Introduction

In the dynamic world we live in, the landscape emerges as a complex tapestry of interconnected ecosystems (Forman 1995; Wu 2013). The landscape serves as a vital bridge between humans and nature, facilitating a profound interaction between them. The concept of landscape sustainability, encompassing the enduring ability to deliver a wide range of ecosystem services (ESs) and enhance human well-being in a consistent and long-lasting manner, has garnered significant attention (Wu 2013; Hák et al. 2016; Zhou et al. 2019; Fan et al. 2021). Currently, the conflicts between human activities and natural systems are intensifying, especially in urban agglomerations (Steffen et al. 2011; Peng et al. 2017). Intensive human activities, including urban expansion (such as infrastructure development and population growth), agricultural development (including intensive farming practices and reclaiming cultivated land), and industrialization (such as manufacturing and resource extraction) have caused substantial land use and land cover changes in the landscape, leading to changes in ecosystem structure and function and reduction in ES provision (Xu et al. 2020; Winkler et al. 2021). Yet the ever-increasing urban and rural populations and the development of economies are accompanied by sharp increases in the demand for ESs (MEA 2005; Carpenter et al. 2009). This imbalance has threatened human survival and the sustainability of urban agglomerations (Ouyang et al. 2016; Wang et al. 2019; Wu 2021; Xu et al. 2020).

As a typical region in which urban and rural landscapes are coupled, the sustainable development of an urban agglomeration should include coordinated development of both urban and rural landscapes (Fang et al. 2018; Wu 2022). In China, promoting rural revitalization and sustainable urban–rural development has risen to the level of a national strategy. This phenomenon emphasizes the great significance of urban–rural fringe and rural landscape sustainability while promoting urban development (Liu 2018; Wu 2022). However, previous studies on ESs have concentrated on highly urbanized areas, neglecting the spatiotemporal heterogeneity and coupling mechanisms in urban–rural fringe and rural areas (Fang et al. 2023; Yin et al. 2022). Moreover, the supply and demand of some services exhibit notable variations across the urban–rural gradient. For example, increased demands for food, water, and air purification may be higher in urban areas due to increased human and industrial activity compared to rural areas (Zhang et al. 2021a). Rural areas, particularly those in mountainous regions, experience more significant challenges related to soil erosion compared to urban areas (Li et al. 2022). These differences mean that sustainable management of urban and rural landscapes should also be tailored to local conditions.

Assessing ES supply and demand can provide a deeper understanding of the interrelationships between human activities and the natural environment, promoting better landscape sustainability (Wu 2021; Sun et al. 2022). Many studies have focused on the spatial mismatches between the supply and demand of ESs (Xu et al. 2021; Chen et al. 2022a). Usually, supply assessment of ESs on small scales, such as cities and communities, is expressed by biophysical indicators such as product output, water volume, pollution interception, and soil loss (Andrew et al. 2015; Schirpke et al. 2019; Vallecillo et al. 2019; Zhang et al. 2021b); demand is often expressed from the perspective of actual human consumption (Peña et al. 2015; Larondelle and Lauf 2016; Bryan et al. 2018). In large-scale systems with regional differences, due to the limitations of parameter and data acquisition, ES supply and demand matrices are usually constructed based on the expert scoring method (Burkhard et al. 2015). However, prior research has regarded the region as a whole and analyzed the ES differences along the urban–rural gradient, or else only focused on urban center areas, rather than considering ES differences under different urbanization levels in the same region, especially for the demand side (Metzger et al. 2021; Xu et al. 2021). With respect to large-scale regional landscape sustainability assessment, such as the urban agglomeration, it is necessary to use delicate quantitative models to explore the ES supply-demand mismatches across urban–rural landscapes.

In the face of the ongoing imbalance between the supply and demand of ESs, adopting an optimized planning approach is paramount. This not only ensures we meet human needs but is also a pivotal stride in advancing regional landscape sustainability (Forman 2008; Wu 2013). Historically, the majority of studies have predominantly placed their emphasis on the ES supply aspect. Recent research underscored the imperative of addressing both the supply and demand and capturing their mismatches holistically (Syrbe and Grunewald 2017; Wei et al. 2017). These methodologies, rooted in natural and socioeconomic criteria, can simulate alternative land use scenarios in regions with serious ES supply-demand mismatches to improve the supply of ESs (Goldstein et al. 2012; Li et al. 2021b; Sun et al. 2022). Additionally, achieving a balance between ESs that share a trade-off relationship is paramount to improving the provision of both services and concurrent enhancement. This trade-off conflict problem in ES management can usually be solved by finding the land use under the pareto optimality using a multi-objective optimization method (Lautenbach et al. 2013; Cord et al. 2017). Yet, it is difficult for mismatch optimization to solve conflicts of ESs, and single trade-off optimization is difficult to optimize in regions with serious ES supply-demand mismatches. Comprehensive studies on how to equilibrate these trade-offs while bridging the ES supply and demand gap remain relatively scant in different urban–rural landscapes (Wang et al. 2019b).

