Spatial–Temporal Variations of the Gross Ecosystem Product under the Influence of the Spatial Spillover Effect of Urbanization and Ecological Construction in the Yangtze River Delta Region of China

Abstract

Rapid development of urbanization and intense human activities had a profound influence on the ecosystem service functions. As an integrated monetary index for the evaluation of final ecosystem services, the gross ecosystem product (GEP) is widely used in the quantification of ecosystem service value (ESV). This study initially assessed and analyzed the spatial distribution of the GEP at the county-level scale using multisource data spanning 2000, 2005, 2010, 2015, and 2020. Then, the spatial transfer characteristics of the GEP were measured. Finally, the study employed spatial panel econometric models and the geographically weighted regression (GWR) model to investigate the spatial effect of urbanization and ecological construction on the GEP. The results indicated that: (1) In 2020, the GEP in the Yangtze River Delta Region was RMB 15.24 trillion, and the GEP per unit area was RMB 42.58 million per square kilometer. It exhibited a cumulative decrease of RMB 298.72 billion from 2000 to 2020. (2) The spatial transfer efficiency of the GEP in urban agglomerations showed a clear decline trend. During the period of 2000–2020, over 96% of county-level units exhibited a decline with RMB 90,076,103.17/km², indicating a consistent downward trend from the central regions towards the periphery. (3) Based on the decomposition effects of the spatial Durbin mode, urbanization and the ecological construction indicator showed spatial spillover effects on the GEP, but their impact mechanisms varied substantially. Among them, the urbanization rate (UR), population density (PD), and the proportion of impervious land (ILP) had the largest negative effect on the GEP, and a 1% rise in ILP locally resulted in a 0.044% decline in the local GEP and a 0.078% rise in the GEP of neighboring units. And the area of ecological land had a positive effect on the GEP of both local and neighboring areas. Those conclusions can offer evidence in favor of encouraging ecologically responsible building practices and sustainable growth in urban agglomerations.

1. Introduction

The ecosystem provides abundant and necessary resources for human activities, and also plays service functions in climate, gas, and hydrological regulation [1]. Currently, the ecosystems are under great environmental pressure, and the imbalance between them and rapid urban development has been highlighted, resulting in a sharp increase in research to assess and quantify ecosystem services (ES). Ecosystem service value (ESV) becomes a trendy topic for evaluating the interactions between commercial society and ecosystems [2,3]. Costanza et al. first created the Global Ecosystem Service Value Assessment System [1] in 1997. Subsequently, Xie et al. [4,5] introduced the equivalent value factor (EVF) approach in China, and proposed the updated global ESV estimation based on those supplied by Costanza et al. [6] in 2014. With low data requirements and simple accounting process, the enhanced model proposed by Xie et al. has been widely adopted within China. However, the EVF approach is mainly applied to macro-regional studies and is unable to adequately reflect the ecosystem values of a specific region [7]. In 2013, drawing on the concept of gross domestic product (GDP), Ouyang et al. [8] proposed the concept of gross ecosystem product (GEP), i.e., the sum of the values of goods and services provided by ecosystems to humans. Since the GEP methodology applies different accounting models and various indicators to estimate the value of each ecosystem service function, the results obtained are more relevant and accurate and the GEP can be a useful complement to the GDP [9]. The pilot studies are currently being conducted at multiple scales and in multiple regions [10,11]. Quantitative evaluation of ecosystem functions and operations through changes in the GEP has become a feasible way [12].

With the development of urbanization, significant changes in urban and ecological construction have occurred, and the value offered by ecosystems is subject to continuous fluctuation [13]. Numerous studies have been undertaken to explore the temporal and spatial evolution of the GEP [14] both globally [15,16] and specifically in China [17,18]. Based on the evaluation of the GEP and analysis of spatial–temporal variations, various quantitative methodologies have been extensively employed in prior research to the driving force analysis of the GEP. These methodologies include the multiple linear regression model and Pearson’s correlation analysis [19,20,21]. But the above approaches ignored some important issues. The GEP spatial heterogeneity [22,23] and the flow of multiple complex components in ecosystems [24] have resulted in the disruption of regional boundaries in the provision of the GEP [25,26]. This implies that the presence of neighbors is likely to influence the GEP, hence introducing a certain degree of bias into the findings [27]. It is necessary to go beyond boundaries to analyze the GEP evolution and analyze the spatial transfer of the GEP.

As a result of urbanization and industrialization, exchanges between regions have become more frequent, especially in urban agglomerations. The spatial transfer of the GEP would lead to the influence of the urban area on the adjacent region [28]. This spatial transfer has the potential to enhance overall regional efficiency and also has advantages to help both economic and social growth. Nevertheless, there is a scarcity of research on the phenomenon of the GEP spatial transfer [29,30]. Additionally, spatial spillover effects of urbanization on the GEP have been a significant issue [31,32]. The spatial econometric model has become a popular method for measuring the spatial spillover effect [33,34]. What is more, the spatially uneven effects and spatial spillover effects of urbanization on the GEP have been demonstrated. Meanwhile, more and more regions realize that urbanization will bring environmental and ecological problems that may have a negative impact on the GEP and began to pay attention to ecological construction projects [35] in order to cope with ecological pressure [36]. It is very interesting to study the spatial spillover effects of the GEP in the comprehensive effects of urbanization and ecological constructions [37,38]. This will provide better support for the formulation of ecological policies. However, such studies focusing on the spatial spillover effects of urbanization and ecological construction on the GEP and their differences are few at present. Hence, much more deserves further research on how to assess the impact of urbanization on the GEP, especially in important economic development zones of China.

