Spatiotemporal evolution characteristics and driving factors of ecological resilience in the Yangtze River Delta: A dual perspective based on impact analysis and configuration analysis Open Access

Since the reform and opening up of China, the country has achieved remarkable economic success, driven by deepening industrialization and urbanization. However, problems such as climate change and ecological degradation are becoming increasingly prominent. Ecosystems are threatened by water pollution and the risk of biodiversity decline and are characterized by spatial and temporal complexity and relative vulnerability to multifactorial influences (Tao et al., 2022). Unchecked and unsustainable production and consumption patterns have severely weakened ecological carrying capacity, threatening both regional and overall ecological security, and leading to a decline in ecological resilience (Zhao & Sun, 2023). The Yangtze River Delta (YRD), which spans Shanghai, Anhui, Jiangsu, and Zhejiang provinces and cities, is one of the metropolitan clusters in China with the largest economy and the greatest potential for development. The 2019 Outline of the Plan for the Integrated Development of the YRD explicitly states that ‘ecological protection should be prioritized, and the protection and restoration of the ecological environment should be placed at an important position, so as to consolidate the ecological basis for green development’, and that the ecological governance of the YRD region is of great significance for the country's sustainable development and for guiding the development trend in the future.

The 14th Five-Year Plan emphasizes the importance of building resilient cities to balance economic growth with ecological protection, thereby promoting high-quality, sustainable development. This provides a fresh perspective on ecological governance. With the development of the times, resilience extends to various research fields such as economics, sociology, and the environment (Sun et al., 2017), and the focus of this paper is on ecological resilience. The term resilience was first applied to the physical disciplines and then first applied to ecology by Holling (1973). In 1996, Holling in Engineering Resilience Versus Ecological Resilience further identified the special features of ‘ecological resilience’ that distinguish it from the traditional concept of ‘engineering resilience’. Subsequently, scholars have come to realize that no form of resilience can be separated from social factors. Holling first introduced ecosystem resilience to human social systems in 2001. Since then, ‘socio-ecological’ resilience has gradually become one of the main directions of research. Folke (2006) also summarizes the three perspectives of ‘engineering resilience’, ‘ecological resilience’, and ‘socio-ecological resilience’. The first two belong to the equilibrium perspective, while the latter belongs to the evolutionary perspective, which provides a dynamic and non-linear explanatory framework for ecological resilience. ‘Socio-ecological resilience’ is a process of adaptive cycles and dynamic interactions, with a greater emphasis on the capacity for transformative learning. Berkes & Folke (1998) explain this concept by arguing that social systems and ecosystems cannot be separated individually and that it is important to focus on the role of human activities in influencing the system. To synthesize the logic of resilience development, we will start from the ‘socio-ecological’ perspective to examine the development law of ecological resilience in the YRD region.

Scholars both domestically and internationally have extensively studied ecological resilience, with research focusing on three main aspects. First, the evaluation models of ecological resilience have been a key area of study. Scholars have developed various models to quantitatively measure ecological resilience. Two prominent models are: the ‘resistance-adaptability-recovery’ model (Li & Liu, 2022; Huang et al., 2023; Wang et al., 2023a). Furthermore, Duo et al. (2022), taking Nanchang City as an example, extended and updated the original model to construct ‘resistance-adaptability-vitality’ to examine its spatio-temporal evolution pattern. The second model is the ‘Driving Force-Pressure-State-Impact-Response’ framework, which Zhao et al. (2021) used to assess the ecological resilience of Chinese cities. Additionally, some scholars have incorporated systems like the energy-economy-environment-society (3E1S) framework to evaluate the coordinated development of ecological civilization (He & Zeng, 2024), further expanding the model to include water resource perspectives (Wang et al., 2023b). Second, the research scope and content of ecological resilience studies have covered various levels, including national (Li et al., 2023), regional (e.g., the Yellow River Basin) (Liu & Liu, 2024), and city-specific studies (Xue et al., 2023). These studies primarily analyze the spatiotemporal evolution of ecological resilience, investigate the coupling between resilience and other variables, and explore regional differentiation in influencing factors. Third, influencing factor content and method selection have been key components of ecological resilience research. Various methods have been employed to analyze these factors. For example, Qiu & Zhou (2020) used geographically weighted regression to analyze how factors like technological innovation and population density influence ecological efficiency. Li et al. (2024) introduced the time dimension, using geographically and temporally weighted regression to reveal significant spatial and temporal differences in the factors influencing Wenzhou's urban ecological resilience, such as forest cover and industrial structure (IS). Additionally, geographic detectors (Wang et al., 2024) and the STIRPAT model (Lyu et al., 2023) are widely used to analyze the spatial differentiation of influencing factors.

This study focuses on 41 cities in the YRD region, using data from 2010 to 2022. The primary sources of data include the China Urban Statistical Yearbook and China Urban and Rural Construction Statistical Yearbook, as well as the national economic and social development statistical yearbooks and bulletins of each city. Missing values were supplemented using linear interpolation.

