refers to the Shannon Diversity Index, FN refers to the Fragmentation of the Whole Ecological Landscape, FN₁ refers to the Fragmentation of Forest Landscapes, and COHESION refers to Cohesion Index of Forest Landscapes.

(3) Ecosystem resilience

ER relates to the capacity of an ecosystem to sustain its overall structure and functionality despite external disturbances [48]. A healthy ecosystem should have sufficient resilience to endure a certain level of external disturbance, with resilience coefficients varying for different ecosystem types [50]. Considering the delicate ecological environment and OUV within the study region, the ER coefficient was determined based on existing research findings [35, 57] (Table 3). The calculation formula is as follows:

$$ {\text{ER}}\, = \,\sum\nolimits_{{{\text{i}}\, = \,1}}^{{\text{n}}} {{\text{A}}_{{\text{i}}} {\text{*RC}}_{{\text{i}}} } $$

(6)

where A_i is the area ratio of ecosystem type i, RC_i is the resilience coefficient of ecosystem type i, and n is the number of ecosystem types.

Table 3 Resilience coefficients of ecosystem types [35, 50, 58]

(4) Ecosystem services

ESs are frequently defined as the capability of ecosystems to provide specific goods and services to human society, representing the direct or indirect benefits humans derive from ecosystems. Costanza [48] proposed that healthy ecosystems can offer sustainable ESs to human society [48], and some studies have regarded ESs as a crucial index of ecosystem health [22, 27]. In this study, the values of ESs were calculated using the equivalent factor method [59, 60], referenced by Xie et al. [61] based on the research findings of Costanza [59, 61]. These values are integrated with the social and ecological context within the study region to establish the value coefficients of various ESs types (Table 4). The calculation formula is as follows:

$$ {\text{ESs}}\, = \,\sum\nolimits_{{{\text{i}} = 1}}^{{\text{k}}} {\sum\nolimits_{{{\text{i}} = 1}}^{{\text{f}}} {{\text{A}}_{{\text{i}}} {\text{*VC}}_{{{\text{ij}}}} } } $$

(7)

Table 4 Ecosystem services value coefficients per unit area for different ecosystem types (yuan/hm²) [61, 62]

ESs is the value of total ecosystem services, A_i is the area of ecosystem type i, and VC_ij is the value of ESs on type j of ecosystem type i.

Spatial autocorrelation analysis

According to the first law of geography, the characteristics of geographic phenomena are determined by their location and distribution [63]. Identifying these spatial relationships provides a scientific basis and decision support for ecosystem management. Spatial autocorrelation, the similarity between variables in different spatial locations, reflects the degree of interdependence and clustering among spatial variables and their neighbors. It is categorized into global spatial autocorrelation (Global Moran's I) and local spatial autocorrelation (Local Moran's I). Global spatial autocorrelation describes the spatial clustering within a study area [64]. In contrast, local spatial autocorrelation reveals the spatial correlation between spatial attribute values and neighboring spatial attribute values [65]. The formula is as follows:

$$ {\text{Global }}\,{\text{Moran}}^{\prime}{\text{s }}\,{\text{I}}\, = \,\frac{{\mathop \sum \nolimits_{{{\text{i}} = 1}}^{{\text{n}}} \mathop \sum \nolimits_{{{\text{j}} = 1}}^{{\text{n}}} {\text{W}}_{{{\text{ij}}}} \left( {{\text{x}}_{{\text{i}}} - \,\overline{{\text{X}}} } \right)\left( {{\text{x}}_{{\text{j}}} - \,\overline{{\text{X}}} } \right)}}{{{\text{S}}^{{2\mathop \sum \nolimits_{{{\text{i}} = 1}}^{{\text{n}}} \mathop \sum \nolimits_{{{\text{j}} = 1}}^{{\text{n}}} {\text{W}}_{{{\text{ij}}}} }} }} $$

(8)

$$ {\text{Local Moran}}^{\prime}{\text{s I}} = \frac{{{\text{n}}\left( {{\text{x}}_{{\text{i}}} - \,\overline{{\text{X}}} } \right)\mathop \sum \nolimits_{{{\text{j}} = 1}}^{{\text{m}}} {\text{W}}_{{{\text{ij}}}} \left( {{\text{x}}_{{\text{j}}} - \,\overline{{\text{X}}} } \right)}}{{\mathop \sum \nolimits_{{{\text{i}} = 1}}^{{\text{n}}} \left( {{\text{x}}_{{\text{i}}} - \,\overline{{\text{X}}} } \right)^{2} }} $$

(9)

where n is the number of spatial units in the study area; x_i is the value of spatial unit i; x_j is the value of spatial unit j. W_ij is the spatial matrix of spatial autocorrelation. And $\overline{\text{x} }$ is the mean value. $\text{S}=\frac{1}{\text{n}}\sum_{\text{i}=1}^{\text{n}}{\left({\text{x}}_{\text{i}}-\overline{\text{x} }\right)}^{2}$, the range of Global Moran's I value is [−1,1], where Global Moran's I > 0 indicates a positive spatial correlation and a high degree of clustering, Global Moran's I < 0 indicates a negative spatial correlation and dispersion, Global Moran's I = 0 indicates no spatial correlation [64].