The Yangtze River Delta Region (YRDR) is one of the important economic development zones in China and is also a mega-city cluster. Since the 1980s, it has witnessed tremendous urbanization and industrialization. Thus, this study takes the Yangtze River Delta Region as an example to study the spatial and temporal variation of the GEP in the YRDR from 2000 to 2020. Firstly, we estimate the GEP in the YRDR by using the methodology in a national standard of the GEP estimation. Then the spatial transfer of the GEP in urban agglomerations were statistically analyzed. Finally, the spatial interactions among urbanization and ecological constructions on the GEP were employed by the spatial panel econometric models and geographically weighted regression (GWR) model. The main purposes of this study are as follows: (1) To account the ESV in the YRDR based on the GEP method. (2) From the perspective of transcending boundaries, analyze the spatial and temporal variation of the GEP in the YRDR and the spatial spillover effect of urbanization development and ecological constructions on the GEP. (3) The research results can provide a more scientific reference for ecosystem construction proposals and provide guidance for the regional formulation of optimized ecological structures for sustainable regional development.

2. Materials and Methods

2.1. Study Area

The Yangtze River Delta Region (YRDR), one of the most important regions in China, consists of 308 county-level units in the Shanghai, Anhui, Zhejiang, and Jiangsu Provinces(Figure 1). It is one of the most modernized and urbanized regions in China. By the end of 2020, it had approximately 35.8 million ha of land occupied by 16.68 million inhabitants. The urbanization level was 70.85%, surpassing the overall national average. Moreover, the industry in the YRDR has developed rapidly with the swift growth of urbanization since the 1980s, and the ratio of industrial added value to GDP is 33.25%. The land use changes and spatial shifts brought about by industrialization lead to numerous and increasingly prominent problems in the region, including the intensification of urban sprawl, the struggle between people and land, and conflicts between the objectives of ecological preservation and economic and social progress, which have also led to dramatic and significant changes in ecosystem service functions. In addition, with the introduction of strategies such as the Yangtze River Delta integration policies, the spatial spillover effects in metropolitan areas are influenced by the intensification of regional linkages and the movement of factors, particularly capital and people migration, as a result of regional economic integration. The county-level units are increasingly likely to be influenced by the surrounding regions.

2.2. Data Sources

The datasets in this study included remote sensing data, digital elevation model (DEM) data, meteorological data, hydrological data, and statistical data, such as price parameter data, and so on.

2.2.1. Remote Sensing Data

This study selected the multisource remote sensing data including Landsat and GF for the main basis in the monitoring of land use and land cover changes which were used for the estimation of the GEP. The remote sensing data in 2000, 2005, 2010, 2015, and 2020 were obtained from the geospatial data cloud (http://www.gscloud.cn/search (accessed on 13 June 2022)). The spatial resolution of Landsat and GF were 30 m and 1 m, respectively. The land use data in 2000, 2005, and 2010 were visually interpreted based on Landsat data, while those in 2015 and 2020 were visually interpreted based on GF data. Firstly, the images were preprocessed by geometric correction, mosaic cropping, and atmospheric correction. Then, depending on the different characteristics of various land use types, the features of remote sensing data were determined. Next, the final land use classification was completed by the visual interpretation method. Meanwhile, we further verified the accuracy of the data through a field survey, and the final accuracy reached 90%. Furthermore, in order to reduce the possible effects of the differences in the spatial resolution, we resampled the remote sensing data in 2015 and 2020, so that the resolution of all the land use data was standardized at 30 m. The final result was defined as a classification system including cropland, forest area, shrub, grassland, water area, snow and ice, barren land, impervious land, and wetland.

2.2.2. Other Data

As for the estimation of the GEP, the following data were mainly involved in addition to remote sensing data. The DEM data derive from the geospatial data cloud platform (http://www.gscloud.cn (accessed on 13 June 2022)). The meteorological data, such as the air temperature and precipitation, are from the China Meteorological Data Network (http://data.cma.cn (accessed on 13 June 2022)). The soil attribute data derive from the National Earth System Science Data Center (http://soil.geodata.cn/data/dataresource (accessed on 13 June 2022)). The Net Primary Production (NPP) data are from the MODIS land standard product of the NASA website (http://www.nasa.gov (accessed on 13 June 2022)). The statistical data come from the Statistical Yearbook, Water Resources Bulletin, and Environmental Quality Information Bulletin of provinces and cities.

2.3. Methods

2.3.1. Methodology for the Estimation of the GEP

The ecosystem provides the ecological products and services [39]. In this study, we constructed a framework of indicators for the GEP estimation in the YRDR. The framework included the material product value, the regulating service value, and the cultural service value. Specifically, the material product (MP) value included agriculture, forestry, animal husbandry, fishery products, and water supply. The regulating services (RS) included water conservation, soil retention, flood regulation and storage, water purification, air purification, carbon sequestration, oxygen release, and climate regulation. The cultural service (CS) included ecotourism. The GEP calculation formula is:

$$GEP=EMPV+ERSV+ECSV$$

where $GEP$ is the gross ecosystem product; $EMPV$ is material product value; $ERSV$ is regulating service value; and $ECSV$ is cultural service value.

In the process of the GEP estimation, based on the National Ecosystem assessment Guidelines for the gross ecosystem product and Technical specifications for the gross ecosystem product (GEP) for Terrestrial ecosystems in Zhejiang Province, the shadow engineering method, the market value method, and the alternative cost method [10,12,40,41,42,43] were used to calculate the value of different ecological products. The regional differences and interannual variations in prices were taken into account for the accuracy of the results and the results of the calculations for different years were comparable.

2.3.2. Location Quotient of the GEP

The GEP may have a spatial clustering effect [44] and strong cross-regional influence. The ecosystem service values between adjacent areas can have obvious interactions [45], and high-value areas are likely to promote the improvement of the GEP in neighboring areas.

To attempt to further reveal the density distribution of primary service value in the YRDR counties from 2000 to 2020, location quotient (LQ) was introduced in order to clarify the dominant service type in different counties and the agglomeration of the GEP in different counties. When LQ > 1, it means that the supply capability of a certain ecosystem service in region I is greater than the average of the supply capacity in the YRDR. And the larger the LQ value, the stronger the supply capacity of a certain ecosystem service in the region. The location quotient calculation formula is:

$${LQ}_{ij}=\frac{{GEP}_{ij}/{GEP}_i}{{GEP}_j/GEP}$$

where ${LQ}_{ij}$ is the location quotient in ecosystem service type $j$ in county $i$; ${GEP}_{ij}$ is the ecosystem service type $j$ value in county $i$; ${GEP}_i$ is the total GEP in county $i$; ${GEP}_j$ is the ecosystem service type $j$ value.

