Introduction

Urbanization and CESs

As one of the most pervasive global trends, urbanization involves population migration, urban land expansion and economic development (Yuan et al. 2018; Elmqvist et al. 2021). At present, 56% of the world’s population live in urban areas, and the figure is projected to rise to 68% by 2050 (United Nations 2022). Such rapid urbanization presents significant challenges to the sustainability of the global environment, including changes in ecosystems and their capacity to provide ecosystem services (Peng et al. 2020; Plieninger et al. 2022). Ecosystem services (ESs) refer to direct and indirect benefits provided by ecosystems for human well-being (de Groot et al. 2010; Costanza et al. 2017). They are divided into four main categories in the Millennium Ecosystem Assessment (2005): provisioning, regulating, supporting and cultural services. Among them, cultural ecosystem services (CESs)—defined as the contributions of ecosystems to nonmaterial benefits (Millennium Ecosystem Assessment 2005; Chan et al. 2012)—play a critical role in rapidly urbanizing landscapes (Kremer et al. 2016). Specifically, urbanization has led to residents’ less reliance on local ecosystems to deliver provisioning and regulating ESs but to a greater appreciation of specific CESs, such as recreation, cultural heritage, and education (La Rosa et al. 2016; Richards et al. 2020). This is because CESs can be easily understood and directly experienced in urban contexts (Andersson et al. 2015). Therefore, CESs are increasingly recognized as important for sustainable landscape management and planning to enhance residents’ well-being in rapidly urbanizing landscapes (Dou et al. 2017; Chen et al. 2019; Gould et al. 2019).

CES supply, demand, and SWB

A critical challenge for CES management against the background of urbanization is to investigate the links between CESs and subjective well-being (SWB; Bieling et al. 2014; Huynh et al. 2022; Nowak-Olejnik et al. 2022), particularly by taking a comprehensive look at the supply and demand sides of CESs (Chen et al. 2019; Kalinauskas et al. 2022). The CES–SWB relationship in changing landscapes is a major topic of landscape sustainability science (Wu 2013, 2021). SWB is an important dimension of human well-being, focusing on people’s life satisfaction and feelings about their circumstances (Summers et al. 2012; King et al. 2014; Aguado et al. 2018). Although the link between CESs and SWB is generally less direct than provisioning and regulating ESs, the low mediation potential of CESs means it is difficult to replace them to meet SWB (Millennium Ecosystem Assessment 2005; Plieninger et al. 2013). The links between CES and SWB are more immediate against the background of urbanization, as the urban environment is strongly shaped by human preferences and social processes (Andersson et al. 2015). Moreover, urbanization can be reflected in urban expansion and consumption of natural resources, but it also brings changes in people’s lifestyles and awareness of ecosystem protection, which leads to changes in the perceived supply and demand of CESs (Peng et al. 2017; Zhou et al. 2018; Meng et al. 2020). Thus, the supply–demand relationship of CES should be considered in studying the link between CES and SWB.

Currently, some hypotheses on socioeconomic development and environmental transformation have been used to understand the relationships between urbanization, ES supply, demand, and SWB. One example is the “green-loop to red-loop transition” model proposed by Cumming et al. (2014). This conceptual model assumes that with urbanization and socioeconomic development, social–ecological systems will shift from a dependence on local ESs to a dependence on local non-ESs (i.e., socioeconomic services) and remote ESs. This model has been used to understand the relationship between ES and well-being in areas with different levels of socioeconomic development (Hamann et al. 2016; Liu et al. 2022), yet their explanations for CESs were unclear. A more systematic assessment and comprehensive understanding of the complex CES–SWB relationship in the context of urbanization is needed. Thus, the present study seeks to adopt an integrative framework to understand human–ecosystem interactions in rapidly urbanizing areas through the lenses of CES and SWB.

Previous research

Despite the importance of CESs in landscape management, research on them in the context of urbanization is scarcer than that on provisioning and regulating ESs (Fan et al. 2022). Currently, most studies dealing with the relationship between urbanization and ESs have only considered CESs as a minor part of many ESs, focusing mainly on recreation and aesthetic CESs that can be more easily assessed (Yuan et al. 2018; Zhang et al. 2018; Zhou et al. 2018). A few studies have explored the association between urbanization and a broader range of CESs. For example, Jaligot et al. (2018) found that the provision of multiple CESs has decreased due to urbanization in peripheral areas. Dou et al. (2017), on the contrary, revealed a greater contribution of CES in more densely populated urban areas due to the additional value of scarce natural resources. Wang et al. (2021b) compared villages at different urbanization stages and observed that land development in the initial stages of urbanization degraded CESs, while subsequent spatial planning and landscape design facilitated the reconstruction of the CESs. Residents’ perceptions of multiple CESs have also shown different responses to urbanization, thus exhibiting varying distributions along the urban–peri-urban–rural gradient (Rall et al. 2017; Riechers et al. 2019).

