Land Use Change and Ecosystem Health Assessment on Shanghai–Hangzhou Bay, Eastern China

Abstract

Reasonable quantitative assessment on urban ecosystem health is conducive to the sustainable development of the economy and human society. This paper quantitatively evaluated the impact of land use change on ecosystem services and ecosystem health by building a comprehensive evaluation system (vigor–organization–resilience–ecosystem services), and then analyzed the spatial-temporal pattern, evolution characteristics, and driving factors in the Shanghai–Hangzhou Bay area (SHB) over the 2000–2015 period. The results show that: the area of cropland and forest accounted for more than 65% and was mainly converted into built-up land in the past 15 years. The overall ESV showed a trend of first increasing and then decreasing. Forest accounted for the largest proportion of the total ESV, more than 60% in each year. The ecosystem health value of SBH decreased from 2000 to 2015. At the city scale, the ecosystem health was significantly deteriorated. All cities reached the lowest value by 2015. At the districts/counties scale, the number with the relatively well or well level decreased from 32 in 2000 to 20 in 2015 by 24.64% of the total area. Overall, inland regions of SBH had better ecosystem health situation than coastal areas. The rapid urbanization of population and economy were driving factors for the decline of the ecosystem health. The indicator system of integrating the vigor, organization, resilience, and ecosystem service for ecosystem health assessment is a potential method which could provide a quantitative and comprehensive way for evaluating ecological and environmental effects in the future.

1. Introduction

To achieve common human well-being and rational natural resource use, the United Nations 2030 Agenda was drafted and the Sustainable Development Goals (SDGs) were issued, including 17 goals and 169 targets [1]. As the object of SDG 15, the terrestrial ecosystem, which is the basic environment for social development and human survival, provides kinds of ecosystem services (ES), such as food supply, material production, climate regulation, and water conservation [2,3,4]. Therefore, terrestrial ecosystems are related to the cycles and development of all ecosystems, and maintaining a healthy ecosystem state is an important way to achieve the sustainable development of the economy and society [5,6,7,8]. However, with the rapid economic and social development of major urban agglomerations of China in recent years, such as the Yangtze River Delta, the Pearl River Delta, and the Beijing–Tianjin–Hebei urban agglomeration, land use has changed dramatically and led to an unprecedented impact on local landscape patterns and urban ecosystem health [9,10]. Presently, researchers and government departments of the whole world pursue the common goal of the harmonious coexistence of human beings and nature, and the way to coordinate between urban development and ecological environment protection is a pressing and difficult issue of common concern [11].

Rapport et al. first proposed the concept of ecosystem health. A healthy urban ecosystem should be able to maintain organizational structure and self-recovery under external pressure and provide resource support and ecological service guarantees for the sustainable development of the economy and human society [12,13]. The urban ecosystem is a huge and complex system composed of multiple interactions. Therefore, establishing evaluation indicators and models is commonly applied to the evaluation of ecosystem health [14]. Costanza argued that a healthy ecosystem could be classically defined as three main indicators: vigor, organization, and resilience, and then the proportional relationship between land use type and economic value of ecosystem services was determined [15,16].

Land use/land cover change not only affects local landscape structure, material circulation, and energy flow but also the ability to provide ecosystem services and the biodiversity that humans depend on [17,18,19]. Therefore, quantitative assessments on land use change could help highlight the essential characteristics of regional land use change and effectively assess the ecological effects [20]. For example, Sharma et al. modeled land use changes and their effects on biodiversity in Central Kalimantan, Indonesia [21]. Woldeyohannes et al. evaluated the effect of land use dynamics on ecosystem services values (ESV) in the Abaya–Chamo basin over the 1985–2050 period [22]. The assessment results of ecosystem health mainly depend on different indicators and models. Indicators covered economic, social, and other ecological attributes, and models mainly included the vigor–organization–resilience, the fuzzy synthetic assessment, the set-pair, and the press–state–response model [23,24,25,26]. This research integrated ideas from the humanities, ecology, social economy, and other fields. However, there did not exist an absolute standard for evaluating the urban ecosystem health due to uncertainty and complexity.

As the key region of the Yangtze River Delta urban agglomeration in China, the Shanghai–Hangzhou Bay area (SHB) has dramatic changes in land use/land cover, which brought many problems to the urban ecosystem health. Therefore, this paper selected SHB as the study case and integrated GIS and remote sensing methods with the comprehensive evaluation systems (vigor–organization–resilience–ecosystem services) to assess the impact of land use change on ecosystem services and health. The main objectives were to: (1) map and investigate the characteristics of land use change based on Landsat images over 2000–2015, (2) evaluate the impact of land use change on the spatial and temporal distribution of ESV and ecosystem health value by building a comprehensive evaluation system, and (3) identify the driving factors affecting ecosystem health change.