2.3.3. Spatial Transfer Intensity of the GEP

The spatial transfer of ecosystem service values is a complex process due to differences in ecosystems and is influenced by multiple factors such as meteorology and hydrology. However, there is a basic consensus that ecosystem products could naturally move across regions [46]. The spatial transfer enables the GEP to transfer out of the habitat and function on a larger scale. To validate this value flow beyond administrative boundaries in the YRDR, we employed the breaking point formula [29,47] to estimate the radiation radius and intensity of the spatial transfer of the GEP.

Taihu Lake is the largest lake in the YRDR, and its area has changed significantly over the study period. As an important part of the Yangtze River system, which is the main channel for the flow of ecosystem service values, Taihu Lake is also the center of water resource transfer in the YRDR. This study designated Wuzhong District, which is the geometric center of Taihu Lake, as the transfer-in area, whereas all county-level units, excluding Wuzhong District, were taken as the transfer-out area in the YRDR. The GEP spatial transfer radius calculation formula is:

$$R_i=\frac{D_{is}}{1+\sqrt{{GEP}_s/{GEP}_i}}$$

where $R_i$ is the GEP spatial transfer radius; $i$ is the transfer-out area; $s$ is the transfer-in area (the geometric center of Taihu Lake); $D_{is}$ is the distance between the central point of the transfer-out area and the central point of the transfer-in area; and ${GEP}_s$ and ${GEP}_i$ are the GEP in the transfer-out area and transfer-in area, respectively. The calculation formula of the GEP spatial transfer intensity and its changes is:

$$I_i=\frac{{ESV}_i}{{R_i}^2}$$

$$C=\frac{I_{t2}-I_{t1}}{I_{t1}}\times 100\mathrm{\%}$$

where $I_i$ is the average GEP transfer intensity from the unit $i$ to the geometric center of Taihu Lake, namely, radiation intensity. $C$ is the change rate of the GEP spatial transfer intensity and $I_{t1}$ and $I_{t2}$ are the GEP spatial transfer intensity at $t_1$ and $t_2$, respectively.

2.3.4. Spatial Spillover Effects of Urbanization and Ecological Constructions on the GEP

(1)

Selection of influencing factors

In this paper, the main focus of attention is on the impact of urbanization and ecological construction indicators on the GEP. As for the urbanization indicators, the urbanization rate (UR, %), population density (PD, people/km²), and the proportion of impervious land (ILP, %) were chosen to describe urbanization, which were considered to be the main influencing factors [48]. As for the ecological construction indicator, land cover transfer is considered a direct way to reflect the ecological construction situation [49]. In this study, the ecological land includes forest area, shrub, grassland, water area, and wetland. The area of ecological land (AE, ha) was chosen to describe the level of ecological construction. The presence of social and economic connections between regions gives rise to a discernible geographical spillover effect under the combined influence of urbanization [50] and ecological construction [49]. The GEP of one county-level unit is influenced not just by its own, but also by the adjacent units. In order to explore this spatial spillover effect and its heterogeneity, spatial panel econometric models and the geographically weighted regression model were introduced to analyze in depth the impact of urbanization and ecological construction indicators on the GEP. In this study, Moran’s I is employed to explore the spatial autocorrelation of the GEP prior to conducting spatial analysis. Spatial correlation is a fundamental basis for assessing the presence of spatial effects [51].

(2)

Spatial panel econometric model

To study the spatial spillover effect of urbanization and ecological construction on the GEP, we employed spatial panel econometric models. These models can be categorized into three fundamental forms: the spatial lagged model (SLM), spatial error model (SEM), and spatial Durbin model (SDM). The calculation of these models requires a spatial weight matrix which was generated by using the software Geoda1.16, with a primary emphasis on the dominant spatial relationship of queen contiguity.

In order to evaluate the spatial spillover effects of urbanization and ecological construction on the GEP, a battery of tests was conducted to determine the most appropriate spatial panel econometric model for our data. Initially, an OLS regression analysis is conducted to examine the relationship between the driving force and the GEP. The importance of expanding the OLS model to incorporate spatial regression models is assessed, along with the evaluation of the superior fixed effect model for our data, through the Lagrange multiplier (LM) test and the robust Lagrange multiplier (RLM) test. Next, we proceeded with the implementation of a more appropriate model and selected several commonly used statistical measures, such as R², Log likelihood, and Akaike information criterion (AIC), to assess and contrast the efficacy of the model. Finally, the Wald test and Likelihood ratio (LR) test were employed to ascertain the most appropriate spatial panel model for fitting purposes.

(3)

Geographically weighted regression (GWR) model

The geographically weighted regression model (GWR) is widely recognized as one of the most effective methods for addressing spatial heterogeneity [52]. There are significant differences in urbanization development and ecological construction in different counties, meaning that the impacts of urbanization and ecological construction indicators on the GEP are spatial–temporal heterogeneous in different counties and in different development periods. To further improve and reflect the differences in the impact mechanism in more detail [53], we employed the GWR models. In this study, GWR describes the impacts of urbanization and ecological construction on the GEP and reflects the spatial heterogeneity and the direction of the impact through the regression coefficients within each unit [20].

3. Results

3.1. Spatial–Temporal Variations of the GEP in the YRDR

3.1.1. Temporal Variation Characteristics of the GEP from 2000 to 2020

The GEP was estimated by the methodology in the National Ecosystem assessment Guidelines and Technical specification for gross ecosystem product (GEP) in Zhejiang Province. As shown in Figure 2, the results showed that the GEP in the YRDR was RMB 15.24 trillion in 2020, and the GEP per unit area was RMB 42.58 million per square kilometer. As seen from 2000 to 2020, the total GEP in the YRDR exhibited an initial upward trajectory followed by a subsequent decline, resulting in an overall declining trend. The total reduction seen over this period amounted to RMB 298.72 billion. From the perspective of the ecosystem service function, RS has the largest value proportions, consistently above 75% of the GEP in 2000–2020, the MP value proportions were around 10%, and the CS value proportions were around 13%. Over these 20 years, all types of ecological service value showed a trend of decline. Among them, the CS exhibited little changes, with a fall rate of 1.07%. In contrast, the RS experienced a substantial decline, amounting to RMB 237.56 billion, which accounted for 79.53% of the counties with a decline trend in the YRDR.