Regarding the studies that have assessed the supply and demand of CESs, several have been conducted based on the widely used cascade model proposed by Haines-Young and Potschin (2010). These studies have adopted an interdisciplinary methodology by applying biophysical approaches to quantify supply while using socio-cultural approaches to quantify social demand on ESs (Castro et al. 2014; Martín-López et al. 2014; Quintas-Soriano et al. 2019) and CESs (Arbieu et al. 2017; Shi et al. 2020; Crouzat et al. 2022). A few studies, on the other hand, have depicted supply and demand from people’s perceptions (Zoderer et al. 2019; Khosravi Mashizi and Sharafatmandrad 2021), using socio-cultural and participatory approaches, such as interviews, questionnaires, and public participation geographic information systems (PPGIS; Cheng et al. 2019). Surprisingly, although CES has gradually become an active research topic in recent years, SWB, which is linked to CES, has been largely overlooked (Wang et al. 2021a). Several case studies have explored the relationship between perceived CES and SWB (Ciftcioglu 2017; Aguado et al. 2018; Zhang et al. 2022a) but neglected the link between the supply–demand relationship of CES and SWB. Some available studies have attempted to connect the supply and demand of ESs with SWB (Wei et al. 2018; Ketema et al. 2021; Khosravi Mashizi and Sharafatmandrad 2021; Tang et al. 2023) but rarely integrated CESs, with only a few exceptions (Zhang et al. 2022b). Furthermore, these studies on the links between CESs and well-being have focused more on rural landscapes (Xie et al. 2022). In comparison, urban and peri-urban landscapes still need more research attention (Kosanic and Petzold 2020).

The present study aims to fill these gaps by developing an integrated framework to comprehensively identify and uncover the interactions between urbanization level, residents’ perceived supply and demand of CESs, and SWB. The study was conducted in Qingpu District, a peri-urban area located in Shanghai, China. Peri-urban areas often have rural/semi-natural landscapes while being strongly influenced by urban development, thus becoming key areas for studying the relationship between urbanization and ESs (Peng et al. 2017; Zhu et al. 2017). Specifically, our study aims to: (1) quantify residents’ perceptions of the supply and demand of CESs and SWB in this area, (2) reveal the relationship between urbanization and CES as well as SWB, and (3) identify the links between the supply–demand of CES and SWB under different urbanization levels. On the basis of these analyses, we seek possible explanations for the results and discuss relevant policy suggestions for landscape sustainability in the context of rapid urbanization.

Study area

Qingpu District is a peri-urban area in western Shanghai, China (Fig. 1). It has an area of 668.54 km2 and is divided into three subdistricts and eight towns, consisting of 184 administrative villages and 88 communities. Located in the Yangtze River Delta, it has a subtropical maritime monsoon climate with an annual precipitation of 1423.1 mm and an average temperature of 17.6 °C. Qingpu is the source of the ancient civilization of Shanghai, and has a long history of agricultural development. The district is also a water conservation area, with Dianshan Lake in the west serving as the main source of the Huangpu River, the largest river passing through Shanghai. The district also has impressive green spaces such as gardens and country parks as well as visitor attractions, including farmhouses and campsites. All these provide residents with a variety of CESs.

Fig. 1
figure 1

Study area (Note: a. Location of Shanghai in China, b. Location of Qingpu District in Shanghai, c. Land cover in Qingpu District and d. Urbanization levels of different villages and communities)

Full size image

During the past two decades, Qingpu District has experienced rapid urbanization (Xia et al. 2023). From 2000 to 2020, the resident population of the area grew by 2.13 times, and the GDP increased by 9.52 times (Qingpu Statistical Yearbook 2021). Urbanization was also reflected in the land cover change; the key ecosystems of the area, such as cropland, forests, and water bodies, were gradually fragmented by the encroachment of urban and rural built-up areas (Ren et al. 2022). In the future, the construction of the central subdistrict and the development of road traffic will further promote the urbanization of the district. Thus, facing rapid urbanization, local public officials are promoting the sustainability of social–ecological systems through landscape planning and ecological restoration. However, residents’ perceptions of CESs and the relationships of such services with SWB remain largely unknown.

Methods

Research framework

The design of the integrated approach and research process is shown in Fig. 2. Following the precedents of Wang et al. (2021b) and Yi and Wang (2021), urbanization levels, CES, and SWB were assessed at the village and community levels—the smallest unit for land use allocation and ecological conservation in the spatial planning system of China (Chen and Lü 2021). After evaluating and classifying urbanization levels, data were collected within each level through participatory approaches to elicit residents’ perspectives on CESs and SWB. Perceived CES supply refers to the use potential attributed by people to an ecosystem’s biophysical traits (Spangenberg et al. 2014; Zoderer et al. 2019). We used the PPGIS approach to capture respondents’ perceptions of CES provision. Then, we applied the social values for ecosystem services (SolVES) model to map the potential CES supply in the study area. Furthermore, CES demand refers to the expression of the social importance or preferences for specific attributes of the CES (Martín-López et al. 2012; Schröter et al. 2014), and SWB refers to people’s life satisfaction (Aguado et al. 2018). We used the Likert scale to obtain perceptions of these two components. Multiple statistical and spatial analyses were conducted to identify the respective relationships between urbanization and CES supply, CES demand, and SWB. Finally, the links between supply and demand of CESs and SWB at different urbanization levels were analyzed by constructing a matrix. The data analysis is detailed in the following sections.

