Some indicators have dimensional differences, and their effects on the corresponding system can be positive or negative. To facilitate comprehensive calculation and comparative analysis, the method of extreme value standardization was adopted to carry out dimensionless processing on the original data (Guo, 2007):

where λ, i, and j are year, region, and indicator respectively, X_max and X_min are the maximum and minimum of the jth indicator in all regions and years. Z_λij and X_λij are the standardized and original values of the jth indicator in the ith area of the λth year. After processing, all indicator values will be in the range of [0, 1]. To reduce the interference of subjective factors, this paper uses the entropy method to calculate the weighting coefficients of each indicator when constructing the comprehensive evaluation system of urbanization and ecological resilience (Bai et al., 2018).

(1) Size resilience

According to the theory of ecological infrastructure and the landscape security pattern (Yu, 2005), the mountains, forests, and water systems surrounding cities are particular types of infrastructure that can restrain urban expansion. When the scale of the urban built-up area exceeds the service scope of ecological infrastructure, both the city’s disaster response and balance recovery ability will be weakened accordingly. Therefore, the size resilience can be defined by the proportional relationship between the urban built-up area and the ecological infrastructure (Xiu et al., 2008).

Ecological infrastructure (EI) includes all naturally protected area systems, forestry and agricultural systems, urban green space systems, water systems, and ecological recreation systems that can provide natural services (Yu et al., 2008). Consistent with recent research (Wang et al., 2017; Ma et al., 2018), this paper identifies a protection red line for important ecologically functioning areas based on the spatial distribution data of ecosystem service value. Then, the protection red line and the nature reserve areas were spatially superimposed to obtain the EI boundary that meets the minimum ecological safety standards.

where R_s is the urban size resilience index, L_s is the suitable construction land area under the constraint of EI, and L_d is the construction land area.

(2) Density resilience

The “density” of human activities in a city directly affects the sustainability of the ecosystem. Human activities can increase pressure on the natural environment, which will lead to ecological disorder. The ecological footprint theory proposes a method to directly measure the “density” of human activities (Wackernagel and Rees, 1996; Xiong et al., 2003). The ecological population footprint refers to the total ecologically productive land required to provide resources and absorb waste (Xu et al., 2000). According to existing studies, the land for providing resources is generally cultivated land, grassland, forest land, fossil fuel land, construction land, and water areas. The actual coverage area of productive land in the city is defined as the ecological carrying capacity. By comparing a city’s ecological footprint and carrying capacity, one can quantitatively assess whether its economic and social development is sustainable. If the ecological footprint overdraws the ecological carrying capacity, the city is in a state of ecological deficit; otherwise, it is in a state of ecological surplus. This article uses density resilience to calculate the gap between supply and demand. The formula of ecological footprint is as follows (Xu et al., 2003):

where E_f is the total ecological footprint of the region, N is the total population of the region, e_f is the per capita ecological footprint of the region, i is the category of consumer goods, C_i is the per capita annual consumption of the ith consumer goods, P_i is the annual average productivity of the global standard land corresponding to the ith consumer goods, and r_i is a balance factor.

Consistent with the Guangdong Statistical Yearbook (2001-2016) and the China Energy Statistical Yearbook (2001-2016), we calculated the per capita annual consumption of grain, oil, meat, eggs, milk, aquatic products, vegetables, fruits, and fossil energy in the Pearl River Delta, and converted them into the corresponding productive land area. Since the productivity of cultivated land, grassland, and forest land varies greatly per unit area, it is necessary to multiply the balance factor r_i by the land area (Wackernagel and Rees, 1996). Balance factor r_i is derived from World Ecological Footprint Report 1996 (Table 1).

The formula of ecological carrying capacity is as follows:

where E_c is the ecological carrying capacity of the region, j is land-use type, S_j is the area of the existing jth productive land in the region, and m_j is yield factor. Considering that the productivity of the same type of productive land varies significantly for different latitudes and longitudes (Liu et al., 2010), this paper introduced the yield factor m_j to convert the land area of the Pearl River Delta into the global standard land area (Table 1). According to the initiative put forth by the World Commission on Environment and Development, 12% of the world’s productive land should be set aside to protect biodiversity (Xu et al., 2000). Therefore, this proportional area was deducted in calculating the ecological carrying capacity of the cities in the Pearl River Delta.

The formula of density resilience is as follows:

where R_d is the urban density resilience index.

