Spatiotemporal Evolution, Driving Mechanisms, and Zoning Optimization Pathways of Ecosystem Health in China

Abstract

The health status of ecosystems is an important prerequisite for ensuring regional ecological security. Exploring the spatiotemporal patterns, driving mechanisms, and zoning regulation pathways of ecosystem health is of great significance for achieving co-ordinated and sustainable regional ecosystems. This study uses China as a case area and applies the InVEST model to measure integrated ecosystem services and incorporates it into an evaluation framework for ecosystem health based on the “Vigor-Organization-Resilience-Ecosystem Services” (VORS) model. It reveals the spatiotemporal evolution characteristics of ecosystem health in China from 2000 to 2020 and employs the geodetector and spatiotemporal geographically weighted regression model to analyze the main influencing factors and spatial differentiation characteristics, thereby exploring ecological management zoning and optimization pathways. The study results show that (1) during the study period, the overall ecosystem health level in China showed a declining trend, dropping from 0.397 in 2000 to 0.377 in 2020. (2) Overall, China’s ecosystem health exhibits strong spatial positive correlation and spatial clustering characteristics, with a basic pattern of lower values in the northwest and higher values in the southeast. (3) Vegetation coverage, population density, density of road network, and per capita GDP are the main influencing factors of ecosystem health in China. (4) China is divided into five types of Ecological Management Zones: Ecological Conservation Zone, Ecological Enhancement Zone, Ecological Buffer Zone, Ecological Remediation Zone, and Ecological Reshaping Zone, with differentiated strategies proposed for optimizing ecosystem health in each zone.

1. Introduction

Ecosystem health (EH) generally refers to the ability of a regional complex, composed of different ecosystems within a specific spatiotemporal range, to sustainably and stably provide ecosystem services (ES) while maintaining the health of each ecosystem [1]. In recent decades, under the dual influence of climate change and human disturbances, EH in some regions of China has been compromised, leading to environmental problems such as land degradation and biodiversity loss. Data show that, by 2024, the areas of soil erosion and deforestation in China are expected to reach 2.6534 million square kilometers and 1572 thousand hectares, respectively. This will seriously threaten the sustainable development of the original regional ecosystems and the long-term assurance of regional ecological security [2,3]. How to promote and ensure regional EH has become a major environmental issue that humanity urgently needs to address. Therefore, identifying the spatiotemporal evolution characteristics and zoning regulation pathways of China’s EH is of urgent theoretical and practical significance for optimizing regional EH and ensuring regional ecological security patterns. This can provide some theoretical lessons for developing refined and differentiated ecological management strategies in China. Currently, existing studies have conducted extensive research on regional ecosystem health from various aspects, such as key research areas [4,5,6,7], quantitative assessment methods [8,9,10], detection of influencing factors [11,12,13], and Ecological Management Zoning (EMZ) [14,15,16]. In terms of key research areas, existing studies mostly focus on small regions or watershed scales. Typically, regional EH assessments are conducted by constructing various EH assessment models for wetlands [4,5], rivers [6], forests [7], coastal areas [17], and pastoral regions [18]. In terms of quantitative assessment of EH, regional EH is mainly evaluated through methods such as the indicator species approach [8] and the indicator system approach [9,10]. The indicator species approach mainly evaluates the health of an ecosystem indirectly through indicators such as the quantity, diversity, and community structure of indicator species, but it is generally applicable only to the health assessment of a single ecosystem [19]. The indicator system method, on the other hand, comprehensively considers multiple indicators of an ecosystem to construct a theoretical framework for EH assessment. Common indicator system methods include the Driving Force-Pressure-State-Impact-Response (DPSIR) model [20], the Pressure-State-Response (PSR) model [21], the Vigor-Organization-Resilience (VOR) model [22], and the Vigor-Organization-Resilience-Services (VORS) model [23]. The DPSIR model and PSR model can measure the state of ecosystems and external disturbances, but they neglect the evaluation of ES provision capacity. In 2012, Costanza and other scholars proposed incorporating ES into the EH assessment framework and built an EH evaluation system based on the Vigor-Organization-Resilience-Services (VORS) model [24]. This model makes the evaluation framework more rational and comprehensive and has the characteristics of easy operation and objective evaluation results, which has led to its widespread use in regional EH assessment studies [25,26,27,28].

Currently, with the continuous refinement of ES assessment research, the InVEST model has gradually become the mainstream model for ES calculations. Through the various calculation models in the InVEST 3.13 software, the spatiotemporal evolution patterns of ES can be simulated and evaluated [29,30]. Therefore, scholars have attempted to measure regional ES based on the InVEST model and incorporate them into EH evaluation models, constructing a comprehensive assessment framework for EH based on the VORS model [31,32,33]. In terms of detecting influencing factors of EH, scholars mainly use methods such as geographic detectors [11], coupled co-ordination models [12], and geographically weighted regression models (GTWR) [13] to explore the mechanisms and process responses of human and natural disturbance factors on regional EH. In terms of EMZ, existing studies on EMZ at meso- and macro-scales often base their work on assessment results of regional ecological security patterns [14], ecological sensitivity [15], and the supply–demand relationships of ES [16]. These studies then carry out EMZ of territorial space and formulate differentiated zoning control strategies.

In summary, domestic and international scholars have conducted quantitative assessments of regional EH at various scales, and significant research achievements have been made in the areas of quantitative assessment methods and the detection of influencing factors. However, in the process of detecting influencing factors of regional EH, there is still a lack of research on the spatiotemporal heterogeneity characteristics of key influencing factors. Furthermore, few studies have constructed EMZ and proposed differentiated zoning regulation strategies based on EH assessments. In view of this, this study used the InVEST model to measure the comprehensive ES of the region and incorporated them into the evaluation system of the VORS model. The study revealed the spatiotemporal evolution patterns of China’s EH from 2000 to 2020 at a grid scale. Using both geographic detectors and GTWR models, the study explored the main influencing factors and their spatial differentiation characteristics. Based on the evaluation results of EH, EMZ was constructed for China. Differentiated and refined zoning regulation paths and optimization strategies were proposed. The aim is to provide a practical reference for the scientific formulation of sustainable management and conservation policies for China’s EH.

