Abstract
Introduction:
Under the framework of the Sustainable Development Goals (SDGs), the development of green economy and the enhancement of ecological resilience are widely regarded as critical pathways for promoting sustainable urban transformation. However, systematic empirical evidence on the synergistic relationship between the two, along with their spatiotemporal evolution patterns and driving mechanisms, remains limited.
Methods:
Based on panel data from 280 prefecture-level cities in China from 2010 to 2023, this study first constructs a comprehensive evaluation indicator system for green economy and ecological resilience. It then employs methods such as the coupling coordination degree model, Dagum Gini coefficient, Markov chain, and geodetector to systematically analyze the spatiotemporal evolution characteristics and driving mechanisms of the coupling coordination degree (CCD) between green economy and ecological resilience.
Results:
The key findings are as follows: (1) During the study period, both urban green economy and ecological resilience levels exhibited a continuous upward trend, with the CCD steadily increasing, indicating a gradual transition toward a stage of coordinated development. (2) The CCD shows significant regional heterogeneity, following a pattern of “convergence first, divergence later”, with inter-regional disparities serving as the primary source of overall inequality. (3) Spatially, the CCD exhibits a clustering pattern characterized by “high–high” and “low–low” agglomerations, with its evolution displaying notable path dependence and neighborhood spillover effects. (4) The geodetector analysis reveals that the level of digital infrastructure is the dominant driver of spatial differentiation in the CCD, with universal synergistic enhancement effects among various factors.
Discussion:
These findings elucidate the spatiotemporal dynamics underlying the coordinated development of green economy and ecological resilience, providing empirical evidence and policy implications for advancing high-quality and sustainable urban development under the SDGs.
1 Introduction
The 21st century is characterized by unprecedented urbanization and escalating ecological pressures, placing cities at the forefront of global sustainability challenges. With the continuous advancement of global urbanization and industrialization, cities play a key role in driving economic growth and technological innovation, but they also bring a series of environmental problems, including resource overconsumption, ecosystem degradation, and climate change (). In this context, how can damaged natural resources be restored and urban ecological baselines be recovered? Rewilding, as an emerging ecological conservation concept, offers an important approach to addressing this issue. This concept can be traced back to 1998, when Soulé and Noss systematically defined it based on the three core elements of “core areas, corridors, and large carnivores ()”, aiming to rebuild complete trophic structures and self-sustaining ecosystems. Since then, scholars have continued to refine the definition of rewilding, and its target space has shifted from “uninhabited wilderness” toward restoring the autonomy of natural processes (). Furthermore, the United Nations Environment Programme () summarized rewilding as “a conservation approach that allows the wild to return to the land”, a definition that captures the essence of rewilding and provides a theoretical possibility for its application in smaller-scale landscapes, including cities. In response, the concept of urban rewilding has emerged. Urban rewilding refers to measures that restore natural ecological processes in urban environments, rebuild biodiversity, and create shared spaces for humans and other species (; ; ). It aims to allow ecological resources to self-recover by reducing human intervention, thereby achieving ecosystem self-maintenance and resilience, opening new pathways for enhancing urban ecological resilience, improving ecosystem services, and fostering new human–nature relationships ().
The development of a green economy and the enhancement of ecological resilience are two important pathways for advancing urban rewilding from concept to practice (; ). Their coordinated development does not represent two isolated dimensions but rather falls within the theoretical framework of socio-ecological system co-evolution. As a coupled socio-ecological system, the sustainability of a city depends on the interactions between its socio-economic subsystem and its natural ecological subsystem. The green economy, as part of the socio-economic subsystem, primarily reduces the pressure of urban operations on natural capital by optimizing industrial structures, improving resource use efficiency, and promoting clean technologies, thereby creating favorable conditions for ecosystem restoration and regeneration. Ecological resilience, on the other hand, refers to the capacity of the natural ecological subsystem to maintain key functions and self-adjust in response to external disturbances. The coupled and coordinated development of these two dimensions is supported by the role of green economy development in enhancing ecological resilience. On one hand, the advancement of the green economy provides necessary financial and technical support for ecological restoration and environmental governance (; ; ), such as green finance, carbon trading mechanisms, and eco-innovation, which play an important role in improving urban ecological infrastructure (). On the other hand, the industrial structure upgrading and spatial optimization brought about by green economic growth help reduce the encroachment of high-pollution (), high-energy industries on ecological space, releasing more ecological land for urban rewilding (; ; ). In comparison, the improvement of ecological resilience also provides an important foundation for the sustainable development of the green economy. In areas with high levels of ecological resilience, socio-economic order can recover more quickly from external shocks such as natural disasters and climate change, reducing economic losses caused by environmental risks and thereby ensuring the stable operation of the green economy. Moreover, ecosystem services—such as water conservation, climate regulation, and biodiversity maintenance—are essential natural resources upon which green economy development depends. Thus, the green economy provides the natural resource base for improving ecological resilience, while ecological resilience provides the carrying capacity for the sustained development of the green economy. The two are interdependent and co-evolving, together constituting the driving force for advancing urban rewilding ().
As the country with the fastest urbanization process in the world, China is actively exploring pathways to deeply integrate ecological concepts into urban development through strategies such as “Ecological Civilization” construction and “Sponge City” initiatives. This study takes China as a typical case, selecting 280 prefecture-level cities as research samples, and constructs a comprehensive assessment framework to explore the coupling and coordination relationship between green economy development and ecological resilience. By incorporating the ecological connectivity and functional diversity advocated by urban rewilding, as well as the restoration of natural processes and the effectiveness of nature-based solutions emphasized by urban rewilding (), into the measurement and analysis of ecological resilience, this study reveals the dynamic mechanism of their co-evolution. The findings not only provide empirical evidence for understanding sustainable transformation in rapidly urbanizing regions but also offer theoretical references and practical pathways for building more resilient future cities globally through urban rewilding practices. Therefore, exploring the interactive relationship between green economy development and ecological resilience from an integrated perspective is of great value for understanding the sustainable transformation of urban socio-ecological systems.
However, existing studies typically investigate urban green economy and ecological resilience independently. For example, some scholars have explored the path dependence of green economy development from the perspectives of industrial transformation and upgrading and green total factor productivity (; ; ). Other studies have focused on landscape pattern indices and infrastructure resilience assessment, constructing ecological security indicator systems for cities to cope with risks such as flooding (, ). While these studies have achieved fruitful results in their respective fields, most treat the “green economy” as a development goal or performance output, and “ecological resilience” as a spatial baseline or constraint condition, without deeply analyzing the intrinsic mutual feedback mechanisms and coupling logic. Although a few studies have begun to examine the synergistic relationship between economic resilience and green development (), the above research has yielded considerable theoretical value and empirical progress. Nevertheless, it generally exhibits a binary dichotomy in which the “green economy” is seen as a development goal or output performance, and “ecological resilience” as a spatial constraint or foundational condition. Few scholars have integrated the two into a unified theoretical analytical framework to systematically reveal their intrinsic coupling mechanisms. Although recent studies have explored the synergistic relationship between economic resilience and green development () and the relationship between green finance and economic resilience (; ), attempting to break through the traditional paradigm of separate analysis, overall, theoretical construction and empirical testing concerning the dynamic interaction and co-evolution of green economy and ecological resilience remain weak and urgently need further investigation. Existing research provides an important theoretical foundation for this paper, but still lacks a comprehensive assessment framework capable of deeply analyzing the co-evolution of the two and their joint contribution to sustainable development.
Based on this, this study selects urban panel data from 2010 to 2023, constructs a comprehensive evaluation indicator system for green economy and ecological resilience, and applies the coupled coordination degree model to measure the interactive relationship between the two. Furthermore, the Dagum Gini coefficient decomposition method is used to analyze the sources of regional differences, and the Markov chain model is employed to explore the dynamic evolution characteristics of the coupling coordination degree and its spatial neighborhood effects, thereby systematically revealing the spatiotemporal evolution patterns of the synergistic development of green economy and ecological resilience. Finally, the geographical detector model is introduced to investigate the driving factors affecting the coupling coordination development of the two and their interaction effects.
The contributions of this study are mainly as follows: First, from the perspective of socio-ecological systems, it integrates green economy development and ecological resilience into a unified analytical framework, providing operational theoretical support for urban rewilding. Second, it comprehensively applies multiple econometric models to analyze the spatiotemporal evolution characteristics and regional differences of the coupling coordination relationship, revealing the heterogeneous patterns of socio-ecological system development in different regions, thus providing a scientific basis for promoting urban rewilding in a differentiated manner. Third, the research results provide empirical evidence for the synergistic pathway of promoting green economy development and ecological resilience enhancement, and offer policy references for the large-scale implementation of urban rewilding.