In existing research, scholars have adopted different approaches to constructing a comprehensive ecological resilience evaluation index. Most create multidimensional evaluation systems that have practical relevance but vary in approach. Sun et al. (2017) view resilience as encompassing both elasticity and recovery, where higher resilience indicates stronger resistance and absorption capacity. Similarly, Li et al. (2022) emphasize that aquatic ecosystem resilience includes defense, adaptive, and transformative capacities during perturbation phases. Adger et al. (2005) further argue that social-ecological systems possess various mechanisms to adapt and learn from changes and unexpected shocks. From these perspectives, ecological resilience can be summarized as having three key characteristics: resistance, adaptability, and recovery. Li et al. (2019) suggest that regional resilience also involves adaptability, innovation, and sustainability, while Boschma (2015) highlights the importance of long-term capacity to create new growth paths in evolutionary resilience.

Based on this, the article starts from a ‘socio-ecological’ perspective and refers to related studies (Ding et al., 2020; Wang & Niu, 2023; Zhang et al., 2024), and incorporates the ‘sustainability’ into the YRD ecological resilience evaluation system. The ecological resilience evaluation index system of ‘resistance, adaptability, recovery and sustainability’ was constructed with 18 indicators in four dimensions (Table 1). Resistance resilience refers to the ability possessed by ecosystems to resist external disturbances, reflecting the vulnerability and sensitivity of the ecology when it is impacted (Lyu et al., 2023), which is specifically manifested in the impacts of human production and life on ecosystems. Adaptive resilience is the ability to focus on external disturbances and adjust to them. It is characterized by the responsiveness of cities and the absorptive capacity of nature. Recovery resilience refers to the resilient ability to improve the greening of urban space through government financial support to achieve the original ecological state. Sustainability reflects the ability of ecosystems to build on resistance, adaptation, and resilience, and to further achieve sustainability through learning and innovation.

Technological factors include two secondary conditions, ED and IS (Luo et al., 2022). The ED base can provide sufficient funds for ecological restoration and ensure ecological stability. Further improvement of the IS can further adjust the allocation of resources to guide production and manufacturing will be developed in a greener direction, providing strong support for ecological resilience. Organizational factors involve two secondary conditions, government support (GS) and human capital (HC) (Walker, 2014). As the largest developing country globally, the Chinese government has contributed to environmental protection by reducing pollution emissions and mitigating climate warming. The demonstration effect brought about by the ‘competition for the top’ among local governments has also led to the enhancement of ecological resilience in neighboring cities (Zhang et al., 2024). Human resources are the core drivers of eco-resilience building and provide intellectual support for eco-resilience building. The environmental factors focus on two conditions: the degree of openness to the outside world (EO) and domestic market demand (DM). Increased openness to the outside world contributes to the refinement of the scale of production, bringing with it advanced concepts and experience in ecological governance, thereby reducing the vulnerability of the economic system. Market demand means that consumer environmental awareness will be improved, which will guide enterprises to increase investment in green transformation, thus promoting the level of ecological resilience construction. Synthesizing the above analysis and drawing on the studies of related scholars (Wang et al., 2018; Tao et al., 2022), the ED foundation, IS, GS, HC, opening up to the outside world and domestic market demand are measured. Each driver variable is defined, as shown in Table 2.

The entropy value method assigns weights to the indicators, which further enhances the scientificity and rationality of the article while reducing the impact of errors.

The steps involved in the entropy value method primarily include:

• (1) Data standardization:

In order to ensure that the results are reliable, this paper adopts the method of polar deviation to standardize the index value, so that it is between [0–1]. Let there exist n years, b cities, c indicators, then the indicator value of city i in year a is denoted as ⁠, and the indicator values are processed as follows:

The larger the score value, the higher the level of ecological resilience; conversely, the lower the level of ecological resilience.

In formula (8)), T represents the overall difference in the level of ecological resilience of China's YRD region Tel index, the size of which is in [0,1], the smaller the Tel index indicates that the overall difference in the level of ecological resilience is smaller, and vice versa, indicates that the overall difference is larger. q denotes the city, k denotes the number of cities, denotes the ecological resilience level of city q, and denotes the average of the ecological resilience development level of the cities within the whole YRD region. Equation (9) where denotes the overall difference Terrell index for region p, denotes the number of cities in region p, denotes the level of ecological resilience development in city q in region p, and denotes the average value of ecological resilience development water in region p. In Equation (10), the overall variation in the level of ecological resilience is further decomposed into the within-region variation and the between-region variation ⁠. In addition, and are defined as the contribution rate of intra-regional and interregional differences to the overall differences, respectively, and is the contribution rate of each region to the overall differences within the region, denotes the sum of the ecological resilience level of each city within the region p, and S denotes the sum of the ecological resilience level of development of the YRD in China.

Geographic detectors were proposed by Wang & Xu (2017) for detecting spatially differentiated features and identifying system driving forces. They can be divided into factor detectors, interaction detectors, risk detectors, and ecological detectors. This article mainly uses factor detectors and interaction detectors to determine whether the conditional variables of the three subsystems under the TOE framework can become driving factors for ecological resilience in the YRD.

• (1) Factor detector: In this study, it is used to explain whether a certain conditional variable is the cause of ecological resilience formation and the extent to which it can explain it. The size of the q-value indicates to what extent the conditional variable can explain the outcome variable. The model is as follows:

  

  (14)

Among them, h = 1, 2, 3,…, L, L is the indicator classification, N_h is the number of evaluation indicators, N is the number of evaluation units; and σ² represent the variance of the indicator layer h and ecological resilience values. The value of q is [0,1], and the larger the value, the stronger the explanatory power of this conditional variable on the ecological resilience level in the YRD region.