2. Materials and Methods

2.1. Overview of the Study Area

China is located in East Asia (3°51′–53°33′ N, 73°33′–135°05′ E) with a diverse topography, predominantly mountainous and hilly (Figure 1), forming a three-step terrace distribution. China’s climate is diverse, and its diverse climate and complex terrain have created various natural features and abundant flora and fauna resources across the country. In 2023, China’s GDP reached CNY 12,605.82 billion, with a permanent population of 1.40967 billion people. As a typical rapidly urbanizing region in the world, China’s urbanization rate has risen rapidly from 17.92% in 1978 to 66.16% in 2023 since the reform and opening up. The rapid advancement of urbanization has posed serious threats to EH, leading to ecological problems such as soil erosion, vegetation degradation, and desertification.

2.2. Construction of an EH Research Framework Based on the VORS Model

Based on the VORS model, a research framework was constructed to assess China’s EH in 2000, 2010, and 2020 (Figure 2), exploring the spatiotemporal patterns, influencing factors, EMZ, and optimization pathways of EH in China. The research framework consists of 4 parts: (1) constructing an EH assessment framework based on the VORS model from the dimensions of vigor, organization, resilience, and ES to reveal the spatial and temporal evolution patterns of EH in China from 2000 to 2020; (2) using spatial auto-correlation models to show the distribution of spatial clusters of China’s EH from both a global and local perspective; (3) applying geographic detectors and GTWR models to investigate the main influencing factors of China’s EH and their spatial differentiation characteristics; (4) based on the results of the EH assessment, dividing China into EMZ and exploring differentiated and refined zoning optimization strategies for improving regional EH in the future.

2.3. Data Sources and Preprocessing

The evaluation index system for assessing China’s EHI is constructed mainly from multiple dimensions such as land use, socioeconomics, hydrological conditions, transportation, administrative divisions, vegetation cover, soil conditions, topography, climate conditions, and food production. Detailed data can be found in

Table 1. Data source.

Data Type	Index	Year	Resolution Ratio/m	Data Source
Land use	Land use types	2000, 2010, 2020	1000	Chinese Academy of Sciences (CAS) Geographic Resource Center (GRC) (http://www.resdc.cn, accessed on 25 March 2022)
Society and economy	Population density	2000, 2010, 2020	1000
	GDP per capita	2000, 2010, 2020	1000
	NPP-VIIRS	2020	500
Hydrological condition	Watershed	2000, 2010, 2020	/
Traffic conditions	Road density	2000, 2010, 2020	/
Administrative division	Provincial and district administrative divisions	2000, 2010, 2020	/
Vegetation coverage	Percentage of forest area	2000, 2010, 2020	1000
Soil status	Soil texture	2009	1000	National Tibetan Plateau Science Data Center World Soil Database (HWSD) Soil Dataset (https://data.tpdc.ac.cn, accessed on 5 October 2018)
Landforms	DEM	2020	250	Geospatial data cloud (https://www.gscloud.cn, accessed on 10 June 2022)
Climatic condition	Average annual humidity	2000, 2010, 2020	30	Monthly dataset of meteorological element station observations in China (https://www.resdc.cn, accessed on 5 August 2022)
	Evapotranspiration	2000, 2010, 2020	30
	Average annual temperatures	2000, 2010, 2020	30
Grain production	Food production	2000, 2010, 2020	/	《China County Statistical Yearbook》

. Before conducting the regional EHI assessment, data preprocessing was performed on each indicator. The InVEST 3.13 software was used to calculate five key ES—carbon sequestration, soil retention, water yield, habitat quality, and food supply—to represent comprehensive ES supply. Nighttime light data were used to represent the level of urbanization [34]. The moving window method of Fragstats 4.2 software was used to calculate landscape pattern indices such as critical ecosystem connectivity (CIE), landscape heterogeneity (LH), and overall landscape connectivity (LC) based on data from six land use types (cropland, woodland, grassland, waters, building land, and unused land) to characterize ecosystem organization. Vegetation cover is characterized by the percentage of forested area in the county, which is calculated by the proportion of the area of forested ecosystems in each county in relation to the total land area of the county. Based on the ArcGIS 10.2 software platform, all the spatial data were uniformly resampled to a spatial resolution of 1 km × 1 km and spatialized to the county-level units using the zoning statistics function, and the projected co-ordinate system of all the spatial data was unified as Krasovsky_1940_Albers to build a GIS spatial database. Additionally, SPSS26.0 software was used to normalize the various EHI to avoid the effects of quantity differences and different dimensions.

2.3.1. Ecosystem Health Evaluation Based on the VORS Model

Referring to previous research practices, the EH status of the region is represented by the mean of the EH status [33]. The formula can be expressed as follows:

$$\mathrm{EHI}=\sqrt{\mathrm{PHI}\times \mathrm{ES}}$$

(1)

$$\mathrm{PHI}=\sqrt[3]{\mathrm{EV}\times \mathrm{EO}\times \mathrm{ER}}$$

(2)

where EHI represents the ecosystem health index; PHI represents the natural EH index; ES represents the comprehensive ecosystem services index; and EV, EO, and ER represent the Ecological Vigor, Ecological Organization, and Ecological Resilience, respectively. The natural breakpoint approach is employed to delineate the indicators above into five levels, low, slightly above low, medium, slightly lower than high, and high.