2. Materials and Methods

2.1. Study Area

The Shanghai–Hangzhou Bay area (SHB) is located in eastern China and the North Pacific coastal area, covering an area of 52,399 km² between the latitudes of 28°51′ and 31°53′ N and longitudes of 118°21′ and 123°25′ E. The spatial scale of SHB is moderate, which is vital for a detailed and in-depth analysis of the internal components and mechanisms of the ecosystem. It comprises 7 cities (Figure 1), including Shanghai, Jiaxing, Huzhou, Hangzhou, Shaoxing, Ningbo, and Zhoushan and 61 districts or counties (47 are studied). There are connected water systems and ecological corridors between cities, where administrative boundaries are broken in the ecological management. The region is affected by a subtropical monsoon humid climate, with an annual average temperature of 16 °C and annual rainfall over 1100 mm. Hills and mountains dominate the terrain of the SHB (in the southwest and southern parts). Plains are mainly located in the northern parts of the study area. As the key element of the Yangtze River Delta urban agglomeration, the SHB has a population of 54.41 million and a gross domestic product (GDP) of CNY 9441.76 billion. The SHB has developed rapidly in policy, economy, society, culture, and ecology and is characterized by sufficient water resources, abundant vegetation coverage and high land ecosystem services. With superior natural conditions and rapid urbanization development, SHB has become a typical compound ecosystem of both natural and artificial components. It can be regarded as an example to visually identify the profit and loss of the ecological environment during the process of economic development and urbanization by exploring changes in land use and ecosystem health of the SHB in recent years.

2.2. Data Source

2.2.1. Vector and Elevation Data

The China map was derived from the National Standard Map Service platform (NSMS, http://bzdt.ch.mnr.gov.cn/ (accessed on 30 January 2019)). Data of region boundary and elevation were from the Resource and Environmental Sciences and Data Center, China (ESDC, http://www.resdc.cn/Default.aspx (accessed on 30 January 2019)). The spatial resolution of elevation data was 30 m.

2.2.2. Remote Sensing Data

Remote sensing images of Landsat-5 TM and the Landsat-8 OLI in 2000, 2005, 2010, and 2015 were used for extracting different land use types. These data were acquired from the United States Geological Survey platform (https://earthexplorer.usgs.gov/ (accessed on 30 January 2019)) with a spatial resolution of 30 m. SHB required 9 remote sensing images to fully splice the whole study area each year, so we select images Path/Row at 117/39, 118/38, 118/39, 118/40, 119/38, 119/39, 119/40, 120/39, and 120/40. Considering the abundant land feature information, the times of images were selected between June and September in 2000, 2005, 2010, and 2015.

2.2.3. Socio-Economic Data

To evaluate the ecosystem services of the SHB, we collected statistical data on permanent resident population, population density, Gross Domestic Product (GDP), and other information from the National Bureau of Statistics, China (http://www.stats.gov.cn/ (accessed on 1 May 2019)), rice, wheat, corn and soybeans from the Shanghai Municipal Statistics Bureau (https://tjj.sh.gov.cn/sjfb/index.html (accessed on 1 May 2019)) and the Zhejiang Province Bureau of Statistics (http://data.tjj.zj.gov.cn/page/systemmanager/admin/homePage.jsp?orgCode=33 (accessed on 1 May 2019)), and their national mean prices from the Chinese Compilation of Agricultural Costs and Benefits. The times of data were between 2000 and 2016.

2.3. Methods

2.3.1. Land Use Change Analysis

By radiation correction, geometric correction, mosaic, cutting, and other preprocessing operations, we divided the land use types of the study area into 6 categories by using the method of supervised classification from Landsat-5 TM and Landsat-8 OLI images in 2000, 2005, 2010, and 2015. According to the Chinese Classification Standard of Land Use Status (GBT 21010-2017) and the prior study [11], 6 categories of land use types were cropland, forest, grassland, water, unused land, and built-up land (

Table 1. Land use classification system in the SHB.

Land Use Type	Detail
Cropland	paddy field, irrigated land, dry land, etc.
Forest	evergreen broad-leaf forest, deciduous and evergreen broadleaved forest, deciduous broad-leaved forest, etc.
Grassland	mainly herbaceous plants
Water	rivers, lakes, ponds, reservoirs, etc.
Built-up land	commercial service land, residential land, transportation land, etc.
Unused land	coastal beaches, marshes, sand islands, bare land, etc.

). Verification points were randomly selected from high-resolution Google Earth images to verify the accuracy of classification results, and we finally obtained the classification accuracy for 2000, 2005, 2010, and 2015. After classification, works on small patch erasure, majority analysis, minority analysis, clustering, filtering, etc., were carried out in classification–results. The land use dynamics degree can quantitatively describe the change speed of a certain land use type in a certain time and show the quantity change in a certain land use type in a certain time. The single land use dynamic degree was for the change rate of a certain land use type in a certain time interval. Its formula is as follows [27,28]:

$$\mathrm{K}=\frac{{\mathrm{U}}_{\mathrm{b}}-{\mathrm{U}}_{\mathrm{a}}}{{\mathrm{U}}_{\mathrm{a}}\times \mathrm{T}}$$

(1)

where $\mathrm{K}$ is the single land use dynamic degree of a certain land use type in the region (%), ${\mathrm{U}}_{\mathrm{a}}$ and ${\mathrm{U}}_{\mathrm{b}}$ are the area of a certain land type at the beginning and end of the study, respectively (km²), and$\mathrm{T}$ is the length of the study period (year).

The comprehensive land use dynamic degree: To describe the overall change in land use in the SHB during the 2000–2015 period, the comprehensive land use dynamic degree was selected to show the change speed of all land use types in the study area within a certain time interval. The formula is as follows [27,28]:

$$\mathrm{LC}=\left[\frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{LU}}_{\mathrm{i}-\mathrm{j}}}{2{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{LU}}_{\mathrm{i}}}\right]\times \frac{1}{\mathrm{T}}\times 100\%$$

(2)

where $\mathrm{LC}$ is the comprehensive land use dynamic degree (%), ${\mathrm{LU}}_{\mathrm{i}}$ is the area of class $\mathrm{i}$ land use type at the beginning of the study (km²), ${\mathrm{LU}}_{\mathrm{i}-\mathrm{j}}$ is the absolute value of the area of class $\mathrm{i}$ land use type transformed into nonclass $\mathrm{i}$ land use type in the study period (km²), and $\mathrm{T}$ is the length of the study period (year).