(3) Morphology resilience

“Morphology” refers to the spatial organization of the built environment and ecological land within a city. According to the theory of “source-sink” in landscape ecology (Chen et al., 2008), the urban landscape can be divided into a “source” and “sink.” A source refers to gray landscapes such as buildings, which harm the ecological environment, and a sink is a blue (water area) or green (greenbelt) landscape that can stop or slow this negative impact. When the water, greenbelt, and built-up areas maintain a balanced layout, they can alleviate the negative impact of urban waterlogging and the tropical island effect, improving urban ecological resilience (Xiu et al., 2008). Therefore, this article used the average distance between the source and sink landscapes in the city to measure the morphology resilience. In ArcGIS, we reclassified the land use raster data in the study area into sources and sinks. Source raster includes construction land, while sink raster includes cultivated land, forest land, grassland, and water area. For each source raster, we calculated its distance to the edge of the nearest sink raster and defined this measurement as the nearest neighbor distance. Next, we calculated the average of the nearest neighbor distances for all source rasters and then obtained the average distance between the source and sink landscapes. The formulas are as follows:

where R_m is the urban morphology resilience index, L_d is the average distance between source and sink rasters, min(d_i) is the nearest neighbor distance from ith source raster to sink raster, m is the number of source rasters in the study area, L is a constant whose value is the average distance of the “source-sink” landscape for the entire Pearl River Delta region in 2000.

This article selected both demographic and economic urbanization as first-level indicators to evaluate urbanization. In accordance with scientific principles and data availability, nine secondary indicators were selected to establish a comprehensive urbanization evaluation indicator system, e.g., urbanization rate, urban population density, and the proportion of non-agricultural industry employees (Table 2). The comprehensive evaluation indicator system of ecological resilience was composed of size resilience, density resilience, and morphology resilience (Table 3).

Coupling coordination is a concept from physics, which refers to the interaction between different systems under the combined influence of themselves and the outside world. This paper introduced a coupling coordination degree model to explore the interaction between urbanization and ecological resilience. The formulas are as follows:

where C is the coupling degree, D is the coupling coordination degree, T is the comprehensive evaluation index, U₁ is the comprehensive evaluation index of the urbanization system, U₂ is the comprehensive evaluation index of the ecological resilience system. The coupling coordination degree, D, can comprehensively evaluate the relationship between the two systems (Wang and Tang, 2018). The parameters α and β can be explained as the relative importance of the two sub-systems (Wang et al., 2015). The authors of this paper believe that urban ecological resilience is equally as important as urbanization, so α=β=0.5. Drawing lessons from the classification standards of physics (Li et al., 2012), the coupling coordination degree between urbanization and ecological resilience was divided into four major categories and 12 sub-categories (Table 4).

To analyze the influence of the three types of resilience on the coupling coordination degree, the coordination influence (CI) was introduced (Wang et al., 2019). The formula is as follows:

where D_x (x=1, 2, 3) is the coupling coordination degree between urbanization and size resilience, density resilience and morphology resilience, respectively, D_y is the coupling coordination degree between urbanization and ecological resilience, and W_x is the weighting coefficient corresponding to the resilience. The coordination influence (CI) can measure the influence of sub-category coordination upon the overall coordination, and its positive and negative values represent the promoting and blocking effects, respectively; the magnitude of these values represents the degree of influence.

The size resilience index of the Pearl River Delta region continued to decline during the research period. The decline was rapid from 2000 to 2010, before it slowed down from 2010 to 2015. Future development of the Pearl River Delta will be constrained by size security (Table 5). In terms of spatial distribution, the size resilience of the Pearl River Delta displayed a pattern that was low in central cities and high in peripheral cities, and the gap between central cities and peripheral cities was distinct. These results were closely related to the extensive spatial development model generally adopted by the Pearl River Delta cities. After 2010, the central cities of the Pearl River Delta faced a severe shortage of development space. In particular, the size resilience index of Shenzhen and Dongguan approached the critical level of 1 (Table 5).

The ecological footprint of the Pearl River Delta was higher than the carrying capacity during the study period, and the ecological deficit was substantial. This fact reflected hidden ecological dangers such as high population density, unreasonable energy consumption, and food insecurity in the Pearl River Delta. The difference in density resilience between cities remained significant, noting that low-density resilience indexes have been typical of central cities for a long time. In light of this, Shenzhen had the most severe ecological deficit problem (Table 6).

As the government prioritized environmental protection measures, the deterioration of the ecological deficit in the Pearl River Delta cities eased from 2010 to 2015. This change was largely dependent upon ‘city’ control of the ecological footprint (Table 6). The developmental practice of the Pearl River Delta showed that the natural environment exerted a strong constraint on ecological carrying capacity in the rapid urbanization area. Therefore, it is not easy to increase the urban ecological carrying capacity by expanding the ecological space. In the future, important actions which favor the successful construction of an ecological civilization in the Pearl River Delta will be to transform production and lifestyles, save resources, and reduce energy consumption.