EV, EO, and ER can be worked out as follows:

• Ecosystem Vigor (EV): it generally refers to the primary productivity or metabolic capacity of a regional ecosystem. Previous studies have found that the normalized vegetation index (NDVI) is widely used to assess the level of ecosystem vigor [23]. Therefore, this study uses NDVI to represent the ecosystem vigor level of the region.
• Ecosystem Organization (EO): it is mainly used to describe the completeness and stability of ecosystem architecture and is commonly represented by critical ecosystem connectivity (CIE), landscape heterogeneity (LH), and overall landscape connectivity (LC). Referring to the calculation methods from previous studies and study area characteristics [12,24,32], the weights for LH, LC, and CIE were assigned as 0.35, 0.35, and 0.3, respectively, based on their importance to ecosystem health. A comprehensive ecosystem organization assessment model was built, calculated as follows:

  $$\begin{array}{l}\mathrm{EO}=0.35\mathrm{LH}+0.35\mathrm{LC}+0.30\mathrm{CIE}=(0.10\mathrm{AWMPFD}+0.25\mathrm{SHDI})+(0.25\mathrm{DIVISION}+0.10\\ \mathrm{CONTAG})+(0.10{\mathrm{DIVISION}}_f+0.05{\mathrm{COHESION}}_f+0.10{\mathrm{DIVISION}}_w+0.05{\mathrm{COHESION}}_w)\end{array}$$

  (3)

  In the formula, LC, LH, and CIE represent overall landscape connectivity, landscape heterogeneity, and critical ecosystem connectivity, respectively. AWMPFD and SHDI represent area-weighted mean patch fractal dimension and Shannon diversity index, respectively, used to represent landscape heterogeneity. DIVISION and CONTAG represent the landscape division index and contagion index, respectively, used to characterize overall landscape connectivity. DIVISION and COHESION represent the patch fragmentation index and patch cohesion index, respectively, used to characterize critical ecosystem connectivity. Among them, DIVISION_f and DIVISION_w represent the patch fragmentation index for forests and water bodies, respectively, while COHESION_f and COHESION_w represent the patch cohesion index for forests and water bodies, respectively.
• Ecosystem Resilience (ER): it represents the ability of an ecosystem to maintain its structural stability under human disturbance. Referring to previous studies [17], habitat quality is used to represent the ecosystem resilience of the region. The settings and selection of model-related parameters were primarily based on existing studies [35] and the InVEST user manual, calculated as follows:

  $$Q_{\mathrm{xj}}=H_i\left[1-\left({\displaystyle \frac{D_{\mathrm{xj}}^z}{D_{\mathrm{xj}}^z+K^z}}\right)\right]$$

  (4)

  In the formula, Q_xj is the habitat quality index of the jth land use type in the xth grid cell. D_xj is the habitat degradation of the jth land use type in the xth grid cell. H_i is the habitat suitability of the land use type. K is the half-saturation constant, set as half of the maximum degradation degree.
• Comprehensive ES: five types of typical ES supply capacity are measured using the InVEST model, as shown in

  Table 2. Quantitative methods for the service capacity of various ecosystems.

  ES Type	Calculation Method	Formula
  Carbon sequestration	The carbon sequestration amount of four carbon pools, belowground biomass, aboveground biomass, dead organic matter, and soil organic matter, is comprehensively considered. It is calculated based on the average carbon density and the area of different land use/cover types [37].	${\mathrm{CS}}_{\mathrm{total}}={\mathrm{CS}}_{\mathrm{above}}+{\mathrm{CS}}_{\mathrm{below}}+{\mathrm{CS}}_{\mathrm{soil}}+{\mathrm{CS}}_{\mathrm{dead}}$	(6)
  		In the formula, CStotal is the total carbon storage of the ecosystem. CSabove is the aboveground carbon storage. CSbelow is the belowground carbon storage; CSsoil is the soil carbon storage. CSdead is the dead organic matter carbon storage.
  Soil retention	The actual soil erosion and potential soil erosion in the region were measured separately. The spatial quantification of soil retention was achieved by calculating the difference between the two [38].	$A_s=A_p-A_a$	(7)
  		$A_p=R_i\times K_i\times L_i{S}_i$	(8)
  		$A_a=R_i\times K_i\times {LS}_i\times C_i\times P_i$	(9)
  		In the formula, Aa is the actual soil erosion per unit area in t/(hm2·a). Ap is the potential soil erosion per unit area in t/(hm2·a). Ri is the rainfall erosivity factor for the ith grid cell (MJ mm hm−2 h−1). Ki is the soil erodibility factor for the ith grid cell. Si and Li are the slope steepness and slope length factors for the ith grid cell, respectively. Ci and Pi are the vegetation cover management factor and the soil conservation practices factor, respectively.
  Water yield	The difference between precipitation and actual evapotranspiration in each grid cell is used to represent water yield capacity.The settings and selection of model-related parameters, such as potential evapotranspiration and actual evapotranspiration, are primarily based on existing studies [39] and the InVEST user manual.	$W_i=(1-{\displaystyle \frac{\mathrm{AET}(i)}{P\left(i\right)}})\times P\left(i\right)$	(10)
  		In the formula, Wi represents the water yield of the ith grid, which is used to indicate the potential water retention function of the grid cell. AET(i) is the annual actual evapotranspiration of the ith grid. p(i) is the annual precipitation of the ith grid.
  Food supply	There is a significant linear correlation between the regional food supply function and food production [40]. This study integrates food production, cropland distribution, and NDVI data for each county to achieve spatial quantification of China’s food supply at a 1 km grid resolution from 2000 to 2020.	$F_i={\displaystyle \frac{{\mathrm{NDVI}}_i}{{\mathrm{NDVI}}_{\mathrm{sum}}}}\times F_{\mathrm{sum}}$	(11)
  		In the formula, Fi represents the total food production of the ith grid. NDVIi is the NDVI value of the ith grid on cropland. NDVIsum is the sum of NDVI values on cropland within a county unit. Fsum is the total food production of a county unit.

  . On that basis, this study adopts the analytic hierarchy process (AHP) for the weight assignment of these five types of ES [36]. Following that, the comprehensive ES index of China from 2000 to 2020 is worked out by the following formula:

  $${\mathrm{ES}}_i={\displaystyle \sum _{j=1}^m{\mathrm{ES}}_{ij}}\times W_j$$

  (5)

  where ES_i represents the composite ecosystem service index of the ith grid cell; m represents the type of ecosystem service (m = 5); ES_ij represents the standardized value of the jth ecosystem service type of the ith grid cell; and W_j represents the weight coefficient of each type of ecosystem service.