• (2) Interaction detector: By calculating the p-value of the superposition of two different conditional variables, it determines whether there is an interaction effect, which indirectly reflects whether a certain conditional variable has an enhanced or weakened explanatory power for ecological resilience.

QCA is able to deal with complex causal relationships resulting from interdependence and co-action between conditioned variables, digging deeper into the grouping of more conditioned variables in action (Douglas et al., 2020). QCA combines the respective advantages of qualitative and quantitative analyses to better answer the problem of asymmetry in causality (Du & Jia, 2017). QCA methods include three forms, namely: clear set qualitative comparative analysis, multi-valued set qualitative comparative analysis, and fuzzy set qualitative comparative analysis (fsQCA). The conditional variables selected in the article to affect the ecological resilience of the Yangtze River Delta region are all continuous variables, and considering that fsQCA is more suitable for dealing with such variables, this article adopts fsQCA. NCA is a new tool that specializes in analyzing necessity conditions and complements QCA, which has advantages in full causal analysis (Du et al., 2020). The article first determines whether the six conditional variables under the TOE framework are necessary for ecological resilience, and then applies the NCA test of necessity to analyze the robustness of the results; Finally, fsQCA will be used to conduct a grouping analysis to explore what grouping paths can promote the level of ecological resilience in the YRD region.

The QCA and NCA methods were used to explore the causal effects of conditioning variables on ecological resilience in the TOE framework for several reasons: First, ecological resilience is a complex system that is influenced by a combination of factors, and traditional regression analyses assume no serious multicollinearity among the study variables. QCA can accommodate data that cannot be accommodated by traditional methods and is able to deal with complex causal relationships resulting from interdependence and co-action between conditioned variables, and to dig deeper into the grouping of more conditioned variables in action (Douglas et al., 2020). Second, the diversity of pathways to higher levels of ecological resilience across cities suggests that there may be multiple ‘equivalent’ causal chains leading to the same outcome (Tao et al., 2021). In this study, although the traditional statistical analysis methods can portray the different ways in which conditioned variables affect ecological resilience through mediating and moderating variables, all conditioned variables are only substitutive or cumulative in explaining the dependent variable, rather than fully equivalent (Wang et al., 2014). This approach, QCA, is able to avoid the above limitations by deeply dissecting the groupings of different influences on ecological resilience, and each group is equivalent. Finally, it is more specialized to use NCA to analyze the necessity. Before conducting the group analysis, the first step is to conduct a one-factor variable necessity analysis. Although QCA is capable of conducting necessity analysis, it can only make a rough judgment from a qualitative point of view. NCA is able to dissect the necessary conditions in terms of degree based on identifying whether the conditional variables are necessary or not, complementing rather than replacing the traditional methods, thus providing a more comprehensive analysis (Dul, 2016).

According to the dimensions and evolutionary characteristics (Figure 4(b)), the resistance, adaptability, recovery, and sustainability resilience of ecological resilience in the YRD region show a fluctuating upward trend. Sustainability resilience plays a dominant role, well beyond the other three dimensions. The YRD has placed greater emphasis on sustainable development in the process of ecological environmental protection. Ecological environmental protection is not a short-term endeavor and should not be rushed into for the sake of ecological ‘achievements’. The YRD region focuses on an innovation-led sustainable development model that promotes ecological resilience over time; Meanwhile, with the continuous improvement of environmental protection policies and the vigorous implementation of ecological restoration measures in the YRDregion, the resistance, adaptability, and recovery of ecological resilience have been enhanced accordingly. The ecological dimension of the development pattern of ‘one force pulling, three forces in tandem’ is presented.

In 2012, excellent ecological resilience was only in Shanghai, good ecological resilience was concentrated in Nanjing, Suzhou, Hangzhou, and Hefei, and medium ecological resilience or above accounted for only 12.2% of the study sample. The medium resilience cities are mainly concentrated in southern Jiangsu and eastern Zhejiang, including 12 cities such as Wuxi, Changzhou, and Nantong. Average ecological resilience and poor ecological low resilience value cities are mainly distributed in contiguous clusters in western Zhejiang and northern Jiangsu. The same is true for all of Anhui except Hefei and Wuhu. Together, they account for 58.54% of the study sample. In 2015, excellent ecological resilience, good ecological resilience, and medium ecological resilience areas remained unchanged, and poor resilient cities decreased from seven in 2012 to five in 2015. This indicates that the development of ecological resilience in the YRD from 2012 to 2015 has shown a ‘steady progress’ trend. The level of ecological resilience in the YRD was further improved in 2018, with excellent resilient cities increasing to five cities from only Shanghai in 2012 and 2015, and good ecological resilient cities increasing to nine cities, with excellent resilient and good resilient cities accounting for 34.15% of the sample. The development of ecological resilience in the YRD region relies on the three provincial capitals of Nanjing, Hangzhou, and Hefei, as well as the economic center of Shanghai, and continues to radiate and spread to the surrounding areas, forming a ‘central high-value corridor’ in the YRD region. The 2022 YRD ecological resilience values are lower than in 2018, and the distribution pattern is similar to that of 2012, but the overall resilience values are higher than in 2012. Notably, the number of cities in northern Anhui's poor ecological resilience values decreased to two.