Over the study period, the morphology resilience index of the Pearl River Delta first decreased and then increased (Table 7). From 2000 to 2010, the morphology resilience index of all cities decreased significantly. However, from 2010 to 2015, the morphology resilience index of Guangzhou, Zhuhai, Foshan, Zhaoqing, and Huizhou rose slightly, while that of Shenzhen, Jiangmen, Dongguan, and Zhongshan continued to decline. Combined with remote sensing images, the change of urban morphology resilience in the Pearl River Delta was closely related to its urban spatial expansion pattern. From 2000 to 2010, the spatial expansion of cities in the Pearl River Delta had the characteristic of an outward spreading core. After 2010, the layout of each city had been optimized, with various forms of urban expansion. Cluster development increased the accessibility of construction land to ecological space. With the gradual exhaustion of construction land in central cities, future spatial expansion of the Pearl River Delta will mainly be in Jiangmen, Zhaoqing, and Huizhou. These three peripheral cities have rich ecological landscape resources. Future development should fully recognize the relationship between construction land and ecological space and actively create a highly-coupled urban morphology representative of a previously discussed “source-sink” landscape (Section 3.3.2).

It can be seen from Table 8 that the ecological resilience index of cities in the Pearl River Delta continued to decrease from 2000 to 2015, posing significant threats to regional ecological security. From 2000 to 2010, the ecological resilience index of all cities decreased sharply, with an average decline of nearly 50%. After 2010, the Pearl River Delta began to attach importance to ecological protection in its urban planning and construction practices. Although rapid urbanization was ongoing, the declining trend of the ecological resilience index for each city weakened. The urban ecological resilience degree and urbanization level of the Pearl River Delta resembled a core-periphery structure. The difference is that the development of urbanization was the fastest in central cities, while the ecological resilience degree was relatively high in peripheral cities (Figure 4). Therefore, in future rounds of urbanization, if the urban expansion scale and development intensity are not actively controlled, the ecological resilience level of cities in the Pearl River Delta will likely decrease further.

In terms of temporal distribution, the coupling coordination degree between urbanization and ecological resilience in the Pearl River Delta declined from basic coordination to basic imbalance. The average urban coupling coordination decreased from 0.51 to 0.45 (Figure 5). In 2000, the coupling coordination degree of each city was between [0.38, 0.61], and they were all in the stage of basic coordination or basic imbalance. In 2010, the coupling coordination degree of each city was between [0.23, 0.61]. Except for Zhaoqing, the coupling coordination degree of the other eight cities showed various degrees of decline. The stage of coupling coordination in Shenzhen decreased from basic imbalance to severe imbalance. The declines in Guangzhou, Foshan, and Zhuhai were relatively small, but they still decreased from the basic coordination to the basic imbalance stage. In 2015, the coupling coordination degree of each city was between [0.09, 0.60].

The coupling coordination degree in Guangzhou, Foshan, Zhuhai, Huizhou, and Dongguan increased, while that in Zhongshan, Jiangmen, and Zhaoqing decreased, and the decline degree in Shenzhen was the largest. Jiangmen, Zhaoqing, and Huizhou maintained their status in the basic coordination stage. This change was caused by the continuous improvement of the urbanization level and a simultaneous reduction of the ecological resilience level, which reflected the fact that urbanization in the Pearl River Delta comes at the expense of the land, resources, and the environment. The Pearl River Delta attracted foreign investment due to its large amounts of cheap land, which resulted in the blind expansion of construction land. The subsequent development of the manufacturing industry caused increasingly serious ecological deficits and environmental pollution problems in the region. As the limiting effect of ecological resilience becomes more and more obvious, the urbanization model that emphasizes speed over quality is no longer sustainable.

In terms of spatial distribution, the coupling coordination degree between urbanization and ecological resilience in the Pearl River Delta region presented as a circular pattern increasing from the core to its periphery (Figure 6). This spatial characteristic is related to the differences in the ecological endowment and developmental timing of each city. Shenzhen, located in the inner circle, was the pilot demonstration area for the economic construction of the Pearl River Delta. However, the foundation of ecological resilience was poor and rapidly degraded during the study period, and this triggered a severe dislocation with the level of urbanization. In the middle circle, Guangzhou, Foshan, Zhuhai, Zhongshan, and Dongguan had relatively high levels of urbanization. The rapid urbanization process also exerted a negative impact on their ecological resilience to varying degrees. Although the cities restrained and guided urban development through policy intervention and planning means, the coupling coordination degree did not rebound significantly. In the outer circle, Jiangmen, Zhaoqing, and Huizhou had good ecological background conditions, with sufficient reserves of ecological space such as cultivated land and grassland; and the process of urbanization was slow. Therefore, ecological resilience supported a robust carrying capacity for the stress of urbanization.