2.3.2. Geodetector Model

Widely applied to the detection of major factors influencing different geological phenomena, the geodetector is a spatial statistical tool that can ensure immunity to multivariate collinearity through the detection of spatial heterogeneity and its influencing factors. Geodetector is essentially a nonlinear model that does not need to solve the problem of multicollinearity [41]. In this study, from natural, social, and economic aspects, 10 indicators, namely mean annual temperature, altitude, vegetation coverage, urbanization, slope, per capita GDP, population density, annual mean humidity, water network density, and road network density, were selected as the driving factors of ecosystem health in China. Based on the regional statistics tool in ArcGIS software, the different indicators and evaluation results were statistically partitioned, the mean values of the indicators within each county unit were calculated and imported into the geoprobe model, the different indicators were graded using the equidistant method and reclassified into scores ranging from 1 to 10, and the main influencing factors of ecosystem health were analyzed by geodetector using the factor detection model [26,27,32,42]. The formula can be expressed as follows:

$$q=1-{\displaystyle \frac{{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{l}}{N}_{\mathrm{i}}{{\sigma }_{\mathrm{i}}}^2}}{N{\sigma }^2}}$$

(12)

where l represents the number of types in the study area; q represents the explanation degree of the driving factor; N_i represents the number of units in layer i; N represents the total number of units; σ_i² represents the variance of the global dependent variable within the layer i; σ² represents the variance of the dependent variable in the entire domain.

2.3.3. Spatial Autocorrelation Analysis

The spatial autocorrelation model is a statistical approach that investigates spatial data distribution features and correlation [43]. This model measures the correlation and differentiation of specific observed variables in an adjacent space, globally and partially. The intensity of correlation of observed variables is represented by the Global Moran’s I, while the intensity of spatial autocorrelation and heterogeneity of local adjacent units by the Local Moran’s I. The formula can be specified as follows:

$$Global {Moran}^{\prime }s I={\displaystyle \frac{n{\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^n{W}_{ij}}}({\mathrm{x}}_i-\overline{x})({\mathrm{x}}_j-\overline{x})}{{\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^n{W}_{ij}{\displaystyle \sum _{i=1}^n{(x_i-\overline{x})}^2}}}}}$$

(13)

$$Local {Moran}^{\prime }s I={\displaystyle \frac{{\displaystyle \sum _{j=1,j\ne i}^n{W}_{ij}}(x_i-\overline{x})(x_j-\overline{x})}{{\displaystyle \sum _{i=1}^{\mathrm{n}}{(x_i-\overline{x})}^2}}}$$

(14)

where Global Moran’s I represents the Global Moran’s Index; Local Moran’s I represents the Local Moran’s Index; n represents the total number of counties; x represents the mean of values observed at the county level; x_j and x_i represent the value observed at j and i, respectively; and W_ij represents the spatial weight between i and j.

2.3.4. Spatiotemporal Geographically Weighted Regression Models (GTWR)

GTWR model incorporates the time dimension into the regression model, which can be used to analyze the influencing factors under the spatiotemporal dimension and effectively solve the problem of spatiotemporal non-stationarity. It has been widely used in the field of ecological environment assessment to explore the spatiotemporal differentiation characteristics of its different driving factors and has achieved good results [44]. The formula can be expressed as follows:

$$Y_i={\beta }_0(u_i,v_i,t_i)+{\displaystyle {\sum }_{l=1}^k{\beta }_l}(u_i,v_i,t_i)X_{il}+{\delta }_i, i=1, 2, \dots , n$$

(15)

where Y_i represents the EHI index; X_il is each influencing factor; i represents the ith sample county; n represents the total number of counties; k represents the number of influencing factors; i represents a random perturbation term; β₀ (u_i, v_i, t_i) represents a spatiotemporal intercept term, where u and v are the co-ordinates of the sample counties, and t represents time, and β_l (u_i, v_i, t_i) represents the estimated coefficient of the lth influencing factor at time t in county i.

3. Results

3.1. Characterization of Spatial and Temporal Patterns of Ecosystem Health

This study constructed a “Vitality-Organization-Resilience-Ecosystem Services Function (VORS)” assessment framework to calculate China’s EHI. The range standardization method was used to normalize each indicator. Based on the equal interval method, EHI was divided into five levels: low-value area [0,0.2), relatively low-value area [0.2,0.4), medium-value area [0.4,0.6), relatively high-value area [0.6,0.8), and high-value area [0.8,1.0].

The overall situation of the various indicators of EH in China in 2000, 2010, and 2020 is shown in Figure 3. The evolution trends of the indicators constituting the EHI showed significant differences. Among them, EV showed a steady annual increase, rising from 0.483 in 2000 to 0.581 in 2020. This may be attributed to the Chinese government actively promoting and implementing ecological protection projects and policies between 2000 and 2020, such as the Grain for Green Program and the Three-North Shelter Forest Program, which, to some extent, enhanced vegetation coverage in regions, thereby gradually boosting ecosystem vitality. In contrast, EO, ER, and ES all showed a significant downward trend. EO decreased from 0.492 in 2000 to 0.460 in 2020, ER dropped from 0.532 in 2000 to 0.414 in 2020, and the overall ES index declined from 0.316 in 2000 to 0.296 in 2020. This could be due to the weakening of ecosystem services, stability, and resilience to disturbances caused by human activities during China’s rapid urbanization, which triggered the multidimensional decline in EO, ER, and ES.

The spatiotemporal pattern changes of the EHI in 2000, 2010, and 2020 are shown in Figure 4. Overall, the average EHI values for 2000, 2010, and 2020 were 0.397, 0.393, and 0.377, respectively, showing a general downward trend, with the areas of declining EHI covering 59.49% of the total land area. Between 2000 and 2020, China’s EHI was mainly in the “medium-value” and “relatively high-value” categories, with cumulative proportions of 55.81%, 57.77%, and 55.98% for 2000, 2010, and 2020, respectively. Additionally, the proportion of areas with “low-value”, “relatively low-value”, and “relatively high-value” EHI increased overall. For example, the proportion of “relatively high” EHI areas increased from 22.82% in 2000 to 32.78% in 2020, while the proportion of “medium-value” and “high-value” areas decreased overall, with the “medium-value” areas dropping from 32.99% in 2000 to 23.21% in 2020. In terms of spatial distribution, China’s EHI from 2000 to 2020 showed significant spatial heterogeneity, with an overall distribution pattern of “low in the northwest, high in the southeast”. High EHI areas were mainly concentrated in southeastern mountainous and hilly regions, such as Fujian, Zhejiang, Jiangxi, and Guangdong. Medium-value areas were primarily located in central and southwestern mountainous regions such as Sichuan, Yunnan, Guizhou, and Anhui. Low-value areas were concentrated in northwestern regions, including Tibet, Xinjiang, Qinghai, and Inner Mongolia.