2 Materials and methods
2.1 Study area description
This study takes 280 prefecture-level cities in China from 2010 to 2023 as the research object, and constructs a city-level panel data set (excluding Hong Kong, Macao and Taiwan). The sample covers my country’s major urban agglomerations and economic corridors, and can more fully reflect the significant differences in ecological environment endowments, green industry foundations, innovation factor agglomeration and governance capabilities of different cities. The research objects include core cities and megacities with a high concentration of innovative resources and a good foundation for green transformation, as well as resource-based, old industrial bases and cities in ecologically sensitive areas. This provides a representative observation sample for describing the temporal evolution and spatial differentiation of urban ecological resilience and green economy coupling synergy. In order to facilitate regional comparison and spatial gradient analysis, this article divides the sample cities into four major regions: eastern, central, western and northeastern regions based on the relevant regional classification standards of the National Bureau of Statistics. The design of this research area not only helps to identify the evolutionary paths and sources of differences in the coupling coordination degree (CCD) of urban ecological resilience and green economy in different regions, but also provides a suitable analytical framework for subsequent examination of their spatial evolution patterns.
2.2 Method Selection
The specific formulas used in this study are presented as Equations 1–18.
2.2.1 CRITIC–entropy weighted TOPSIS method
In order to comprehensively evaluate the level of urban ecological resilience, this paper uses the TOPSIS method of CRITIC-Entropy combined weighting to construct a comprehensive evaluation index. Compared with a single subjective weighting or a single objective weighting method, CRITIC weight can simultaneously consider the degree of discreteness of indicators and the conflict between indicators. The entropy weight law reflects the information amount and difference intensity of indicators based on information entropy. Combining the two can reduce weight bias to a certain extent and improve the robustness of the comprehensive evaluation results; then the TOPSIS method is used to sort and score the evaluation objects based on the relative closeness of the sample to the ideal solution. The specific calculation steps are as follows:
First, we use the range method to non-dimensionalize the original indicators. Suppose there are m evaluation objects. n evaluation indicators, is the original value, is the normalized value. The calculation formula is as follows:
The entropy weight method can determine the weight based on the degree of variation of the observed values of each indicator. By calculating the feature proportion , information entropy and information utility value , the entropy weight is finally obtained:
The CRITIC method can measure the contrast intensity through the standard deviation of the indicator , The correlation coefficient is used to measure the conflict of indicators, and the information amount is calculated to determine the weight :
Next, this article combines the above two objective weights with equal weight to obtain the final weight . Then construct a weighted normalized decision matrix :
Furthermore, according to the Euclidean space distance formula, the degree of deviation between each evaluation object and the ideal solution is calculated. Determine the positive ideal solution vector and the negative ideal solution vector , and calculate the distance and from each sample to the ideal solution:
Finally, the comprehensive score is calculated through the relative closeness formula
2.2.2 Super-efficiency SBM model
Since the traditional DEA model cannot re-evaluate effective units with an efficiency value of 1, and does not consider the “slack variable” problem of input and output, this article uses the Super-efficiency SBM model that considers undesired output to measure the comprehensive efficiency value of the urban green economy. Assume that under variable returns to scale, there are n decision-making units (DMUs), and each DMU contains m input indicators and q output indicators . The model of the k-th decision-making unit from the input perspective is constructed as follows:
In the formula, is the comprehensive efficiency value of green economy. The larger the value, the higher the efficiency; if , the decision-making unit DEA is considered effective, and vice versa, the non-DEA is effective. and are the slack variables of input and output respectively; m and q correspond to the number of indicators of input and output; is the weight vector; means that when evaluating the k -th decision-making unit, it is eliminated from the reference set, so that the efficiency value of the effective unit can be greater than 1 to achieve further sorting.
2.2.3 Coupling coordination degree model
The preceding sections have successfully developed comprehensive metrics for urban ecological resilience and green economy, establishing a foundation for further analysis of their synergistic relationships. To this end, this study introduces a coupling coordination degree model to quantify the level of coordinated development between these two systems.
The degree of coupling C is used to characterize the degree of interdependence between different variables in economics. In this article, g(a) and g(b) represent the comprehensive evaluation index of urban ecological resilience and urban green economy respectively; T is the total system score, which is used to measure the overall development level of the two systems. In view of the equal importance of urban ecological resilience and urban green economy, α=β=0.5 is taken, and the coupling coordination index D is used to reflect the development coordination between urban ecological resilience and urban green economy. The D value is usually between 0-1: the closer it is to 1, the higher the degree of coupling and coordination between the two; the closer it is to 0, the weaker the coordination. In order to further reveal the coupling relationship between the two, it is necessary to grade the degree of coupling coordination. The grade classification standards are shown in Table 1:
Table 1
| Coordination type | CCD | Classification |
|---|---|---|
| Dissonance Phase | [0.0, 0.1) | Extreme non-coordination |
| [0.1, 0.2) | Severe non-coordination | |
| [0.2, 0.3) | Moderate non-coordination | |
| [0.3, 0.4) | Mild non-coordination | |
| Transitional Phase | [0.4, 0.5) | Near non-coordination |
| [0.5, 0.6) | Slight coordination | |
| Coordination Phase | [0.6, 0.7) | Primary coordination |
| [0.7, 0.8) | Intermediate coordination | |
| [0.8, 0.9) | Good coordination | |
| [0.9, 1.0] | Excellent coordination |
Coordination level classification criteria.
2.2.4 Dagum Gini coefficient decomposition method
The coupling coordination degree model can measure the absolute levels of CCD across cities, but fails to reveal the internal structure and sources of regional disparities. To further characterize the extent and origins of regional differences in CCD, this study adopts Dagum’s decomposition framework by utilizing the Dagum Gini coefficient for measuring and decomposing overall inequality. The calculation formula for the aggregate Gini coefficient G is as follows:
On this basis, according to Dagum’s decomposition method, the overall Gini coefficient G can be decomposed into three parts: intra-regional difference contribution , inter-regional net value difference contribution and hypervariable density contribution , that is, . The calculation formulas for each part are as follows:
In the above formula, k is the number of regional divisions, n is the total number of cities, represents the CCD value of the -th city in region , represents the average CCD of all cities; is the Gini coefficient in region j, is the Gini coefficient between region j and region h; is region j city number share, is the CCD share of region j; represents the relative influence between regions j and h, which is calculated by the inter-regional CCD difference expectation and the hypervariable first-order moment , and is used to measure the mutual overlap of sample values between regions.
2.2.5 Nuclear density estimation method
Dagum Gini coefficient decomposition focuses on static difference structure identification. To further characterize the temporal evolution trends of CCD from the perspective of dynamic distribution evolution, this paper introduces kernel density estimation methods. Kernel estimation is a non-parametric estimation method that can depict the overall trend of coupling coordination from the perspective of distribution patterns. To reveal the overall characteristics of China and its four major regions (spatiotemporal differences and dynamic evolution features of CCD), we employ this method to further reflect the relative positions and evolutionary trajectories between different regions. Its calculation formula is as follows:
In the formula, f(x) is the estimated value of Kernel density function, n represents the number of samples, h is the optimal bandwidth, is the observation value, is the mean value, and K(x) represents the Kernel function.
2.2.6 Spatial autocorrelation analysis
The aforementioned analysis primarily focuses on the temporal evolution and regional variation characteristics of CCD, without examining their spatial correlation patterns. To test whether CCD exhibit spatial clustering or dependency features, this study further incorporates spatial autocorrelation analysis. Specifically, we employ Moran’s I to measure the spatial clustering direction and intensity of CCD, with the calculation formula as follows:
In the formula, , z represents the level of urban ecological resilience and green economy coupling coordination (CCD), i and o are cities, and n is the number of cities. is the spatial weight. We use a binary spatial adjacency weight matrix based on the Queen criterion. If two cities i and o are adjacent, then =1, otherwise, =0.The aforementioned analysis primarily focuses on the temporal evolution and regional variation characteristics of CCD, without examining their spatial correlation patterns. To test whether CCD exhibit spatial clustering or dependency features, this study further incorporates spatial autocorrelation analysis. Specifically, we employ Moran’s I to measure the spatial clustering direction and intensity of CCD, with the calculation formula as follows:
2.2.7 Markov chains
Markov chain method is an effective tool for analyzing state transition and dynamic evolution. This paper analyzes the dynamic evolution characteristics of urban ecological resilience and green economy CCD based on the Markov transition probability matrix to identify the transfer rules of CCD between different levels of status and reveal its stability and convergence in the time dimension. In order to characterize the transfer differences under different development gradients, this article divides CCD into four categories of states according to sample quantiles: Weak, Basic, Good and Excellent. Specifically, cities in the 0–25% quantile range of the sample CCD distribution are defined as Weak, 25%–50% as Basic, 50%–75% as Good, and 75%–100% as Excellent. On this basis, the frequency of city transitions between the four types of states was counted under one lag period and three lag periods respectively, and normalized processing was performed to construct a transition probability matrix, thereby depicting the evolution path of CCD and the probability characteristics of maintaining, rising, and declining states at each level. The specific calculation formula is as follows:
In the formula, represents the probability that the CCD level of a certain city moves from the level of category i in year t to the level of category j in year t+d, (where d=1 represents a lag of one period, and d=3 represents a lag of three periods).