From the perspective of the evolution of the coupling coordination types between urbanization and ecological resilience (Figure 7), it was found that the coupling coordination types of the three peripheral cities have not changed during the study period, showing that urbanization lagged behind ecological resilience. The coupling coordination types of the other six central cities regressed to various degrees, and their ecological resilience was blocked. The evolution characteristics of urban coupling coordination types in the Pearl River Delta were still consistent with three types of circular structures.

Shenzhen, in the inner circle, regressed to a severe imbalance stage during the study period. Its ecological resilience continued to be blocked, showing the characteristics of a low coupling-low coordination condition. Shenzhen completed its demographic urbanization in 2004, becoming the first city in China without a rural area. Motivated by the momentum of the urbanization process, a more radical policy of agricultural land nationalization was implemented to promote the demand for industrial land (SASS, 2015). Subsequent rapid industrialization caused a severe imbalance in the ratio of construction land and ecological space in Shenzhen. By 2015, construction land in Shenzhen was almost exhausted. In addition, decreases in cultivated land area and ecological carrying capacity led to an extreme decoupling between urbanization and ecological resilience. In the middle circle, the coupling coordination types of Guangzhou, Foshan, and Zhuhai changed from basic coordination-ecological resilience lag to basic imbalance-ecological resilience blocked, respectively. Zhuhai, in particular, regressed from a harmonious state of high coupling-high coordination to a separate state of low coupling-low coordination. This happened because Zhuhai was mainly positioned as a livable city in its early stage of urban development and did not vigorously develop a high-polluting manufacturing industry, thus maintaining a high level of ecological resilience.

Since the strategy of strengthening cities by industry was put forward in the 11th Five-Year Plan (2006-2010), the process of economic urbanization significantly stressed ecological resilience. Due to lacking ecological space in Dongguan and Zhongshan, these cities were already in the stage of basic imbalance in 2000. After 2010, they also changed to basic imbalance-ecological resilience blocked. In the outer circle cities of Jiangmen, Zhaoqing, and Huizhou, the urbanization was relatively slow due to location factors, so their coupling coordination type stabilized at basic coordination-urbanization lag during the study period. However, as the double-transfer strategy of Guangdong province was put forth in 2008, Jiangmen, Zhaoqing, and Huizhou successively established high-standard industrial transfer parks. These cities continuously promoted the transfer of the rural labor force, effectively driving the urbanization process. As a result, Huizhou entered the basic coordination stage at the end of the study period.

This paper used the CI index to reflect upon the direction and magnitude of the ecological resilience sub-systems’ effect on the coupling coordination degree (Figure 8).

From 2000 to 2015, the CI of the urban size resilience in the Pearl River Delta was generally negative, indicating that it had a negative effect on the coupling coordination degree. In contrast, the morphology resilience of each city, with the notable exception of Shenzhen, had a positive effect on the coupling coordination degree; and the intensity of this effect increased over time. The effect of density resilience on coupling coordination degree exerted both positive and negative effects during the study period. Early density resilience generally played a positive role, but after 2010, its CI continued to decline; and after falling below a zero value, it declined further, obtaining increasingly negative values. In general, the size resilience and morphology resilience strongly influenced the urban coupling coordination degree. Changes in the coupling coordination degree of cities were largely determined by the checks and balances of these two resilient sub-systems. One reason for this is that the scale expansion of urban construction land had not been strictly restrained. From 2000 to 2015, the Pearl River Delta entered a period of accelerated industrialization and a real estate boom, leading to a rapid increase in the demand for construction land. Due to political achievements and financial pressures, local governments had accelerated the construction of suburban industrial parks to stimulate urban expansion. In addition, after the 2008 financial crisis, cities in the Pearl River Delta began to emphasize the quality of urbanization; and the urban spatial organization was also adjusted accordingly. By planning and guiding the accessibility of ecological space, the urban “source-sink” landscape pattern was constantly improved. Shenzhen is a classic example of this process. Since the 21st century, the government has taken the construction of a garden city as a strategic goal. With the advancement of urban renewal and the improvement of the green space system, gardens have become the urban business card of Shenzhen, and the negative impact of morphology resilience has decreased accordingly. It is worth noting that the directional change of CI and density resilience has sounded the alarm for the Pearl River Delta region. The ecological deficit problem has become a significant threat to the Pearl River Delta, and the adaption to low-carbon life and the commensurate transformation industry should become the focus of future urban development.