2.3.2. Ecosystem Services Valuation

The equivalent coefficient method has been widely used to evaluate ESV due to its easy use and applicability to the terrestrial ecosystem in China. Xie et al. revised the equivalent coefficients of 6 land uses for 9 ecosystem functions based on the research results of Costanza et al. [16]. This equivalent coefficient table reflects an average state of China (values per unit area of ecosystem services in China,

Table 2. The equivalent coefficient table of land use types in China and the SHB (CNY/hm² a).

China a	Land Use Type
Ecosystem Services	Cropland	Forest	Unused Land	Water	Grassland
Gas exchange	442.4	3097	0	0	707.9
Climate regulation	787.5	2389.1	0	407	796.4
Water conservation	530.9	2831.5	26.5	18,033.2	707.9
Soil formation and conservation	1291.9	3450.9	17.7	8.8	1725.5
Waste treatment	1451.2	1159.2	8.8	16,086.6	1159.2
Biodiversity	628.2	2884.6	300.8	2203.3	964.5
Food production	884.9	88.5	8.8	88.5	265.5
Raw material	88.5	2300.6	0	8.8	44.2
Entertainment culture	8.8	1132.6	8.8	3840.2	35.2
Total	6114.3	19334	371.4	40,676.4	6406.5
SHB b	Land Use Type
Ecosystem Services	Cropland	Forest	Unused Land	Water	Grassland
Gas exchange	490.6	3434.6	0.0	0.0	785.1
Climate regulation	873.3	2649.5	0.0	451.4	883.2
Water conservation	588.8	3140.1	29.4	19,998.8	785.1
Soil formation and conservation	1432.7	3827.0	19.6	9.8	1913.6
Waste treatment	1609.4	1285.6	9.8	17,840.0	1285.6
Biodiversity	696.7	3199.0	333.6	2443.5	1069.6
Food production	981.4	98.1	9.8	98.1	294.4
Raw material	98.1	2551.4	0.0	9.8	49.0
Entertainment culture	9.8	1256.1	9.8	4258.8	39.0
Total	6780.8	21,441.4	411.9	45,110.1	7104.8

^a is the equivalent coefficient table of China [29]. ^b is the equivalent coefficient table of the SHB after revising.

), so it needs to be revised according to the specific situation of the SHB. The conversion formula is as follows [29]:

$$\mathsf{\alpha }=\mathrm{M}\times \frac{\mathrm{N}}{7}$$

(3)

$${\mathsf{\beta }}_{\mathrm{i}}=\mathsf{\alpha }\times {\mathsf{\gamma }}_{\mathrm{i}}$$

(4)

where $\mathsf{\alpha }$ is the revised coefficient of ESV, $\mathrm{M}$ is the grain yield per unit area of farmland, $\mathrm{N}$ is the grain unit price during the study period, ${\mathsf{\beta }}_{\mathrm{i}}$ is the equivalent of the ecosystem service after the revision of class $\mathrm{i}$ land use type, and ${\mathsf{\gamma }}_{\mathrm{i}}$ is the equivalent of ecosystem service of t class $\mathrm{i}$ land use type from Xie et al. (the average state in China).

After revising the equivalent coefficient, we obtained the new equivalent coefficient table for the SHB (Table 2). In this study, the ESV of built-up lands of roads, commercial land, residential land, and others was not calculated.

Then, the ESV of each land use type, each ecosystem service, and total value of the SHB were calculated according to the following formulas [4]:

$${\mathrm{ESV}}_{\mathrm{i}}={\displaystyle \sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{A}}_{\mathrm{i}}\times {\mathrm{VC}}_{\mathrm{i},\mathrm{j}}$$

(5)

$${\mathrm{ESV}}_{\mathrm{j}}={\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{A}}_{\mathrm{i}}\times {\mathrm{VC}}_{\mathrm{i},\mathrm{j}}$$

(6)

$$\mathrm{ESV}={\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{n}}({\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{VC}}_{\mathrm{i},\mathrm{j}})\times {\mathrm{A}}_{\mathrm{i}}$$

(7)

where ${\mathrm{ESV}}_{\mathrm{i}}$, ${\mathrm{ESV}}_{\mathrm{j}}$, $\mathrm{ESV},$ are the ESV of class $\mathrm{i}$ land use type, $\mathrm{j}$ ecosystem service, and total ESV, respectively. ${\mathrm{A}}_{\mathrm{i}}$ is the area of class $\mathrm{i}$ land use type. ${\mathrm{VC}}_{\mathrm{i},\mathrm{j}}$ is the unit price of class $\mathrm{i}$ land use type of class $\mathrm{j}$ ecosystem service.