The spatial autocorrelation model was used to analyze the characteristics of the spatial correlation pattern of ecological resilience in the Yangtze River Delta region, as shown in Table 4. The global Moran's I indexes of the YRD ecological resilience index from 2012 to 2022 are all positive and all pass the significance test at the 1% level. It indicates that there are significant positive spatial correlations and spatial clustering effects of ecological resilience in the YRD. Specifically, from 2012–2017, the global Moran's I index showed a downward trend, from 0.256 in 2012 to 0.189 in 2017. There was an upward trend in 2017–2019, indicating that the spatial agglomeration effect was increasing at a steady rate in this period. After 2019, there is a downward trend again. Overall, it shows a reverse ‘N’ direction of ‘decreasing-increasing-decreasing’.

The difference between the ecological resilience of the H–H cities and that of their neighboring cities is small and all of them are at a high level of resilience. Spatially, they mainly include: Shanghai, Nanjing, Wuxi, and Hangzhou. These cities have superior geographic locations and high economic levels, which can lay a good foundation for resisting urban ecological risks; with the continuous deepening of the integration strategy of the YRD, the continuous improvement of the IS, the continuous improvement of the infrastructure makes the ecological resilience level of the cities increase rapidly. The spatial spillover effect of the cities has been continuously revealed, forming a development pattern of ‘connectivity and sharing’ through the development of radiation.

L–L-type cities have little difference between their own ecological resilience and that of their neighboring cities and are all at a lower level of resilience. During the study period, this type of city occupied more than 50% of the samples and was mainly distributed in Jinhua, Zhoushan, and Fuyang. The distribution is ‘concentrated’, which is corroborated by the previous time-series analysis. These cities are economically backward, lack scientific research and high technology among cities, and have less experience in green development, which makes it difficult to provide technical and experiential support for the formulation of effective protection and restoration measures. At the same time, most of these cities are in the third or fourth tier, and may neglect green technology innovation in pursuit of rapid ED. Traditional industries account for a high proportion of the total, and the resulting emissions, wastewater, and pollutants will seriously damage the ecological environment, making it difficult to sustain the ecological resistance.

During the study period, the ecological resilience of the H–L cities has a large gap, i.e., their own ecological resilience is high, but the ecological resilience of the surrounding cities is at a lower level, and spatially there is only one city, Hefei. In recent years, Hefei has experienced rapid economic development (ED), while also committing to ecological and environmental protection through efforts such as wetland restoration and the integrated protection of mountains, water, forests, fields, lakes, grasslands, and sands. These initiatives have led to remarkable achievements. Due to the strong siphon effect, the regional coordinated development mechanism is not yet sound, which makes the ecological resilience of neighboring cities relatively lagging behind, and still poses certain challenges in promoting the development of neighboring cities.

L–H-type cities have relatively low ecological resilience, while the surrounding cities exhibit higher resilience. These cities are mainly distributed in Zhenjiang, Taizhou, Zhoushan, and other areas. Due to their relatively closed environmental development, they face greater difficulty in securing the financial and human resource support required for ecological advancement. Zhoushan City, for example, is located in the East China Sea and consists of many islands, which are vulnerable to natural disasters such as typhoons and have fragile ecosystem resistance, limiting the enhancement of ecological resilience. The next step should be to take advantage of the YRD integration strategy as a platform to actively learn from the experience of regions with high values of ecological resilience and accelerate the development of ecological resilience in regional cities.

In the process of applying geoprobes, there is a need to filter the optimal scale of spatial data discretization. Different discretization methods and different numbers of intervals will affect the size and significance level of the q-value. Drawing on related studies, the data were discretized using methods such as K-mean clustering and quantile classification. In this paper, the way and interval with the largest q-value are selected as the optimal parameters for the geodetectors of this study, based on satisfying the level of significance (all p-values are less than 0.1). Take the 2012 ED as an example: when using the K-mean clustering method, the q-value is maximized when the clustering interval is 2. Therefore 2012 ED should be clustered in geodetectors with a K-mean clustering of two classes as the optimal parameter choice.

Based on the results of ecological resilience detection in the YRD region from 2012 to 2022, the variables are mainly categorized into four grades according to the different degrees of influence(Fu et al., 2024), which are: primary driver (q greater than 0.7), key driver (0.5–0.7), important driver (0.3–0.5), and basic driver (q less than 0.3). As can be seen from Table 5, there are changes in the q-value and rank of the driving factors in different years.

Primary drivers include IS, GS, HC, EO, and DM. Among them, the q-value of DM, with an average value of 0.942 for the period 2012–2022, is in the first place. This suggests that the rising consumer demand brought about by expanding market capacity is driving the continuous development of the environmental protection industry and enhancing the level of ecological resilience. The q-value of HC had the second-highest mean value of 0.939 over the study period. This reflects the fact that the emphasis on higher education and the development of innovative human resources is an important cornerstone for the improvement of the level of ecological resilience. The mean value of q-value for GS was 0.921, which ranked third. A series of environmental protection measures have been implemented and environmental protection expenditures have been increased in various parts of the YRD, and GS has played a positive role in the enhancement of ecological resilience. The mean q-value of EO is 0.874, which is the fourth highest. It indicates that opening up to the outside world plays a significant role in the ecological resilience of the YRD region, with strong explanatory power. The mean q-value of IS is 0.730, which is the fifth highest. Optimization of IS will reduce the total amount of industrial ‘three wastes’ emissions, thus increasing ecological resistance, and this variable has a significant impact on ecological resilience. The important driver is ED. Although this factor is ranked sixth, the q-value changes from 0.239 in 2012 to 0.484 in 2022 in a fluctuating upward trend. It can be seen that the factor of ED has an important impact on the ecological resilience of the YRD region. The higher the level of ED, the better the foundation for risk resistance, and the higher the level of ecological resilience.