Local Moran's I reflects the heterogeneity and instability of spatial variable distribution, categorized into four patterns: When Local Moran's I > 0, it suggests that the spatial units exhibit high-high or low-low clustering. Conversely, when Local Moran's I < 0, it suggests that the spatial units are clustered in high-low or low–high clustering [65].

Geo-detectors

Geo-detectors uncover the underlying principles of geospatial phenomena by examining the interactions between explanatory factors and analyzed variables. The core principle is that if a clear causal relationship exists between two variables, their spatial distributions should also converge [66]. Compared to multiple linear regression and gray correlation models, the geo-detector model effectively captures and analyzes nonlinear relationships between variables, identifies key drivers of ecosystem health, and yields statistically significant results even with complex ecosystem data. In contrast to the Geographically Weighted Regression (GWR) model, the geo-detector model requires less data, handles small sample sizes effectively, and is not constrained by data distribution types. Accordingly, as an exploratory spatial data analysis tool, the geo-detectors model effectively elucidates the nonlinear relationships between factors and analyzed variables and is well-suited for identifying drivers of ecosystem health. Moreover, it facilitates the analysis of spatial data heterogeneity, thereby simplifying the translation of scientific discoveries into practical ecological protection and management strategies. The calculation formula is as follows:

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

(10)

The q represents the explanatory capacity of the influence factor on EHA, with a value range of [0, 1]. The more significant value indicates a greater explanation capacity of the factor; the q-value needs to pass the significance level test (less than 0.1); h denotes the subregion associated with the influencing factor. L represents the number of classifications and categorizations of EHA along with the influencing factor. N_h and N represent the number of units in different classified regions and the entire region, respectively. ${\sigma }_{h}^{2}$ and ${\sigma }^{2}$ are the variances of the EHA in the regions of different classifications and the whole region.

To further analyze the intensity of various influencing factors, factor detection was conducted utilizing the EHA framework and the geo-detector model. Over the period from 2004 to 2020, the EHA of Shibing Karst was utilized as the dependent variable, while five indexes, LUCC, EV, EO, ER, and ESs, were selected as the independent variables. Factor detection allowed for the quantitative measurement of the independent variables' impact strength on the dependent variable. Initially, the data of both dependent and independent variables from 2004 to 2020 were divided into five levels using the intermittent method in ArcGIS software. Subsequently, a grid comprising 300*300 units was established, with the center point as the sample point for analysis, resulting in a total of 3135 units. Finally, the capacity of each detector factor that influences EHA was calculated. Due to the geo-detector resilience to covariate covariance, there was no requirement for diagnosing covariance within the analysis results.

Results

Changes in ecosystem landscape types

From 2004 to 2020, the predominant landscape types in the Shibing Karst ecosystem remained forest land, shrubland, and dryland, maintaining a nearly stable landscape structure. However, differences in the changes in these landscape types were observed during the study period. The areas of forest land, construction land, and water continued to expand, with forest land experiencing the most significant growth and construction land exhibiting the fastest increase. Conversely, shrubland, grassland, paddy land, and dryland continued diminishing, with shrubland displaying the most substantial decline. Firstly, forest land consistently occupied a predominant role within the geographical scope, constituting 70.14%, 76.58%, 80.35%, and 82.56% of the entire area in 2004, 2010, 2016, and 2020, respectively. The area of forest land in 2020 showing a sustained growth trend, increasing by 17.71% compared to 2004. Secondly, the proportion of shrubland area continued to decrease over the study period, declining by 11.41%, 9.51%, 7.61%, and 5.71%. The area of shrubland in 2020 decreased by 49.96% compared to 2004, resulting in a drop from the second to the third position among landscape types. Similarly, the proportion of dryland area exhibited a decreasing trend, albeit lower than that of shrubland, with reductions of 10.66%, 9.48%, 8.29%, and 7.11%. By 2020, the dryland area surpassed shrublands, rising from the third to the second position. Moreover, grassland and paddy land areas demonstrated a gradual but marginal decrease compared to their already limited proportions. Finally, the construction land area increased from 0.58% in 2004 to 0.89% in 2020, representing the most rapid growth rate of 53.45% among all landscape types. With the smallest area share among landscape types, water displayed a gradual but consistent upward trend, with percentages of 0.05%, 0.05%, 0.07%, and 0.11% in 2004, 2010, 2016, and 2020, respectively (Fig. 3).