2.3.3. Ecosystem Health Assessment

The ecosystem physical health mainly depends on three indicators: ecosystem vigor, ecosystem organization, and ecosystem resilience. However, this physical health emphasizes the condition of the ecosystem itself, lacking the ecosystem services for spatial entities that meet human health needs [30]. Therefore, based on the ecosystem vigor–organization–resilience system, we added the ESV into the ecosystem health assessment. The formula is as follows [31]:

$$\mathrm{H}=\sqrt{\mathrm{PH}\times \mathrm{ESV}}$$

(8)

where $\mathrm{H}$ is the ecosystem health, $\mathrm{PH}$ is the ecosystem physical health, and $\mathrm{ESV}$ is the ecosystem services values.

$$\mathrm{PH}=\sqrt[3]{\mathrm{V}\times \mathrm{O}\times \mathrm{R}}$$

(9)

where $\mathrm{PH}$ is the ecosystem physical health (the closer it is to 1, the better ecosystem health it is) and $\mathrm{V}$, $\mathrm{O}$, and $\mathrm{R}$ are ecosystem vigor, ecosystem organization, and ecosystem resilience, respectively.

Ecosystem vigor refers to the primary productivity. To reflect the ecosystem vitality in this study, we quantitatively evaluate the change in vegetation coverage caused by land use change based on land use type. According to the Chinese Technical Specifications for Ecological Environment Assessment (HJ/T192-2006), the vegetation coverage index (LC) was computed by setting the weights of each land use type (

Table 3. Relative coefficients for ecosystem vigor, ecosystem resilience, and ecosystem services of each land use type.

Coefficient Type	Cropland	Forest	Grassland	Built-Up Land	Water	Unused Land
LC	0.190	0.380	0.340	0.070	0	0.020
RC	0.300	0.800	0.700	0.200	0.800	1
ESC	0.325	1	0.426	0.015	0.932	0.035

). The formula is as follows [32]:

$$\mathrm{V}={\mathrm{A}}_{\mathrm{VEG}}\times \frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{LA}}_{\mathrm{i}}\times {\mathrm{LC}}_{\mathrm{i}}}{\mathrm{SA}}$$

(10)

where $\mathrm{V}$ is the ecosystem vigor, ${\mathrm{A}}_{\mathrm{VEG}}$ is the vegetation coverage normalization index, ${\mathrm{LA}}_{\mathrm{i}}$ is the area of class $\mathrm{i}$ land use type, ${\mathrm{LC}}_{\mathrm{i}}$ is the ecosystem vigor coefficient of class $i$ land use type, $\mathrm{SA}$ is the total area, and $\mathrm{n}$ is the number of land use type.

Ecosystem organization refers to the structural stability of ecosystems. From the perspective of the landscape, the ecosystem organization is determined by landscape patterns of the landscape heterogeneity and the landscape connectivity [33]. The landscape heterogeneity can be measured by the Shannon diversity index (SHDI) and the area-weighted mean patch fractal dimension (AWMPFD) [34]. The landscape connectivity can be quantified by the landscape fragmentation index (FN1), the landscape contagion index (CONT), the fragmentation index of water (FN2), the patch cohesion of water (COHESION1), the fragmentation index of forest (FN3), and the patch cohesion of forest (COHESION2). The weight setting and formula is as follows [33,35]:

$$\mathrm{O}=0.25\times \mathrm{SHDI}+0.1\times \mathrm{AWMPFD}+0.25\times \mathrm{FN}1+0.1\times \mathrm{CONT}+0.1\times \mathrm{FN}2+0.05\times \mathrm{COHESION}1+0.1\times \mathrm{FN}3+0.05\times \mathrm{COHESION}2$$

(11)

where $\mathrm{O}$ is the ecosystem organization, $\mathrm{SHDI}$ is the Shannon diversity index, $\mathrm{AWMPFD}$ is the area-weighted mean patch fractal dimension, $\mathrm{FN}1$ is the landscape fragmentation index, $\mathrm{CONT}$ is the landscape contagion index, $\mathrm{FN}2$ is the fragmentation index of water, $\mathrm{COHESION}1$ is the cohesion index of water, $\mathrm{FN}3$ is the fragmentation index of forest,$\mathrm{and} \mathrm{COHESION}2$ is the cohesion index of forest.

Ecosystem resilience refers to the ability for natural ecosystems to recover their original structure and functions after external disturbances. Considering the significance of land use to ecosystem resilience, the ecosystem resilience was quantified by the total area-weighted ecosystem resilience coefficients for all land use types [36]. Generally, human-dominated land use types are likely to be less resilient to external pressures, while water and unused land are highly resistant to natural disasters. Combined with the specific situation of the SHB, the resilience coefficients of different land types were set by the evaluating method of expert marking. The formula is as follows [33]:

$$\mathrm{R}={{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{A}}_{\mathrm{i}}\times {\mathrm{RC}}_{\mathrm{i}}$$

(12)

where $\mathrm{R}$ is the ecosystem resilience, ${\mathrm{A}}_{\mathrm{i}}$ is the area ratio of class $\mathrm{i}$ land use type, ${\mathrm{RC}}_{\mathrm{i}}$ is the ecosystem resilience coefficient of class $\mathrm{i}$ land use type (Table 3), and class $\mathrm{n}$ is the number of land use type.

3. Results

3.1. Analysis of Land Use Changes in the SHB from 2000 to 2015

By verifying from points that were randomly selected from high-resolution Google Earth images, the overall accuracy of land use classification in 2000, 2005, 2010, and 2015 all exceed 78%, meeting the needs of follow-up research. From the spatial patterns of land use in the SHB from 2000 to 2015, the area of each land use type had dramatic changes (Figure 2). The cropland and forest were dominant land use types, exceeding 35% over the whole study period. Grassland, water, and unused land shared a smaller proportion, no more than 5% (

Table 4. The area of each land use type in the SHB from 2000 to 2015 (km²).