Fig. 2
figure 2

Research framework to assess social perspectives on supply and demand of CESs and SWB in rapidly urbanizing landscapes

Full size image

Evaluating urbanization level

Urbanization is a complex process that can be understood through three aspects: population urbanization, spatial urbanization, and economic urbanization (Yuan et al. 2018; Xu et al. 2022). Considering the availability, measurability, and completeness of data at the village and community levels, we selected six indicators across these aspects (Table 1), based on empirical studies by Cai et al. (2021), Peng et al. (2017), Yi and Wang (2021) and Zhou et al. (2018). Data for calculating the indicators, including resident population, employees and output value, were obtained from the Qingpu Statistical Yearbook (2021) and the statistical yearbooks of each township. Built-up areas were derived from the global land cover product of the European Space Agency (ESA) WorldCover 2020 (Zanaga et al. 2021). Road data were sourced from OpenStreetMap (https://www.openstreetmap.org). GDP data were obtained from the Resource and Environmental Science and Data Center, Chinese Academy of Sciences (https://www.resdc.cn). These latter three datasets were extracted based on village and community boundaries.

Table 1 Indicators and weights for evaluating urbanization level of villages and communities
Full size table

To determine the weights of each indicator, we employed the entropy weight method (Cai et al. 2021). The concept of entropy, originating from thermodynamics and introduced to information theory by Shannon (1948), measures the information conveyed by indicators. The entropy weight method is based on the idea that the weight of an indicator depends on the dispersion or variation of its values (Wang et al. 2019). A greater dispersion of values leads to lower entropy, indicating less uncertainty and more informative data, resulting in a higher weight (“entropy weight”) for the indicator (Yu 2021). This method has been widely used in evaluating urbanization levels due to its ability to objectively reflect the information represented by the indicators (Yuan et al. 2018; Cai et al. 2021; Yu 2021). The weights obtained through the entropy weight method are presented in Table 1, and detailed calculations can be found in Appendix B of the Supplementary Material. The evaluation results of urbanization were categorized into four levels (low, relatively low, relatively high, and high) using the quantile method with ArcGIS 10.5.

Quantifying CES supply, demand, and SWB

Selection of CES and SWB

We chose six representative CESs based on the categories presented in the Millennium Ecosystem Assessment (2005) with local managers and experts in Qingpu District. These six CESs were as follows: (1) recreation (outdoor and leisure activities), (2) aesthetic (beautiful and attractive scenery), (3) cultural heritage (local historical and cultural traditions, wisdom, and ways of life, as services and not the same as cultural heritage sites), (4) education (learning about the natural environment through observation and study), (5) spiritual (special meanings such as reverence and respect for nature), and (6) social interaction (spending time together with other people to maintain social connections). We developed one indicator question per CES following the study of Plieninger et al. (2013). Next, the subjective perceptions of the main dimensions that influence life satisfaction were understood as indicators of SWB. We selected the following six indicators based on the Millennium Ecosystem Assessment (2005) and the study of Aguado et al. (2018): basic materials for a good life, health, security, good social relations, freedom of choice and action, and leisure time. One indicator question per SWB was also developed (Aguado et al. 2018).

Survey implementation and questionnaire development

A questionnaire including a PPGIS was used to gather social perceptions of CES supply, CES demand, and SWB. The survey was conducted via face-to-face interviews in August 2022. To ensure representation, we employed a stratified random sampling method. The four urbanization level areas determined by the urbanization evaluation served as the strata (Fig. 1d). Within each stratum, we randomly selected sampling points and residents as the sample population. We collected a total of 223 valid questionnaires from 44 sampling points across the study area. This sample size is comparable to other studies using PPGIS for ES mapping at the local scale, such as Fagerholm et al. (2019), Ho Huu et al. (2018), and Wang et al. (2021b).

The interviews began with a brief introduction, followed by showing the respondents an A3-sheet satellite image of Qingpu District. We helped them in identifying the direction and location of their homes. The questionnaire consisted of four sections. The first section involved participatory mapping of CES supply. Respondents were asked to locate points on the maps that represented locations providing each of the six CESs, using markers with the help of interviewers. They were allowed to mark up to five points for each CES. If they felt that a certain CES did not have a corresponding point, they could leave it unmarked. The second and third sections focused on CES demand and SWB ratings, respectively. Respondents were asked to rate the importance level of CES and the satisfaction level of SWB using a 5-point Likert scale, ranging from “1 = not important/satisfied at all” to “5 = very important/satisfied”. The fourth section collected socio-demographic information, including gender, age, education level, and household disposable income per capita. The complete questionnaire can be found in Appendix A.