In summary, from the perspective of impact analysis, all six variables selected for the study are able to provide good explanatory power for the ecological resilience of the YRD, which is further supported by the results of the interaction probes. Meanwhile, it was found that the interaction between variables had a greater explanatory effect than that of a single variable, providing a basic idea for the subsequent group analysis. This leads to the consideration of whether there are multiple combinations of factors that work together to promote ecological resilience in the YRD region. Next, fsQCA is applied to explore the group path of ecological resilience enhancement in the YRD and to deeply analyze the complex causal relationship between each conditional variable and ecological resilience under the TOE framework.

A direct calibration method was used to transform the data into fuzzy set affiliation scores. Due to the lack of external standards and empirical evidence (Zhang & Du, 2019), drawing on existing research (Yang et al., 2022), the quartile method was used to determine the thresholds for each condition variable and the outcome variable, i.e., three anchor points were set. They are the threshold for full affiliation (0.75), the threshold for full non-affiliation (0.25), and the crossover point (0.5). The calibration anchor points and affiliation thresholds for each variable are shown in Table 6.

In the necessity test performed by QCA, the degree of consistency is used to identify the necessity conditions. It is generally accepted that when the consistency of a conditional variable is greater than 0.9, it indicates that the condition is necessary to constitute the outcome. The results are shown in Table 7. The conditional variable consistency of individual influences on ecological resilience in the YRD region is generally low, all less than 0.9. This suggests that the necessary conditions for ecological resilience do not exist among the six factors, such as ED, i.e., the development of ecological resilience is not the result of a single antecedent condition.

NCA is a specialized method for determining necessity conditions, which identifies these conditions by analyzing the effect size and significance of the conditional variables. It further assesses the necessity level values of the judging condition variables at the bottleneck level (Du et al., 2022). NCA serves as a valuable complement to necessity condition analysis in qualitative comparative analysis (QCA). The implementation of NCA involves using the NCA package in R to analyze the necessity of single factors through two estimation methods: ceiling envelopment (CE) and ceiling regression (CR). CE is applicable to discrete variables, while CR is suited for continuous variables (Dul et al., 2020). The utility value ranges from 0 to 1, with larger values indicating greater utility. A variable constitutes a necessary condition for the outcome variable only if the effect size exceeds 0.1 and passes the Monte Carlo simulation replacement test (p < 0.05) (Dul, 2016; Du et al., 2020). Since all the conditional variables selected for this study are continuous, CR was employed as the basis for analysis. The results are shown in Table 8. Although the p-values for ED, GS, HC, EO, and DM are all below 0.05, their effect sizes are less than 0.1, indicating that they do not constitute necessary conditions for the outcome variable. Additionally, neither the effect size nor the p-value for the IS meet the necessary criteria, thus it also does not constitute a necessary condition.

The bottleneck level analysis is able to explain the necessary level of conditions required to reach a given level result (Li & Zhang, 2023). The results are shown in Table 9. To reach the 100% level value of ecological resilience in the YRD region, a combination of ED level of 9.9%, GS of 5.5%, HC of 1.0%, openness to the outside world of 5.0%, and domestic market demand of 6.7% of the antecedent conditions are required to achieve the desired situation, which is consistent with the results of high ecological resilience level in fsQCA without a single condition.

Building on the necessity analysis, the fsQCA software is utilized to analyze the configurations of drivers influencing ecological resilience in the YRD region. Different configurations of drivers indicate varying combinations of factors that can lead to a high level of ecological resilience. Each configuration is named according to the core conditions involved. Before conducting the configuration analysis, it is essential to construct a truth table to identify any contradictory configurations. The raw consistency threshold is set to 0.80, while the probability of reversed inclusions (PRI) consistency threshold is established at 0.70. Any outcome variable with a consistency less than 0.7 is set to 0, and the frequency of occurrence threshold is set to 1. This process leads to the derivation of complex, intermediate, and parsimonious solutions, with the intermediate solution serving as the primary focus and the parsimonious solution as a supplementary option. The conditional variables that appear in both the intermediate and parsimonious solutions are classified as core variables, while those that appear only in the intermediate solution are regarded as auxiliary conditions.

After constructing the truth table, it was found that there are no contradictory configurations. The subsequent configuration analysis yielded results presented in Table 10. The overall consistency level of the conditional configuration stands at 95.08%, which exceeds the threshold of 90%. Additionally, the consistency level for each individual configuration is also above this threshold. The overall coverage is an impressive 78.01%, indicating a 95.08% probability of achieving a high level of ecological resilience in areas characterized by these conditional configurations. This means that these configurations can explain 78.01% of the cases exhibiting high ecological resilience, demonstrating that the five identified configurations effectively account for the occurrence of high ecological resilience in the region.

