Data collection
Selection of the case cities and city functional components
To make the research results more universal, we set the criteria for the selection of case cities as follows. (1) Large cities: cities in which the built-up area exceeded 1000 km2. We chose Beijing, Shanghai, and Tianjin. Beijing is China’s capital and political centre, Shanghai is China’s largest economic centre, and Tianjin is one of China’s four municipalities directly governed by the Central Government; (2) medium cities: cities in which the built-up area varied between 400 and 1000 km2. We chose two provincial capital cities in central China, Wuhan and Hefei, and an economically developed coastal city, Ningbo; (3) small cities: cities in which the built-up area was smaller than 400 km2. Small cities need to have a complete urban form and functions. We selected three economically developed small cities Changzhou, Nantong and Jiaxing.
The selection of city functional component types should cover typical city functional components related to the coupling between humans and the city in urban systems, including production, processing, circulation, decomposition and other functions: Kentucky Fried Chicken (KFC) and McDonald’s (McD), two of the most popular western fast-food restaurants in China; Lanzhou Noodles (LZN) and Shaxian Snacks (SXS), two of the most popular Chinese fast-food restaurants in China; Agricultural Bank of China (ABC), one of the four most widely distributed banks in China; swimming pool (SP), a type of indoor sports venue popular in recent years; Shunfeng (SF) and Shentong (STO) express outlets, two of the most commonly used express service components in China; China National Petroleum Corporation (CNPC) and China Petroleum and Chemical Corporation (Sinopec) gas stations, two gas station enterprises accounting for more than half of the total number of gas stations in China; WTP, a type of waste treatment component; GH, a type of primary biological production component; and DF, a type of secondary biological processing component.
Acquisition of city functional component data
Latitude and longitude data of the above city functional components were obtained through electronic maps and remote sensing images and verified through field investigation. AutoNavi and Baidu electronic maps are the two most widely used map suppliers in China due to their high accuracy and practicality46. In particular, the location of service city functional components can be accurately obtained through electronic maps. WTPs have detailed lists and location data on the government websites, and GHs can be accurately identified in Google Earth images due to their unique appearance31. Therefore, these three types of raw data are listed as the main sources of location data for functional components.
Latitude and longitude data of the KFC, McD, ABC, SP, LZN, SXS, SF, STO, CNPC, Sinopec and DF locations were retrieved from AutoNavi and Baidu historical electronic maps through Python 3.5 software (https://www.python.org/). The 2012 and 2015 historical electronic map data originated from the East China Normal University Humanities and Social Sciences Big Data Platform47, and the 2018 historical electronic map data originated from the Peking University Open Research Data Platform48. Based on AutoNavi and Baidu, each individual component was strictly filtered by name and type. Please refer to the Supplementary Table S3 for a summary of the detailed filtering conditions.
Accurate WTP latitude and longitude data were obtained by using the WTP name and address to query the AutoNavi map coordinate picking system [the WTP name and address were acquired from the Ministry of Ecology and Environment of the People’s Republic of China (www.mee.gov.cn), China Environment Network (www.cenews.com.cn) and Beijing Municipal Ecology and Environment Bureau (sthjj.beijing.gov.cn)]. GH latitude and longitude data were determined via a method commonly used in community ecology, which has previously been reported31. Briefly, ArcGIS 10.3 software was employed to generate grids covering the entire city (the size of each grid was 0.5 × 0.5 km), and these grids were then converted into the keyhole markup language (KML) format and imported into Google Earth for GH visual interpretation. The GHs were characterized as (a) bright white or bluish-white, (b) rectangular-shaped objects, (c) oriented in rows or separated by paths or bare areas. If a GH occurred in a specific grid, the centre of the grid was marked with the landmark tool to obtain the corresponding latitude and longitude data.
Land price and housing price are affected by location factors such as population, employment, transportation, and amenities and are important indicators to determine whether a city is monocentric or polycentric49,50. Land price was also used as a determining indicator in our study. The concentric circle model was first established by Von Thünen51 to study the order of agricultural land use from urban to rural areas, and it is still an important method to explore research topics along the urban–rural gradient32,52.
To obtain the land price distribution curve along the urban–rural gradient, all the standard land parcel information in each case city through the real-time land price query function provided by the China Land Price Information Service Platform (www.landvalue.com.cn), including land price, latitude and longitude, was obtained, and the parcel with highest land price was defined as the city centre. Concentric circles with an increasing radius of 1-km intervals were generated by adopting the city centre as the circle centre, and the average land price of all standard land parcels in each concentric ring was considered as the land price of the ring. We found that in all the case cities, the land price exhibited an obvious monotonous downward trend from the centre to the edge of the city (Supplementary Fig. S7). Therefore, we assumed a monocentric city model and used the concentric circles to define the urban–rural gradient.
To acquire density distribution curves of the city functional components along urban–rural gradients, the latitude and longitude data of the KFC, McD, ABC, SP, LZN, SXS, SF, STO, CNPC, Sinopec, GH, WTP and DF components were applied for map labelling purposes. Concentric circles with the increasing radius of 1-km intervals were generated by adopting the city centre as the circle centre, and the number of each type of component in each concentric ring was counted. Since the overall number of WTPs and DFs was smaller, the concentric circle radius was increased at 5- and 10-km intervals, respectively, and the number of WTPs or DFs in each concentric ring was determined, while the component density in each ring was calculated by dividing the number by the area of the ring.