2.2.8 Geographic detector
In this study, geographic detectors were applied to quantify the explanatory power of each driver factor on the spatial disparities of CCD across cities in China, and to test whether synergistic or multiplicative enhancement effects occur when factors interact. The factor detector module evaluates the explanatory degree of each candidate variable on CCD spatial variation, while the interaction detector module assesses whether the combined explanatory power of two factors exceeds, equals, or falls below their independent explanatory power, thereby revealing the compound driving mechanisms underlying CCD spatial inequality. The calculation of q-statistics is as follows:
where q denotes the explanatory power of a given driving factor, with values ranging from 0 to 1, a higher q value indicates a stronger spatial association between the factor and the CCD distribution. h is the category, h=1,2,…,n; L is the number of strata formed by partitioning the driving factor; and are the sample size and variance of stratum h, respectively; and represent the total sample size and overall variance of CCD across all cities.
To further examine interaction effects between different risk factors X, for example, whether factors X1 and X2 work together increase or decrease the explanatory power of the dependent variable Y, or the effects of these factors on Y are independent of each other. The evaluation method is to first calculate the q values of the two factors X1 and X2 against Y: q(X1) and q(X2), and calculate the q value of their interaction (the new polygon distribution formed by the tangent of the two layers of variables X1 and X2): q(X1∩X2), compare the sizes of q(X1), q(X2) and q(X1∩X2), and further use an interaction detector to analyze the impact of different factors on the CCD, it constitutes bivariate enhancement.
2.3 Variable settings
Urban ecological resilience is an important indicator of an urban ecosystem’s ability to maintain structural and functional stability, achieve self-regulation, and recover quickly under external disturbances and uncertain shocks. It is directly related to ecological security, environmental quality improvement, and sustainable urban development. In this study, ecological resilience serves as the core benchmark for evaluating the effectiveness of urban rewilding practices. Urban rewilding aims to restore natural ecological processes, and the improvement of ecological resilience directly reflects the success of this process: the higher the level of ecological resilience, the better the ecosystem recovery, and the more significant the effectiveness of urban rewilding. Based on the logical framework of “ecological resource supply – environmental pressure stress – governance response regulation”, and drawing on relevant studies (; ), this paper constructs an urban ecological resilience evaluation indicator system from three dimensions: state resilience, pressure resilience, and response resilience. This indicator system systematically characterizes the supply and support capacity of ecological space, the environmental stress caused by socio-economic activities, and the mitigation and restoration capacity of pollution control and environmental infrastructure. The specific evaluation indicator system is shown in Table 2.
Table 2
| First level indicator | Secondary indicators | Property | Entropy | CRITIC | Combined |
|---|---|---|---|---|---|
| State resilience | Per capita water resources | positive | 0.068 | 0.082 | 0.075 |
| Green coverage rate of built-up areas | positive | 0.073 | 0.112 | 0.092 | |
| Park green space per capita | positive | 0.429 | 0.086 | 0.258 | |
| Stress resilience | Per capita industrial wastewater discharge | negative | 0.044 | 0.081 | 0.062 |
| Industrial sulfur oxide emissions per capita | negative | 0.027 | 0.105 | 0.066 | |
| Industrial smoke and dust emissions per capita | negative | 0.014 | 0.055 | 0.035 | |
| Carbon emissions per capita | negative | 0.017 | 0.083 | 0.050 | |
| PM2.5 annual average concentration | negative | 0.022 | 0.103 | 0.063 | |
| Responsiveness to resilience | Industrial sulfur dioxide removal rate | positive | 0.080 | 0.020 | 0.050 |
| Industrial smoke and dust removal rate | positive | 0.004 | 0.027 | 0.016 | |
| Harmless treatment rate of domestic waste | positive | 0.121 | 0.099 | 0.110 | |
| Centralized treatment rate of sewage treatment plants | positive | 0.101 | 0.145 | 0.123 |
Evaluation index system for urban ecological resilience.
This study establishes an urban green economy evaluation index system based on the “input-output” analytical framework under green development and resource-environmental constraints, incorporating characteristics of factor inputs in urban economic operations. The system aims to comprehensively and objectively characterize urban green economic development levels by integrating economic output capacity with environmental costs. Urban green economic development relies on multiple factor inputs to generate desired outputs such as economic growth while simultaneously producing unintended outputs like pollution emissions. Building upon existing research (), this study establishes indicators across three dimensions: factor inputs, desired outputs, and unintended outputs, with environmental pollution included in the unintended outputs category. Specifically, at the factor input level, the classical production function framework is adopted, using fixed capital stock, total electricity consumption, and year-end employment per capita to represent capital, energy, and labor inputs respectively. Fixed capital stock better reflects cumulative production capacity than annual investment amounts; total electricity consumption covers multi-sector energy use, offering more comprehensive data coverage and unified statistical standards compared to single-industry energy consumption metrics; employment figures more accurately capture the actual workforce engaged in production than permanent resident populations. For desired outputs, price-adjusted real GDP is selected to eliminate inflationary distortions, while using single output indicators avoids multicollinearity issues and enhances interpretability in efficiency evaluations (). At the level of unintended outputs, industrial wastewater discharge volume, industrial particulate matter emissions, and industrial sulfur dioxide emissions were selected to cover three dimensions: water pollution, atmospheric particulate matter, and acidic gases, respectively. Industrial pollutants were chosen because industrial emissions are the primary source of environmental pressure in China’s cities, with their emission intensity directly correlated to production methods, better reflecting the core essence of green transformation. Moreover, the aforementioned indicators have continuous and complete statistical data at the prefecture-level city level. Relevant indicators and measurement methods are shown in Table 3.
Table 3
| First level indicator | Secondary indicators | Measurement | Unit |
|---|---|---|---|
| factor input index | capital investment | fixed capital stock | Ten thousand yuan |
| energy input | Electricity consumption of the whole society | Thousands of kilowatt hours | |
| labor input | Number of employees in the unit at the end of the year | Thousands of people | |
| expected output indicators | economic output | real GDP | billion |
| Unexpected output indicators | environmental pollution | Industrial wastewater discharge | ton |
| Industrial smoke and dust emissions | ton | ||
| Industrial sulfur dioxide emissions | 10,000 tons |
Evaluation index system for urban green economy.
2.4 Data resources
In order to ensure the representativeness of the sample and the continuous availability of indicator data, this article takes 280 cities at the prefecture level and above in my country from 2010 to 2023 as the research objects (excluding Hong Kong, Macao and Taiwan), and constructs a city-level panel data set. The regional division follows the relevant standards of the National Bureau of Statistics, and the sample cities are grouped into four major regions: eastern, central, western and northeastern regions. The data mainly comes from the National Bureau of Statistics, “China Statistical Yearbook”, “China Cities Statistical Yearbook”, “China Environmental Statistical Yearbook”, “China Energy Statistical Yearbook” and other publications, and is also combined with relevant data from various city statistical yearbooks, statistical bulletins and government information disclosure platforms; some indicators are supplemented and verified through authoritative databases such as EPS and CSMAR. For missing observations in individual years or cities, this article uses interpolation to complete the data to ensure the integrity and comparability of the data.
3 Result analysis
3.1 Analysis of measurement results
As can be seen from Figures 1a–d, urban ecological resilience and green economic levels are both on the rise from 2011 to 2023, but there are significant differences in their evolution paths and spatial distribution characteristics. Ecological resilience shows a continuous, smooth and steady increase in the time dimension, indicating that the carrying capacity, risk resistance and recovery capabilities of China’s urban ecosystems continue to increase; in terms of regions, the east has maintained a long-term lead, the central and western parts are in a medium gradient, and the northeast is relatively low but the catch-up in the later period is more obvious. The distribution results further show that the urban ecological resilience level in the eastern region is higher and relatively stable, while the dispersion is greater in the west, reflecting that the improvement of ecological resilience is universal but the intra-regional differences are still prominent. In contrast, the green economy shows stronger periodic fluctuations and later acceleration characteristics. It is at a low level in the early stage, gradually turns to an upward channel in the middle and late stages, and shows a significant jump at the end of the period, indicating that green growth momentum is concentrated in the later stages of the study. Spatially, the overall level of the east and northeast is higher and the upper tail is longer, indicating that they have more significant advantages in the agglomeration of innovative factors, industrial green upgrading and technology diffusion, while the central region is overall low and the distribution is relatively concentrated, indicating that the accumulation of green momentum and the efficiency of structural transformation in this region are still constrained.
Figure 1
It can be further observed from Figures 1e, f that the CCD of ecological resilience and green economy was generally in a high range and increased significantly from 2010 to 2023, and there was a simultaneous jump at the end of the period, indicating that the synergistic relationship between the two continued to deepen and the quality of synergy improved significantly. In time, CCD shows a phased rise and accelerates upward in the later period, reflecting the steady-state improvement of ecological resilience and the later acceleration of the green economy forming a co-directional strengthening effect, promoting the evolution of the system from a lower coordination level to a higher coordination level. Spatially, the CCD in the northeastern region is generally at a high level and the distribution has moved upward as a whole, reflecting a strong stability advantage; the eastern region has seen a more prominent growth rate in the later period and achieved overtake, showing that its coupling mechanism between green transformation efficiency and resilience improvement is smoother; while the CCD level in the central and western regions is low and its improvement is relatively slow, indicating that there are still constraints in terms of technology diffusion efficiency, industrial structure optimization, and green governance performance transformation. Overall, the national CCD continues to rise clearly, but regional gradient differences are still prominent, and coordinated improvement is more reflected in the overall improvement driven by high-level regions.