For promoting urban–rural landscape sustainability, we developed a methodological framework from the perspectives of assessing and optimizing ES supply and demand across urban–rural landscapes for one of the largest urban agglomerations in China. Our objective was to identify the key ESs and their priority regions to invest in and manage across urban–rural landscapes in the Beijing-Tianjin-Hebei (BTH) urban agglomeration. Toward this objective, Restoration Opportunities Optimization Tool (ROOT) was applied in our study. Our approach takes into account both the trade-off and deficit facets, and the ROOT as a comprehensive tool is used to achieve ES optimization in our study (Beatty et al. 2018).

Materials and methods

Study areas

The BTH urban agglomeration (36.03°N–42.62°N, 113.52°E–119.85°E) is located in North China (Fig. 1a), with the northern part transitioning to the Inner Mongolia Plateau through the Yanshan Mountains, bordering the Loess Plateau with the Taihang Mountains in the west, the North China Plain in the center and south, and the Bohai Sea in the east (Fig. 1b). Covering an area of approximately 218,000 km2, the BTH consists of three provincial-level administrative regions, Beijing Municipality, Tianjin Municipality and Hebei Province, including 11 prefecture-level cites (Fig. 1c). With a monsoon-influenced humid continental climate, the temperature of the BTH region averages 4–13 °C, and the annual average precipitation ranges from 400 to 800 mm, decreasing from southeast to northwest.

The BTH is an ideal area to study landscape sustainability from the perspective of urban and rural landscape ES supply and demand, for the following two reasons. First, the mismatch of ES supply and demand in BTH is growing. Factors brought about by the long-term economic development of the BTH such as urban expansion, industrialization, and agricultural modernization, have become non-negligible threats to the supply of ESs and are accompanied by a dramatic increase in the demand for ESs (Chen et al. 2022b; Xiang et al. 2022). Second, the imbalance between urban and rural landscapes in BTH is very significant and has its own idiosyncrasies. In the BTH, there is not only a huge development gap between urban and rural landscapes, but also unbalanced and inadequate development within urban and rural areas (Yang et al. 2018). Achieving urban–rural integration and coordinated development in the BTH has been one of the crucial goals in China’s national development strategy (Zhou et al. 2018).

Fig. 1
figure 1

Overview of the Beijing-Tianjin-Hebei (BTH) region: a Geographical location of BTH in China. b Elevation map of BTH and surrounding areas. c Land use map of BTH in 2020, including two municipalities (Beijing and Tianjin) and eleven prefecture-level cites (Zhangjiakou, Chengde, Qinhuangdao, Tangshan, Langfang, Baoding, Cangzhou, Shijiazhuang, Xingtai, and Handan)

Data sources

The data used in this study included land use/land cover data, meteorological data, remote sensing inversion data, socioeconomic data and statistical data of the BTH region for 2000, 2010, and 2020 (Table 1). All of the data were reprojected to the same coordinate system, and the spatial resolution of all raster data was unified to 30 m to ensure consistency.

Table 1 Data sources and detailed descriptions

A methodological framework for ES supply-demand optimization in urban–rural landscapes

The theoretical framework for ES supply-demand optimization in urban–rural landscapes mainly consisted of four steps (Fig. 2).

Step 1: Identifying urban–rural landscapes. According to population, economic and land indicators, the urban, urban–rural fringe, and rural areas of the BTH from 2000 to 2020 were identified.

Step 2: Quantifying the supply and demand of multiple ESs. The spatiotemporal patterns of supply and demand for six typical ES indicators in the BTH from 2000 to 2020 were quantified.

Step 3: Exploring ES supply-demand relationships across urban–rural landscapes. The dynamic evolution of ES supply and demand in mismatch and trade-off relationships across urban–rural landscapes in the BTH was analyzed.

Step 4: Identifying priority regions for ES optimization in urban–rural landscapes. This step involved the optimization of identification of the priority regions for key ESs and generating their trade-off curves in urban–rural landscapes.