Land Use Type	2000		2005		2010		2015
	Area	%	Area	%	Area	%	Area	%
Cropland	23,371.02	44.11	19,035.66	35.93	21,807.92	41.15	15,979.74	30.16
Forest	20,890.73	39.43	23,680.78	44.70	20,815.80	39.28	19,189.37	36.22
Grassland	1302.44	2.46	1258.35	2.38	1548.01	2.92	1280.01	2.42
Water	2327.04	4.39	2420.94	4.57	2109.35	4.02	1582.82	2.99
Built-up land	4458.76	8.42	6076.20	11.47	6390.91	12.06	14,592.57	27.54
Unused land	627.55	1.18	505.60	0.95	305.53	0.58	355.01	0.67

). Forest and grassland concentrated in the southwest and southern of the SHB; cropland and built-up land mostly focus on eastern coastal areas.

From 2000 to 2005, area changes of cropland, grassland, and unused land in SHB revealed different levels of decline: cropland showed a significant reduction in the amount of 867.07 km² (reduction rate at 0.037), the decrement of grassland and unused land area was relatively slight, with an annual change of −8.82 km² (0.007) and −24.39 km² (0.039). Built-up land (annual change reached 323.34 km²) and forest (558.01 km²) showed rapid growth (

Table 5. The transition matrix and dynamic changes of land use in the SHB from 2000 to 2015 (km²).

2000/2005	Cropland	Forest	Grassland	Water	Built-Up Land	Unused Land	Area Change	The Dynamic Degree (%)
Cropland	13,931.43	5828.90	249.27	475.98	2719.71	165.73	−867.07	−0.037
Forest	2088.06	16,908.29	991.88	189.52	693.66	19.32	558.01	0.027
Grassland	953.80	316.72	7.72	0.99	4.24	18.97	−8.82	−0.007
Water	378.23	269.69	1.06	1509.78	83.80	84.48	18.79	0.008
Built-up land	1608.21	248.20	7.75	54.33	2529.46	10.80	323.34	0.073
Unused land	75.93	108.97	0.67	190.33	45.33	206.31	−24.39	−0.039
2005/2010	Cropland	Forest	Grassland	Water	Built-Up Land	Unused Land	Area Change	The Dynamic Degree (%)
Cropland	14,415.12	1273.63	517.68	347.64	2448.57	33.03	554.45	0.029
Forest	4787.41	18,013.81	460.40	143.83	269.12	6.21	−573.00	−0.024
Grassland	122.25	1012.64	110.41	3.54	9.29	0.22	57.93	0.046
Water	360.81	210.84	448.32	1228.86	54.91	117.20	−58.00	−0.024
Built-up land	1911.25	262.36	11.02	210.92	3568.19	112.46	63.09	0.010
Unused land	211.08	42.53	0.18	174.57	40.83	36.41	−40.01	−0.079
2010/2015	Cropland	Forest	Grassland	Water	Built-Up Land	Unused Land	Area Change	The Dynamic Degree (%)
Cropland	10,934.20	2454.39	276.69	121.65	7975.90	45.08	−1165.67	−0.053
Forest	3306.48	15,540.63	871.35	149.71	934.55	13.09	−325.29	−0.016
Grassland	641.40	716.31	94.14	45.01	50.97	0.18	−53.60	−0.035
Water	334.95	104.20	3.22	1226.39	310.91	129.69	−109.62	−0.051
Built-up land	710.51	343.04	13.98	25.53	5290.23	7.63	1640.33	0.257
Unused land	50.20	30.81	20.63	14.53	30.01	159.35	9.90	0.032

). The increment of water was unobvious. From 2005 to 2010, forest and water also showed a declining trend with an amount of 573 km² and 58 km²; unused land decreased by 40.01 km². Cropland, grassland, and built-up land increased by 554.45 km², 57.93 km², and 63.09 km², with the dynamic degrees of 0.029%, 0.046%, 0.01%, respectively (Table 5). During the 2010–2015 period, the declination of cropland was the most significant, with an annual average decrease of 1165.67 km². Grassland, forest, and water dropped by 53.60 km², 325.29 km², and 109.62 km². However, built-up land increased dramatically in all cities (reaching 1640.33 km²), especially in Shanghai. The unused land expanded slowly.

According to the transition matrix of land use over the past 15 years (Table 5), large areas of forest, grassland, and water had been converted into cropland. There were two rapid area increase stages in the SHB: between 2000 and 2005, the period for the start of the development of urbanization in cities; many roads were built up and led to the cropland around towns being replaced by land for industrial and commercial use or road construction. Between 2010 and 2015, the period for the rapid development of urbanization, the original housing area was far from meeting the demand caused by the rural population pouring into cities. Moreover, the surge of urban residential area and the relocation of the original urban industry to the surrounding cities and towns resulted in the area increase in built-up land, mainly by occupying cropland and forest.

3.2. Changes in Ecosystem Services Values from 2000 to 2015

Based on the area of land use type and the equivalent coefficient table of the SHB (Table 2), we calculated the ESV of different land use and cities in four periods by Equations (3)–(7). The total ESV of the SHB decreased from 2000 to 2015, with a gain of about CNY 72, 76, 70, 60 billion in 2000, 2005, 2010, 2015, respectively (

Table 6. ESV of each land use type and city in the SHB from 2000 to 2015 (10⁹ CNY).