3.2 Analysis of regional disparity results
We use the Dagum Gini coefficient to calculate and decompose the regional differences in CCD levels in China as a whole and in each region. The results are shown in Figure 2.
Figure 2
As can be seen from Figure 2, the intra-regional differences in CCD from 2010 to 2023 show the phased characteristics of “first convergence and then differentiation”. The national overall situation dropped significantly from 2011 to 2014 and maintained low fluctuations, indicating that the CCD gap between provinces in the early stage was narrowed, which was more reflected in the general improvement of low-level areas driven by unified environmental protection constraints, green governance investment and industrial upgrading policies. The national Gini coefficient rose significantly after 2019 and reached a high in 2023, indicating that the later CCD increase did not simultaneously bring about equalization, but instead widened intra-regional differences. This change may be related to the fact that green transformation has entered a stage of rapid development: this stage relies more on digital infrastructure, green technology innovation and high-end industrial chain support. There are agglomeration and scale effects in related factors, resulting in faster improvement in leading cities and limited follow-up in tail regions, thus strengthening internal stratification.
From a regional perspective, the Gini coefficient in the eastern and central areas is generally low and relatively balanced internally, but it increased at the end of the period, which shows that in the high-level coordination stage, the leading provinces have accelerated their improvement by relying on their advantages in innovative resources, green finance and industrial chains, as shown in Figure 2. The early differences in the western region were large and then significantly converged, indicating that infrastructure improvements and policy support had alleviated internal imbalances; but the rapid rise after 2020 reflects the intensified differentiation within the western region in terms of undertaking new energy industries, factor attraction capabilities, and digital transformation conditions. The Northeast is more volatile and the difference remains high after 2019, which may be related to the high proportion of traditional industries and the imbalance of structural adjustment and green innovation momentum.
From the perspective of inter-regional differences, the CCD of urban ecological resilience and green economy showed significant spatial differentiation and staged evolution characteristics during the study period, as shown in Figure 3. Overall, the difference between cities in the west and northeast is the most prominent. The average Dagum Gini coefficient is 0.0496. It fell significantly from 2010 to 2014 and then increased rapidly starting in 2020, reaching a peak of 0 in 2022.0676, falling back to 0.0561 in 2023; the difference between central and northeastern China is second, with an average value of 0.0474, which also shows the phased characteristics of “early convergence - later expansion” and will remain at a relatively high level after 2020. In contrast, the difference between the east and the central region is relatively smallest, with an average value of 0.0400, and has been fluctuating at a low level for a long time, indicating that the cities in adjacent regions are closer in CCD level; the differences between the east and the west and the east and the northeast are in the middle, and there is a clear upward trend at the end of the sample, indicating that there are signs of further expansion of the coupling coordination gap between cross-regional cities. Overall, inter-regional differences also showed the structural characteristics of “first convergence, then differentiation” during the study period, and the differences between the central and western regions and the northeast were more obvious. The fundamental reason for this formation lies in the systematic differences in economic foundation, innovation capabilities, resource endowments and policy environments between regions.
Figure 3
As shown in Figure 4, inter-regional differences were the main source of CCD spatial imbalance during the study period, and the contribution rate was higher than intra-regional differences for a long time, indicating that cross-regional development gaps still dominate the overall pattern. The contribution of intra-regional differences is relatively stable and has a low proportion, indicating that urban differentiation within the same region has a limited role in promoting overall differences. The contribution rate of hypervariable density is generally small, but has increased in recent years, reflecting the increased overlap and overlap of CCD distribution in cities in different regions, and the differentiation between cities showing a certain structural complexity trend. In short, the key to the balanced improvement of CCD still lies in the gap between regional cities in ecological governance capabilities, green industry foundation and innovation factor agglomeration. As the diffusion of green technology accelerates, the flow of factors becomes smoother, and cross-regional collaborative governance continues to advance, the level of coupling and coordination between urban ecological resilience and green economy is expected to be further improved, and the spatial pattern will also tend to be more coordinated and balanced.
Figure 4
3.3 Analysis of kernel density estimation results
The Gini coefficient decomposition of Dagum reveals the structural characteristics and origins of CCD regional disparities, yet fails to intuitively demonstrate the dynamic evolution of CCD distribution patterns. To further analyze overall distribution shifts, dispersion level variations, and polarization trends of CCD, this study employs Gaussian kernel density estimation to generate three-dimensional kernel density maps, as illustrated below:
The three-dimensional kernel density curve distribution of urban ecological resilience and green economy CCD at the national level showed clear temporal evolution characteristics during the study period, as shown in Figure 5. First of all, judging from the distribution position and curve shape, the main peak of core density moves to the right as a whole, indicating that the overall level of CCD continues to improve, and the collaborative development capabilities of urban ecological resilience and green economy continue to increase. At the same time, the peak height fluctuates repeatedly and the curve width expands, indicating that the CCD dispersion between cities is increasing, and urban differences are showing a certain expansion trend, reflecting that there are still structural gaps in different cities in terms of green industry foundation, innovation factor aggregation, and ecological governance capabilities. Secondly, from the perspective of distribution ductility, the tail on the right side of the high-value range gradually lengthens and tends to thicken, indicating that some cities are the first to enter a higher CCD level and form a leading echelon, which has a significant driving effect on the overall improvement. However, it also reveals that the CCD improvement process is uneven, which is closely related to differences in cities’ green technology diffusion capabilities, infrastructure resilience construction, greening of industrial structures, and policy implementation intensity. Finally, from the perspective of polarization characteristics, the distribution curve shows certain shoulder and multi-peak shapes, indicating that urban CCD have hierarchical grouping characteristics and show an echelon evolution trend.
Figure 5
The evolution of core density at the national level reveals that the overall improvement of urban CCD coexists with structural differentiation, but this trend may show significant heterogeneity in different regions. In order to further identify regional differences and their evolution paths, this paper uses the four major regions of eastern, central, western and northeastern China as grouping units to conduct a regional comparative analysis of the core density curves, as shown in Figure 6.
Figure 6
As shown in Figure 6, the core density distributions of urban ecological resilience and green economy CCD in the four major regions all show varying degrees of rightward shifts, but there are significant differences in the pace of improvement and the evolution of internal structures. Figure 6a shows that the main peak in the eastern region has been stable in the mid-to-high value area for a long time and has moved slightly to the right, indicating that the CCD of eastern cities continues to improve on a higher basis. At the same time, the peak value has dropped, the curve has broadened, and the high-value end has trailed and extended, indicating that the dispersion within the region has increased, and high-level urban groups have expanded and promoted the deepening of gradient stratification. The above differences are mainly due to the inconsistencies in the depth of green industry upgrading, intensity of innovation factor agglomeration, ecological governance investment and resilient infrastructure construction process in eastern cities. Figure 6b shows that the main peak in the central region steadily moves to the right and the curve is more concentrated. The width expansion is limited and the right tail is not prominent in the middle and late stages. This shows that the CCD of central cities is improving more synchronously and the internal differences are relatively small. The overall trend is a balanced advancement. This is consistent with the general improvement of the central region under the influence of industrial inheritance, factor return and policy diffusion. Figure 6c shows that the main peak in the western region shifts to the right more slowly and fluctuates more obviously. The distribution is more dispersed and accompanied by a certain right tail and shoulder. It reflects that although the CCD of western cities has improved, it is overall slow and has strong internal differences. A few cities are the first to break through, while most cities are still in the low-to-medium level range. This pattern is related to the differences between western cities in terms of factor endowment and marketization degree, ecological vulnerability constraints, and resilience shortcomings of public services and infrastructure. It leads to uneven marginal returns from promoting green economy and improving ecological resilience, thus exacerbating internal gradient differentiation. Figure 6d further shows that while the main peak in the Northeast slowly moves to the right, the right tail of the high-value end is more obvious and there are signs of secondary peaks in the later period, indicating that some cities in the Northeast are improving faster and entering the higher CCD range, driving the expansion of the high-value end, but also making the stratification and differences within the region begin to deepen. The reason may be that there are differences in the intensity of industrial structural adjustment and green transformation promotion among northeastern cities: resource-based and old industrial base cities have different processes in industrial renewal, green technology application and governance investment. At the same time, differences in population mobility and factor attraction capabilities have led to the formation of relatively prominent improvement channels in some cities, while other cities have relatively slowly improved.