Fig. 2
figure 2

Methodological framework of this study

Step 1: identifying urban–rural landscapes

In this study, construction land intensity, population density, and gross domestic product were chosen as indicators to represent the three most commonly used dimensions of urbanization (Fig. S1): land urbanization, population urbanization, and economic urbanization (Li et al. 2021a; Zhang et al. 2021b). These indicators were selected to identify and characterize urban–rural landscapes. We first clustered three indicators using the Iso Cluster Classification tool in ArcMap 10.6, and then corrected the initial result by comparing it with other related research results and adding field sampling results for validation to adjust the number of iterations (Peng et al. 2018; Wang et al. 2021). The analysis yielded the partition of urban, urban–rural fringe, and rural areas of BTH in 2000, 2010, and 2020 as the result of urban–rural landscape identification. Finally, our results were compared with the results obtained using the urban–rural identification method developed by Peng et al. (2018) to test the degree of agreement (Table S1).

Step 2: quantifying the supply and demand of multiple ESs

In this study, according to ecological problems faced in the BTH (Chen et al. 2022b), the importance of ESs to urban and rural residents (Li et al. 2016; Hou et al. 2020; Zhang et al. 2021b), and the opinions of 12 experts in the fields of remote sensing, landscape ecology, and sociology from the national key universities or institutes in Beijing, Tianjin, Hebei and other provinces, six typical ESs were selected. The list and information about these experts are presented in Table S2. These indicators included grain production, water yield, carbon sequestration, air purification, soil retention, and recreational opportunity. The ES supply were then quantified by using biophysical models, the unit area method, and empirical formulas, and the demand were quantified based on actual consumption, target expectations, and risk mitigation in 2000, 2010, and 2020.

  1. (1)

    Grain production

Grain production supply was spatially mapped using the grain yield statistical data and was spatially assigned to the cropland pixels according to the potential crop yield data from the GAEZ model (Liu et al. 2015). Grain production demand was spatially mapped based on population density data and per capita grain consumption statistics (Table S3) (Cui et al. 2019; Zhang et al. 2021b).

$$\begin{array}{*{20}c} {S_{{FP}} = \sum\limits_{{x = 1}}^{n} {\frac{{GP}}{{\sum\limits_{{x = 1}}^{n} {PY_{x} } }}} \times PY_{x} } \\ \end{array}$$
(1)

where\({S}_{FP}\) is grain production supply; \(GP\) is the total grain production; and \({PY}_{x}\) is the potential crop yield data value of cropland pixel “\(x\)”.

$$\begin{array}{*{20}c} {D_{{FP}} = \sum\limits_{{x = 1}}^{n} {\rho _{x} } \times GD} \\ \end{array}$$
(2)

where \({D}_{FP}\) is the grain production demand; \({\rho }_{x}\) is the population density in pixel \(x\); and \(GD\) is the grain consumption per capita.

  1. (2)

    Water yield

Water yield supply was calculated using the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) model that is based on the water balance equation (Sharp et al. 2020). Water yield demand was composed of four parts: environmental water, domestic water, agricultural water and industrial water. Among these, environmental water consumption was equally allocated to urban parks, green spaces and water bodies, domestic water consumption was allocated according to population density data, agricultural water consumption was allocated according to crop yield data, and industrial water consumption was equally allocated to industrial land for spatial mapping (Chen et al. 2019; González-García et al. 2020; Zhang et al. 2021b).

$$\begin{array}{*{20}c} {S_{{WY}} = \sum\limits_{{x = 1}}^{n} {(1 - \frac{{AET_{x} }}{{P_{x} }})} \times P_{x} } \\ \end{array}$$
(3)

where \({S}_{WY}\) is the water yield supply; \({AET}_{x}\)is the annual evapotranspiration in pixel \(x\); and \({P}_{x}\) is the annual precipitation in pixel \(x\).

$$\begin{array}{*{20}c} {D_{{WY}} = W_{{environment}} + W_{{living}} + W_{{agriculture}} + W_{{industry}} } \\ \end{array}$$
(4)

where \({D}_{WY}\) is the water yield demand; and \({W}_{environment}\), \({W}_{living}\), \({W}_{agriculture}\) and \({W}_{industry}\) are the environmental, domestic, agricultural, and industrial water consumption, respectively.

  1. (3)

    Carbon sequestration

For carbon sequestration supply, we used NPP data to quantify the net CO2 absorption (Pan et al. 2014; Chen et al. 2019). Carbon sequestration demand was calculated based on agricultural consumption, industrial consumption, services consumption and household consumption (Schirpke et al. 2019). Specifically, agricultural energy consumption was spatially mapped by crop yield, industrial energy consumption was equally distributed to industrial land, and service industry energy consumption and living energy consumption were spatially mapped by population density data (Chen et al. 2019; González-García et al. 2020).

$$\begin{array}{*{20}c} {S_{{CS}} = \sum\limits_{{x = 1}}^{n} {NPP_{x} } \times 1.63} \\ \end{array}$$
(5)

where \({S}_{CS}\) is the carbon sequestration supply; \({NPP}_{x}\) is the net primary productivity in pixel \(x\), and 1.63 is the carbon fixation coefficient of plant dry matter, which means plants absorb 1.63 g of CO2 for every 1.0 g of dry matter accumulated (Chen et al. 2019).

$$\begin{array}{*{20}c} {D_{{CS}} = C_{{agriculture}} + C_{{industry}} + C_{{service}} + C_{{household}} } \\ \end{array}$$
(6)

where \({D}_{CS}\) is the carbon sequestration demand; \({C}_{agriculture}\), \({C}_{industry}\), \({C}_{service}\), and \({C}_{household}\) are agricultural, industrial, service, and household energy consumption, respectively.