The decline in this changing trend of the GEP is in line with transformations in land use patterns. During the period 2000–2020, the cropland in the YRDR decreased the most, with a reduction of 946,708.2 ha, followed by forest area and water area with the decrease of 291,051.45 and 149,148.45 ha, respectively. The ecological land decrease leads to the reduction of the GEP. The extent of impervious land witnessed a substantial increase from 2,411,344.53 ha in 2000 to 3,822,845.49 ha in 2020, reflecting a growth rate of 58.54%. The phenomenon of land use conversion is observed as a consequence of the rapid process of urbanization. The phenomenon can be attributed to the proliferation of construction activities, resulting in the encroachment of significant portions of cropland and forest area.

3.1.2. Spatial Variation of the GEP in the YRDR

The distribution of the GEP exhibited significant disparities among the various counties within the YRDR (Figure 3). As shown in Figure 3, the spatial distribution of the GEP in 2020 is highly similar to that in 2000 except that five county-level units in northwest Anhui Province have been upgraded. Spatial variations in the distribution of the GEP within the YRDR are readily apparent, with distinct disparities observed between different regions. The southwestern and northeastern countries exhibit a notable prevalence of high GEP, whereas the eastern countries serve as the primary area characterized by low GEP. Among them, Wuzhong District has the highest GEP because of widespread cover by Taihu Lake, and the proportion of water area in Wuzhong District is ranked first, accounting for 7% of the water area in the whole YRDR, while Jing’an and Changning District in Shanghai City have the lowest GEP. These districts are characterized by a predominant land use of impervious areas, accounting for nearly 90% of the total land area in these county-level divisions.

In terms of change rates of the GEP, the majority of county-level units experienced a decrease in the GEP, especially in Jiangsu Province and Shanghai City. These counties were widely distributed in the YRDR and generally exhibited lower GEP. Meanwhile, 118 counties showed an increase in the GEP, which generally had higher GEP values. This result indicated that county-level units with high GEPs have a greater potential for rising GEPs. To further analyze the changes of primary service functions value during the study period, their LQ values and change rates were also calculated (Figure 4 and Figure 5). The results revealed that the number of counties with LQ over 1 in primary ecosystem services were 135, 229, and 88, respectively.

To be more detailed, it is worth noting that the LQ values pertaining to the MP exhibited a range spanning from 0.13 to 1.39. The spatial distribution with LQ over 1 was mainly located in the northern region and those with low LQ in the southern region. In addition, with the development of urbanization, about a third of counties showed an increase in supply service value, and they were mainly concentrated in the central region of Anhui Province and the western region of Zhejiang Province. Meanwhile, the quantity of counties showing an increase in the regulating service function value was the highest, indicating that the ecosystems in the county level have stronger dominance in the provision of regulating service functions. The LQ values of cultural service ranged from 0.59 to 1.44, with a regional distribution of higher LQ values in the southern regions and lower values in the northern regions. The counties that showed an increase in cultural service value were mainly located in Zhejiang Province, which tended to have higher LQ values. It showed that the supply capacity of cultural services in the counties of Zhejiang Province is higher than that of other areas in the region, and this supply capacity is still increasing. Zhejiang Province occupied an absolute dominant position in the provision of cultural service functions in the whole region.

3.2. Spatial Transfer Variations of the GEP from 2000 to 2020

3.2.1. Spatial Transfer Radius of the GEP during the Period of 2000–2020

The service value of counties may be affected by the surrounding area, both in terms of the GEP and primary ecosystem services. Therefore, we further analyzed this impact by neighboring counties with spatial transfer (Figure 6). The results showed that the spatial transfer radius of the GEP in the YRDR exhibited a range of 0.76–148.01 km from 2000 to 2020. Susong County demonstrated the highest spatial transfer radius, ranging from 146.45 to 148.01 km, while Jing’an District exhibited the smaller radius, 0.76 km; 63% of the counties had a transfer radius of over 25 km, and 39% of the counties had a transfer radius of over 50 km, which is beyond the administrative boundaries of the counties. It can be presumed that there is a flow of the GEP beyond the administrative boundaries. Generally, the distribution characteristics exhibited a lower concentration in the central areas and a higher concentration in the peripheral regions during the study period. This indicated that counties with greater distance to Taihu Lake and higher GEP had a greater GEP transfer range and had the potential to regulate the ecosystem service value over a larger area.

In contrast to 2000, the spatial transfer radius of the GEP exhibited a declining pattern throughout the majority of counties in 2020. This might be related to the decline in the GEP over the study period in the YRDR. The counties with increased transfer radius were mainly concentrated in the southern region of Zhejiang Province and the northern region of Anhui Province, which was highly similar to the spatial distribution of the GEP that showed an increase. Among them, Xiangshan District increased the most by 40 km, with an increase of 25.58%. It can be concluded that in the YRDR, cross-regional transfer ranges of the GEP became smaller, and more counties tended to conduct spatial transfer activities of the GEP at smaller ranges.

3.2.2. Spatial Transfer Intensity of the GEP

This study estimated the spatial transfer intensity of the GEP (Figure 7), which exhibited a range of RMB 161,918.90–175,533,783.2/km² from 2000 to 2020. It is noteworthy that over 90% of counties displayed a spatial transfer intensity falling within the range of hundreds of thousands to millions of RMB. The difference between the transfer intensity of counties was very obvious. Huqiu District, which is nearest to Wuzhong District, had the largest transfer intensity at RMB 175,533,783.20–182,809,301.10/km². Contrary to the spatial distributions of the transfer radius, the transfer intensity increased from peripheral counties to central counties, and the counties closer to Taihu Lake had greater transfer intensity. The findings of the investigation revealed a pattern of high distribution in the middle and low distribution in the surrounding counties throughout the study period. It was revealed that there was a greater magnitude of cross-regional economic spatial variation in the central county-level units.