3.2. Spatial Clustering of Ecosystem Health

Based on the assessment results of EHI, the global Moran’s I tool from the spatial analysis module in ArcGIS software was used to analyze the spatial clustering characteristics of China’s EHI. The detection results, shown in

Table 3. Moran’s Index of ecosystem health in China.

Years	Moran’s I	Z Value	p Value
2000	0.368	102.545	0.000
2010	0.317	88.596	0.000
2020	0.324	90.381	0.000

, indicate that, between 2000 and 2020, the global Moran’s I index for China’s EHI was positive, with values of 0.368, 0.317, and 0.324, and p < 0.01. This indicates that China’s EHI exhibited significant spatial autocorrelation and positive spatial correlation in 2000, 2010, and 2020.

This study further used the local Moran’s I tool in ArcGIS software to analyze the local spatial clustering characteristics of EHI from 2000 to 2020. The LISA clustering of EHI is shown in Figure 5. Overall, from 2000 to 2020, there was significant local spatial autocorrelation in China’s EHI, mainly characterized by high–high clustering and low–low clustering. Among these, high–high clustering had the largest number of counties, accounting for 48.04%, 40.54%, and 38.15% of the total number of counties in 2000, 2010, and 2020, respectively. These were mainly distributed along the southeastern coast, the middle and lower reaches of the Yangtze River plain, and in regions like Hainan. In contrast, low–low clustering areas accounted for 27.98%, 26.16%, and 31.73% of the total number of counties in 2000, 2010, and 2020, respectively, mainly distributed in the northwestern regions of China, such as Tibet, Xinjiang, Inner Mongolia, and Heilongjiang. Additionally, high–low and low–high clustering areas were scattered throughout the study area. The number of counties with high–low clustering gradually increased, accounting for 1.86%, 3.21%, and 5.60% of the total counties in 2000, 2010, and 2020, respectively, and were mainly distributed in areas such as Beijing, Hebei, and Shaanxi. Meanwhile, the number of counties with low–high clustering remained relatively stable, making up 4.64%, 3.53%, and 3.53% of the total counties in 2000, 2010, and 2020, respectively, and were primarily distributed in Hubei, Anhui, and Sichuan.

3.3. Analysis of Driving Mechanisms for Ecosystem Health

3.3.1. Detection of Main Influencing Factors

To better reveal the driving mechanisms of EHI in China, this study used the geographic detector model to explore the main influencing factors of EHI. Ten indicators were selected from various dimensions. Using the factor detection module of the geographic detector model, the impact strength of these 10 driving factors on EHI was measured at the county scale based on q-values, allowing for the selection of the main driving factors affecting EHI in China [2,17,45].

As shown in

Table 4. Geodetector results for the initial selection factors.

q Value	X1	X2	X3	X4	X5	X6	X7	X8	X9	X10
2000	0.281	0.206	0.436	0.203	0.094	0.262	0.291	0.246	0.185	0.287
2010	0.246	0.231	0.472	0.229	0.103	0.285	0.325	0.253	0.126	0.309
2020	0.208	0.213	0.497	0.233	0.152	0.296	0.338	0.215	0.154	0.321
p value	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000

Note: X1 to X10 are the mean annual average temperature, elevation, vegetation coverage, urbanization, slope, per capita GDP, population density, mean annual average humidity, water network density, and density of road network, respectively.

, based on the q-value ranking, vegetation coverage, population density, road network, and per capita GDP were the four main driving factors influencing China’s ecosystem health. Additionally, the changes in the q-values of different driving factors from 2000 to 2020 indicate that, over time, the influencing factors of China’s EHI have gradually shifted from natural factors to socio-economic factors. The negative impact of human activities on EHI has become increasingly prominent, with growing effects on EHI.

3.3.2. Spatiotemporal Differentiation Characteristics of Key Influencing Factors

This study further applied GTWR to explore the spatiotemporal differentiation characteristics of the effects of these four main influencing factors on EHI. The detection results are shown in Figure 6. Overall, the major influencing factors in China show significant spatiotemporal heterogeneity. Vegetation coverage primarily exhibits positive impacts, while population density road network and per capita GDP mainly show negative impacts. The specific spatial differentiation characteristics of the various influencing factors are as follows:

• Vegetation coverage showed a significant positive effect on EHI in China during the study period. Vegetation cover is an important factor in improving the regional carbon cycle, water cycle, and ecosystem stability, thus playing an important role in promoting the healthy development of regional ecosystems. Specifically, vegetation coverage has the important function of retaining soil and water, and its increased coverage can significantly improve the regional soil structure and reduce the intensity of soil erosion, thus enhancing the ecosystem’s soil retention and carbon sequestration capacity. At the same time, increasing vegetation coverage is conducive to the development of regional biodiversity, which, in turn, enhances the self-repairing capacity of ecosystems, thereby promoting the vitality and resilience of ecosystems. Overall, from 2000 to 2020, the impact of vegetation coverage on China’s EHI generally shows a gradually increasing trend and, especially during the period of 2000 to 2010, the positive driving effect of vegetation coverage on the EHI increases especially significantly. Additionally, from 2000 to 2020, the influence of vegetation coverage on EHI showed a general decreasing trend from southeast to northwest. The impact of vegetation coverage on EHI was strongest in the southeastern coastal areas of China. This may be due to the good vegetation coverage and relatively healthy ecosystems in the southeastern coastal areas of China. In addition, China’s afforestation policy over the past 20 years has, to a certain extent, effectively promoted the recovery of regional vegetation coverage, which, in turn, has contributed to the effective enhancement of regional ecosystem health. As a result, vegetation coverage and regional ecosystem health have gradually increased over the past 20 years.
• During the study period, the effect of population density on EHI in China mainly showed a negative impact. Generally, population density reflects the level of urbanization in a region. Higher population density often accompanies rapid socio-economic development, and the increase in population density places greater demands on the natural–social system, leading to higher resource and environmental pressures. Thus, the effect of population density on EHI is mainly inhibitory. Overall, population density in eastern China predominantly has a negative impact on EHI, especially in some provinces in northeastern and southwestern China, where heavy industrial areas and fragmented landscapes are prevalent. The effect of population density in these regions is particularly severe. As the population continues to grow, the pressure within cities increases, threatening the stability of regional ecosystems. Additionally, the rise in human activities can undermine the resilience of regional ecosystems. In contrast, in sparsely populated regions of western China, such as Xinjiang, Tibet, population density has a positive effect. These areas are characterized by vast land and sparse populations, where an increase in population can actually promote vegetation recovery and ecosystem restoration, showing a positive effect.
• Road network density primarily exerts a negative inhibitory effect on China’s EHI. Generally, road network density reflects the level of road accessibility in a region. Higher road network density indicates more intense economic activity, leading to more severe damage to the regional ecosystem. As a result, the negative impact on ecosystem health becomes more pronounced, making road network density a significant suppressor of EHI. Overall, from 2000 to 2020, road network density predominantly showed a significant negative influence on EHI in China. This negative impact was mainly concentrated in provinces like Sichuan, Guizhou, and Chongqing, where hilly and mountainous terrain is widespread, leading to a high degree of landscape fragmentation and ecological sensitivity. Road construction in these regions significantly affects ecosystem stability and health. Additionally, in parts of eastern China, such as Fujian, Zhejiang, Jilin, and Shandong, where there are abundant natural resources and stable ecosystem structures, road construction has a relatively smaller negative impact on ecosystem health, as these areas are less susceptible to external disturbances.
• The impact of per capita GDP on the health of China’s ecosystems shows a coexistence of positive facilitating and negative inhibiting effects. Generally speaking, per capita GDP reflects the level of economic development of a region and, along with the high rate of regional socio-economic growth, the increasing needs of human society often bring about significant changes in the spatial pattern of the national territory, leading to a significant decline in regional landscape connectivity and ecosystem services, which, in turn, has a certain negative impact on the overall ecosystem health of the region. Specifically, during the study period, the high values of negative impacts of per capita GDP on ecosystem health were mainly concentrated in the northeastern and eastern coastal regions of China, of which the northeastern region is the region where China’s old industrial bases are concentrated, and socio-economic development is accompanied by the destruction of natural ecological environment, which, in turn, causes damage to the ecosystem health within the region; meanwhile, the eastern coastal region, represented by the Yangtze River Delta, is the typical representative of China’s rapid urbanization, with high socio-economic growth in the region, which has led to the rapid expansion of urban construction land, and the conversion of a large amount of ecological land (such as forests, meadows, wetlands, and watersheds) into construction land and, to a certain extent, reduced the connectivity of the regional ecosystems, biodiversity, ecosystem vitality, resilience, and ecosystem services, leading to a reduction in the health of the regional ecosystems. In addition, per capita GDP in the western regions of China also shows a negative impact on ecosystem health, probably because the original ecological environment in these regions is more fragile, the ecosystems themselves have a weaker self-repairing capacity, and the impact of socio-economic development on ecosystem health in the region is more sensitive. In the southwestern, central, southern, and southeastern regions of China, per capita GDP has a positive impact on ecosystem health, such as Guizhou Province, Yunnan Province, Fujian Province, Zhejiang Province, and other southeastern coastal counties, where the original ecological environment is relatively good, and the ecosystems themselves have a strong ability to resist interference and self-repair, so that the damage to the ecosystems of the region caused by urbanization and socio-economic development at the appropriate intensity is relatively limited. Urbanization and socio-economic development at an appropriate intensity have a relatively limited impact on regional ecosystems, and ecosystem self-repair has compensated for the impact of ecological damage brought about by urban expansion. These regions are also demonstration zones for China’s ecological civilization, and local governments and the public are gradually becoming more aware of the protection of ecological resources and green development and focus on ecological protection, restoration, and sustainable development of the region and a promotion effect.

4. Discussion

4.1. Collinearity Test of Driving Factors and Model Accuracy Evaluation

This study uses a GTWR to explore the spatiotemporal differentiation characteristics of the impact and these four main factors on EHI. To effectively avoid multicollinearity issues among the different influencing factors, a multicollinearity test was conducted on these four factors before applying GTWR. The test results showed that the variance inflation factors (VIF) for vegetation coverage, per capita GDP, population density, and road were 2.642, 2.534, 1.826, and 1.461, respectively, all of which were less than 10, indicating that these four factors passed the multicollinearity test. Additionally, this study compared the fitting effects of the OLS, GWR, and GTWR models in revealing the spatiotemporal differentiation characteristics of the factors affecting EHI, primarily using R², adjusted R², and AICc values to represent the fitting degree of the three models. As shown in

Table 5. Comparison results of GTWR with OLS and GWR.

Model	R2	Correction R2	AICc
OLS	0.502	0.501	−1848.4
GWR	0.617	0.616	−2133.6
GTWR	0.688	0.687	−3509.9

, the R² value for the GTWR model was 0.688, with an adjusted R² of 0.687 and an AICc value of −3509.9, all of which outperformed the corresponding values of the GWR and OLS models. This indicates that the GTWR model is the most effective in measuring the spatiotemporal differentiation characteristics of the main factors influencing EHI in China.

4.2. Response of Ecosystem Health to Land Use Type

Changes in land use types in a region often have significant impacts on the structure and functions of ecosystems. Therefore, this study explores the response mechanisms of EHI to land use types based on the spatial overlap of land use types and ecosystem health, with the results presented in

Table 6. Response of land use types in China to ecosystem health.