  1. (4)

    Air purification

Air purification supply was calculated according to the air purification potential supply, which is the PM2.5 (particulate matter in the air less than 2.5 μm in diameter) removal capacity per unit area (Table S4). For the demand, we referred to the standard value of PM2.5 allowed by the air quality guidelines from World Health Organization and the actual concentration of PM2.5 in the atmosphere (WHO 2006, Chen et al. 2019). When the concentration of PM2.5 in the atmosphere is greater than the standard value, the difference between the two was used as the demand; otherwise, the demand was set to zero.

$$\begin{array}{*{20}c} {S_{{AP}} = \sum\limits_{{i = 1}}^{j} {RC_{i} } \times A_{i} } \\ \end{array}$$
(7)

where \({S}_{AP}\) is the air purification supply; \({RC}_{i}\) is the removal capacity of PM2.5 of land use type \(i\); and \({A}_{i}\) is the area of land use type \(i\).

$$\begin{array}{*{20}c} {D_{{AP}} = \left\{ {\begin{array}{*{20}c} {\sum\limits_{{x = 1}}^{n} {\left( {PM_{x} - 10} \right)} \times H,if\rho _{{PM}} > 10} \\ {0,if\rho _{{PM}} \le 10} \\ \end{array} } \right.} \\ \end{array}$$
(8)

where \({D}_{AP}\) is the air purification demand; \({PM}_{x}\) is the atmospheric PM2.5 concentration of pixel \(x\); 10 is the standard value of PM2.5 concentration that is recommended by the World Health Organization; and \(H\) is the height of the air column used to convert the PM2.5 concentration to the surface area capacity.

  1. (5)

    Soil retention

Soil retention supply and demand were calculated using the revised universal soil loss equation (RULSE). Soil retention supply was calculated as follows: the potential soil erosion minus the actual soil erosion (Ouyang et al. 2016). The difference between the potential soil erosion and the official maximum allowable soil loss value was used as the soil retention demand (Li et al. 2022).

$$\begin{array}{*{20}c} {S_{{SR}} = R \times K \times LS \times \left( {1 - CP} \right)} \\ \end{array}$$
(9)

where \({S}_{SR}\) is the soil retention supply; \(R\) is the rainfall erosivity factor; \(K\) is the soil erodibility factor; \(LS\) is the topographic factor; and \(CP\) is the vegetation and management factor.

$$\begin{array}{*{20}c} {D_{{SR}} = \left\{ {\begin{array}{*{20}c} {R \times K \times LS - ASE,if \, R \times K \times LS - ASE > ASE} \\ {0,if\, R \times K \times LS - ASE \le ASE} \\ \end{array} } \right.} \\ \end{array}$$
(10)

where \({D}_{SR}\) is the soil retention demand, and \(ASE\) is the allowed soil erosion value from the Chinese Standards for Classification and Gradation of Soil Erosion (2008), which is 200 tons/km2 in North China.

  1. (6)

    Recreational opportunity

Recreational opportunity supply was represented by the area of green space. Specifically, the areas of forest parks, grasslands, parks, and natural reserves in the township were used as the supply of recreation opportunity in this township (Tao et al. 2022). Recreation opportunity demand was determined by residents’ needs for green space, which was mapped by the population density and the ideal green space per capita (Baró et al. 2016; Chen et al. 2019; Zhang et al. 2021b).

$$\begin{array}{*{20}c} {S_{{RO}} = A_{{forest}} + A_{{grass}} + A_{{park}} + A_{{reserve}} } \\ \end{array}$$
(11)

where \({S}_{RO}\) is recreational opportunity supply; \({A}_{forest}\), \({A}_{grass}\), \({A}_{park}\), and \({A}_{reserve}\) are forest park, grassland, park and natural reserve areas of the township, respectively.

$$\begin{array}{*{20}c} {D_{{RO}} = \rho \times A_{{guided}} } \\ \end{array}$$
(12)

where \({D}_{RO}\) is the recreational opportunity demand; \(\rho\) is the population density; and \({A}_{guided}\) is the government-guided green space per capita from the Statistical Bulletin of Beijing Municipality on national economic and social development in 2020, 16.5 m2/person.