As seen from 2000 to 2020, there was a notable decline in the transfer intensity from neighboring counties to Taihu Lake, especially in the eastern region of the YRDR. Compared with 2000, over 96% of counties in 2020 showed a decline in the transfer intensity, while 10 counties demonstrated improvement. Spatially, the counties showing an increase in GEP spatial transfer intensity were concentrated in the Anhui Province, while the counties showing severe decay in GEP spatial transfer intensity were concentrated around Taihu Lake. This indicated that the changes in the spatial transfer intensity of the GEP were correlated with the decline of the GEP and the expansion of the transfer radius. In addition, this changing characterization illustrated the weakening of the GEP support of peripheral counties to Taihu Lake, and the increasing importance of peripheral counties as an essential component of the cross-regional GEP.

Regarding the spatial transfer intensity of the primary service functions (Figure 8), the majority of the counties showed a decay whether in the material products, regulating services, or cultural services. This indicated that the decay of spatial transfer intensity covered a multitude of service functions in an all-round decay. Each service function should be emphasized in the future development. Specifically, compared with the regulating services and cultural services, there were more counties whose spatial transfer intensity of supply services showed a declining trend, which might also be related to the result that the largest number of counties showed a decay in the supply service value. If the supply services of counties are not regulated, they may not be able to meet regional development needs in the future.

3.3. Effects of Urbanization and Ecological Construction on the GEP

3.3.1. Tests of Spatial Autocorrelation

This study applied the Moran’s I to analyze the spatial correlation of urbanization, ecological construction, and the GEP. The Moran’s I for the GEP in the YRDR in 2000, 2010, and 2020 were 0.25, 0.26, and 0.26, respectively, among of which pass the 1% level of significance test. This showed that the GEP has significant positive spatial autocorrelation (

Table 1. Regression results of OLS.

Determinants	Pooled OLS
UR	−0.0927059 ***
ILP	0.2053717 ***
PD	−0.1057907 ***
AE	0.1004327 ***
R2	0.9475
LM spatial error	224.201 ***
RLM spatial error	190.291 ***
LM spatial lag	48.421 ***
RLM spatial lag	14.512 ***

Note: The *** distributions is significant at the 1% levels.

), which indicated that urbanization and ecological constructions have a statistically significant effect on the GEP when employing the OLS model without accounting for spillover effects. However, it is important to note that the OLS does not capture any spatial non-stationary consequences of urbanization and ecological constructions on the GEP.

Prior to conducting spatial analysis, a number of tests were performed to determine the optimal model for fitting the data. To examine the presence of the spatial spillover effects, we employed the LM and RLM tests. The rejection of the null hypothesis regarding the absence of a spatial spillover effect suggested that the application of spatial econometric models was warranted for this study. The statistical significance of the test results was seen at a confidence level of 1%. The preliminary assessment suggests that the SDM may be the most appropriate choice.

Subsequently, the Wald test and LR test were employed to ascertain the optimal model for fitting the given data. The statistical analysis revealed that the SEM and the SLM cannot be simplified to the SDM. This conclusion is supported by the results of the Wald tests, which yielded a test statistic of 73.18 (p = 0.0000) for the comparison between the SDM and SEM, and a test statistic of 10.40 (p = 0.0000) for the comparison between the SDM and SLM. It indicated that the SDM was the most fitting model. Additionally, the Hausman test yielded evidence of the rejection of random effects at a significance level of 1% (with a chi-squared statistic of 73.58, 11 degrees of freedom, p = 0.0000), indicating that fixed effect models were suitable specifications. Based on the evaluation of R², Log likelihood, and AIC (

Table 2. Regression results of spatial panel model.

Determinants	Spatial Fixed Effects	Time Fixed Effects	Spatial and Time Fixed Effects
UR	−0.018 ***	0.002	−0.011 **
ILP	−0.038 ***	0.022	−0.048 ***
PD	−0.020 ***	−0.012	−0.020 ***
AE	0.011 ***	0.045 ***	0.012 ***
W × UR	−0.019 ***	0.042	0.012
W × ILP	0.114 ***	0.094 ***	0.068 ***
W × PD	0.006	0.077 ***	0.007
W × AE	0.009	0.027	0.011
R2	0.465	0.873	0.520
Log likelihood	1532.4089	−127.3157	1575.6361
AIC	−3040.818	276.6313	−3127.272

Note: The ** and *** distributions are significant at the 5% and 1% levels.

), it can be concluded that the SDM with spatial and time fixed effects demonstrated the most optimal fitting specification.

3.3.2. Spatial Spillover Analysis of the Urbanization and Ecological Construction on the GEP

According to the above test results, the SDM fits the data best, both statistically and substantively. Based on the selected model, it was determined that the spatial lag term had a coefficient of 0.459. This implies that a 1% rise in the local GEP results in a 0.459% increase in the GEP of neighboring counties. Therefore, the spatial lag effects of the GEP are significant at the county-level scale, a finding corroborated by the related research of Lu et al. [54], who determined that the GEP has spatial spillover effects, similar to others. To further decompose the spatial spillover effects of urbanization and ecological construction on the GEP to clarify the effects of each part, we studied the direct, indirect, and total effects of variables (

Table 3. Decomposition results of the urbanization and ecological construction effects on the GEP.

Variables	Direct Effect	Indirect Effect	Total Effect
UR	−0.010 *	0.012	0.002
ILP	−0.044 ***	0.078 ***	0.034
PD	−0.020 ***	−0.005	−0.025
AE	0.014 ***	0.030 **	0.044 ***

Note: The *, **, and *** distributions are significant at the 10%, 5%, and 1% levels.

). Both local urbanization and ecological construction indicators have a significant effect on local GEP, and neighboring ILP and AE also have a significant effect on local GEP. As for total Effect, the impact of ecological construction was more significant compared to the urbanization indicator. It indicated that urbanization had a significant impact on local GEP, whereas ecological construction had a significant impact not only on the local GEP but also on the GEP of nearby areas. Especially, the indirect effects of ILP and AE were far greater than the direct effects, resulting in a strong spatial spillover effect.