Year	Land Use	Low Health/km2	Lower Health/km2	Medium Health/km2	Higher Health/km2	High Health/km2
2000	Cropland	6522	220	1,022,080	736,312	/
	Woodland	9463	50,507	1,696,769	1,696,236	418,743
	Grasslands	6922	50	2,400,459	570,358	/
	Water	1814	182,151	66,137	294	/
	Construction Land	1325	59,351	101,697	/	/
	Unused Land	1,984,994	/	/	/	/
Year	Land use	Low health/km2	Lower health/km2	Medium health/km2	Higher health/km2	High health/km2
2010	Cropland	6818	1008	783,408	962,341	/
	Woodland	10,108	2	21,670	1,669,422	477,935
	Grasslands	7174	172,054	2,400,037	387,252	/
	Water	26,387	161,979	64,298	97	/
	Construction Land	2202	68,986	114,659	/	/
	Unused Land	1,979,458	/	/	/	/
Year	Land use	Low health/km2	Lower health/km2	Medium health/km2	Higher health/km2	High health/km2
2020	Cropland	16,506	7970	792,168	912,143	11,433
	Woodland	28,836	2756	84,257	1,589,297	476,617
	Grasslands	60,644	161,036	1,871,174	565,339	5411
	Water	48,851	147,989	65,893	5923	716
	Construction Land	3351	103,086	135,152	12,585	422
	Unused Land	2,099,486	13,778	39,458	7301	371

.

In 2000, arable land was primarily classified as having high and moderate health, with areas of 736,312 km² and 1,022,080 km², respectively; forests were primarily classified as having moderate and high health, while grasslands were primarily classified as having moderate and high health, with areas of 2,400,459 km² and 570,358 km², respectively; water bodies were primarily classified as having low and moderate health, with areas of 182,151 km² and 66,137 km², respectively; built-up land was primarily classified as having low and moderate health, with areas of 59,351 km² and 101,697 km², respectively; other land types were primarily classified as low health, covering an area of 1,984,994 km². From the perspective of ecosystem health types, low health types were mainly concentrated in other land types, accounting for 98.70%; lower health was primarily concentrated in water bodies, accounting for 62.32%; moderate health was mainly concentrated in grasslands and forests, accounting for 45.41% and 32.09%, respectively; high health was primarily concentrated in forests and arable land, accounting for 56.48% and 24.52%, respectively; high health was basically concentrated in forests.

In 2010, arable land was mainly categorized as having high and moderate health, with areas of 962,341 km² and 783,408 km², respectively; forests were primarily classified as having high and higher health, with areas of 477,935 km² and 1,669,422 km², respectively; grasslands were mainly classified as having moderate and high health, with areas of 2,400,037 km² and 387,252 km², respectively; water bodies were primarily classified as having low and moderate health, with areas of 161,979 km² and 64,298 km², respectively; built-up land was mainly classified as having low and moderate health, with areas of 68,986 km² and 114,659 km², respectively; other land types were mainly classified as low health, covering an area of 1,979,458 km². From the perspective of ecosystem health types, low health was primarily concentrated in other land types, accounting for 97.41%; lower health was mainly concentrated in grasslands, accounting for 42.58%; moderate health was primarily concentrated in grasslands, accounting for 70.92%; higher health was mainly concentrated in forests, accounting for 55.30%; high health was basically concentrated in forests.

In 2020, arable land was mainly classified as having high and moderate health, with areas of 912,143 km² and 792,168 km², respectively; forests were primarily classified as having high and higher health, with areas of 476,617 km² and 1,589,297 km², respectively; grasslands were primarily classified as having moderate and high health, with areas of 1,871,174 km² and 565,339 km², respectively; water bodies were primarily classified as having low and moderate health, with areas of 147,989 km² and 65,893 km², respectively; built-up land was mainly classified as having low and moderate health, with areas of 103,086 km² and 135,152 km², respectively.

4.3. Ecological Management Zones and Optimization Paths

To effectively optimize the spatial pattern of land use in China and implement refined and differentiated land control strategies, this study is based on the EHI of China from 2000 to 2020, using county units as the evaluation unit for zoning. A method combining the current status and evolution trends of ecosystem health is employed to scientifically divide Ecological Management Zones. The division process of Ecological Management Zones can be mainly divided into three steps: (1) based on the ArcGIS 10.1 platform, the current status of EHI in China is classified into three levels: poor [0,0.4), moderate [0.4,0.6), and good [0.6,1.0]. Additionally, based on the differences in EHI from 2000 to 2020, the evolution trends are categorized into three types: EHI deterioration (<0), stable EHI [0,0.05], and improved EHI (>0.05). (2) The levels of current EHI are spatially overlaid with the types of evolution trends to preliminarily divide the sub-zones for ecological management, resulting in a total of nine sub-zone types: I-1, I-2, and I-3 (indicating a low level of EHI that has deteriorated, stabilized, or improved from 2000 to 2020); II-1, II-2, and II-3 (indicating a moderate level of EHI that has deteriorated, stabilized, or improved from 2000 to 2020); and III-1, III-2, and III-3 (indicating a high level of EHI that has deteriorated, stabilized, or improved from 2000 to 2020) [42]. (3) Based on the characteristics and functional differences of the nine sub-zones, these sub-zones are further combined into five major categories of Ecological Management Zones: Ecological Conservation, Ecological Enhancement, Ecological Buffer, Ecological Remediation, and Ecological Reshaping [27]. The basis of the delineation of different Ecological Management Zones, control objectives, and zoning characteristics are shown in

Table 7. China’s ecosystem health management zones.

Type	Include Regions	Management Objective
Ecological Conservation	II-3, III-3	Nature should play the main role, allowing ecosystems to evolve and develop on their own while restricting human interference with ecological spaces.
Ecological Enhancement	I-3, III-2	Leverage the natural resilience of ecosystems to optimize the spatial pattern of ecological spaces.
Ecological Buffer	II-2	Supplemented by ecological design, ecological buffer zones, and restoration areas should be established to create an environment for natural recovery.
Ecological Remediation	I-2, III-1	Supplement with necessary artificial measures, focusing on ecological remediation.
Ecological Reshaping	I-1, II-1	Directly implement proactive human interventions for ecosystem reconstruction or reshaping.

.

To better achieve ecosystem stability and socio-economic sustainability, this study proposes differentiated and refined land use optimization pathways and control strategies based on ecological management zoning. The specific optimization paths for the five types of Ecological Management Zones—Ecological Conservation, Ecological Enhancement, Ecological Buffer, Ecological Remediation, and Ecological Reshaping—are as follows (Figure 7).