Although we used the word “nature” in our descriptions of selected CESs, respondents might relate them to any location the satellite map provided, including non-natural or social systems. In this case, two different interpretations of urban and peri-urban CESs are involved, which were “ecosystem services in urban areas” and “services of urban ecosystems” (Tan et al. 2020). The former refers to CESs provided by natural and semi-natural ecosystems in urban areas, such as trees, rivers, and lakes (Haase et al. 2014). The latter conceived urban areas as systems themselves and includes not only the former types of services but also social services provided by non-natural systems (Antognelli and Vizzari 2016). While the latter interpretation has been used in recent CES assessments (Kalinauskas et al. 2022), our study relies on the former interpretation commonly employed in the Millennium Ecosystem Assessment (2005) and the literature on ESs in urbanizing areas. This choice is due to the fact that our study was conducted in peri-urban areas, which have more natural and semi-natural ecosystems than urban areas, and the natural environment serves as the ultimate source of all social services and human well-being (Wu 2013). Our aim was to emphasize the social values and cultural significance of these natural and semi-natural ecosystems and utilize them more effectively in landscape planning and policymaking. Therefore, during our interviews, we suggested that residents mark the locations of natural landscapes on the map.

CES supply

We used the SolVES 3.0 model to quantify CES supply (Meng et al. 2020). The model is a GIS application developed by the United States Geological Survey (USGS) to provide quantitative and spatially explicit assessments of the social perceptions of ESs (Sherrouse et al. 2011, 2014). We digitized and entered the survey points obtained from PPGIS into the model along with the five environmental layers required for the model, as listed in Table 2 (all resampled to a spatial resolution of 10 × 10 m). Although SolVES 3.0 offers the option to apply weights to survey points (e.g., allow respondents to allocate the hypothetical 100 RMB to six CES during the survey), we did not use it in this study. This is because the use of weights in PPGIS does not yield sufficient additional analytical benefit (Nielsen-Pincus 2011), and it also represents the relative importance of CES duplicated with the subsequent demand analysis.

Table 2 Description and sources of environmental data used in the SolVES model
Full size table

The SolVES model runs in a three-step process:

  1. (1)

    First, the model calculated the average nearest neighbor statistics to describe the spatial arrangement of marked points (i.e., dispersion, clustering, or randomness). As the average nearest neighbor indexes, R-Value less than 1 and large negative Z scores indicate spatial clustering (Zhang et al. 2020).

  2. (2)

    Then, the model integrated kernel density analysis and the maximum entropy (Maxent) model to generate the value index maps. Compared with the information entropy weighting of existing urbanization indicators based on their certainty, the reason for using information entropy here is that the information about the potential supply distribution of CESs within the study area is unknown. Without knowing more information, the probability distribution should be the most uncertain state, that is, the state with maximum entropy (Jaynes 1957). When the probability distribution is uniform, it has the maximum entropy under the known layer constraint of variables (Phillips et al. 2006). Therefore, this step of the Maxent model applied a machine learning approach to estimate the probability distribution of maximum entropy (closest to uniform) under the constraints of the five environmental variable layers (Sherrouse et al. 2014). In this way, we can generate maps of the value index (VI) ranging from 0 to 10 for CESs.

  3. (3)

    Finally, the Maxent model calculated the area under the curve (AUC) of the receiver operating characteristic curve to validate the reliability of the model. An AUC value greater than 0.7 indicates that the model is valid (Swets 1988). The mean value of the study area for each generated VI layer (i.e., Mean-VI) was further calculated, representing the potential supply of each CES within the study area.

CES demand and SWB

To test the reliability of the CES demand and SWB, we first calculated Cronbach’s α for the Likert scale data. In this study, Cronbach’s α value of CES demand and SWB were 0.799 and 0.840, respectively, both indicating good internal consistency of the questionnaire (Taber 2018). Then, the mean ratings of CES demand and SWB were elicited, which represent the level of CES demand and SWB. In addition, as dimensions contribute differently to SWB (Wang et al. 2017), principal component analysis (PCA) was performed to calculate the weights of each SWB and generate the total score of SWB. The detailed calculation can be found in Appendix C. We followed the Kaiser criterion (eigenvalue > 1) to determine the number of components. PCA was conducted using the package “psych” (Revelle 2022) in R 4.2.2 (R Core Team 2022).

Identifying the relationships between urbanization and CES supply, demand, and SWB

Regarding the supply side, we calculated CESs within each village and community unit using the Zonal Statistics tool in ArcGIS. For each unit, we examined the relationships between urbanization level and CES supply through quantitative correlation analysis (Pearson correlation; R package “psych”, Revelle 2022) and spatial correlation analysis (bivariate spatial autocorrelation). The latter involved measuring global and local spatial autocorrelation using Global Moran’s I and Local Moran’s I (LISA) in GeoDa 1.20.0, respectively. We tested the relevant significance of Global Moran’s I through Monte Carlo simulation with 999 permutations (Zhou et al. 2018). The statistical significance level was set at p < 0.05.

Regarding the demand side, we categorized CES demand and SWB based on the urbanization level of respondents’ places of residence. We calculated the mean rating for the demand side within each of the four urbanization levels. To identify differences in CES demand and SWB across the urbanization levels, we performed one-way analysis of variance (ANOVA) and conducted post hoc pairwise comparisons of means using the least significant difference (LSD) method with the R package “agricolae” (de Mendiburu 2021).