Specifically, among the urbanization indicators, UR, ILP, and PD all have significant negative effects on local GEP, and only ILP had a significant spatial spillover effect. Firstly, the proportion of impervious land had the largest negative direct and positive indirect impacts, which implied that local land development reduced the development of local ecosystem services but relieved ecological pressure on neighboring areas. In addition, the total effect of ILP on the GEP was 0.034 but it was not statistically significant. Secondly, the urbanization rate influence on the GEP had a direct effect with the coefficient of −0.010, while those for the indirect and the total effect were insignificant. The results meant that UR had negative direct effects on the GEP, while it was not helpful for neighboring units. It indicated that the urbanization in the YRDR was likely dominated by local urbanization and long-distance urbanization, and the increase in local urbanization rate does not have a significant impact on the neighboring areas, such as the supply of the GEP. Lastly, the population density had a significant negative effect on the GEP. The coefficients for the direct effects of PD were −0.020. This implied that a 1% rise in PD locally resulted in a 0.020% decline in the local GEP. With the rapid growth of the region, the population density in the YRDR was increasing rapidly, reducing the area of ecological land use and increasing the area of hardening throughout the region, thus affecting the supply of the GEP. Among the urbanization indicators, the direct, indirect, and total effect coefficients of the ecological construction indicators passed the 5% test. This indicated that ecological construction has a significant effect on the GEP of both local and neighboring areas. The area of ecological construction has a positive effect on both local and surrounding GEP, which means that strengthening ecological construction is not only beneficial to the supply of local ecological products but also to the supply of ecological products in the surrounding areas.

Overall, based on the decomposition effects of the SDM model, urbanization and ecological construction showed a significant effect on the GEP but their impact mechanisms varied substantially. Among them, the proportion of impervious land and the area of ecological land have spatial spillover effects on the GEP. The results meant that both local GEP and neighboring GEP tended to decline with the development of urbanization and increase with the ecological construction investment.

3.3.3. The Spatial Heterogeneity of the Effect of the Urbanization and Ecological Construction on the GEP

To further analyze the spatial heterogeneity in the degree of influence of the impact factors on the GEP, we introduced the GWR model. The value of AICc of the GWR model was 367.499, which was larger than that of OLS, meaning that the fitting result of THE GWR is better. Meanwhile, comparing the R² and the adjusted R², the results of the GWR model were found to be better, which again verified that the GWR model should be chosen to analyze the influencing factors of the GEP in the YRDR. Figure 9 shows the spatial variation of the regression coefficients of each influence factor based on the GWR results.

Based on the results of the GWR model, the relationship between the four influencing factors and GEP changed over space and time. Both positive and negative effects of the influencing factors on the GEP were observed. Specifically, the correlation coefficients between the ecological construction indicators and GEP were all positive during the study period, indicating that ecological construction always maintains a positive impact on the GEP, which is consistent with the SDM model’s results. Spatially, the southern region of the YRDR, especially Zhejiang Province and Shanghai City, were affected by ecological construction to a greater extent. In the past decade, these areas have vigorously pursued the construction of ecological civilization, which plays an important role in improving the level of ecosystem services. As for the indicators of urbanization, the regression coefficients of the urbanization rate were all less than 0, showing that the urbanization rate is always negatively correlated with GEP. Among them, the GEP of Anhui Province was most negatively affected by the urbanization rate. Thus, it is necessary to pay attention to the ecological pressure while increasing the urbanization rate. In terms of the proportion of impervious land, more than 200 counties have a correlation coefficient between the ILP and GEP greater than zero, a number that is more than double the number of counties with coefficients less than zero. This meant that the ILP, overall, exhibited a positive effect on the GEP. Spatially, the counties showing positive effects were mainly distributed in southern Anhui and northern Zhejiang, while the counties showing negative effects were mainly distributed in the northern area, which was similar to the distribution of the results of UR. In terms of the population density, the correlation coefficient between the PD and GEP was negative in 90% of the counties, and the coverage proportion of the negative impact expanded overall. It indicated that almost the whole region was suffering from ecological pressure caused by the larger population pressure.

According to the aforementioned, based on the results of the GWR model, significant spatial heterogeneity existed in the impact of the influencing factors on the GEP. The area of ecological land showed mainly positive impacts on the GEP, the urbanization rate and population density showed mainly negative impacts, and the proportion of impervious land showed positive impacts in more counties. These results were consistent with the results of the SDM model. Spatially, the northern region, especially Anhui Province, was negatively affected by UR, ILP, and PD to a greater extent, while the southern region, mainly Zhejiang Province, was positively affected by EA to a greater extent. Each region should pay attention to the common problem of ecological pressure due to population pressure but also focus on alleviating the individual problems of the region.

4. Discussion

4.1. Effect of Land Use Change on the GEP Variation

Elements such as climate elements [55,56], urbanization [57,58], and land cover transfer [59,60,61] are considered to be the main factors leading to changes in the GEP. Combined with previous studies [62], we can find that in developed regions, the influence of climate elements on the GEP is gradually decreasing, and the dominant role of external factors on the ecosystem is becoming increasingly obvious. More studies focus on the relationship between land cover transfer and GEP changes [63]. In recent years, the conversion from forest area, grasslands, shrubs, and wetlands to cropland and impervious areas has led to predominantly shrinking ecosystem services [56,64]. The decline in the GEP during rapid urbanization has been repeatedly illustrated around the world [16], including China [65]. Similar conclusions were obtained in the YRDR in both this study and others which were based on the EVF approach [66], the InVEST model [67], and similar approaches [62].

In the estimation of the GEP, this study chose the land use data, climate data, the NPP, and other major data as data sources. Among them, the annual average climate data are used for all the climate data involved. Thus, the influence of climate fluctuation was minimized to a certain extent. In the exploration of the influencing factors of the GEP, this study focused on the influence of urbanization, population, construction land, and ecological construction on the GEP and introduced the SDM and GWR models for in-depth analysis of the spatial spillover effect and spatial heterogeneity of the influence. Through the calculation, the GEP in the YRDR overall decreased by RMB 298.72 billion from 2000 to 2020. Based on the above, we mainly discuss the change in GEP from the perspective of land use change (

Table 4. Changes in the area of land use types in the YRDR.