Fig. 3

Landscape types and changes in the Shibing Karst, 2004–2020

Measurement of ecosystem health indexes

From 2004 to 2020, the EV in Shibing Karst exhibited a consistently high level, demonstrating a gradual evolutionary trend. Spatially, high EV areas were concentrated primarily within the Shibing Karst, with moderate EV values in the northern portion of the buffer zone and lower levels in the eastern and southeastern regions. EV serves as an index of ecosystem cycling efficiency and energy input. Throughout the study period, the average EV value in Shibing Karst increased from 0.9018 to 0.9625 (Fig. 4). Between 2004 and 2010, substantial low EV value areas were prevalent in the southeastern, eastern, and northern portions of the buffer zone, and sporadic distribution within the WHS. During 2010–2020, a few areas within the buffer zone exhibited high EV values. The EV level in the study region improved during this period, significantly reducing low-value areas, with only small clusters remaining in the eastern and southeastern regions. The EV distribution within the WHS appeared homogeneous, with a predominant presence of high-value areas. Since the Shibing Karst was inscribed as a WHS in 2014, the region has experienced accelerated tourism development, increased human activities, and expanded construction land. This development has led to an EV-level decline in the eastern and southeastern areas of the buffer zone.

Fig. 4

Ecosystem health indexes spatial and temporal changes in the Shibing Karst, 2004–2020

The EO values in Shibing Karst exhibited noticeable improvement, with a spatial pattern revealing better conditions in the northern region of the buffer zone and WHS, while poorer conditions were present in the southeastern region. EO serves as an indicator of ecosystem stability. Figure 4 reveals that between 2004 and 2020, the overall level of the Shibing Karst EO values remained nearly high, with marginal increases in the mean values of 0.8707, 0.8866, 0.8628, and 0.8867, respectively. At the spatial pattern level, regions with high EO values are predominantly within the WHS, while low-value areas are concentrated in the southeastern portion of the buffer zone. A notable correlation between the intensity of human activity and EO was observed, with the density of villages and population in the buffer zone being significantly lower than that in the WHS and external disturbances caused by human activities being more limited in the WHS. Additionally, the stringent heritage protection policies and ongoing ecological restoration initiatives implemented by local governments have led to a rapid expansion in forest land, fostering the material circulation of the ecosystem and resulting in a high concentration of EO value within the region. Conversely, the southeastern part of the buffer zone features comparatively flat terrain. It is adjacent to the county, with prevalent paddy land and construction land, higher population density, and increased human activity intensity. These factors have the potential to induce unavoidable damage to the stability and biodiversity of the ecosystem, consequently leading to a relatively lower level of EO within the area.

The ER of the Shibing Karst increased, indicating an enhanced capacity for self-recovery in response to external disturbances. The spatial pattern of ER in the Shibing Karst shows a concentration of high-value areas within the WHS, with comparatively lower indices observed in the southeastern portion of the buffer zone (Fig. 4). Between 2004 and 2016, significant low-value ER areas were primarily distributed along the border zone between the buffer zone and the WHS, as well as in the southeastern region. By 2020, the high-value ER area had gradually expanded, forming spatial clusters in the eastern and northern regions of the buffer zone and the WHS. Conversely, the low-value area remained confined to a small distribution range within the southeastern region. Significantly, the rocky desertification control and ecological restoration projects implemented in Shibing Karst have yielded remarkable results. Extensive paddy land and dryland areas have transitioned into forest land, remarkably enhancing vegetation cover and biodiversity. These efforts have facilitated ecosystem stability and its capacity to withstand disturbances.

The ESs of the Shibing Karst have exhibited a continual enhancement. At the level of spatial pattern, high ESs values are predominantly distributed in areas characterized by abundant vegetation cover and minimal human disturbances. Conversely, ESs indices were lower in buffer zone characterized by ongoing construction land and paddyland use (Fig. 4). Over time, the ESs value within the WHS has demonstrated stability at a high level. In contrast, the buffer zone exhibited some fluctuations, indicating an emerging trend of polarization. In 2004, areas with low ESs values were concentrated in the study area's northern, eastern, and southeastern segments. By 2010 and 2016, the ESs value consistently improved, as regions with previously low values in the northern and eastern segments transitioned to high-value areas, with low values sporadically distributed solely in the southeastern segment. In 2020, the ESs value within the WHS is projected to continue to increase, confirming its position as a high-value concentration area. Simultaneously, most buffer zone areas are anticipated to gradually improve their ESs value. Nevertheless, in the southeastern and eastern segments, the persistent expansion of construction land will contribute to a ecrease in the ESs value.