Integrating CES supply, demand, and SWB at different urbanization levels

We developed a matrix to identify the links between CES supply, demand, and SWB at different urbanization levels, based on studies by Albert et al. (2016), Burkhard et al. (2012), and Castillo-Eguskitza et al. (2019). First, we calculated CES supply within each urbanization level using the Zonal Statistic tool in ArcGIS to ensure its consistency with the demand-side analysis units. Then, we identified possible matches and mismatches between CES supply and demand within the four urbanization levels. High supply/high demand was represented by values higher than the mean of the total study area, while low supply/low demand represented values lower than the mean. Thus, we generated four types of outcomes: high supply–high demand (H–H), low supply–low demand (L–L), low supply–high demand (L–H), and high supply–low demand (H–L). The former two types were considered matches and referred to as high-degree and low-degree matches (Castillo-Eguskitza et al. 2019). The latter two types were identified as mismatches, with L–H categorized as deficient and H–L as plentiful. Finally, we ranked the mean values of SWB for the four urbanization levels and compared them with the supply–demand relationship of CES to analyze the influence of supply–demand matches/mismatches on SWB (Wei et al. 2018; Ketema et al. 2021).

Results

Urbanization levels and respondent characteristics

The spatial pattern of urbanization levels of villages and communities in Qingpu District is shown in Fig. 1d. Generally, the distribution of urbanization levels showed a spatial pattern of an east–west gradient. Areas with high urbanization levels were mainly distributed in the northeast and central-east areas, closer to the central urban area of Shanghai, while those with low urbanization levels were located in the southwest and central-west areas. The different urbanization dimensions of population, spatial, and economic urbanization showed similar patterns (Appendix D).

Using the levels delineated by the evaluation results of urbanization as strata for sampling, a total of 223 questionnaires were obtained. The characteristics of the respondents are shown in Table 3. There were more men than women in the sample (63.2% and 36.8%, respectively). The largest group of respondents (40.8%) was aged 36–55 years old, followed by the group aged 16–35 years old (36.8%). More than half of the respondents (57.0%) had a college degree. The largest groups of household disposable income per capita were 50,001–100,000 RMB (34.1%) and 10,001–50,000 RMB (33.2%).

Table 3 Characteristics of the respondents
Full size table

Perceived CES supply, demand, and SWB

The average nearest neighbor statistics calculated from respondents’ marked points for each CES are shown in Table 4. All CES points were significant spatial clustering (p < 0.01). Moreover, the AUC values for all CES were greater than 0.7, indicating the validity of the SolVES model.

Table 4 Average nearest neighbor statistics and performance of the SolVES model
Full size table

The spatial distribution of CES supply derived from the SolVES model is displayed in Fig. 3. The highest potential supply of CES was cultural heritage (Mean-VI = 2.372), which was mainly distributed in the central and northeast area, as well as Dianshan Lake in the west area. This was closely followed by education (Mean-VI = 2.299), with the higher VI in the west area. Recreation ranked in third place (Mean-VI = 1.610) and had the higher VI in Dianshan Lake, rivers, and wetlands. The lower potential supply of CES were found in spiritual (Mean-VI = 0.928), aesthetic (Mean-VI = 0.894), and social interaction (Mean-VI = 0.676), with predominantly low value distributions.

Fig. 3
figure 3

Spatial distribution of CES supply. Mean-VI represents mean value index

Full size image

The mean values of residents’ demand for each CES obtained using the 5-point Likert scale are displayed in Fig. 4a. CES demand differed from the perceived supply, with the most important CES assigned to aesthetic (4.184). This was followed by recreation (4.055), cultural heritage (4.000), social interaction (3.981), and education (3.899) in descending order of importance. Spiritual was perceived as the least important CES (3.770) by the respondents.

Fig. 4
figure 4

Importance ratings of CES demand and satisfaction ratings and weights of SWB. Ratings are based on the 5-point Likert scale

Full size image

The mean values of residents’ SWB are displayed in Fig. 4b. All SWB received high scores, and the highest level of satisfaction was security (4.309). This was followed by freedom of choice and action (4.253), health (4.171), good social relations (4.088), leisure time (4.055), and basic materials for a good life (4.009) in descending order of satisfaction. By summing these SWB values through the weights calculated by PCA (Fig. 4b, Appendix E), the score of residents’ total SWB was 4.154.

CES supply, demand, and SWB in relation to urbanization

Regarding the supply side, the results of Pearson correlation analysis showed significant negative correlations between the level of urbanization and recreation, aesthetic, and education (p < 0.01). Conversely, a significant positive correlation was observed between the level of urbanization and the supply of cultural heritage (p < 0.01). The results of bivariate spatial autocorrelation analysis were generally consistent with the results of Pearson correlation analysis (Fig. 5). They also revealed an additional significant negative correlation between the level of urbanization and social interaction (p < 0.01). Among the different urbanization dimensions (Appendix F), a significant negative correlation was found between the level of population urbanization and the supply of recreation, aesthetic, education, and social interaction. In comparison, significant positive correlations were found between the level of spatial urbanization and cultural heritage and spiritual. The level of economic urbanization, on the other hand, not only showed a significant negative correlation with recreation, aesthetic, education, and social interaction but also a significant positive correlation with cultural heritage (Appendix F).