For the Ecological Conservation zone, this area mainly includes two subtypes, II-3 and III-3, with counties in this zone comprising 29.7% of the total number of counties. The zone is primarily located in central and northern China, encompassing vast natural forests and nature reserves, including the Qinba Mountains, Qinling Mountains, and Wuyi Mountains. The primary management goal for this zone should focus on natural conservation, maximizing the ecosystem’s self-restoration capacity and resistance to disturbances. The objective is to protect and maintain a high-quality natural ecological environment, minimize human interference to allow for the ecosystem’s natural regeneration and evolution, maintain ecological functions and biodiversity, and avoid large-scale economic development activities. The focus should be on natural recovery, enhancing forest quality through measures like forest tending, constructing multi-dimensional ecological corridors to maintain biodiversity, and ensuring the area serves as an important ecological barrier.

For the Ecological Enhancement zone, this area mainly includes subtypes I-3 and III-2, accounting for 4.96% of all counties, primarily located in southeastern coastal areas and some counties in central and northern China. The ecosystems in this zone demonstrate strong vitality and service capacity, with good structural integrity, high to medium levels of ecological resilience, rich species diversity, and good self-recovery ability under external pressures. The primary management goal in this zone should fully utilize the ecosystem’s natural recovery capacity to optimize the layout and structure of ecological spaces. By enhancing the ecosystem’s self-repair mechanisms, we can effectively improve ecosystem quality and promote biodiversity. Additionally, the unique vitality, resilience, and service potential of the ecosystem in this area should be leveraged to prevent further degradation, aiming to create a better environment for ecological development. Measures like afforestation and forest tending can improve ecological benefits, optimize the spatial layout of ecological spaces, and enhance the ecosystem’s self-recovery ability by increasing landscape diversity and biodiversity. While maintaining the integrity of the original ecological environment, reasonable green resource development should be pursued, with localized or low-intensity development activities kept within rational limits.

For the Ecological Buffer zone, this area mainly includes the II-2 subtype, accounting for 18.31% of all counties. It is primarily distributed in Hunan, Sichuan, eastern Inner Mongolia, and surrounding areas of Tibet. The ecosystem exhibits moderate health across multiple dimensions, with a relatively unstable structure that is vulnerable to natural disasters and human activities. Some regions even show significant ecological degradation. The management goal of this zone should incorporate ecological design principles, establishing ecological buffer zones and protection and restoration areas to provide favorable environmental conditions for natural recovery, effectively protect ecosystems, promote self-repair, and enhance the overall health and stability of the ecological environment. Additionally, ecological landscape design, restoration of native vegetation, wetland creation, and water source protection projects should be implemented to enrich landscape diversity and improve ecosystem integrity and stability. Actively constructing ecological barriers to reduce external disturbances, promotes orderly protection and utilization of ecological resources and effectively enhances the system’s ability to withstand external pressures.

For the Ecological Remediation zone, this area mainly includes subtypes I-2 and III-1, accounting for 5.21% of all counties. It is primarily located in provinces such as Fujian and southern Gansu, where the vitality of the ecosystem is generally average and the overall organizational capacity, ecological resilience, and service capacity are relatively weak. The management goal of this zone should be supplemented with appropriate human interventions, focusing on ecological remediation and reconstruction. Through these artificial measures, the functions and structures of the ecosystem can be effectively enhanced, promoting its recovery and development. Urban land expansion should be moderately controlled, and idle, vacant, and abandoned lands should be utilized as much as possible to reduce further pressure on the ecological environment. Additionally, special funds can be established, and social capital can be attracted to ensure the sustainability of ecological restoration projects.

For the Ecological Reshaping zone, this area mainly includes subtypes I-1 and II-1, accounting for 42.46% of all counties, primarily located in the northwest of China, surrounding the southeastern coastline, Shandong, and Jiangsu. The vegetation in this zone is sparse, and species diversity is low, with a significant lack of biodiversity. Besides the low vitality of the ecosystem, the health in other dimensions is generally poor. In addition, the ecosystem health in the southeastern coastal areas has been severely affected by intensified human activities in recent years. The primary management goal for this zone should involve direct and proactive human interventions, aiming to rebuild or reshape the ecosystem by reinforcing basic ecological restoration efforts. Active human intervention measures should be combined with landscape ecological design to optimize the landscape structure of urban green spaces. The protection and construction of regional ecosystems such as forests, green spaces, wetlands, and parks should be strengthened to effectively improve the ecological environment, restore ecosystem functions, and achieve sustainable development and balanced ecosystem health. Additionally, the concept of ecological civilization should be incorporated into the comprehensive land management of cities. This includes revitalizing existing land resources, improving the quality of newly added land, and increasing the input–output efficiency per unit of land. Natural and artificial measures should be taken scientifically and reasonably, thereby achieving comprehensive ecosystem reshaping.

5. Conclusions

This study uses China as a case study and constructs an assessment framework for EHI based on the VORS model, revealing the temporal and spatial patterns of EHI in China from 2000 to 2020. Using geographic detectors and GTWR, which explores the main influencing factors of EHI in China and their spatial differentiation characteristics, Ecological Management Zones were further divided and optimization paths explored for these zones. The main conclusions are as follows:

• From 2000 to 2020, the level of EHI in China exhibited an overall declining trend, decreasing from 0.397 in 2000 to 0.377 in 2020, with the area of declining EHI accounting for 59.49% of the total land area. Furthermore, the distribution pattern of EHI exhibited a spatial tendency of “lower in the northwest and higher in the southeast”.
• The four main driving factors affecting EHI in China are vegetation coverage, population density, road networks, and per capita GDP. Additionally, from 2000 to 2020, the degree of influence on ecosystem health shifted from natural factors to socioeconomic factors, with human activities increasingly negatively impacting EHI.
• The land space of China is divided into five Ecological Management Zones: Ecological Conservation Zones, Ecological Enhancement Zones, Ecological Buffer Zones, Ecological Remediation Zones, and Ecological Reshaping Zones, proposing refined and differentiated ecological management paths to provide references for future land space planning and ecological management policy formulation in China.