Land Use Type	2000	%	2005	%	2010	%	2015	%
Cropland	15.847	22.07	12.908	17.09	14.787	21.11	10.835	18.05
Forest	44.364	61.79	50.775	67.24	44.632	63.72	41.145	68.52
Grassland	1.067	1.49	0.894	1.18	1.100	1.57	0.909	1.51
Water	10.497	14.62	10.921	14.46	9.515	13.58	7.140	11.89
Unused land	0.026	0.04	0.021	0.03	0.013	0.02	0.015	0.02
Total	71.802	100	75.518	100	70.047	100	60.044	100
City	2000	%	2005	%	2010	%	2015	%
Hangzhou	29.765	41.45	32.595	43.16	29.669	42.38	28.517	47.49
Huzhou	7.481	10.42	7.159	9.48	6.887	9.84	6.072	10.11
Jiaxing	3.826	5.33	3.734	4.94	3.679	5.26	2.557	4.26
Ningbo	11.116	15.48	11.553	15.30	10.785	15.41	9.074	15.11
Shanghai	4.750	6.62	4.669	6.18	4.290	6.13	2.602	4.33
Shaoxing	12.897	17.96	13.721	18.17	12.754	18.22	9.453	15.74
Zhoushan	1.968	2.74	2.086	2.76	1.942	2.77	1.773	2.95
Total	71.803	100	75.517	100	70.006	100	60.048	100

). It showed a downward trend. In terms of land use types, forest, cropland, and water provided most of the ESV, which contributed more than 90% of the total. The ESV of forest was the largest. Temporally, cropland, water, and unused land exhibited decreasing trends from 2000 to 2015, while forest and grassland showed clearly increasing trends. For cropland and grassland, their ESV decreased slightly from 2000 to 2005 but increased slightly from 2005 to 2010 and decreased slightly from 2010 to 2015 again. For forest and water, their ESV increased slightly from 2000 to 2005 but decreased slightly from 2005 to 2015. Moreover, the unused land decreased slightly from 2000 to 2010 but increased slightly from 2010 to 2015. Among seven cities, Hangzhou City is the top ESV in the SHB, contributing more than 41% of the total. Zhoushan City was the smallest of the total ESV. The ESV ranked: Hangzhou > Shaoxing > Ningbo > Huzhou > Shanghai > Jiaxing Zhoushan.

From 2000 to 2005 in the SHB, the increase in forest area was obvious with the ESV increasement of CNY 6.391 billion, mainly because of the implementation of the national policy of returning farmland to forest, soil, and water conservation after 2000. The forest area began to grow and reached the maximum value in 2005, and the value of ecosystem services generally showed an increased state. From 2005 to 2010, the built-up land area increased rapidly for the first time; the main reason was that a lot of housing and roads were put up because of the rapid development of urbanization in this period. So, industrial land, road construction land, and residential land occupied a large amount of cropland and forest, and the ESV had decreased in each city. Between 2010 and 2015, the built-up land area increased rapidly again. With the rapid development of urbanization and the continuous migration of original urban industries to surrounding towns, a large amount of forest and grassland were occupied by built-up land, resulting in the overall decrease in ESV in the SHB.

3.3. Ecosystem Health Variation and Assessment Results in the SHB

3.3.1. Spatial-Temporal Variation of Ecosystem Health Value

By the improved indicator system of ecosystem health assessment, the ecosystem health value was calculated by Equations (8)–(12), including indicators of the ecosystem vigor, ecosystem organization, ecosystem resilience, and ESV. The result of the ecosystem health assessment was the relative value. Ecosystem health values of the SHB were approximately the normal distribution from 2000 to 2015, and they could be symmetrically divided according to prior research [33,35]. The mean value of ecosystem health in the SHB is 0.59 from 2000 to 2015, and values are generally between 0.4 and 0.7, rarely lower than 0.4. Therefore, we set 0.50–0.59 as the ordinary condition, and ecosystem health values were divided into five levels: 0–0.39 (weak), 0.40–0.49 (relatively weak), 0.50–0.59 (ordinary), 0.60–0.69 (relatively well), and 0.70–1 (well).

• At the city scale.

The ecosystem health value of each city was declining from 2000 to 2015 (Figure 3). For Hangzhou City, the value was decreasing at first, increasing slightly next, and then decreasing again. Jiaxing City showed a downtrend with the same value in 2005 and 2010. Huzhou, Shanghai, Shaoxing, Ningbo, and Zhoushan City went straight down from 2000 to 2015. At the city scale, there was a significant deterioration in the level of ecosystem health in the SHB from 2000 to 2015 (Figure 3). For Shanghai City, its health deterioration was the most obvious, from an ordinary level in 2000 to a relatively weak level in 2005 and 2010, and finally to a weak level in 2015. Followed by Jiaxing City, its health level had changed from ordinary in 2000 to relatively weak in 2015. However, Hangzhou and Zhoushan City had maintained well or relatively well levels of ecosystem health. The area of well and relatively well level decreased by 44.95%, and relatively weak and weak level increased by 21.06%, showing that rapid urbanization posed a serious threat to the balance of the ecosystem of the surrounding areas of cities, which adversely caused a downward trend in the overall urban ecosystem health of the SHB.

The period from 2010 to 2015 could be regarded as the turning point of the deterioration in ecosystem health, which the area of built-up land was increasing significantly. With the acceleration of unhealthy urbanization processes and the large-scale influx of rural populations, the built-up land area rose significantly, caused by the continuous migration of the original urban industry to the surrounding cities and the increase in urban residences. Furthermore, areas of forest, cropland, grassland, and water were decreasing. The balance of the whole urban ecosystem was broken, leading to the deterioration of the ecosystem health in the SHB.