Fig. 5
figure 5

Bivariate spatial autocorrelation and Pearson’s correlation analysis for urbanization and CES supply at village and community level. **Indicate significant differences at p < 0.01.)

Full size image

The spatial heterogeneity of the relationships between urbanization level and CES, which we derived from the local spatial autocorrelation analysis, can be found in Fig. 5. The results included two main types: (1) Urbanization level and recreation, aesthetic, education, spiritual, and social interaction showed relatively consistent spatial patterns. The H–H areas were mainly located in the central area and scattered around Dianshan Lake in the west. The L–L areas were scattered in the northeast and southwest areas. The L–H areas were concentrated around Dianshan Lake, whereas the H–L areas were concentrated in the eastern part closer to the central urban area of Shanghai. (2) The relationship between urbanization level and cultural heritage showed different patterns, with H–H and L–H areas in the central and east areas, and L–L and H–L areas in the west. The spatial heterogeneity of the relationships between different urbanization dimensions and CES can be found in Appendix F.

Regarding the demand side, the one-way ANOVA results indicated significant differences (p < 0.05) for the CESs of recreation and social interaction, as well as the SWB of basic materials for a good life across the four urbanization levels (Table 5). In such cases, the demands in villages and communities with moderate (relatively low or relatively high) urbanization levels were significantly lower than those with low or high urbanization levels. The demand for these three first decreased and then increased as the urbanization level increased. Other CES demands and SWB indicators showed no significant differences across urbanization levels.

Table 5 One-way ANOVA and post hoc multiple comparisons (LSD method) of residents’ CES demand and SWB ratings across different urbanization levels
Full size table

Links between CES supply, demand, and SWB at different urbanization levels

The supply–demand relationship of CES and their links with SWB indicated varying characteristics in areas with different urbanization levels (Table 6). In particular, areas with high urbanization levels had the largest number of deficient CESs, with low supply and high demand, and these areas ranked first in terms of SWB (4.202). In areas with low urbanization levels, half of the CESs showed a high degree sufficient of high supply and high demand, and these areas had the second highest SWB (4.193, only 0.21% lower than the highest areas). In areas with moderate urbanization levels, the areas with relatively low urbanization levels were dominated by plentiful CESs, with high supply and low demand. Meanwhile, the areas with relatively high urbanization levels were dominated by low-degree sufficient CES, with low supply and low demand. Lower levels of SWB were found in these two types of areas (4.128 and 4.071, 1.76% and 3.12% lower than the highest areas, respectively).

Table 6 Matrix of CES supply, demand, and SWB across urbanization levels
Full size table

Discussion

Social perspectives on CES supply, demand, and SWB in the context of urbanization

Our results reveal that residents identified different supply and demand of CESs in rapidly urbanizing landscapes. For example, residents identified cultural heritage and education as the highest potential supply in the district, while aesthetic and recreation were deemed the most important. The current widely used cascade model in the ES framework depicts ES as a unidirectional flow from biophysical supply to socioeconomic demand (Haines-Young and Potschin 2010). Recent studies, however, have argued that the dichotomy of this framework oversimplifies the complex and nonlinear nature of the process (Costanza et al. 2017). Zoderer et al. (2019) surveyed social perspectives on the supply and demand of ESs in the Central Alps and found that stakeholders identified different supply and demand. Our study yielded similar findings for CESs, suggesting that CES is a relational entity that people actively express through their interactions with ecosystems (Fish et al. 2016). CES is shaped by the interaction between natural capital and human-derived capital (Costanza et al. 2017; Tan et al. 2020), which occurs in environmental spaces and cultural practices. In rapidly urbanizing landscapes, natural capital tends to decrease, while human-derived capital increases (Wu 2013). Thus, residents actively contribute to the co-production of CESs and perceive diverse supply that differs from their demand.

The relationships between the supply and demand of CESs and the urbanization level also showed different characteristics. On the supply side, most CESs, including recreation, aesthetic, education, and social interaction, showed a significant negative correlation with urbanization level, primarily determined by population and economic urbanization. Among these, recreation and aesthetic are considered the more studied and more easily quantifiable CESs (Cheng et al. 2019). Existing studies in the Chinese context have observed that they have been eroded by rapid urbanization, with the decrease in their supply in response to urbanization (Zhang et al. 2018; Zhou et al. 2018; Chen and Chi 2022). The same results were obtained in our study using the participatory approach and extended to the relationship between urbanization and education as well as social interaction. However, cultural heritage services exhibited contrasting characteristics, with a significant positive correlation with urbanization level, mainly determined by spatial and economic urbanization. Consistent with our findings, several PPGIS studies on CESs have found that cultural heritage services are concentrated in inner city areas with high population density (Fagerholm et al. 2016; Rall et al. 2017). Zoderer et al. (2016, 2019) also found the exceptional role of cultural heritage compared to other CESs provided by the Central Alps, where people are more likely to perceive its supply in intensively managed landscapes with strong human influence. In addition, scholars noticed that the relationship between cultural heritage and well-being differs from other CESs, as cultural heritage is associated with public welfare, while recreation, aesthetic, and education are linked to personal welfare (Swapan et al. 2017; Wang et al. 2022). Therefore, we conclude that areas with higher urbanization levels are strongly shaped by human activities, possess more human-derived capital (such as economic and infrastructure support) for intensive landscape management to maintain public welfare, and are likely to have higher cultural heritage services.