Step 3: exploring ES supply-demand mismatches and trade-offs

In this study, we used the ecological supply-demand ratio (ESDR) to describe the mismatch between ES supply and demand (Li et al. 2016).

$$\begin{array}{*{20}c} {ESDR = \frac{{S - D}}{{(S_{{max}} + D_{{max}} )/2}}} \\ \end{array}$$
(13)

where \(S\) and \(D\) are ES supply and demand, respectively, and \({S}_{max}\) and \({D}_{max}\) are the maximum values of ES supply and demand, respectively. ESDR > 0 indicates a surplus, ESDR = 0 indicates a balance, and ESDR < 0 indicates a deficit.

Spearman’s correlation analysis was used to describe the spatial correlations among the supply and demand of ESs in urban–rural landscapes in 2020. According to the result of urban–rural identification and computational constraints, a 2 km \(\times\) 2 km raster net was created to extract the values for calculating correlations of ESs. If the correlation coefficient was > 0, the two ESs showed a synergistic relationship; otherwise, the two ESs showed a trade-off relationship.

Step 4: identifying priority regions for ES optimization in urban–rural landscapes

The optimization of ESs was divided into four steps. First, the types of optimization and corresponding key ESs for optimization were determined. Second, detailed optimization objectives and areas for optimization were determined. Third, priority restoration regions for ES optimization in urban–rural landscapes were identified. Fourth, the spatial changes in priority regions were explored and the relevant curves for different solutions in the optimization were generated. To determine the key ESs that require optimization and the types of optimization (Three types of optimization: deficit optimization only, trade-off optimization only, and both deficit and trade-off optimization), we established following rules: (1) the key ESs that needs to optimized and their optimization types were chosen according to the ES deficits and trade-offs characteristics; (2) the optimization objectives in priority regions for optimizing key ESs in urban, urban–rural fringe, and rural areas were determined according to the physical and socioeconomic conditions and policy guidelines; (3) the generation of curves must be based on ES pairs that trade-off optimization existed.

In this study, ROOT was used to identify priority regions for ES optimization. This software was developed by the Natural Capital Project of Stanford University and was designed to promote multi-objective landscape planning (https://natcap.github.io/ROOT/index.html) (Hawthorne et al. 2017; Beatty et al. 2018). It includes two modules: preprocessing module and optimization module (Fig. 3). Input data: spatial changes of ESs from 2000 to 2020 (Impact Potential Raster), spatial distribution of ES deficit and trade-offs (Spatial Weight Map), spatial distributions of urban areas, urban–rural fringe, and rural areas (Activity area); Spatial processing: creating spatial decision unit, combining changes of key ESs with their deficit or trade-offs (Composite Factors); Optimization input: optimization types, detailed objectives, and area constraints; Optimization: single optimization using specific weights for deficit optimization and multi-objective optimization using random weights for trade-off optimization and both deficit and trade-off optimization (Analysis types). The output of ROOT included all of the optimization solutions for the three optimization types with different weights and maps showing the number of times that the smallest decision units were selected in all of the solutions (Fig. 3).

After the identification of priority regions, the trade-off curves were generated for trade-off optimization and both deficit and trade-off optimization. The curves draw on the idea of the production possibility frontier (Polasky et al. 2008; Bennett et al. 2009), and included the contribution values of each solution to key ESs under different weights and their spatial distribution of priority regions.

Fig. 3
figure 3

Workflow of the Restoration Opportunity Optimization Tool (ROOT)

Results

Spatial patterns of urban–rural transitions from 2000 to 2020 in BTH

From 2000 to 2020, the areas of urban and urban–rural fringe increased from 2,844 km2 and 9,826 km2 to 6,268 km2 and 21,719 km2, respectively, while the area of rural areas decreased from 203,376 km2 to 188,082 km2. Urban areas mainly included the city centers of prefecture-level cities and most county towns. The urban–rural fringe surrounded the city and extended outward as the urban area expanded. On the periphery of the metropolis, the urban–rural fringe connects two spatially disjoint urban areas to form urban agglomerations (Fig. 4). The results of urban–rural identification are in good agreement with previous research results (Table S1).

Fig. 4
figure 4

Spatial changes in urban areas, urban–rural fringe, and rural areas in 2000, 2010, and 2020

Spatiotemporal mismatches of ES supply and demand across urban–rural landscapes in BTH

Overall, from 2000 to 2020, the demand for most ESs far exceeded the supply, except for grain production and recreational opportunity (Table 2). Among them, the supply of grain production, water yield, and carbon sequestration increased markedly over the years, while the demand for carbon sequestration increased most significantly. For the spatial distributions, the difference in spatial change over the two decades was relatively small. The supply was mostly high in the northwest and low in the southeast, while the high-value areas of demand were mainly located in the eastern and southern regions (Fig. 5).