Overall, the EO, EV, ER, and ESs levels of the Shibing Karst ecosystem demonstrated a gradual improvement trend between 2004 and 2020, with 2016 as a significant node. The spatial distribution patterns of these four indexes exhibited relative proximity. Low-value areas were predominantly clustered in the southeastern region of the buffer zone, characterized by a relatively flat topography and adjacency to the urban area of Shibing county, benefiting from favorable socio-economic conditions. In contrast, high-value areas were concentrated within the WHS, characterized by minimal human activity and a continuous increase in forest land area.

Spatial and temporal evolution of ecosystem health assessment

Between 2004 and 2020, the mean EHA values in the Shibing Karst region were 0.8303, 0.7689, 0.6976, and 0.7824. The assessment exhibited a fluctuating trend, initially decreasing and then increasing, with a noticeable inflection point in 2016. As illustrated in Fig. 5, the spatial distribution of EHA in Shibing Karst remained relatively constant, characterized by higher values in the central and western regions and lower values in the eastern and southern parts. The rigorous limitations imposed by the protection regulations on the OUV and integrity of the NWH, coupled with the rugged topography and landscape in the western buffer zone, significantly restrict the mode and intensity of human activities. The EHA levels V (highest healthy) and IV (suboptimal healthy) are prominently concentrated in the WHS and the western portion of the buffer zone, representing the highest levels of a constant and healthy ecosystem structure. In contrast, the southeastern region lies close to the densely populated Shibing county. This area serves multiple spatial purposes, functioning as a buffer for tourism activities related to the WHS and facilitating regional economic development. Additionally, it serves as a prominent tourism area within the Shibing Karst region and functions as both a residential and production zone for residents. Consequently, this region experiences a concentrated distribution of EHA levels II (unhealthy) and I (degraded), with human activities significantly impacting the organizational structure and stability of the local ecosystem.

Fig. 5

Ecosystem health assessment spatial and temporal changes in the Shibing Karst, 2004–2020

During the study period, the EHA in the Shibing Karst region remained comparatively stable, with most results falling within the highest and suboptimal healthy levels. The combined area share of these two grades consistently remained above 90% for an extended duration. Changes in the area covered by EHA levels I (degraded) and II (unhealthy) were minimal. The regional ecosystem health level transitions primarily occurred between neighboring levels, with movements from lower to higher levels being the most prevalent, particularly from level III to IV and from level IV to V (Fig. 6). EHA level V (highest healthy) exhibited the most significant area percentage, accounting for 69.39%, 70.77%, 74.80%, and 73.54% in sequential order, indicating a consistent overall increase over time. The area covered by EHA level IV (suboptimal healthy) ranked second, constituting 23.41%, 22.42%, 17.96%, and 19.69%, respectively, with a slight area decline in 2020 compared to 2004. EHA level III (average healthy) ranked third, covering 5.85%, 6.13%, 2.97%, and 5.57%, respectively. The area share exhibited a decreasing and then increasing trend over the study period. EHA level II (unhealthy) occupied the fourth position, representing 0.95%, 0.65%, 3.09%, and 1.09%, respectively, reaching its peak in 2016 and subsequently experiencing a rapid decline. By 2020, the area share had returned to a value close to that of 2004, illustrating an overall inverted U-shaped trend. In contrast, EHA level I (degraded) had the lowest percentage, accounting for 0.40%, 0.03%, 1.18%, and 0.11%, respectively (Fig. 6). Overall, the sizes of the various EHA grades in the Shibing Karst area exhibited some variation during the study period. However, the ranking of these grades remained consistent, with the order being highest healthy > suboptimal healthy > average healthy > unhealthy > degraded.

Fig. 6

Areas of ecosystem health assessment at different levels in the Shibing Karst, 2004–2020

Generally, the Chinese government's policy of returning farmland to forests, the rocky desertification control projects, and proactive WHS management at various government levels have significantly contributed to the continuous improvement and stabilization of ecosystem health at the WHS. The southeastern part of the buffer zone is a crucial area for future ecosystem restoration. Effective measures must be developed to protect and improve the health of its ecosystems. This includes maintaining and restoring the coherence of ecological processes and landscape patterns, preventing ecosystem degradation from undermining the WHS's OUV, and achieving synergistic development in the conservation and development of the KNWH.