Land Use Type	Area in 2020/ha	Proportion in 2020	Area Change/ha	Area Change Rate
Cropland	18,852,166.62	53.51%	−946,708	−4.78%
Forest area	10,499,589.54	29.80%	−291,051	−2.70%
Shrub	113.04	0.00%	−1136.7	−90.95%
Grassland	2881.80	0.01%	−15,504	−84.33%
Water area	2,051,576.82	5.82%	−149,148	−6.78%
Snow and ice	0.00	0.00%	−0.27	−100.00%
Barren land	190.26	0.00%	−1743.3	−90.16%
Impervious land	3,822,845.49	10.85%	1,411,501	58.54%
Wetland	0.00	0.00%	−0.09	−100.00%

).

Forest areas are usually considered as the primary provider of the GEP [68]. During the period of 2000–2020, the forest area decreased by 291,051 ha and it is one of the reasons for the decline in the GEP. Meanwhile, water areas are also one of the most dominant ecosystem product value providers in this study because of their large area and high service value, especially hydrological regulation, flood regulation, and storage. From 2000 to 2020, the water areas decreased by 149,148.45 ha and the GEP in 2020 was also seriously affected in the YRDR. In addition, in order to deeply analyze the impact of ecological land and construction land changes on the GEP, the SDM and GWR models were introduced. Through them, it was found that impervious land and ecological land have significant spatial spillover effects on the GEP, and their impacts were spatially heterogeneous. Combining the model results, it was noticed that urbanization had a significant negative impact on the GEP, while ecological construction land had a significant positive impact on the GEP. This further proves that the land use conversion from cropland, forest, and other ecological land to construction land will lead to a significant decline in the GEP, which to some extent explains the phenomenon of the GEP declining in 2020. During the period of study, the rate of urbanization and economic growth in the YRDR increased rapidly, and large-scale and continuous urban expansion led to encroachment on cropland and forests [69]. The reduction of ecological land has led to a decrease in the regional capacity to provide ecological products, especially supplying and regulating services. Such decline in the GEP due to conversion from ecological to built-up land has been frequently justified [70].

The overall land use changes and its effect on the GEP provide important information for landscape planning. Based on the results obtained, the land use shifts caused by urbanization should attract our attention. It becomes important to incorporate land use factors when considering social development decisions, and measures need to be taken to expand the extent of water bodies, arable land, and forest areas that possess significant service value.

4.2. Regional Integration and Spatial Transfer of the GEP in the YRDR

Environmental issues that extend across regional boundaries are prevalent in densely populated areas, and collaborative efforts across regions play a significant role in addressing these issues comprehensively. Scholars have conducted some research on transboundary ecological problems such as water pollution [71] and ecosystem services [72]. The ecosystem product transfer includes material goods, air purification, water conservation, and other ecosystem services. And their spatial transfer characteristics are different due to their different generators and functions. For example, air purification is affected by atmospheric circulation. Comprehensive research of the spatial transfer of the GEP is a massive and complex process. In this study, the supply service in the GEP had broken a regional boundary and the spatial transfer of ecosystem service exists within urban agglomeration. We only considered the spatial transfer of the total GEP. Thus, this work introduces the breaking point formula, which is primarily applied to transfer and the GEP flows beyond the boundaries of administrative areas [73]. And Taihu Lake, as the largest lake in the region, is the most important component of the Yangtze River system, which is the main channel for the flow of the GEP. Therefore, we take the geometric center of Taihu Lake as the transfer area.

Based on research results, 39% of the counties have a transfer radius of over 50 km, which is beyond the administrative boundaries of the counties. It can be presumed that there is a flow of the GEP beyond the administrative boundaries. Overall, over 96% of counties in 2020 showed a decline in the transfer intensity, directly owned by the decrease in the GEP in counties. Measures to enhance the ecosystem service value urgently need to be taken. This can be achieved through the implementation of ecological engineering construction and the establishment of natural reserves or national parks, tailored to local conditions, among other measures. Meanwhile, the spatial transfer radius and intensity of the GEP showed a decreasing trend. Similar results were found in the Yellow River Basin [29]. Promoting regional ecosystem product mobility was also an important means. It is also necessary to establish a diversified cooperation mechanism, promote ecosystem service flow and integration, and strengthen regional cooperation. Several potential actions can be undertaken to address this issue, such as the establishment of ecological corridors and the implementation of an urban greening infrastructure network development and landscape connection.

In addition, the spatial transfer characteristics of different service functions were different. The decrease in spatial transfer efficiency of the three primary service functions was a common feature, with the spatial transfer efficiency of the supply service function being the lowest and promoting the flow of primary service functions in regional ecosystems, especially the supply service function. In the process of regional ecological integration construction, counties in the YRDR were guided to integrate regional ecological resources, leverage their comparative advantages in ecosystem product management, promote the flow of various service functions in regions, and achieve the goal of a regional win–win.

4.3. Spatial Spillover Effect of Urbanization and Ecological Construction on the GEP

As one of the regions with the highest urbanization rate in China, the rapid expansion of cities in the YRDR has an impact on ecosystem products and it has become a consensus among all scholars. Urbanization and the GEP can be positively [74], negatively [75], U-shaped [32], or inverted U-shaped [76]. During the period of study, impervious area increased 1,411,501 ha due to urbanization, which somewhat led to a decrease in the GEP. Ecological construction is considered to be a favorable measure to alleviate the environmental pressure [35,77] caused by urbanization, and the YRDR has also made a large investment in it, which has led to the improvement of the regional ecological environment. As urban agglomerations grow, these factors not only affect the local GEP, but also spread to the GEP of neighboring areas [60]. The existence of the GEP that flows beyond administrative boundaries was further verified in this study. However, studies on assessing the spillover effect are still relatively few and focus on ESV estimated by the EVF approach. This study focused on the spatial spillover effect of the impact of urbanization and ecological construction on the GEP from 2000 to 2020, taking counties in the Yangtze River Delta Region as the object of study.