Fig. 5
figure 5

Spatial patterns of ecosystem service supply, demand, and ESDR in the BTH region in 2020. GP = grain production; CS = carbon sequestration; WY = water yield; AP = air purification; SR = soil retention; and RO = recreational opportunity

Table 2 The supply and demand of ecosystem services in 2000, 2010, and 2020

The supply of most ESs was the highest in rural areas, followed by urban–rural fringe, and the lowest in urban areas, except for grain production and water yield. Most of the demand was the highest in urban areas and the lowest in rural areas, except for soil retention. The demand for grain production and water yield were much greater in urban areas than in urban–rural fringe and rural areas (Fig. 6a and b). The carbon sequestration, air purification, and recreational opportunity demands in urban areas were similar to those in urban and rural fringe areas (Fig. 6c and d, and 6f), and soil retention demand was much greater in rural areas than in urban areas and urban–rural fringe (Fig. 6e).

Overall, for the surplus and deficit of ESs, all of the services had a spatial mismatch between supply and demand, and most were generally in deficit. The deficit of air purification service was the most serious (Fig. 6g). Urban areas were the worst-deficit areas among all ESs. In urban–rural fringe, there were also relatively serious deficits in carbon sequestration and air purification. The deficits in water yield, air purification, and recreational opportunity were also in the southeast rural areas. Surpluses in grain production were mainly located in the eastern and southern rural areas, while surpluses in carbon sequestration, air purification, and recreational opportunity were located in northwest rural areas (Fig. 5). From 2000 to 2020, deficits of most services in urban areas declined. In addition, deficits in water yield and air purification in southeast rural areas has eased, while the surplus of carbon sequestration in the rural northwest decreased, and the deficit of recreational opportunity in southeast rural areas increased (Fig. 6h).

Fig. 6
figure 6

Distributions and changes in ecosystem service supply, demand, and the ecological supply-demand ratio (ESDR) in urban areas, urban–rural fringe, and rural areas: a–f Ecosystem services supply and demand of urban areas, urban–rural fringe, and rural areas in 2000, 2010, and 2020. g ESDR of urban areas, urban–rural fringe, and rural areas in 2020. h Changes in ESDR in urban areas, urban–rural fringe, and rural areas from 2000 to 2020. GP = grain production; CS = carbon sequestration; WY = water yield; AP = air purification; SR = soil retention; and RO = recreational opportunity

Trade-offs and synergies of ES supply and demand across urban–rural landscapes in BTH

Across the urban-to-rural transitional landscapes, correlations between the supply and demand of ESs, showed different patterns. For supply, most ESs showed positive correlations in urban areas, but the correlations were weak. As the urban–rural fringe transitioned to rural areas, provisioning services and regulating services began to show negative correlations, and the correlations between all services became more significant. For demand, the strongest positive correlations were among grain production, water yield, and carbon sequestration while others had weaker positive correlations in urban areas and urban–rural fringe. In rural areas, the demand for all of the services exhibited significant positive correlations except for soil retention and other services (Fig. 7).

Fig. 7
figure 7

Correlation coefficients for pairs of ecosystem service supply and demand. GP = grain production; WY = water yield; CS = carbon storage; AP = air purification; SR = soil retention; and RO = recreational opportunity

Priority regions and trade-off curves for ESs optimization of urban–rural landscapes in BTH

According to ES supply-demand mismatch and trade-off characteristics in urban areas, urban–rural fringe, and rural areas, optimization types, corresponding key ESs, and objectives were determined for the identification of priority regions (Table 3).

Table 3 Objectives and details for ecosystem services optimization in urban–rural landscapes

For urban areas, a total of 2,000 km2 of priority regions were implemented to maximize the goal of achieving a 48% urban greening rate (Beijing Green Space System Planning). Most of these regions were located in large cities. In the large cities, priority regions were primarily distributed in both the center and periphery, while priority regions were only distributed in the periphery in small and medium-sized cities (Fig. 8a). For urban–rural fringe, 1,592 km2 of priority regions were implemented to improve both deficits and trade-offs (General Plan for Land Use in Hebei Province). Priority regions were distributed in the periphery of urban–rural fringe, close to the urban side (Fig. 8b). For rural areas, a total of 13,175 km2 of priority regions were implemented according to General Plan for Land Use in Hebei, Beijing, and Tianjin, and they were distributed in a T shape. In the north, there was a horizontal band distribution, including Zhangjiakou, Beijing, Chengde, and Qinhuangdao. From Beijing to the south, there was a vertical band distribution, concentrated in Baoding, Shijiazhuang, and Handan (Figs. 1 and 8c).