Spatial correlation analysis of ecosystem health assessment

Spatial autocorrelation is used to analyze the spatial correlation of EHA in Shibing Karst. The Global Moran's I index values for 2004 to 2020 were 0.666, 0.586, 0.602, and 0.665, respectively (Fig. 7). Notably, all four indexes exceeded 0.5, with the P-values passing the 0.01 significance level test. This indicates a significant positive spatial correlation of EHA in the Shibing Karst region, signifying that areas with high EHA values are spatially clustered, just as areas with low values are closely grouped. Examining the trend over time, the spatial dependence of EHA exhibited an inverted "U" shape, with a notable decrease in the degree of concentration of the Global Moran's I index from 2004 to 2010, followed by a consistent increase from 2010 to 2020.

Fig. 7

Global Moran's I of ecosystem health assessment in the Shibing Karst, 2004–2020

The local spatial clustering characteristics of the EHA in Shibing Karst are primarily significant (Fig. 8). Both the high-high and low-low clustering areas exhibit widespread distribution patterns within the region. Regarding temporal changes, the high-high clustering areas experienced an initial decrease followed by an increase, predominantly concentrated in the western section of the WHS and the buffer zone. The high-value areas demonstrated an overall growth trend across the study region. Conversely, the low-low cluster areas increased during the study period, with the spatial distribution gradually shifting toward the southern region, eventually forming a contiguous area densely populated near Shibing county. Within these regions, the polarization of EHA patterns became increasingly prominent.

Fig. 8

Spatial clustering of ecosystem health assessment in the Shibing Karst, 2004–2020

The spatial concentration of ecosystem health in the Shibing Karst is closely related to human activities. Increased human activity intensity triggers changes in the landscape, particularly exerting a detrimental influence on the structure and stability of the delicate Karst ecosystem. Although the distribution of high-high concentration areas beyond the WHS remains sporadic, the continuous development of "heritage tourism fever" and its related industries may expand the low-low concentration area. As a result, the buffer zone distribution range may also increase. The EHA condition in the southeastern part of the buffer zone is likely to encounter significant challenges. The degree of spatial clustering aligns with previous studies, indicating that the spatial distribution of ecosystem health at different levels tends to exhibit clustering tendencies. High-value areas are continuously distributed within the WHS, while low-value areas are clustered primarily in the southeastern buffer zone.

Influencing factors of ecosystem health assessment

The results from the geo-detector analysis indicated the significance of all five detector factors at the 0.01 level throughout the 2004–2020 period, with q-values ranging between 0.378 and 0.785 (Table 5). Each detector factor effectively explained the spatial pattern of EHA and its evolutionary trajectory. Explicitly, two factors, LUCC and ESs, consistently exhibited q-values exceeding 0.6 across all four study periods, highlighting their pivotal roles in influencing the spatial and temporal evolution of EHA within the designated research region. Conversely, the impact of EV, EO, and ER were comparably minor, although discernible variations in the evolution of the q-values of these factors were observed. Specifically, the q-values of LUCC and EV demonstrated a consistent upward trend across the four study periods. In contrast, ESs, EO, and ER indexes, with 2016 as the turning point, demonstrated an evolutionary pattern characterized by initial growth followed by subsequent decline.

Table 5 The EHA single-factor detection of Shibing Karst

From 2004 to 2010, the influence ranking of each factor remained consistent, with ESs demonstrating the most significant correlation with EHA, followed by LUCC, ER, EO, and EV. The q-values for the five factors remained relatively invariable. The ecological substrate of the karst area is delicate, and its rugged terrain challenges traditional agricultural practices. Despite these constraints, agricultural production in Shibing Karst during this period remained predominantly cultivated. The region featured widespread distribution of poor-quality agricultural land, mainly sloping arable land, thereby facing an elevated risk of rocky desertification, which endangers the provisioning function for maintaining ESs integrity. Between 2010 and 2016, the ranking of each factor's influence degree remained unchanged, yet the q-value of each influencing factor increased significantly compared to 2010. In adherence to the NWH integrity protection guidelines and to enhance the ecosystem, the local government implemented extensive farmland reforestation and facilitated the relocation of migrants within the Shibing Karst. During this period, significant alterations in land use patterns occurred, with a substantial portion of sloping cultivated and agricultural land being converted into forest land, consequently enhancing the supply capacity of ESs. In 2020, the q-value of LUCC surpassed ESs as the critical factor influencing EHA. The rankings of ER, EO, and EV remained unchanged, although the q-values of ESs, ER, and EO decreased. From 2016 to 2020, the 'World Heritage' branding effectively promoted tourism development in the Shibing Karst. Infrastructure construction and the establishment of scenic attractions altered the land use structure within the buffer zone, leading to the continuous expansion of construction land in the region. The intensified human activities resulted in the spatial concentration of production and residential patterns, significantly impacting EHA within the Shibing Karst. The findings of this study corroborate the perspective held by some scholars, emphasizing the pivotal role of preserving the LUCC ecological functions and the ecological supply capacity of ESs in maintaining ecosystem health [41, 67].