In this study, the spatial spillover effects of urbanization and ecological construction on the GEP and the spatial heterogeneity of the extent and direction of the effects of the influencing factors are demonstrated through the SDM and GWR models. Based on the results of the SDM and GWR models, urbanization had a generally negative effect on the GEP, thereby imposing a negative externality. This conclusion was generally in agreement with most existing studies [57,65]. To be specific, local urbanization rate and population density have negative impacts on local GEP. In the YRDR, land conversion from other uses to constructed land occurred considerably due to population urbanization. This high concentration of population leads to significant ecological pressures throughout the region. Next, the local proportion of impervious land area presents a negative effect on local GEP and a positive effect on neighboring GEP. This is because regional land construction often needs to absorb capital from neighboring areas in the YRDR, and this siphonic effect hinders the development of local GEP and relieves ecological pressure on neighboring areas. Moreover, the area of ecological construction has a positive effect on both local and surrounding GEP, which means that strengthening ecological construction is not only beneficial to the supply of local ecological products but also to the supply of ecological products in the surrounding areas.

Based on the results of this study, urbanization has a negative effect on local GEP and ecological construction has a positive effect on both local and surrounding GEP in the YRDR. The coordination of the link between urbanization and the ecological environment holds great importance in ensuring the sustainable growth of urban agglomerations. Promoting urban-intensive development can help promote the development of regional ecosystem services. This study indicates that it is imperative to consider the spillover impact between counties and emphasize the importance of collaborative efforts among neighboring counties and regions in order to effectively conserve regional ecosystems. Enhancing the equilibrium of ecosystem services within a given region necessitates the establishment of coordination and collaborative efforts with adjacent regions. Therefore, ecological protection is a common pursuit of regional development. During the course of urbanization, blindly pursuing development efficiency often comes at the cost of huge ecological damage. In urban expansion, some effective land use management and policy-making procedures should be taken to coordinate the ecological environment with urbanization. Furthermore, the enhancement of collaboration among various tiers of government is necessary for efficient land use management and the formulation of policies aimed at safeguarding the ecosystem in urban agglomerations.

4.4. Limitations and Future Directions

This article selected the YRDR in China as the study area, which has the fastest urbanization development, and focused on county-level ecosystem service value from 2000 to 2020. In this study, we chose remote sensing data including Landsat and GF for land use data interpretation and then validated the accuracy of the land use data through field research. Meanwhile, the gross ecosystem product (GEP) method was applied to quantify ESV to make it more suitable for ESV evaluation in the YRDR. Next, the spatial transfer characteristics of the GEP were estimated and the spatial panel model was used to analyze the spatial spillover effects. This study will enhance the comprehension of the GEP temporal variable characteristics resulting from LUCC beyond administrative boundaries. The findings will offer valuable insights and assistance to decision-makers in the YRDR for the protection and integrated management of ecosystems.

Nevertheless, there remains scope for greater enhancement in future study efforts. Firstly, in the ESV quantification, more data and methods are used for different ecological services in the GEP method than in the EVF method, which is conducive to more scientific results. However, the complexity of the methodology also means that different scholars will make different choices when calculating values. The assessments are highly dependent on the evaluation methods and data used. Because no higher resolution data were available in China before 2015, Landsat data were used in 2000, 2005, and 2010, resulting in the use of different remote sensing data sources. Although the data used have been verified in the field to have high accuracy, a few effects will still be present. Second, this study focused only on the GEP every five years, a relatively short period of time that made it difficult to analyze continuous changes in the GEP over long periods of time. In addition, the driving factors of the GEP are multifaceted. This study solely focuses on analyzing the spatial spillover effects of urbanization and ecological construction on the GEP. It does not consider other significant aspects such as economic development, policies, and climate. To reveal the ecological processes and response mechanisms during urbanization development, more accurate GEP accounting results and more diversified driving factors will be further discussed in future studies.

5. Conclusions

The Yangtze River Delta Region, as the most rapidly urbanized and densely populated area in China, exhibits significant changes in the structure of land use. This region exhibits more prominent ecosystem service evolution characteristics compared with the whole country. In this paper, a more comprehensive and scientific GEP methodology was chosen for the ESV evaluation, based on guidance documents such as the National Ecosystem assessment Guidelines for gross ecosystem product and Technical specification for accounting gross ecosystem product (GEP) for Terrestrial ecosystems in Zhejiang Province. This provided more appropriate accounting results for the YRDR for the subsequent analysis in this study. During the study period, the GEP exhibited a discernible pattern of decline in the YRDR because of land use conversion. Furthermore, this reduction surpassed the average national decline. The GEP exhibited distinct spatial clustering patterns, with the southwestern and northeastern regions displaying large concentrations of the GEP, whereas the northern region predominantly exhibited low GEP distribution. Afterward, the breaking point formula was proposed as a means to quantitatively assess the GEP spatial transfer intensity and radius. The results identified that more than half of the counties could have GEP flows beyond administrative boundaries. Nevertheless, as the regional integration process advances, there is a noticeable decline in both the spatial transfer radius and intensity of the GEP from peripheral counties to Taihu Lake. This is highly correlated with the overall decline in the GEP. Furthermore, it is worth noting that there are spatial spillover effects of urbanization and ecological construction on the GEP. The urbanization indicator has negative effects on the local GEP and the proportion of impervious land has positive effects on the neighboring GEP. The ecological land area has strong positive impacts on the GEP, both directly and indirectly. Under the development of new urbanization, it is imperative for urban agglomerations to adopt the notion of regional ecological integration development. This entails striking a harmonious balance between urban growth and the preservation of the ecological environment. Existing policies and plans must be firmly implemented, such as the three designation zones (agricultural, ecological, and urban) and three lines (urban development boundary, ecological red line, and permanent basic farmland red line), which play an important role in mitigating the ecological problems caused by the shrinking of cropland and the decline of the GEP due to the expansion of construction land. This will contribute to the ultimate goal of regional sustainable development.