Economic dominance – market prosperity dual wheel drive type: Configuration 1 indicates that high economic development, high DM, and non-high IS serve as core conditions, while GS functions as an edge condition. Together, these contribute to high ecological resilience in the YRD. The raw coverage of this grouping is 0.413741, indicating that 41.37% of the cases in the set of cases in which this configuration achieves a high level of ecological resilience can be explained by Configuration 1; the unique coverage is 0.0428727, indicating that 4.29% of the cases can only achieve a high level of ecologically resilient development through this grouping pathway. Organizational leadership-domestic environmental linkage type: Configuration 2a uses high GS, HC, and DM as core conditions and non-high IS as edge conditions to jointly advance ecological resilience in the YRD region. The configuration has a raw coverage of 0.436797 and a unique coverage of 0.0384734. This suggests that 43.68% of the cases with high ecological resilience levels in the YRD can be explained by this grouping, of which 3.85% can be explained only by this grouping. Configuration 2b has high GS, HC, and DM as core conditions and high EO as edge conditions. The raw coverage is 0.625617 and the unique coverage is 0.0614364, indicating that 62.56% of the cases with high ecological resilience can be explained by this configuration, of which 6.14% can be explained only by this configuration. This configuration has the highest universal adaptation and is superior to the other four configurations. Technology organizational integration development type: Configuration 3 shows that high technological and organizational conditions as core conditions can lead to high ecological resilience development. The original coverage of this configuration is 0.325921 and the unique coverage 0.0113737. In the Achieving High Levels of Ecological Resilience case set, 32.59% could be explained by Configuration 3, and 1.14% could be explained by that grouping only. Compared to the other configurations, this configuration has the lowest original coverage and is less pervasive. All-factor collaborative development type: Configuration 4 with high ED, HC, and DM as core conditions and IS and EO as edge conditions can contribute to the occurrence of high ecological resilience levels in the YRD. It suggests that the integrated development of technological, organizational, and environmental elements plays an integrated driving role for the level of ecological resilience in the YRD region. The raw coverage in this configuration is 0.350222 and the unique coverage is 0.0361634. This indicates that 35.02% of the high ecological resilience cases can be explained by this configuration, and 3.62% only by this configuration.

Although the same consistency is used to measure the adequacy of the grouping, the minimum acceptable standard and the method of calculation are different from the necessity analysis. Therefore, drawing on the research ideas of related scholars (Zhang & Du, 2019), this paper uses the method of adjusting the level of consistency of adequacy (the level of consistency is increased from 0.80 to 0.81) to conduct the robustness analysis of conditional grouping. The findings of this paper remain robust through empirical findings.

In terms of spatial differences, the overall differences in ecological resilience in the YRD region decreased in the pre-study period. Since 2012, the YRD region has continued to implement the scientific concept of development, and the region has paid more attention to synergistic development and win–win cooperation in environmental governance, forming a synergy of ecological resilience building and development. However, there is a slight ‘tailing off’ of ecological resilience in the YRD after 2019, consistent with the kernel density estimates. The reasons for the discrepancies need to be explored in the later stages of development to provide a continuous boost to the ongoing YRD integration strategy. At the same time, it is not difficult to find that the ecological resilience value of provincial administrative centers is basically higher than that of neighboring cities. Taking the Yangtze River Economic Belt as the object of their study, Wang & Liang (2025) found that the ecological resilience of cities shows a distinctive ‘core-periphery’ spatial distribution pattern. This study further extends this finding.

Within the region, intra-regional differences between Jiangsu and Anhui are much greater than in Zhejiang. There are significant differences in the specific topographic location and distribution of water resources among cities in Jiangsu Province. Coastal cities rely on abundant natural resources and have a high capacity for ecological restoration. For example, the ecological resilience levels of Yancheng and Nantong are higher than those of Taizhou and Suqian. In addition to this, cities such as Nanjing and Wuxi have been able to respond effectively to environmental challenges by leading ecological governance through scientific and technological innovation. The level of ecological resilience is higher than that of other cities in the province. Anhui Province shows a differential distribution of ecological resilience, with ‘Hefei standing out, and southern Anhui higher than northern Anhui’. As the capital of the province, Hefei has policy advantages in building ecological resilience and can effectively curb polluting behavior. The southern Anhui region has a humid climate, high natural green coverage, and more stable ecosystem resistance. The northern part of the Anhui Province is mainly dominated by plains, while its ED is relatively backward, making it difficult to respond effectively to environmental fluctuations and external disturbances.

In terms of spatial dynamics, the level of ecological resilience in the YRD region was generally low at the beginning of the study. Since 2012, the construction of socialism with Chinese characteristics has entered a new era. The YRD region is gradually focusing on the importance of ecological civilization, gradually improving and upgrading its ability to respond to pollution, and steadily increasing its level of ecological resilience. In 2015, the introduction of the five development concepts provided new ideas for ecological construction. The deep integration of the concepts of green and innovative development and the incorporation of innovative development into the construction of ecological civilization have injected a new era of connotation into the sustainable development of the YRD region. As a result, the ecological resilience of the YRD region has improved significantly in the years since. It is worth noting that the number of areas with low ecological resilience values in the northern Anhui region in 2019–2020 has been reduced to two. This is due to the fact that the municipalities in northern Anhui Province, according to their own development characteristics, have pushed forward the prevention and control of air, water, and soil pollution, and improved their resistance to ecological and environmental damage. Meanwhile, 2019 saw a solid start to the integrated development strategy of the YRD. Taking the integration into the YRD as an opportunity, the northern Anhui region has vigorously developed a low-carbon economy, built a green industrial system, and further improved the ecological environment.