3.4 Analysis of spatial autocorrelation results
The preceding analysis revealed the evolutionary characteristics of CCD from temporal and regional perspective. This study then shifts focus to spatial dimensions to examine whether geographic spatial clustering effects and dependencies exist among CCD. We employed the global Moran’s I index to conduct spatial autocorrelation tests using three spatial weight matrices: W1 (0–1 adjacency matrix based on Queen’s rule), W2 (inverse square of geographic distance matrix), and W3 (economic geography weight matrix). Results are presented in Table 4. Under W1, Moran’s I values remained positive throughout the study period, demonstrating statistical significance at least at the 5% level. These findings indicate that CCD exhibit persistent positive spatial clustering patterns rather than random distribution across geographic regions.
Table 4
| Year | W1 | W2 | W3 | |||
|---|---|---|---|---|---|---|
| Moran’s I | p-value | Moran’s I | p-value | Moran’s I | p-value | |
| 2010 | 0.1084 | 0.0049 | 0.0538 | 0.0018 | 0.1017 | 0.0004 |
| 2011 | 0.0955 | 0.0134 | 0.0712 | 0.0001 | 0.0296 | 0.2680 |
| 2012 | 0.1315 | 0.0007 | 0.0732 | 0.0000 | 0.0315 | 0.2411 |
| 2013 | 0.1422 | 0.0002 | 0.0736 | 0.0000 | 0.0279 | 0.2885 |
| 2014 | 0.0979 | 0.0104 | 0.0539 | 0.0017 | 0.0296 | 0.2635 |
| 2015 | 0.1084 | 0.0050 | 0.0650 | 0.0002 | 0.0535 | 0.0560 |
| 2016 | 0.0723 | 0.0563 | 0.0368 | 0.0279 | 0.0382 | 0.1597 |
| 2017 | 0.2388 | 0.0000 | 0.1352 | 0.0000 | 0.1138 | 0.0001 |
| 2018 | 0.1774 | 0.0000 | 0.1028 | 0.0000 | 0.0915 | 0.0015 |
| 2019 | 0.2947 | 0.0000 | 0.1926 | 0.0000 | 0.1344 | 0.0000 |
| 2020 | 0.2512 | 0.0000 | 0.1750 | 0.0000 | 0.1234 | 0.0000 |
| 2021 | 0.2977 | 0.0000 | 0.1920 | 0.0000 | 0.1482 | 0.0000 |
| 2022 | 0.2600 | 0.0000 | 0.1709 | 0.0000 | 0.1328 | 0.0000 |
| 2023 | 0.1188 | 0.0024 | 0.0813 | 0.0000 | 0.0977 | 0.0008 |
Global Moran’s I results for CCD in China, 2010–2023.
Notably, Moran’s I value exhibits distinct phase-specific characteristics closely aligned with major environmental policies. Before 2017, the values remained low with significant fluctuations, reflecting that regional green transitions were still in spontaneous exploration stages without established spatial coordination mechanisms. From 2017 onward, a marked surge occurred coinciding with pivotal policy milestones: the 19th National Congress of the Communist Party of China incorporated ecological civilization into national strategies, and cross-regional environmental supervision systems were fully implemented. These unified policy measures facilitated synchronized green development trajectories among neighboring cities. Between 2019 and 2022, the value maintained above 0.25, corresponding to the institutionalized advancement of regional green governance following the “dual carbon” goals. In 2023, the value declined to 0.1188 yet remained statistically significant, likely attributable to cities adopting differentiated decarbonization strategies and diverging policy-driven synchronization effects.
The validation using W2 matrix revealed that Moran’s I values remained significantly positive across all years, with temporal trends consistent with W1 but generally lower overall values—a finding consistent with the geographical principle of spatial correlation attenuation with distance. Further analysis using W3 matrix demonstrated that most years before 2017 failed to pass significance tests, while all post-2017 years showed highly significant results at the 1% level. This indicates that economic linkages gradually emerge as drivers of CCD spatial agglomeration alongside regional integration and the promotion of economic policy instruments such as green finance and carbon trading. Overall, the three matrices exhibit high consistency in agglomeration directions and temporal characteristics, confirming that the conclusions remain robust regardless of weight matrix selection, demonstrating strong statistical robustness.
It should be pointed out that the global Moran’s I can only reflect the spatial correlation strength at the overall level, and it is difficult to reveal the local agglomeration type and its spatial location. To this end, we further combine local Moran scatter plots for identification. As shown in Figure 7, the fitted regression lines in each year show a positive slope, which is consistent with the global Moran’s I result, indicating that CCD has a stable positive spatial correlation. More importantly, the scattered points are mainly concentrated in the first and third quadrants, with high-high agglomeration and low-low agglomeration as the main characteristics. Although there are still a small number of observation points in the second and fourth quadrants, reflecting that there are still high-low or low-high heterogeneous adjacency and transition phenomena at the local level, it does not change the overall spatial agglomeration pattern with “high-high” and “low-low” types as the core. Overall, the global and local tests corroborate each other, indicating that during the study period, CCD showed evolutionary characteristics dominated by spatial agglomeration, accompanied by dynamic differentiation and staged adjustment.
Figure 7
3.5 Analysis of spatiotemporal evolution results
Spatial autocorrelation analysis confirmed the significant spatial clustering characteristics of CCD, but failed to reveal the dynamic transfer patterns between different hierarchical levels of urban CCD or the moderating role of neighboring environments. To address this, this study constructed both traditional Markov transition matrices (Table 5) and spatial Markov transition matrices (Figure 8), while simultaneously examining transfer characteristics at one-period and three-period lag levels.
Table 5
| Type | t/(t+1) | Weak | Basic | Good | Excellent | N |
|---|---|---|---|---|---|---|
| Traditional (t+1) | Weak | 0.779 | 0.186 | 0.034 | 0.001 | 968 |
| Basic | 0.104 | 0.605 | 0.261 | 0.029 | 958 | |
| Good | 0.025 | 0.130 | 0.597 | 0.248 | 929 | |
| Excellent | 0.009 | 0.038 | 0.103 | 0.850 | 785 | |
| Traditional (t+3) | Weak | 0.590 | 0.283 | 0.098 | 0.030 | 880 |
| Basic | 0.143 | 0.373 | 0.366 | 0.118 | 863 | |
| Good | 0.056 | 0.192 | 0.376 | 0.377 | 756 | |
| Excellent | 0.031 | 0.091 | 0.194 | 0.683 | 581 |
Traditional Markov transition probability matrix results.
Figure 8
First, traditional Markov results show that CCD has significant state stickiness and path dependence. Regardless of whether it is lagged by one period or lagged by three periods, the probability of the main diagonal is significantly higher than that of the off-diagonal items, indicating that cities at all levels are more inclined to maintain the original state and exhibit the characteristics of “club convergence”. Specifically, under the one-period lag condition, the retention probabilities of Weak, Basic, Good, and Excellent are 0.779, 0.605, 0.597, and 0.850 respectively, and the Excellent group has the highest stability. At the same time, transitions mainly manifest as step-by-step changes in adjacent levels, and the upward probability is generally greater than the downward probability: for example, the transfer from Weak to Basic is 0.186, Basic to Good is 0.261, Good to Excellent is 0.248, while the downward transfer is generally small, indicating that the CCD evolution has an overall direction of “stable with improvement”. In contrast, the diagonal probability under the three-period lag condition generally decreases (for example, Weak drops to 0.590 and Excellent drops to 0.683), indicating that as the time span increases, urban cross-level mobility increases, but the system still maintains the characteristics of stable evolution.
The spatial Markov results further show that the level of neighborhood development will significantly affect the evolution path of urban CCD. The results are shown in Figure 8. In general, when the neighborhood is at a low level, it is more likely to form a low lock-in, while when the neighborhood level is high, a positive drive is more likely to occur. Taking the first lag period as an example, when the neighborhood is Weak, the probability of Weak cities remaining in their original state increases to 0.855, and the room for improvement is obviously limited. This shows that under the background of relatively insufficient resource factors, weak green industry foundation, and insufficient cross-city coordination, low-level states are more likely to continue. As the neighborhood is upgraded to Basic or Good, the probability of transition from Weak to Basic increases significantly, and the upward channel from Basic to Good also widens, indicating that the diffusion of green technology, industrial chain linkage and ecological governance collaboration in neighboring cities can promote urban upgrading through spatial spillover. Under the three-period lag, the neighborhood effect is more sufficient: the overall maintenance probability decreases and the upward probability increases. In particular, mid- and high-level cities are more likely to transition to Excellent, reflecting that the cumulative effects of regional integration, factor flow, and policy coordination are more likely to be transformed into sustained improvement over a longer period. But at the same time, low-level cities still face strong lock-in constraints in low-level neighborhoods, suggesting that late-developing regions need to make up for the shortcomings of resilient infrastructure and public services, enhance investment in green innovation, and improve cross-regional collaboration capabilities in order to effectively break through low-level solidification and achieve a continued jump in CCD.