Discussion

Changes in landscape patterns for ecosystem health assessment

Human socio-economic activities and natural environmental factors are the primary driving forces behind landscape pattern changes [68]. Landscape patterns and their conservation significantly influence the supply function of ESs and the state of EHA. Investigating changes in landscape patterns provides a comprehensive understanding of the spatiotemporal evolutionary dynamics of ecosystem health [27, 69]. Ecosystems within the karst regions of Guizhou Province are inherently vulnerable and sensitive to external disturbances. These areas possess limited landscape restoration capacity, rendering them vulnerable to significant risks from unsustainable human activities. Research findings indicate that areas with compromised ecosystem health are primarily concentrated in the southeastern buffer zone, which is characterized by construction land. In contrast, healthy areas are mainly clustered within vegetation restoration zones, such as forest land in WHS. This observation aligns with research conducted by scholars on Lijiang city in China, Kolkata in India, and the southwestern mountainous regions of China [2, 70, 71]. The intensified human activities leading to changes in landscape patterns are identified as the principal cause of ecosystem health degradation.

The Chinese government has continuously advocated for the comprehensive management of mountains, waters, forests, fields, and lakes, emphasizing the control of rocky desertification rehabilitation projects. They promoted a transition from high-speed to high-quality development in the national economy. In approximately 2010, the Guizhou provincial government prioritized the "South China Karst" declarations as a critical project. To strengthen WHS conservation, the local government in Shibing Karst implemented large-scale ecological transformation projects, leading to specific alterations in the landscape pattern within the WHS. Throughout the study period, there was a notable expansion of forest land in the WHS, accompanied by a decrease in the extent of paddy land and dryland. Additionally, there was an observable increase in ecological corridors and landscape connectivity. The state of ecosystem health exhibited relative stability and demonstrated a positive developmental trajectory. Government policies have significantly facilitated the improvement of ecosystem health, aligning with the conclusions drawn by several scholars [72,73,74]. Buffer zones function as protective barriers for WHS and play a significant role in facilitating human socio-economic development. Consequently, human activities can significantly impact the landscape pattern, particularly with the continuous expansion of construction land, which may threaten the conservation of the WHS’s landscape integrity. Future research could focus on analyzing changes in landscape structure, strengthening the development of an ecological security network, and identifying the 'sources,' 'nodes,' and 'bottlenecks' of ecosystem health. Emphasizing that ecological landscape connectivity plays an essential role in maintaining ecosystem structure and function.

Differences in ecosystem health assessment between whs and buffer zone

Significant spatial disparities exist in the ecosystem health status between the WHS and buffer zone. The ecosystem health status of the WHS surpasses the buffer zone, consistent with the findings of Bayanbulak, China [50]. The WHS has consistently maintained a healthy state, demonstrating a uniformly high spatial distribution of ecosystem health. However, clusters of sub-healthy or unhealthy ecosystem status areas are evident in the densely populated southeastern part of the buffer zone. Areas with lower population density and activity intensity exhibit comparatively favorable ecosystem health status, corroborating the EHA research outcomes of Gannan grasslands, China [75]. The ecosystem is the OUV physical foundation of the Shibing Karst, with its integrity strictly safeguarded and monitored. The buffer zone fulfills a dual role in heritage conservation and economic development. Due to local developmental demands, human activities in the buffer zone significantly surpass those within the WHS, facilitating various economic activities, particularly agricultural production. The tourism industry, considered a primary sector for economic growth, has led to establishing various tourist reception facilities, such as parking lots, hotels, and restaurants, primarily within the buffer zone. This has resulted in a continuous expansion of the construction land in the buffer zone, exerting notable pressure on the ecosystem. The area where Shibing Karst is located also faces economic underdevelopment, magnifying the inherent conflict between resource conservation and socioeconomic advancement. Achieving synergistic effects is crucial in balancing nature conservation with socioeconomic development [76]. Several studies suggest a symbiotic relationship between WHS conservation and the development of the buffer zone tourism industry. Heritage conservation can amplify the appeal of tourism resources, providing tangible support for advancing the tourism industry in the buffer zone. Furthermore, developing the tourism industry in the buffer zone can generate new values and social interactions for the WHS, effectively reducing poverty and fostering sustainable development. Coordinating WHS conservation with sustainable resource utilization and buffer zone economic development has become a critical concern in the management of KNWH [77].