2.

At the district/county scale

From 2000 to 2015, the ecosystem health value in all regions showed a decreasing trend, but the reduction rate was different (Figure 3). There were 16 districts or counties higher than 0.7 (well level) in 2000, whose land use types mostly were the forest, so their ecosystem was less affected by human activities. Up to 2015, there were only six districts or counties higher than 0.7. Except for Zhoushan and Jiaxing City, the values of urban areas in Shanghai, Hangzhou, Huzhou, Shaoxing, and Ningbo were all lower than other districts or counties within the corresponding administrative area. The urban areas of each city were always the political and economic center, so the rapid economic development resulted in a strong attraction to immigration. However, large areas of cropland, forest, and water were occupied by the expanding construction land, leading to an ecosystem imbalance in the city. Therefore, the ecosystem health values in urban areas were lower than those in other counties in the same city.

At the district/county scale, the level of ecosystem health also deteriorated obviously. For Shanghai City, the area of weak and relatively weak level changed from 2.79% in 2000 to 69.7% in 2015. From 2000 to 2015, the rapid urbanization promoted a large number of people to flood into various districts and counties in Shanghai. In addition, most cropland and forest were occupied by built-up land, which posed a serious threat to the balance of the urban ecosystem and caused a significant deterioration of the ecosystem health. From 2010 to 2015, the ordinary, relatively weak, and weak levels of ecosystem health began to move slowly to Hangzhou, Huzhou, Shaoxing, and Ningbo City.

Spatially, taking Tongxiang as the demarcation line, the health levels of well and relatively well were mainly located in the southwest and south districts or counties of the SHB from 2000 to 2005, and the same level districts or counties were mainly located in the southwestern of Huzhou and Hangzhou, the south of Shaoxing and Ningbo, and Zhoushan City from 2010 to 2015 (Figure 3). From 2000 to 2015, only Anji, Chun’an, Zhuji, Xinchang, Putuo, Daishan, and Shengsi had reached the relatively well level. The level of economic development in these counties was at a backward level compared with other counties in the same city and time. Urban areas with smaller populations were mainly occupied by forest, where the human activity was not so violent as to exert serious effects on the natural ecosystem. Even if there was, the degree of damage to the ecosystem was lighter than other counties.

3.3.2. Driving Factors of Ecosystem Health Changes

By exploring the correlation between the built-up land area, the population density, GDP, and the ecosystem health value in the SHB, we found that they are significantly negatively correlated using the correlation analysis through the software SPSS.

From Figure 4a,b, the health value in most districts and counties is negatively correlated with the built-up land and the population density. The faster the economy developed in the SHB, the more workers would be attracted, which would also lead to the increase in built-up land, resulting in the destruction of the ecological environment. Except for the Pudong District, the urban areas of Shanghai and Hangzhou City, the ecosystem health value in most regions is negatively correlated with GDP (Figure 4c). As the economic center of China, Shanghai led the country in GDP level. Its economic industry was mainly in the tertiary industry, which resulted in a big gap in GDP compared with other districts. In addition, as the provincial capital of Zhejiang Province, Hangzhou City was followed by Shanghai City in economic development in the SHB. It had a strong economic radiation effect on the surrounding districts and counties. Therefore, the rapid urbanization of population, economy, and land were the driving factors for the decline of the ecosystem health value in the SHB.

3.3.3. Influence of Ecosystem Health Indicators

The value change in ecosystem health indicators had a certain impact on results (Figure 4d). We used the vegetation coverage index to reflect the ecosystem vigor. Overall, the vegetation coverage index showed a downward trend from 2000 to 2015. Remote sensing images collected in this study were all in the same season. So, a difference in vegetation area caused by different seasons could not be considered as the main reason for this index, but different types of land use can be taken into consideration. From Table 4, the total reduction area of forest and grassland was about 1700 km² during the 2000–2015 period, and the increased area of built-up land was about 10,000 km². With the increase of built-up land area, the vegetation coverage index decreased slightly, and the ecosystem health also showed a falling down trend. Therefore, rapid urbanization was one of the causes of the deterioration of ecosystem health. However, the highest value of ecosystem vigor occurred in 2000 and that of ecosystem health appeared in 2005. From this aspect, it could be inferred that the contribution of ecosystem vigor to urban ecosystem health was insignificant.

The ecosystem organization increased first and then decreased with a little fluctuation. This basic stability indicated that the overall landscape heterogeneity and connectivity in the SHB did not change significantly during the 2000–2015 period. Due to the water area of the SHB being relatively small, it could be deduced that the ratio of forest and built-up land area was an important factor affecting the ecosystem resilience. The overall ecosystem resilience changed greatly from 2000 to 2015, with the highest in 2005 and the lowest in 2015. A large amount of land was turned into built-up land, and the negative effects of urbanization on ecosystem resilience were gradually exposed. From Table 3, the resilience coefficient of built-up land was much lower than other land types, slightly higher than the cropland. According to statistical yearbooks in recent years, the total registered population of the SHB was 39.07 million, accounting for one third of the population of the Yangtze River Delta urban agglomeration by the end of 2015. Therefore, the human activity level was relatively high, which would have an extensive effect on results.