3.6 Analysis of driving factor results
The aforementioned analytical framework delineates the spatiotemporal evolution patterns, regional disparity structures, and spatial clustering characteristics of CCD (Centralized Commercial Districts), yet fails to uncover the underlying drivers of their spatial differentiation. To address this gap, this study employs the geodetector model to identify key determinants and their interactive effects influencing CCD spatial heterogeneity. Considering the availability of long-term panel data at prefecture-level city level and the geodetector model’s sensitivity to missing values, four driving factors were systematically selected for analysis: digital infrastructure level (X1, measured by annual mobile phone subscriber count), education expenditure level (X2, calculated as education spending as a percentage of local fiscal general expenditures), financial development level (X3, represented by financial institution deposit balances relative to regional GDP), and government intervention intensity (X4, measured by fiscal expenditure as a percentage of GDP). Table 6 presents descriptive statistical results for each variable.
Table 6
| VarName | Obs | Mean | SD | Min | Median | Max |
|---|---|---|---|---|---|---|
| CCD | 3920 | 0.612 | 0.055 | 0.449 | 0.604 | 0.970 |
| X1 | 3920 | 488.400 | 519.408 | 37.000 | 340.000 | 4594.000 |
| X2 | 3920 | 0.176 | 0.039 | 0.044 | 0.175 | 0.356 |
| X3 | 3920 | 1.558 | 0.697 | 0.000 | 1.412 | 20.103 |
| X4 | 3920 | 0.205 | 0.097 | 0.056 | 0.180 | 0.809 |
Descriptive statistics of variables.
Table 7 presents the factor detection results based on full-sample panel data. All four driving factors passed the significance test at the 1% level, indicating that their explanatory role for the spatiotemporal variation of CCD is statistically significant.
Table 7
| Driving Factor | X1 | X2 | X3 | X4 |
|---|---|---|---|---|
| q-value | 0.0585 | 0.0360 | 0.0052 | 0.0324 |
| p-value | 0.0000 | 0.0000 | 0.0026 | 0.0000 |
| Rank | first | second | fourth | third |
Single-factor detection results of CCD development level, 2010–2023.
In terms of explanatory power ranking among factors, digital infrastructure level (X1) demonstrated the highest q-value, serving as the primary driver of CCD spatial differentiation. This indicates that the digitalization level represented by mobile communication infrastructure plays a pivotal role in shaping the coordinated development pattern of urban green economy and ecological resilience. Cities with more comprehensive digital infrastructure are better positioned to achieve efficient diffusion of green technologies, precise ecological monitoring, and enhanced environmental governance capabilities, thereby promoting higher-level coupling coordination. Education expenditure level (X2, q = 0.0360) ranked second, reflecting the significant impact of human capital accumulation on CCD spatial differentiation: higher education investment helps cultivate green development awareness and ecological governance capabilities, providing intellectual support for continuous improvement in coupling coordination. Government intervention intensity (X4, q = 0.0324) ranked third, indicating that proactive fiscal policy interventions play a substantial role in driving green transformation and bridging regional coordination gaps. Financial development level (X3) exhibited the weakest single-factor explanatory power, which may be attributed to uneven distribution of financial resources across cities and varying capacities to guide capital flows toward green sectors, resulting in relatively limited direct driving effects.
Figure 9 further presents the interaction detection results for four representative cross-sectional years. Overall, the interaction q-values for all pairwise combinations of factors were significantly higher than those for corresponding single factors, indicating that the driving factors exerting effects on CCD are not mutually independent but rather demonstrate a widespread synergistic enhancement effect.
Figure 9
A horizontal comparison of interaction intensity reveals that interaction combinations involving X1 consistently ranked at the top. In 2010, the interaction q-value for X1∩X2 reached 0.3985, followed closely by X1∩X3 (0.3707) and X1∩X4 (0.3610), indicating that the synergy between digital infrastructure and other factors exerted strong shaping power on CCD spatial patterns during the early research phase. The interaction effects of X4 with other factors remained statistically significant, with X4∩X2 reaching 0.2421 in 2015 and X4∩X3 rising to 0.3031 by 2023, revealing enduring complementary enhancement relationships between fiscal policy, human capital, and financial development.
From a temporal perspective, interaction q-values exhibited significant declines across all variables in 2015. For instance, the X1∩X2 value dropped from 0.3985 to 0.1391, while X1∩X4 decreased from 0.3610 to 0.1402. These trends align with the characteristics of slowing CCD growth rates and narrowing regional disparities during the same period, indicating that the driving force structure was moving toward equilibrium. After 2020, interaction effects rebounded overall, with combinations like X1∩X3 and X1∩X2 showing renewed strength. This aligns with the reality of accelerated digital-green transition and widening regional disparities following the “dual carbon” goals. Comprehensive analysis reveals that CCD’s spatiotemporal differentiation results from multiple driving forces working synergistically: digital infrastructure serves as the core driver, though its impact is significantly amplified through interactions with education, finance, and government policies. Government intervention acts as a crucial institutional factor, continuously shaping CCD’s spatial evolution patterns through coordinated efforts with market forces.
4 Discussion
We research reveals several important characteristics of the coupling coordination relationship between green economy and ecological resilience across 280 prefecture-level cities in China. From a temporal perspective, both urban ecological resilience and green economy levels show a continuous upward trend during the sample period, but their evolution processes are not synchronized: the improvement of ecological resilience is more stable and continuous, while the green economy exhibits more pronounced stage-specific characteristics and shows an accelerated improvement trend in the later period. This finding is consistent with the study by on the Yangtze River Delta region, which found that green innovation efficiency jumped from 0.252 to 0.692, while ecological resilience steadily increased from 0.228 to 0.395, jointly driving the coupling coordination degree from mild disorder to primary coordination. The accelerated improvement of the green economy in the later period may be related to the concentrated release of green technology innovation investments following the introduction of the “dual carbon” goals, while the steady improvement of ecological resilience reflects the cumulative and lagged nature of ecosystem responses. This finding has important implications for urban rewilding practices.
Urban rewilding emphasizes the design and management of urban green spaces with high connectivity and functional diversity, the implementation of which requires sustained policy and financial support—exactly what the accelerated development of the green economy provides. Meanwhile, the steady improvement of ecological resilience confirms the core logic of urban rewilding practices: the restoration of ecological processes has its own intrinsic temporal rhythm and cannot achieve leapfrog progress through short-term investments. The “asynchrony” of the two evolution processes precisely reveals the basic pattern of rewilding practices: changes in socio-economic systems can be relatively rapid, while ecosystem responses require patience. This provides an important warning for urban decision-makers: when advancing urban rewilding, it is necessary to have both the “fast variable” drive of green economy inputs and to respect the “slow variable” rhythm of ecological restoration, avoiding the abandonment of long-term commitment due to the lack of short-term results. On this basis, the overall coupling coordination degree between the two has been rising and shows a synchronous leap in the later period of the study, indicating that urban development in China is gradually shifting from a state of relative separation between green growth and ecological governance toward a higher stage of synergistic evolution. This trend is consistent with the findings of scholars on the five major urban agglomerations, who found that the synergy between economic resilience and green economy efficiency has been continuously improving during the study period, indicating a benign interaction and synergistic evolution trend between the two systems (). The study by also confirms that the coupling coordination degree between China’s economic resilience and green low-carbon development is generally on the rise, with the eastern region having the highest level and the central region experiencing the fastest growth. Due to data availability constraints, our measurement of ecological resilience has not yet fully covered key dimensions such as biodiversity and the integrity of ecosystem service functions, which may to some extent affect the comprehensive characterization of the evolutionary features of ecological resilience. Future research could be further refined by integrating remote sensing inversion and ecological monitoring data.
From a spatial perspective, there exists regional heterogeneity and spatial imbalance, and the coupling coordination between the green economy and ecological resilience is not advancing evenly but shows significant regional heterogeneity. The spatial imbalance of urban CCD overall exhibits an evolution trend of ‘convergence first, divergence later,’ but differences between regions have always been the main source of overall differences. Our research findings are mutually corroborated by , who found that spatial differences in the coordination of the five major city clusters primarily arise from super-variable density, with regional differences being a secondary factor. Other scholars studying the efficiency of industrial green technology innovation in China have similarly used the Dagum Gini coefficient decomposition, confirming that inter-regional differences are the dominant source of overall differences (). This suggests that in promoting the synergistic development of the green economy and ecological resilience, narrowing inter-regional gaps should be a key policy focus. Kernel density estimation results further indicate that the national-level CCD distribution peak continues to shift to the right, reflecting a steady improvement in overall coordination level; meanwhile, fluctuations in the peak, widening of the distribution, and extension of the right tail also show that the gap between cities does not disappear synchronously with overall improvement. Instead, some cities leap ahead first, exhibiting characteristics of a tiered evolution of the overall pattern. Additionally, studies on the Huaihe Ecological Economic Belt confirm a clear trend of urban ecological resilience shifting towards a higher level, but significant inter-regional imbalance remains. Empirical results from the spatial dimension further reveal the substantial externality characteristics of urban coordinated transformation. Spatial autocorrelation analysis shows that urban CCD has significant and stable positive spatial correlation, and the ‘high–high’ clustering pattern indicates that leading regions in the synergistic development of the green economy and ecological resilience (such as eastern coastal city clusters) have already formed institutional environments and resource networks conducive to rewilding practices. These regions are better positioned to advance higher-standard urban wilderness restoration projects, such as establishing regional-scale ecological corridors and restoring critical ecological processes. In addition, the ‘low–low’ lock-in effect reveals challenges faced by another group of cities. In areas with both low green economy and ecological resilience, rewilding practices may face dual constraints of resource scarcity and institutional lag, potentially trapping them in a vicious cycle of ‘the less you develop, the harder it is to protect.’ Finally, spillover effects exist, providing a theoretical basis for promoting rewilding through regional coordination—the radiation effect of high-level neighboring areas can serve as external motivation for low-level cities to break development lock-ins. This suggests that urban rewilding should not be an isolated action by a single city but integrated into the overall framework of regional ecological networks and coordinated development strategies. similarly found that coupling coordination exhibits positive spatial autocorrelation, with most provinces showing high–high or low–low clustering characteristics. This indicates that the development status of neighboring cities can significantly influence local cities’ coordinated transformation through technology diffusion, industrial linkage, and policy imitation.