The buffer zone serves as a safeguarding spatial barrier for the WHS. Its ecological degradation threatens the OUV of WHS, undermining its ecological health conservation [31, 78]. Coordinating the preservation of environmental resources with the pursuit of high-quality development in the local economy necessitates a more stringent approach to heritage conservation. Furthermore, heritage conservation has surpassed the conventional boundaries of isolated protection, evolving into a comprehensive approach that favors outward and regionally encyclopedic preservation strategies. Local governments should prioritize areas with inferior ecological health, manage the expansion of construction land at a controlled pace, and enhance the environmental health of the buffer zone. This can be achieved by optimizing the landscape structure in the buffer zone and augmenting vegetation cover in less healthy regions. Future research should concentrate on understanding the social-ecological function of the buffer zone, investigating the complex interplay between the conservation of WHS and the development of supplementary industries in the buffer zone. The imperative preservation of heritage alongside the necessary development requirements of buffer zone residents should be considered. From the perspective of the Sustainable Development Goals (SDGs), it is imperative to formulate a high-quality, coordinated development model that encompasses ecological, industrial, and livelihood benefits for safeguarding the WHS and expanding industry development within the buffer zone. This will foster a harmonious relationship between economic growth in the buffer zone and the environmental protection of the WHS.

Implications for local governance policy

Based on the designated roles of WHS and its buffer zone, namely the protection of the WHS and the development of the buffer zone, local governments can establish differentiated governance policies. These policies respond to the evolving focus in WHS management, shifting from "balancing protection and development" to prioritizing "protection for development."

Targeted conservation based on ecosystem health

Initially, WHS, as custodians of OUV and core conservation areas, should prioritize conservation measures over development claims. Local governments should ensure the ecosystem health and OUV of WHS are protected. Governance policies must explicitly prohibit or strictly limit new construction projects and high-intensity human activities while strengthening the supervision of existing activities. Second, focus on optimizing areas with poor ecosystem health, particularly the expanding construction lands in the southeastern buffer zone. Implement ecological restoration projects, such as reforesting farmland, restoring aquatic ecosystems, and establishing ecological corridors. Continuously monitor restoration effects using technologies like drones and satellite remote sensing, and adjusting management strategies promptly based on monitoring results to prevent ecosystem health degradation from adversely affecting the WHS. Finally, implement green building and low-impact development strategies in the buffer zone, supported by government subsidies and technical assistance. This approach encourages the use of eco-friendly materials and the adoption of environmental protection measures like solar energy and green roofs. It also optimizes land-use patterns and preserves green spaces and water bodies during construction to enhance ecological connectivity and landscape integrity.

Sustainable tourism in buffer zone

Sustainable tourism is an effective pathway for achieving WHS conservation and economic development. First, local governments should uphold the concept of green development and promote regional economic growth through eco-tourism. Second, the tourism industry should be diversified to strengthen its economic impact. This includes enhancing tourism services, cultural experiences, agricultural tourism, and other forms, reducing reliance on a single project, expanding employment and livelihood opportunities for community residents, and decreasing their dependence on natural resources. Finally, community residents play a crucial role in the management of WHS. Residents who participate in tourism development can benefit economically and socially. This participation fosters local attachment and cultural self-awareness, encouraging active involvement in conservation, management, and promotional activities. Such engagement contributes significantly to WHS preservation.

Limitations and future research

This study, based on the changes in the landscape patterns of the WHS, selected four indexes, EV, EO, ER, and ESs, to measure the spatiotemporal evolution trend of the Shibing Karst ecosystem health. Although this research can provide specific references for similar heritage conservation and monitoring, it also has certain limitations.

In this study, remote sensing data was used to estimate the ecosystem health of the Shibing Karst. However, the resolution of the remote sensing images impacts the accuracy of the ecosystem landscape classification. Furthermore, the weights of the indicators and the calculation process of the VORS model are primarily based on results from previous studies, and the model's assumptions and premises may not fully align with the ecosystem conditions of other KNWH. To address the limitations of a single data source and insufficient resolution, future research should aim for higher resolution and more diversified data resources. This can be achieved by strengthening engagement with local governments and third-party organizations, employing modern technologies such as uncrewed aerial vehicles and the Google Earth Engine (GEE), and establishing vital monitoring stations for long-term dynamic data collection. These steps will significantly improve the reliability and accuracy of the research results. To address the limitations of the VORS model, future research will optimize the indicators of the evaluation framework through comparative analysis. It will also refine the indicator weights and adjust the calculation methods through interdisciplinary cooperation and expert consultation, ensuring that the evaluation framework more accurately reflects the health of the KNWH ecosystems.