Typical examples include: Shanghai, Hangzhou, Nanjing, and Hefei. All of these cities are first-tier or new first-tier cities, as well as the capitals of the provinces. This category of cities has sufficient market capacity and purchasing demand, and consumers' concept of green consumption is growing in the context of high-quality development. Coupled with the fact that these regions can easily attract various factors such as resources and talents, they have a solid economic foundation. It is easy to form a development model in which ED and market demand support and promote each other. On the premise of solid ED and huge market demand as the endogenous driving force, it effectively improves the ecological resilience of the YRD region, which to a certain extent compensates for the negative impacts of the underdevelopment of the IS. Lyu et al. (2023) similarly confirmed the importance of economic factors on urban ecological resilience through the STIRPAT model, which further supports the conclusions of this paper.

Most of the typical cases are in regions such as Jiangsu, specifically including cities such as Nanjing and Suzhou. The Jiangsu authorities are actively promoting laws and regulations related to ecological protection and increasing financial support for ecological protection while attracting high-caliber talents from all sides to devote themselves to ecological resilience building. Higher organizational conditions can provide financial and human support for the development of regional ecological resilience. In conjunction with the stable outlook and dynamic development of the domestic market, we will form a pattern of organization-led-domestic-environmental linkage development, and continue to promote the level of ecologically resilient development. It is worth noting that a comparison between histogram 2a and 2b reveals that they share the same core conditions, while the marginal conditions differ. In this grouping, with well-organized conditions as well as domestic market demand, ecological resilience improves even more significantly if a little more openness to the outside world is used as an aid. This is consistent with the study of Tao et al. (2022): an appropriate degree of openness to the outside world is positively correlated with ecological resilience. It also adds to the research that organizational conditions have an integral role in building ecological resilience as a developmental pedestal.

Typical examples are: Yangzhou, Changzhou, and other cities. With the continuous deepening of the supply-side structural reform, the ED of this type of city is active and the industry is more developed. Taking Changzhou as an example, the city has a strong industrial base, and the manufacturing industry, especially the new energy industry, has an important position in the country. The upgrading of emerging industries further attracts more skilled people to the city, providing a talent pool for eco-resilience building. Not only that, on the one hand, local governments have taken the optimization of economic structure and the promotion of industrial revitalization as an opportunity to promote the industrialization and application of green technologies. On the one hand, efforts should focus on accelerating the transformation of green achievements and improving resource efficiency. On the other hand, the government – considering economic system development, resource allocation, and environmental inputs – plays an active role in building ecological resilience by forming a region-specific development path that integrates technological and organizational innovation.

Typical cases include: Wuxi, Shaoxing, Nantong, and other cities. At the level of core conditions, ED is the material foundation, HC is the core driving force, and DM is the directional support, all three of which are interdependent and together constitute a solid foundation for building ecological resilience. First, the rapid ED of the YRD region has provided the necessary material foundation for ecological resilience building. The gradual transformation of the IS from high pollution to low pollution further promotes investment in green technology innovation, which helps to improve the stability of the ecosystem through the application of advanced environmental protection technologies. Technological innovation driven by ED provides a developmental basis for ecological resilience building. Second, highly qualified talents with environmental protection skills will be attracted and gathered in large numbers, and they will combine the actual situation of each place with the local conditions, and make continuous efforts for the construction of ecological resilience resistance and resilience. Third, with the changes in the domestic market demand, coupled with the further increase in the level of consumption brought about by ED, the public demand for greening products is gradually increasing. This shift in consumption patterns will provide an effective and invisible regulation of the market. The YRD region is cooperating with the continuous influx of innovative talents, constantly upgrading its products, guiding industries in a greener direction, and promoting the in-depth development of ecological resilience.

The following recommendations are made to promote ecological resilience in the YRD region:

First, it is essential to coordinate ecological integration system governance and develop a differentiated management model for ecological resilience areas. The level of ecological resilience varies greatly among cities in the YRD region, so it is important to strengthen the exchange of experience in ecological governance across the region and continue to promote integrated environmental governance. For regions with high values of ecological resilience, such as Shanghai and Nanjing, higher levels of green innovative technologies should be pursued. This category of cities plays a leading role in ecological construction in the process of ecological governance, which helps to improve the level of ecological governance of the whole territory. Areas with low and medium values of ecological resilience have a weak risk-tolerance capacity to cope with natural or man-made disasters. The improvement of the IS should be accelerated and more funds should be invested in infrastructure development. Additionally, talent mobility should be enhanced to comprehensively improve the quality of the ecological environment.