Meanwhile, Markov chain analysis reveals that the evolution of CCD exhibits significant path dependence and state stickiness, along with a certain degree of club convergence. This “state stickiness” originates from the self-reinforcing mechanism of urban green transition: the stock of infrastructure formed by ecological governance investments, the scale effects brought about by the agglomeration of green industries, and the inertia of environmental governance institutions together constitute the intrinsic driving force that keeps cities on their existing development trajectories, making short-term cross-tier jumps face high transition costs. A study by on the ecological welfare performance of Chinese cities reached a consistent conclusion, finding that “club convergence” constrains the short-term cross-state transition of ecological welfare performance. The spatial Markov results further indicate that the neighboring development environment significantly influences the evolutionary direction of local cities. When the neighborhood is at a low level, the probability of weak cities remaining in their original state increases from 0.779 under traditional Markov to 0.855, indicating a significantly intensified low-level lock-in effect. As the neighborhood level improves, the probability of upward transitions for cities increases markedly. This neighborhood effect operates primarily through three pathways: first, green technology diffusion—technological spillovers from neighboring high-level cities lower the threshold for green innovation in local cities; second, industrial chain linkage—the green upgrading of upstream and downstream industries within the region creates demonstration and pulling effects on adjacent cities; third, governance synergy—cross-regional environmental regulation and ecological compensation mechanisms encourage neighboring cities to align their governance standards. Conversely, when the neighborhood is generally at a low level, these positive spillover channels are blocked, and cities are more likely to fall into a “low-level equilibrium trap”. This finding is mutually supportive with the conclusions of , who found that neighborhood state affects transition probabilities, but high-level provinces are not influenced by the neighborhood environment. Studies on the ecological efficiency of coastal urban agglomerations have also confirmed that spatial spillover effects help promote dynamic transitions and gradient leaps.
At the driver level, geospatial analysis provides crucial insights into the underlying mechanisms of CCD spatiotemporal differentiation. Factor detection results indicate that digital infrastructure development (X1) serves as the primary driver of CCD spatial divergence, highlighting the pivotal role of digital transformation in shaping coordinated urban green economy and ecological resilience development patterns. Cities with well-established digital infrastructure not only facilitate efficient diffusion of green technologies and precise ecological monitoring but also enhance environmental governance efficiency and resource allocation accuracy through digital platforms, thereby promoting higher-level synergistic coordination. Educational expenditure levels (X2) rank second, demonstrating how human capital accumulation fosters green development awareness and improves ecological governance capabilities, providing essential intellectual support for CCD continuous improvement. Government intervention intensity (X4) ranks third, underscoring the institutional significance of fiscal policies in driving green transition and bridging regional coordination gaps. Notably, financial development level (X3) exhibits relatively weak single-factor explanatory power, likely attributable to uneven distribution of financial resources across cities and disparities in capital allocation capacity for green sectors, with its direct driving effects requiring further exploration. More importantly, the interaction analysis revealed a widespread synergistic enhancement effect among driving factors. The interaction q-values for all pairwise factor combinations were significantly higher than those of corresponding single factors, indicating that the spatiotemporal differentiation of CCD is not dominated by a single factor but rather the result of multiple driving forces working in concert. This finding deepens our understanding of spatiotemporal patterns at the causal mechanism level: The previously observed “convergence followed by differentiation” trend essentially reflects the structural transition of driving forces from equilibrium to differentiated reorganization. The formation of spatial spillover effects and path dependence characteristics is also closely related to the uneven spatial allocation of digital infrastructure and institutional resources. These findings provide crucial policy insights for urban rewilding and refined wilderness practices: Urban rewilding initiatives should not rely on a single approach but require systematic integration of digital infrastructure development, educational investment, and fiscal policies to establish a multi-factor collaborative governance framework. This approach can more effectively narrow regional development gaps and break low-level lock-in patterns.
Although this study comprehensively employs multiple models to investigate the spatiotemporal characteristics of CCD, the following limitations remain. First, the research sample is confined to prefecture-level cities in China, and the generalizability of the findings to other countries or regions requires cautious evaluation, particularly considering the institutional environment and development stage unique to China. Second, the construction of the indicator system is constrained by data availability, and certain dimensions of ecological resilience (such as biodiversity and ecosystem service function integrity) were not adequately captured. Third, the selection of driving factors for geographic detectors is limited by the availability of long-time series panel data at the prefecture-level city level, and some key factors (such as technological innovation levels, urbanization quality, and industrial structure upgrading) were not included in the analysis due to data gaps, which can be further expanded in future studies with improved data conditions. Fourth, this study primarily employs cross-sectional analysis and state transition models to characterize spatiotemporal patterns without establishing a rigorous causal identification framework among variables. Future research could focus on the following directions: first, conducting transnational comparative studies to explore similarities and differences in green economy-ecological resilience coupling patterns under different institutional contexts; second, integrating remote sensing data with field surveys to develop a more refined ecological resilience measurement system; third, introducing causal inference methods such as dynamic panel models, instrumental variable techniques, or quasi-natural experiment designs to further identify the causal effects of driving factors on CCD; fourth, exploring the combination of geographic detectors with machine learning methods (e.g., random forests, XGboost) to comprehensively characterize the complex impact mechanisms of multi-factor nonlinear interactions on CCD spatiotemporal evolution.
5 Conclusion
Based on panel data from 280 prefecture-level cities in China from 2010 to 2023, this study systematically investigates the coupling coordination relationship between green economy and ecological resilience, along with its spatiotemporal evolution characteristics. The results show that during the study period, both urban ecological resilience and green economy levels exhibited a sustained upward trend; however, their evolutionary processes were not synchronized. The improvement of ecological resilience was relatively stable and continuous, while the green economy displayed more pronounced phased characteristics, with an accelerated improvement trend in the later period. On this basis, the coupling coordination degree (CCD) of the two showed an overall upward trend, with a synchronized leap occurring in the later stage, indicating that urban development in China is gradually shifting from a state of relative separation between green growth and ecological governance toward a higher stage of coordinated evolution.
Further analysis reveals that the coupling coordination between green economy and ecological resilience is not progressing in a balanced manner but exhibits significant regional heterogeneity. The decomposition results of the Dagum Gini coefficient indicate that the spatial imbalance of urban CCD follows an evolutionary pattern of “convergence first, divergence later,” with inter-regional differences remaining the primary source of overall disparity. The kernel density estimation results further show that the main peak of the CCD distribution at the national level continues to shift to the right, reflecting a steady improvement in the overall level of coordination. Meanwhile, fluctuations in the peak, broadening of the distribution, and extension of the right tail suggest that disparities among cities have not disappeared in tandem with overall improvement; rather, some cities have taken the lead in making leaps, presenting a pattern of echelon-like evolution in the overall landscape.
The empirical results from the spatial dimension reveal significant externalities in urban synergistic transformation. Spatial autocorrelation analysis indicates that the CCD of cities exhibits a significant and stable positive spatial correlation, characterized by the coexistence of “high-high” and “low-low” agglomeration patterns. This suggests that the synergistic improvement of green economy and ecological resilience depends not only on internal urban development conditions but is also heavily influenced by geographical proximity and spatial spillover effects. Meanwhile, Markov chain analysis shows that the evolution of the CCD exhibits significant path dependence and state stickiness, along with a certain pattern of club convergence. The spatial Markov results further indicate that the development environment of neighboring areas significantly influences the evolutionary direction of local cities—low-level neighborhoods are more likely to lead to low-level lock-in, while high-level neighborhoods are more conducive to local cities achieving upward transitions through technology diffusion, industrial chain linkages, and governance synergy.
Analysis of geographic detectors reveals multiple driving mechanisms behind the spatiotemporal differentiation of CCD. Factor detection results indicate that digital infrastructure level serves as the primary driver of CCD spatial differentiation, followed by education expenditure levels and government intervention intensity, while financial development level demonstrates relatively weak single-factor explanatory power. Interaction analysis further reveals that pairwise interactions among all factors exhibit significantly higher explanatory power than single factors, demonstrating widespread synergistic enhancement effects. Notably, the interaction effects between digital infrastructure and education, finance, and government policies stand out as the most prominent, highlighting the core amplifying role of digital empowerment in multi-factor coupling. The temporal fluctuations in interaction effects align closely with the phased evolution of regional CCD disparities, further confirming that structural reorganization of driving forces constitutes the intrinsic mechanism underlying spatial pattern transformation.