To calculate the ecosystem services per unit area for each type of city functional component, the revenue of each component in the current year was determined. KFC and McD revenue data were retrieved from Yum China Holdings and Askci Corporation, respectively. ABC revenue data originated from the Agricultural Bank of China, Ltd., and SF and STO revenue data were acquired from SF Holding Corporation, Ltd., and STO Express Corporation, Ltd., respectively, while CNPC and Sinopec revenue data were retrieved from PetroChina Company, Ltd., and Sinopec Corporation, respectively. Moreover, LZN and SXS revenue data were obtained via field investigation. Environmental impact data of the KFC, McD, CNPC and Sinopec components originated from the Ministry of Ecology and Environment of the People’s Republic of China (www.mee.gov.cn), while LZN and SXS environmental impact data were obtained via field investigation. The costs of the KFC, McD, LZN, SXS, CNPC and Sinopec environmental impacts were converted according to the Environmental Protection Tax Law, 2018. The WTP ecosystem services were retrieved from Liu et al.53, and the GH ecosystem services originated from Chang et al.54, while the DF ecosystem services were obtained from Fan et al.55. The cultural services of all components were determined through field investigation.
Data processing
To intuitively describe the density changes of city functional components along the urban–rural gradient, the density of the components in the above concentric rings were adopted as the ordinate, the distance from the city centre to the edge of the ring was adopted as the abscissa, and scatter plots were created. To compare the characteristic values of the density distribution of each type of component more clearly, a distribution model was used to fit the scatter plots35,36.
Fitting of the density distribution curve of the city functional components
Through the nonlinear fitting function in OriginPro 2019 software (https://www.originlab.com/), the Gumbel model56,57 was considered to fit the above scatter plots to generate density distribution curves of all city functional components. The goodness-of-fit (choosing the 13 types of components in Beijing as examples) is shown in the Supplementary Fig. S2.
The component density (P, individual components km−2) at a given distance from the city centre (d, km) along the urban–rural gradient is calculated as follows:
$${P} = {P_{max}} {cdot} {{e^{-{e}}}^{-frac{{{d}}-{d^{*}}}{{w}} , – , frac{{{d}}-{d^{*}}}{{w}} , + , {1}}}$$
(1)
where Pmax (individual components km−2) is the peak value of the curve, d* (km) is the peak position of the curve, and w (km) is a parameter controlling the width of the curve.
Calculation of the niche width of the density distribution curve of the city functional components
To intuitively compare the distance spanned by the density distribution curve of the city functional components, the difference in the abscissa between a density value of 10% of the Pmax value on the density distribution curve was adopted as the niche width W (km).
Calculation of the skewness and kurtosis of the density distribution curve of the city functional components
The skewness and kurtosis are calculated according to the following equation58:
$$text{skewness } = frac{frac{1}{{{n}}}{sum }_{{{i}}= {1}}^{{n}} ,{left({{x}}_{{i}}-{bar{x}}right)}^{3}}{{left(frac{1}{{{n}}}{sum}_{{{i}}= {1} }^{{n}} ,{left({{x}}_{{i}}-{bar{x}}right)}^{2}right)}^{frac{3}{{2}}}}$$
(2)
$$text{kurtosis } = frac{frac{1}{{{n}}}{sum }_{{{i}}= {1}}^{{n}} ,{left({{x}}_{{i}}-{bar{x}}right)}^{4}}{left(frac{1}{n}sumnolimits_{i=1}^{n} ,{left({{x}}_{{i}}-{bar{x}}right)}^{2}right)^2}-3$$
(3)
where xi (km) is the distance from each individual type of component to the city centre, and ‾x (km) is the average of the distances from all individual types of components to the city centre.
Correlation analysis between the characteristic values of the density distribution curve
Linear and nonlinear regression analyses in Microsoft Excel 2019 were implemented to study the relationship between the characteristic values of the density distribution curve, and the regression form with the best R2 value was selected.
Correlation analysis between the characteristic values of the density distribution curve and the city size
Linear and nonlinear regression analyses in Microsoft Excel 2019 were implemented to study the relationship between the characteristic values of the density distribution curve and the city size, and the regression form with the best R2 value was selected.
Framework for ecosystem service assessment of the city functional components
According to the classification of the Millennium Ecosystem Assessment (MA), ecosystem services include provisioning, regulating, cultural and supporting services59. In this study, the ecosystem services (goods and services) provided by the city functional components (artificial ecosystems) were divided into target and accompanied services (Supplementary Fig. S6), both of which may include provisioning, regulating and cultural services.
In this study, the target services of the KFC, McD, LZN, SXS, CNPC, Sinopec, GH, and DF components were provisioning services, the target services of the ABC, SF, STO, and WTP components were regulating services, and the target services of component SP were cultural services. According to the guidance of Liu et al.53, the above regulating and cultural services were divided into positive and negative services (dis-services).
The net service (NES, USD m−2 yr−1) is the sum of the positive services (target services + positive regulating services + positive cultural services) and dis-services (negative regulating services + negative cultural services):
$${NES} = sum_{{i} = 1}^{n}{ES}_{i}$$
(4)
where ESi (USD m−2 yr−1) is the value of a given type of ecosystem service involved in this study, and n is the number of ecosystem service types involved in this study.
The ecological index (γ) is calculated as follows:
$${gamma } = {TGS}/ |EDS|$$
(5)
where TGS (USD m−2 yr−1) denotes the target services of the city functional components, and EDS (USD m−2 yr−1) denotes the dis-services of the city functional components.
Calculation of the ecosystem services of the city functional components
The calculation methods are provided in the supplementary materials.
Source: Ecology - nature.com