On the demand side, recreation, social interaction, and the SWB related to basic materials for a good life exhibited significant differences at different urbanization levels. In these cases, demand was higher in areas with low urbanization levels, decreased in areas with moderate urbanization levels, and then increased in areas with high urbanization levels. A study in the peri-urban areas of Beijing, China by Wang et al. (2021b) found a similar relationship. They explained that during China’s urbanization process, collectively owned rural land undergoes transformation into state-owned urban land, accompanied by land expropriation, house demolition, and financial compensation to local residents. They found that in this context, residents prioritized government compensation through increased collective built-up areas, leading to decreased demand for the CESs associated with natural characteristics. After that, as urbanization levels increased, there is a growing need for a better quality of life and increased demand to improve CES through the development of parks and other green infrastructure (Wang et al. 2021b). This causal relationship could similarly explain the changes in CES demand and SWB observed in our study, as land expropriation and demolition are pivotal steps in China’s rural–urban shift impacting ESs and well-being.

Understanding the variations of the CES–SWB relationship at different urbanization levels

Taking a comprehensive look, our study revealed mismatches between CES supply and demand, as well as divergent characteristics of SWB under different urbanization levels. Our results indicate that in areas with high urbanization levels, most CESs had low supply and high demand, while residents experienced high SWB. In areas with moderate (relatively low or high) urbanization levels, most CESs had high supply and low demand, as well as low-degree matches of low supply and low demand, but residents’ SWB was lower. This is similar to the findings of Zhang et al. (2022b), who observed lower SWB in areas with high supply-low demand and low supply-low demand of CESs, as well as lower GDP and transportation network density compared to central areas.

A possible explanation for these relationships can be referred to the “green-loop to red-loop transition” model proposed by Cumming et al. (2014), as depicted in Fig. 6. The relationships between residents’ perceived CES supply, demand, and SWB in areas with low urbanization, moderate (relatively low or high) urbanization, and high urbanization levels can be understood as being dominated by the “green loop”, “transition”, and “red loop”, respectively. In areas with low urbanization levels, corresponding to the “green loop” stage of urbanization, semi-natural ecosystems provide certain CESs. Residents could satisfy their high demand and obtain relatively high SWB from local CESs, which is a state close to self-sufficiency. In areas with high urbanization levels, corresponding to the “red loop” stage of urbanization, residents rely on non-ESs (socioeconomic services) but also have a growing demand for the value of scarce natural landscapes and a better quality of life (Cumming et al. 2014; Dou et al. 2017). For most CESs, their spatial characteristics are shaped by the movement of people (Costanza 2008; Villamagna et al. 2013). Although some CESs may be undersupplied in areas with high urbanization levels, good road transportation and other infrastructure enable residents to travel or commute to other supply areas with unique landscape characteristics and access remote CESs. These remote CESs, combined with local non-ESs, fulfill their demand and thus contribute to high SWB. Furthermore, areas with moderate urbanization levels represent villages and communities transitioning from the “green loop” to the “red loop”. Residents’ perceptions of CESs in these areas gradually shift from high to low supply, indicating a decline in the supply of most CESs during urbanization. At the same time, residents have lower demand for CESs and experience lower SWB. In these areas, residents could share similar perceptions of local CESs, local non-ESs and remote CESs, indicating a gradual change in their livelihoods and dependence on natural resources. Since the CES–SWB relationship during transition is closely linked to land development and infrastructure construction in China’s urbanization process (Wang et al. 2021b), these areas require attention in policymaking.

Fig. 6
figure 6

Summary and illustration of the respective relationships between CES supply, demand, and SWB for areas with different urbanization levels. Possible explanations for these links can be referred to the “green-loop to red-loop transition” hypothesis, with the thickness of the arrows in the figure representing the relative degree of dependence (Cumming et al. 2014; Liu et al. 2022)

Full size image

Policy implications for landscape sustainability

Understanding the links between residents’ perceived CES supply, demand, and SWB is crucial for landscape sustainability, which emphasizes “a landscape to consistently provide long-term, landscape-specific ecosystem services essential for maintaining and improving human well-being” (Wu 2013, 2021). Assessing landscape sustainability requires engaging in dialogue with residents who inhabit the landscape (Opdam et al. 2018). Hence, the guiding role of our results is to allow residents to recognize their living environments and provide the possibility for coming up with bottom-up policy suggestions (Xin et al. 2021). Given the diverse relationships between residents’ perceived CES and SWB in areas with different levels of urbanization, targeted policies are needed to offer zoning management suggestions (Fig. 6).