The ESV of the SHB declined from 2000 to 2015. The forest ecosystem is of great significance in urban landscapes and is known as the most valuable ecosystem service type of land use [36]. From Table 3, the proportion of forest and water area was one of the important factors affecting ecosystem services. The largest proportion of forest and water area appeared in 2005, and the smallest appeared in 2015. This trend was the same for the ESV. While the built-up land had increased largely in 2015, the rapid urbanization did not necessarily lead to a large loss of ESV. If different land use types are planned reasonably in a certain area, the adjacency effect can promote ecosystem services. Overall, change characteristics between ESV and ecosystem health value were consistent during this period.

4. Discussion

Due to a dramatic change in land use type, the ecosystem health situation of Shanghai and Jiaxing City are worse. A large amount of cropland, forest, and water area have been converted into built-up land, and the disturbance of human activities has been significant. Therefore, their ecosystem health has declined sharply in the past 15 years. This study reflects more objective evaluation results and is consistent with Ou’s research [35]. From the overall assessment result, there was a turning point of ecosystem health in 2005, suggesting that the Eleventh Five-Year Plan (2005–2010) of China has indeed played a guiding role in ecological protection. This result is consistent with some studies [11], and many researchers argue that national policy guidance has played a significant role in ecosystem health improvement [37,38,39].

In this study, we found that change characteristics between ESV and ecosystem health value were consistent, and the proportion of forest and water area was one of the important factors affecting ecosystem services. Due to high ESV per unit area in forest and water, the slight degradation of their area can cause obvious losses of total ESV. Therefore, in order to promote ESV, decision makers should pay more attention to the protection of the wetland ecosystem and avoid encroachment by built-up land and cropland [4]. Moreover, cropland can bring in a good deal of ESV. Policy departments only consider the production values of cropland in land use planning but ignore their ESV, which will lead to considerable losses of total ESV [40].

Although much effort has been made, there are still some unavoidable limitations in the evaluation of ecosystem health. Some researchers have previously argued that human factors have a more significant impact than natural elements on the relationship between man and land [26,41]. However, we did not specifically discuss the impact of urbanization in this study because the relationship between urbanization and ecological environment is an open and complex giant system involving many elements. The urbanization system includes many subsystems such as population, economy, society, information, infrastructure, etc. [42]. Many studies have selected more relevant indicators to refer to these elements, such as urban and rural poverty headcount [43], access to public transport [44], water quality [45], etc. These selected indicators have policy applicability.

In addition, this study used land use reclassification results to calculate values of each evaluation index, so the result of the ecosystem health assessment was affected by the land use data. Due to the limitation of the spatial resolution of remote sensing images, there is inevitable uncertainty in the interpretation of images, which would have some influence on the results. Moreover, this paper selected the vegetation coverage index based on land use type as the indicator of ecosystem vigor, but the vegetation coverage index could not completely characterize the ecosystem vigor. The use of weights based on the expert scoring method in the calculation of ecosystem resilience and services would result in higher subjectivity. In addition, the impact mechanism of land use change on ecosystem health was difficult to determine.

With the rapid development of science and technology, methods of data acquisition have greatly increased. At present, the most significant limitation of urban ecosystem health assessment is not the lack of data but how to integrate multiple data to construct the new evaluation model. Simultaneously, global ecological problems such as forest degradation, watershed pollution, and climate warming are mainly solved by a specific single discipline [46]. The answer to many ecological problems such as environmental degradation, species protection, and human survival in a more comprehensive way could start from the interdisciplinary perspective of integrated ecosystem health.

5. Conclusions

This study used the Shanghai–Hangzhou Bay area of China as a study case, including 7 cities and 47 districts/counties. Based on Landsat remote sensing images and the GIS platform, we mapped and investigated the dynamic characteristics of land use change over the 2000–2015 period. According to land use change, we quantitatively evaluated the impact of land use change on ecosystem services values and ecosystem health values by building the comprehensive evaluation system (vigor–organization–resilience–ecosystem services). The spatial-temporal pattern, evolution characteristics, and driving factors of assessment results were analyzed and explored.

The main conclusions are as follows:

The land use/land cover types in the SHB were rich, areas of cropland and forest accounted for more than 65%. Cropland, forest, grassland, unused land, and water area decreased, while built-up land increased from 2000 to 2015. Moreover, cropland and forest were mainly converted into built-up land. With the rapid development of the economy, built-up land had occupied a large amount of cropland and forest.

The overall ESV showed a trend of first increasing and then decreasing. The ESV of forest accounted for a large proportion, more than 60% in each year. The EVS in Hangzhou, Ningbo, Shaoxing, and Zhoushan City increased first and then decreased, while in Shanghai, Jiaxing, and Huzhou City continued to decrease in the past 15 years.

The ecosystem health value of the SBH decreased from 2000 to 2015. In cities, the ecosystem health status was significantly deteriorated. It mainly concentrated in 0.5~0.7, and all cities reached the lowest value by 2015. In districts or counties, the number with the relatively well or well level decreased from 32 in 2000 to 20 in 2015 by 24.64% of the total area. In contrast, the number with relatively weak or weak level increased from 1 in 2000 to 13 in 2015 by 16% of the total area. Overall, inland regions of the SBH had a better ecosystem health situation than coastal areas in the past 15 years.

Ecosystem health is the fundamental guarantee for the sustainable development of the human economy and society. Ecosystem health assessment is the basis for exploring the ecological effects of global environmental change. This study established a system (vigor–organization–resilience–service) to evaluate the ecosystem health. It not only emphasizes the physical health of an ecosystem itself but this study also reflects ecosystem services for spatial entities that meet human health needs. Landscape patterns and ecosystem services as indicators can provide a quantitative method for evaluating the ecological effect from the perspective of land use change in the future.