Statements
Data availability statement
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.
Author contributions
CY: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. YY: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. WY: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This research was funded by Soft Science Project of Henan Province in 2026 (262400411158).
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1
ChenY.ShiJ.GaoW.GuoQ. (2023). Dynamic evolution and differentiation of China’s green economic development efficiency. Stat Decision39, 94–99. doi: 10.13546/j.cnki.tjyjc.2023.11.016
2
DingS.FanZ. (2024). Spatiotemporal evolution and obstacle factor analysis of coupling coordination between economic resilience and green, low-carbon development in China. Sustainability16, 11006. doi: 10.3390/su162411006
3
FinnertyP. B.CartheyA. J.BanksP. B.BrewsterR.GrueberC. E.HoustonD.et al. (2025). Urban rewilding to combat global biodiversity decline. BioScience75, 545–558. doi: 10.1093/biosci/biaf062
4
GaiM.YangQ. (2023). Synergistic study of the green efficiency and economic resilience of agriculture from a sustainable development perspective: evidence from Northeast China. Environ. Sci. pollut. Res.30, 77568–77592. doi: 10.1007/s11356-023-27697-5
5
HwangY. H. (2020). “ Rewilding urban landscapes: Attributes, types, and application as a greening policy,” in The Routledge handbook of urban ecology ( Routledge), 762–772.
6
JinX.QianS.YuanJ. (2024). Identifying urban rewilding opportunity spaces in a metropolis: Chongqing as an example. Ecol. Indic.160, 111778. doi: 10.1016/j.ecolind.2024.111778
7
LehmannS. (2021). Growing biodiverse urban futures: Renaturalization and rewilding as strategies to strengthen urban resilience. Sustainability13, 2932. doi: 10.3390/su13052932
8
LiY.HuangY. (2023). Enhancing resources efficiency: Studying economic development in resource-rich regions for long-term sustainability of China. Resour. Policy86, 104234. doi: 10.1016/j.resourpol.2023.104234
9
LiF.LiuX.ZhangX.ZhaoD.LiuH.ZhouC.et al. (2017). Urban ecological infrastructure: an integrated network for ecosystem services and sustainable urban systems. J. Cleaner Prod.163, S12–S18. doi: 10.1016/j.jclepro.2016.02.079
10
MaS.HuangJ. (2023). Analysis of the spatio-temporal coupling coordination mechanism supporting economic resilience and high-quality economic development in the urban agglomeration in the middle reaches of the Yangtze River. PloS One18, e0281643. doi: 10.1371/journal.pone.0281643
11
MehroushI.ShaheenW. A.ShabirM.TalhaM. (2025). Pathways to ecological resilience: exploring green energy and finance for sustainable development. Environ. Dev. Sustainability27, 17861–17885. doi: 10.1007/s10668-024-04662-x
12
MetawaN.DoganE.TaskinD. (2022). Analyzing the nexus of green economy, clean and financial technology. Economic Anal. Policy76, 385–396. doi: 10.1016/j.eap.2022.08.023
13
PriorJ.WardK. J. (2016). Rethinking rewilding: A response to Jørgensen. Geoforum69, 132–135. doi: 10.1016/j.geoforum.2015.12.003
14
PutriR. L.WerastutiD. N. S.WahyonoE.HastutyE. D. (2023). Green economy and low carbon development drive national economic growth and improve social welfare. J. Finance Econ Business1, 157–185. doi: 10.59827/jfeb.v1i2.67
15
QinQ.HeW.YuanL.DegefuD. M.RamseyT. S. (2025). Coupled and coordinated development of water-energy-food-ecology-land system in the Yangtze River Delta, China. NPJ Clean Water8, 38. doi: 10.1038/s41545-025-00472-4
16
RuanY.WangW.AbubakarM.AhmadN. (2025). Global economic resilience: Developing green growth strategies, renewable energy integration, and environmental economics for sustainability. Renewable Energy251, 123341. doi: 10.1016/j.renene.2025.123341
17
ShiC.ZhuX.WuH.LiZ. (2022). Assessment of urban ecological resilience and its influencing factors: A case study of the Beijing-Tianjin-Hebei urban agglomeration of China. Land11, 921. doi: 10.3390/land11060921
18
SöderholmP. (2020). The green economy transition: the challenges of technological change for sustainability. Sustain. Earth3, 6.. doi: 10.1186/s42055-020-00029-y
19
SouléM.NossR. (1998). Complementary goals for continental conservation. Wild Earth8, 39–64.
20
SwansonR. A.McKinneyM. L. (2026). Effects of green roof management, age, and substrate on plant and invertebrate communities. Front. Ecol. Evol.14, 1748660. doi: 10.3389/fevo.2026.1748660
21
ToneK. (2002). A slacks-based measure of super-efficiency in data envelopment analysis. Eur. J. Oper. Res.143 (1), 32–41. doi: 10.1016/S0377-2217(01)00324-1
22
TripplM.FastenrathS.IsaksenA. (2024). Rethinking regional economic resilience: Preconditions and processes shaping transformative resilience. Eur. Urban Reg. Stud.31, 101–115. doi: 10.1177/09697764231172326
23
UNEP (2019). Rewilding: A conservation approach for the 21st century (Nairobi: UNEP).
24
WangW.DengX.WangY.PengL.YuZ. (2022). Impacts of infrastructure construction on ecosystem services in new-type urbanization area of North China Plain. Resour. Conserv. Recycl.185, 106376. doi: 10.1016/j.resconrec.2022.106376
25
WangY.HanZ.WangB. (2025b). How to coordinate regional economic and ecological resilience: Evidence from the Yangtze River economic belt. J. Resilient Economies5, 24–45. doi: 10.2139/ssrn.5194273
26
WangH.LiuZ.ZhouY. (2023). Assessing urban resilience in China from the perspective of socioeconomic and ecological sustainability. Environ. Impact Assess. Rev.102, 107163. doi: 10.1016/j.eiar.2023.107163
27
WangQ.ZhangY.LiR. (2025a). Achieving sustainable development: Bridging economic growth and environmental resilience through EKC and LCC. Sustain. Dev.33, 7629–7656. doi: 10.1002/sd.3528
28
WangJ.ZhouX. (2024). Measurement and synergistic evolution analysis of economic resilience and green economic efficiency: Evidence from five major urban agglomerations, China. Appl. Geogr.168, 103302. doi: 10.1016/j.apgeog.2024.103302
29
WenB.HeY.JingX.HaroonM. (2025). Advancing renewable energy and green finance for economic growth and ecological resilience. Energy Strategy Rev.59, 101747. doi: 10.1016/j.esr.2025.101747
30
YangS. (2025). Spatiotemporal evolution and driving mechanisms of coupling coordination between green innovation efficiency and urban ecological resilience: Evidence from Yangtze River Delta, China. Sustainability17, 8528. doi: 10.3390/su17198528
31
YuT.JiaS.CuiX. (2024). From efficiency to resilience: unraveling the dynamic coupling of land use economic efficiency and urban ecological resilience in Yellow River Basin. Sci. Rep.14, 16518. doi: 10.1038/s41598-024-67364-4
32
YuQ.WangS.ZhuX. (2025). Greening through growth: Industrial upgrading, environmental governance, and urban green space in China. Int. Rev. Econ Finance, 104854. doi: 10.1016/j.iref.2025.104854
33
YuanL.LiR.WuX.HeW.KongY.RamseyT. S.et al. (2023). Decoupling of economic growth and resources-environmental pressure in the Yangtze River Economic Belt, China. Ecol. Indic.153, 110399. doi: 10.1016/j.ecolind.2023.110399
34
YuanR.RenM.LiZ. (2024). Spatial network structure and shift path prediction of ecological welfare performance in Chinese cities—Evidence from 284 cities. Sustainability16, 5282. doi: 10.3390/su16135282
35
ZhangL.MaX.OckY. S.QingL. (2022). Research on regional differences and influencing factors of Chinese industrial green technology innovation efficiency based on dagum gini coefficient decomposition. Land11, 122. doi: 10.3390/land11010122
Summary
Keywords
coupling coordination, ecological resilience, SDGs, urban green economy, urban rewilding
Citation
Yang C, Yongkang Y and Yuan W (2026) Coupling coordination between urban green economy and ecological resilience in the context of urban rewilding: evidence from Chinese cities. Front. Ecol. Evol. 14:1829217. doi: 10.3389/fevo.2026.1829217
Received
12 March 2026
Revised
08 April 2026
Accepted
09 April 2026
Published
30 April 2026
Volume
14 - 2026
Edited by
Melissa Vogt, Independent Researcher, Sydney, NSW, Australia
Updates
Copyright
© 2026 Yang, Yongkang and Yuan.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Chen Yang, cheny3598@gmail.com
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.