Specifically, in areas with low urbanization levels similar to the “green loop” stage, there are several high-degree sufficient CESs of high supply and high demand. To maintain a balanced state, it is essential to protect the unique natural and cultural characteristics of these areas. Meanwhile, minimizing inappropriate human pressures and potential rapid urbanization is important in sustainably managing natural resources and preventing these areas from falling into a “green trap” (Cumming et al. 2014). Conversely, in areas with high urbanization levels, similar to the “red loop” stage, there are multiple CESs with low supply and high demand. These areas have more human activities and the risk of a “red trap” occurrence. Therefore, these areas warrant environmental assessment and improvement (Meng et al. 2020), as well as measures of “environmental austerity” (Seppelt and Cumming 2016), to reduce externalization or pressure from other areas to provide CESs. For example, constructing green infrastructure and promoting ecological restoration can meet residents’ demand and offer more opportunities to experience CESs in these areas. The high cultural heritage services of these areas should also be preserved, as cultural heritage is the only CES that increases in supply with urbanization. Areas with moderate urbanization levels in the “transition” stage have the highest number of CESs with high supply and low demand or low supply and low demand. It is crucial to control the extent of land development and the expansion of unmanaged built-up areas (Wang et al. 2021b), guide residents’ livelihood transition in adaption to urbanization, and raise awareness about important CESs. Notably, human-derived capital can increase and natural capital can decrease in these aforementioned areas during the changing urbanization dynamics, yet natural capital serves as the ultimate source of all other capital and well-being (Wu 2013; Tan et al. 2020). Therefore, safeguarding the essential natural resources in these rapidly urbanizing landscapes is crucial in maintaining the capacity to provide CESs.

Limitations and future directions

Several limitations exist in our study. First, the accuracy of the questionnaire and PPGIS still requires further improvement. Residents have different understandings of the given closed-ended items (Wang et al. 2022). Although we recommended residents to choose natural landscapes as marker points in our interviews and informed them that they could also provide place names, some residents may still associate some CESs with human-made environments, such as recreation to sports facilities and cultural heritage to monuments. In addition, since respondents may also be unfamiliar with these maps, they may mark any place on the map, regardless of whether the place was a natural feature or not. These may have led to uncertainties in the quantification of the CES supply and their relationships with urbanization in our study. To minimize these uncertainties, future research could combine multiple approaches in the survey. For example, photo-elicitation could be applied by showing photos of natural landscape features (Xia et al. 2023), thus enabling respondents to better understand our questions. Meanwhile, in-depth interviews and qualitative analysis could also be integrated into PPGIS by asking residents why they chose to mark certain places (Cheng et al. 2022). This allows for more accurate analyses of the characteristics of the chosen landscape features and the selection of appropriate places for further digitization. Furthermore, survey data can be complemented by social media data to obtain more robust results for the CESs assessment (Wang et al. 2022).

Second, perceptions of CESs and SWB differ among beneficiary groups. While our study distinguished perceptions of residents in areas with different urbanization levels, it did not further categorize resident groups according to socio-demographic characteristics, such as gender, age, education level and household income. Follow-up studies should consider the effects of these characteristics.

Third, we studied supply and demand of CESs in a peri-urban district and did not address people and environmental factors outside the area. These factors may influence CESs, as CESs can flow from/to areas outside the district with the movement of people. We will conduct follow-up assessments of how CESs flow across districts to better guide policymaking.

Finally, our study only conducted a cross-sectional survey and analyzed the current state of the supply and demand of CESs without considering the temporal scale, which has been a challenge for CES assessment (Martin et al. 2020; Zhang et al. 2020). Considering urbanization as a process, apart from the horizontal comparison of areas with different urbanization levels that we used in this study, another important option in future studies is to assess the change of CESs through longitudinal studies.

Conclusions

Through a participatory-based integrated approach, our study analyzed interactions between urbanization level, residents’ perceived supply and demand of CESs, and their SWB. The results showed that residents identified different supply and demand of CESs, indicating the crucial role of human–ecosystem interactions in the construction of CESs. The relationships between urbanization and multiple supply and demand of CESs also differed. Furthermore, the variations in the CES–SWB relationship at different urbanization levels were largely consistent with the conceptual model of “green-loop to red-loop transition”, which describes the development trajectory of social–ecological systems. In areas with high urbanization levels dominated by the “red loop” stage, CESs were dominated by low supply and high demand, whereas residents’ SWB was the highest. This could be due to the essential characteristics of CES enabling people with high demand to travel to other supply areas to obtain services and thus meet the SWB. In contrast, areas with moderate urbanization levels were in the transition stage from the “green loop” to the “red loop”, where residents had lower demand for CES and lower SWB. These complex and diverse relationships provide a lens for zoning management in rapidly urbanizing areas. We recommend that landscape management integrate CES and SWB assessment against the background of urbanization and involve residents’ perspectives at the outset for the identification of supply and demand. In this way, bottom-up and local condition-adapted policymaking can be facilitated to promote sustainability in rapidly urbanizing landscapes.