Fig. 8
figure 8

Priority regions for ecosystem services in the (a) urban areas, b urban–rural fringe, and c rural areas

Furthermore, urban–rural fringe and rural areas in which trade-off optimization existed were chosen to generate their trade-off curves. In urban–rural fringe, changes in weights had a greater impact on priority regions of water yield than that of air purification (Fig. 9a). In rural areas, changes in weights had a greater impact on priority regions of carbon sequestration than those of grain production (Fig. 9b). The most ideal situation for optimization of carbon sequestration could be achieved by changing the weights to the allocated priority regions.

Fig. 9
figure 9

Trade-off curves of ESs in (a) urban–rural fringe and b rural areas. Point A means maximizing the weight of air purification; point B means giving equal importance to air purification and water yield; point C means maximizing the weight of the water yield. The x and y axis are the contribution values of two ESs. The three land use maps in the figure correspond to points A, B, and C from left to right

Discussion

Improving the matching of ES supply and demand through effective land use management is an important component of landscape sustainability (Musacchio 2013; Wang et al. 2019; Liao et al. 2020). Compared with the general landscape sustainability science framework, we focused more on the heterogeneity of the complex urban–rural systems in landscape restoration (Wu 2013, 2021). Current studies have mainly focused on biophysical and socioeconomic changes along urban–rural gradient (Baró et al. 2016; Hou et al. 2020). However, the distributions of urban–rural regions were diverse in regions with different urban hierarchical levels. For a single center city, the city expanded outward from a particular center (Liu and Wang 2016). Gradient research effectively reflected its urban–rural characteristics. While in multi-center cities and urban agglomerations, gradient research failed to quantitatively contrast the differences from urban to rural areas (Wu 2022). This problem was effectively addressed via ES zoning management for dynamic urban–rural regions in our study (Cavender-Bares et al. 2015; Wang et al. 2019).

Considering the potential of ESs in assessing the status quo, setting targets, and prioritizing improvements in landscape sustainability (Ramyar et al. 2019), our core research idea was to translate theoretical knowledge related to ES supply and human demand into practical actions for ES optimization-based policy-related landscape planning that can advance landscape sustainability (Wu 2021). Thus, we considered the unstainable problems facing the management of ESs in different urbanization gradient landscapes. Compared with previous studies that only considered deficit optimization and trade-off optimization, the third optimization type in this study integrates ES trade-offs into ES mismatches to simultaneously reduce trade-offs and satisfy demand (Boithias et al. 2014; Turner et al. 2014; Geijzendorffer et al. 2015).

Integrating ES supply-demand mismatches and trade-offs into urban–rural landscape optimization

For BTH, the deficit in urban areas was primarily due to human needs, while the deficits in urban–rural peripheries and rural areas were largely contributed by industrial or agricultural needs (Zhang et al. 2021b). Although there were still serious deficits in grain production and water yield in urban areas, ESDRs for these two services have eased with advances in agricultural technology and increased awareness of water conservation (Pérez-Blanco et al. 2020). As the most important industrial region of China, the industrial structure of BTH is dominated by high energy consumption and high pollution entities, such as the steel and petrochemical industries (Shi et al. 2022; Wang et al. 2023). Many industrial concerns have transferred from urban areas to urban–rural fringe, resulting in high industrial emissions and serious pollution. This has led to serious deficits and ESDR decline of carbon sequestration and air purification in urban–rural fringe and rural areas, especially in the periphery of industrial cities like Tianjin, Tangshan, Shijiazhuang, and Handan. The demand for soil retention was independent of human concerns and was determined by the actual soil erosion situation, which led to its supply and demand balance and trade-off characteristics being different from other services (Morri et al. 2014; Li et al. 2022).

The differences in ES trade-offs in urban–rural landscapes could be attributed to socio-ecological factors, such as resource use pressure and land use differences between urban and rural areas. The combined effect of these factors has led to differences in the correlation patterns between the supply and demand of ESs in different regions. Urban areas often undergo large-scale urbanization processes, including land development, construction, and infrastructure development. These activities lead to ecosystem fragmentation and land use diversification, making the linkages among different ESs supply relatively weak (Morri et al. 2014). Less fragmented and more stable ecosystems have consistently led to a more significant relationship among ESs in rural areas.