The evolution of ecosystem health in WHS is a multifaceted and dynamic process involving various stakeholders, including the ecosystem, tourists, residents, governmental authorities, and nearby businesses. Future research should integrate a diverse range of data sources, such as nightlight data, GDP, population density metrics, policy support indexes, climate change data, and air pollution data. Establishing a comprehensive evaluation index system that encompasses both subjective and objective elements of ecosystem health will enhance the practicality of the research framework and increase the analytical validity of the outcomes. Additionally, future research needs to consider potential risks such as rocky desertification, human activities, environmental pollution, soil erosion, and biodiversity. An EHA database grounded in interdisciplinary studies should be established to address these risks. This database should regularly update a range of health indexes, enhance the explanatory and predictive capacity of the outcomes, accurately assess ecosystem health status and critical concerns, and offer scientific and technological guidance for government planning and regional conservation efforts.

Conclusion

Ecosystem health reflects the adequate protection of the KNWH’s integrity. Based on the VORS model, this study develops a framework for evaluating KNWH’s ecosystem health. This framework is customized to the Shibing Karst, where socio-economic activities are restricted, and the ecosystems are fragile. It offers new perspectives and instruments for WHS environmental management. The primary conclusions of this study are as follows:

• (1) The general landscape structure of the Shibing Karst exhibited relative stability. The predominant land type in the study region was forest land, with its area proportion increasing from 70.14% in 2004 to 82.56% in 2020. Construction land and water witnessed continual expansion throughout the study period, while shrubland, grassland, paddy land, and dryland areas demonstrated declining trends.

• (2) Between 2004 and 2020, the ESs, EO, EV, and ER of the Shibing Karst gradually improved. Regarding spatial distribution, areas with lower values for the four indexes tended to aggregate in the southern region, contiguous to the relatively prosperous social and economic conditions of Shibing county. Moreover, the high-value areas were primarily concentrated within the WHS. Human activity intensity emerged as the primary factor influencing the spatial pattern evolution of ESs, EO, EV, and ER.

• (3) From 2004 to 2020, the EHA of the Shibing Karst maintained a comparable better condition. Following a trend of initial decline and subsequent increase, with 2016 as a turning point, the ecosystem health level displayed a spatial distribution pattern characterized by higher levels in the central and western areas and relatively lower levels in the southeast. The percentage of level V (highest healthy) in the ecosystem health grade increased from 69.39 to 73.54%, while the proportion of level IV (suboptimal healthy) decreased from 23.41 to 19.69%. Moreover, the proportion of level III (average healthy) remained approximately invariable, experiencing a slight decline from 5.85 to 5.57%. The level II (unhealthy) percentage initially increased and subsequently decreased, increased from 0.95 to 1.09%. Finally, the proportion of level I (degraded) was the lowest, declining from 0.40 to 0.11%. Transitions in the ecosystem health level within the study area were primarily observed from lower levels to adjacent higher levels.

• (4) The ecosystem health of the Shibing Karst demonstrates significant spatial clustering characteristics, featuring a relative aggregation clustering distribution. The degree of ecosystem health spatial dependence in the study region exhibited a reverse 'U' shaped evolutionary characteristic, with an initial decrease followed by an increase. High-high aggregation areas are primarily concentrated within the WHS, while low-low aggregation areas are predominantly situated in the buffer zone. Spatial aggregation demonstrates a polarized characteristic and becomes increasingly apparent.

• (5) LUCC, alongside ESs, represents the primary drivers influencing the evolution of ecosystem health in the Shibing Karst. The results of the geo-detector analysis indicate that, among the five indexes, ESs and LUCC have the greatest explanatory power concerning the spatial differentiation characteristics of ecosystem health. Compared with the traditional VOR model, incorporating ESs into the EHA framework enhances the accuracy of the assessment.

The main contributions of this paper include identifying the spatial clustering effect, near-neighbor effects, and driving factors influencing the Shibing Karst's ecosystem health. These findings enhancing the explanatory power of the research framework, comprehensively measuring the evolution of ecosystem health in the study area, and promoting effective management and integrity conservation of the KNWH. An efficient ecosystem health monitoring and assessment framework provides valuable information for socio-economic development planning in the buffer zone. It helps the local government formulate sustainable policies that balance conservation and development in Shibing Karst. Future research should explore the long-term impacts of landscape pattern changes on ecosystem health, particularly in the context of climate change and increasing human activities. It is also essential to establish a more comprehensive research dataset and enhance the calculation methods for the indicators. Conduct longitudinal studies on the KNWH over an extended period to monitor and predict the dynamic trends of ecosystem health from a micro and multidimensional perspective, thereby enhancing the scientific rigor and specificity of the KWHS’s integrity protection and effective management.