Second, leveraging the regional spillover effect is vital for establishing a linked green development pattern within the YRD. On the one hand, regions with strong ecological resilience should establish ecological protection cooperation mechanisms with neighboring regions to fully leverage interregional spillover effects. On the other hand, efforts should be made to promote the YRD integration strategy and reduce the spatial disparities in ecological development. On the other hand, the ecological lead role of provincial capitals should also be utilized within each region to radiate the ecologically lagging cities in the region. Among them, Hefei City should continue to promote the development of green and low-carbon industries by leveraging its existing industrial advantages. It should also establish a synergistic eco-industrial chain with neighboring cities, address regional development imbalances caused by the ‘siphon effect,’ and promote the construction of ecological resilience.

Third, resource allocation should be tailored to local conditions to expedite the establishment of a modernization system for ecological resilience. As can be seen in the aforementioned driver analysis, the drivers ED, IS, GS, HC, EO, and DM have different effects on ecological resilience. Each city should build an ecological resilience enhancement program that suits its own characteristics. For example, in more economically developed cities, on the one hand, the development of domestic markets should be actively promoted, and mutual integration should be jointly promoted to build ecological resilience (Configuration 1). On the other hand, it should be based on ED, combined with organizational, environmental and other factors, and all factors should be developed simultaneously (Configuration 4). For cities with better organizational conditions, the promotion of stable development of local markets should be accelerated to provide a broad space for green development. At the same time, we continue to improve industrial institutions, lay a solid economic foundation for development, and accelerate the construction of a modernized system of ecological resilience (Configuration 2, 3).

In recent years, global warming, urban flooding, and other disasters have occurred with increasing frequency. As the engine of China's ED, enhancing the ecological resilience of the YRD region is essential for promoting high-quality economic and social development. This study delves into the spatial and temporal evolution of ecological resilience in the YRD region and examines its driving factors. Based on the findings, targeted recommendations are provided to address the current development situation. However, there are areas that require improvement. First, due to limited access to relevant data, there are gaps in the construction of indicators that fail to incorporate ecological indicators such as biological and landscape diversity into an integrated assessment framework. Meanwhile, relevant scholars have conducted in-depth research on factors such as extreme rainfall and river flow by constructing different effective models (Agarwal et al., 2022, 2023a, 2023b, 2025), which provides reference for further refining the ecological resilience indicator system. Second, while the study analyzed the impact on ecological resilience using the TOE framework – considering factors like ED, IS, GS, HC, degree of openness to the outside world, and domestic market demand – it did not address the public's subjective perspective. Factors such as environmental protection awareness and the degree of ecological consciousness within the region were not considered. Additionally, the study identified a lack of both quantitative and qualitative data to effectively assess ecological resilience. Future research could benefit from employing questionnaires and other relevant quantitative methods for a more in-depth analysis.

First, the overall ecological resilience level of the YRD region is low; however, the development trend is positive, exhibiting a fluctuating upward trajectory. The average ecological resilience values for each region, ranked from high to low, are as follows: Shanghai, Jiangsu, Zhejiang, and Anhui. Notably, the power of sustainability in this region surpasses that of resistance, adaptability, and recovery, indicating an ecological development pattern characterized by ‘one force leading and three forces advancing together.’

Second, in terms of spatial differences, the Terrell Index for the YRD region decreased from 0.149 in 2012 to 0.139 in 2019, experiencing a slight ‘tailing off’ thereafter. The overall differences in ecological resilience first decreased and then increased, primarily due to inter-regional disparities. In terms of spatial dynamics, the ecological resilience level in the YRD demonstrates a ‘seeking progress while stabilizing’ pattern, with a spatial trend favoring the development of a ‘central high-value corridor.’ The low-value areas in the northern part of Anhui Province are gradually diminishing, with noticeable improvements in development trends.

Third, the spatial agglomeration effect is evident, showcasing a reverse ‘N’ direction characterized by ‘decreasing-increasing-decreasing.’ Localized L–L type cities are predominant, resulting in an overall spatial distribution pattern of ‘low-value agglomeration, with high-value areas tending to be continuous.’

Fourth, impact analysis indicates that ED serves as a crucial driving factor for ecological resilience. The primary factors contributing to this resilience include IS, GS, HC, the degree of openness to the outside world, and domestic market demand. Each of these variables effectively explains the ecological resilience of the YRD region, demonstrating a two-factor augmentation relationship. Configuration analysis further reveals that a single driving factor is insufficient to enhance ecological resilience on its own. Instead, four distinct driving pathways have been identified: the Economic Dominance – Market Prosperity Dual Wheel Drive Type, the Organizational Leadership – Domestic Environmental Linkage Type, the Technology Organizational Integration Development Type, and the All Factor Collaborative Development Type.

An in-depth study of ecological resilience in the YRD region is conducted with a view to providing theoretical support and empirical evidence for ecological governance and green transformation in the YRD region and the country as a whole.

This research is funded by the National Natural Science Foundation of China (42301357); Humanities and Social Sciences Youth Foundation, Ministry of Education (23YJC790120); Anhui Office of Philosophy and Social Science (AHSKY2023D109); Fuyang Social Science Fund (FSK2024011; FSK2024012); Fuyang Normal University Teaching and Research Project (2024JYXM0004); Anhui quality engineering project (2022zyxwjxalk177; 2023xscx120; 2024xscx129); Humanities and Social Science Project of Anhui Provincial Education Department (2022AH051274; 2024AH052995); and Production and education cooperative Project (220904978265230).

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.