The results of priority regions are related to deficits, trade-offs, and the spatial ES changes. In urban areas, deficits in grain production, carbon sequestration, and air purification were concentrated in the centers of large cities that were identified as priority regions. In addition, urban expansions resulted in ES losses in the periphery, especially in small and medium-sized cities where the ES demand was not very large, replacing deficits as the most important factor affecting priority regions (Fig. S9, Supplementary material). In urban–rural fringe, the intensity of both deficits and trade-offs was stronger with the urbanization level from urban–rural fringe to urban areas (Fig. S8, Supplementary material). Thus, priority regions of urban–rural fringe were located peripherally near urban areas. In rural areas, trade-offs of grain production and carbon sequestration gradually strengthened southward along the North China Plain. Over the past two decades, grain production supply has decreased in the northern Yanshan region. These two factors have led to a T-shaped priority region (Fig. S8 and S9, Supplementary material).

Trade-off curves for ES supply-demand optimization in urban–rural landscapes

The trade-off curves can help to identify the optimal combination of two ESs, providing more effective solutions for implementing regional optimization of ESs (Vallet et al. 2018). In urban–rural fringe, facing the trade-off between water yield and air purification, more resources can be directed toward improving the water yield as it is more sensitive. Taking urban–rural fringe as a node to compare the river ecological corridors, the connection between urban and rural water sources would be strengthened, and the reasonable spatiotemporal distribution of water resources can be achieved (Becker et al. 2022; Zhang et al. 2022). In rural areas, to ease the trade-off between grain production and carbon sequestration, natural restoration for carbon should be focused on the northern regions. The red line of ecological protection should be strictly adhered to in order to build a security pattern of ecological space in BTH (Bai et al. 2018; MEP 2015; Xu et al. 2018; Sutherland et al. 2022). In priority regions in the south, natural restoration should be based on the protection of croplands to ensure that permanent basic farmland is not encroached upon (Song and Pijanowski 2014; Zhou et al. 2021). In addition, improving the sustainable use of croplands has benefits in mitigating trade-offs between grain production and carbon sequestration. By adopting advanced agricultural technologies and practices to optimize land management, crop yields and the carbon sequestration capacity of cropland systems can be improved (Zomer et al. 2017; Bailey-Serres et al. 2019; Sha et al. 2022). The same is true for trade-offs among other services. In BTH and other urban agglomerations, fully understanding the relationship between ESs and identifying the most reasonable priority regions to improve ESs by allocating reasonable target weights can ensure that the efficient use of resources and the improvement of landscape sustainability complement each other (Srivathsa et al. 2023).

Limitations and future perspectives

In this study, considering BTH as a whole to identify urban–rural landscapes has its limitations. The development levels of Beijing, Tianjin, and Hebei are different, and this may lead to different cities being in various stages of urban–rural development. This may affect the results of urban–rural identification. The InVEST model was used in the calculation of supply for some ESs. Although the results of InVEST have uncertainties and cannot fully reflect reality, most studies have confirmed the relative reliability of the results, including in BTH (Bagstad et al. 2013; Redhead et al. 2016; Sun et al. 2018). The supply of recreational opportunity should focus not only on the area of green and blue spaces, but also on the quality of these ecological lands (Tao et al. 2022). The differences between urban and rural residents were not considered when calculating some of the services due to the difficulty in reflecting the spatial distribution of the urban and rural residents (Zhang et al. 2021b). For some services, the of potential supply-demand and actual supply-demand were not entirely consistent (Ma et al. 2017). In further research, the potential supply-demand needs to be reconciled with actual supply-demand so that ESs can better serve human well-being. In addition, we will design alternative scenarios to optimize landscape patterns based on the results of ES trade-off curves, exploring the maximization of ecological protection, grain production, and economic benefits.

Conclusions

In this study, we developed a framework for promoting urban–rural landscape sustainability from the perspective of ES supply and demand: in order to identify urban–rural landscapes, quantify the supply and demand of ESs, analyze the mismatches and trade-offs of ES supply and demand, and identify priority regions for optimization in urban–rural landscapes. From 2000 to 2020, urban areas and urban–rural fringe expanded significantly. With the improvement of natural environment and the development of society and economy, the ES demand continued to increase, but the supply of ESs did not increase fast enough to match the demand. The BTH was in a supply-demand deficit that was most serious in urban areas and involved water yield, carbon sequestration, and air purification. Trade-offs of ESs occurred in supply and were distributed in urban–rural fringe and rural areas, while there were synergistic relationships with demand. According to the differences in urban–rural landscapes, we selected corresponding priority regions of ESs for optimization. Both deficits and trade-offs in ES supply and demand varied across urban–rural landscapes and were often related to differences in the socio-ecological attributes in the development of urban–rural landscapes. The difference in spatial distributions of ESDR and trade-offs in urban–rural landscapes profoundly affected the results for priority regions. Based on ES supply-demand mismatches and trade-off relationships, the priority regions can be determined by allocating target weights. This trial can simultaneously ensure the efficient use of resources and the improvement of landscape sustainability.