Trade-Offs and Synergies Between Ecosystem Services and Their Ecological Security Patterns in the Guanzhong–Tianshui Economic Zone

Abstract

The Guanzhong–Tianshui economic zone is a strategic link in China’s Belt and Road network, faces the contradiction between ecological protection and economic development, and urgently needs to construct an ecological security pattern based on ecosystem services to permit sustainable development. In this study, we evaluated the ecological services of net primary productivity (NPP), water yield (WY), soil conservation (SC), habitat quality (HQ), and food production (FP). We explored the trade-offs and synergies between services using correlation analysis and geographically weighted regression and constructed an ecological security pattern through circuit theory. NPP, WY, SC, and FP increased during the study period, whereas HQ decreased. The NPP × WY, WY × SC, and WY × HQ shifted from synergies to trade-offs; NPP × SC, NPP × HQ, and SC × HQ were always synergies; NPP × FP, SC × FP, and FP × HQ were always trade-offs; and WY × FP shifted from trade-offs to synergies. We selected service bundles with significant synergies among NPP, SC, and HQ as ecological sources, which were mainly in the Qinling and Weibei mountains, comprising 47 ecological patches. We identified 58 ecological corridors, 330.83 km² of pinch points, and 401.30 km² of barriers, which form a mesh structure covering the study area, proposing a development pattern of six zones and one belt. Our results provide a framework for ecological protection and restoration, which may serve as a scientific foundation for upcoming regional land management initiatives.

1. Introduction

Ecosystem services are the various direct or indirect benefits that humans derive from ecosystems; they are, therefore, closely related to human well-being [1,2]. Since the 20th century, we have seriously damaged the ecological environment (e.g., soil erosion, desertification, biodiversity loss) and overexploited natural resources (e.g., shrinking lakes, decreasing carbon stocks, habitat loss) while transforming nature to meet our own needs [3,4]. As a result, more than 60% of the world’s ecosystem services are deteriorating or being overused [5].

These increasingly severe ecological and environmental problems are the focus of international attention. Many scholars have assessed the value of ecosystem services [6], explored the relationships among services [7], and discerned the factors that influence services [8]. However, these studies did not accurately identify the priority areas for protection and restoration of the ecological environment. Researchers have discovered that ecological security patterns can reveal the most significant ecological elements, so we can use them to identify priority protection areas, thereby helping land managers to stabilize the ecosystem structure and function [9]. In the present study, we constructed these patterns in order to optimize the allocation of ecological elements.

Ecological security patterns are composed of patches of important ecological functions [10]. Research on these patterns has gradually matured and formed a paradigm of identifying ecological sources, constructing resistance surfaces, and identifying ecological corridors [11,12].

Ecological sources are patches where ecosystems provide services, products, and ecological flows [13]. Many scholars have directly used forests (nature reserves) with good habitat quality as ecological sources [14] or used the morphological spatial pattern analysis models to identify ecological sources [15], but these methods are too dependent on land use data. Some scholars have also used indicators of ecological sensitivity [16], landscape connectivity [17], and the importance of services to identify ecological sources based on an assessment of the ecosystem services. However, they all ignored the trade-offs and synergies between services. Therefore, Yu et al. (2021) [18], Guo et al. (2022) [19], and Yu et al. (2022) [20] all attempted to discriminate the best combination of trade-offs among services using an ordered weighted-average model to screen the optimal protected areas as ecological sources. Although this model can integrate trade-off relationships between multiple services, its essence is to assign different ordering weights to each service, and the parameters of the assignment were mostly the same in different cases; that is, they did not reflect the differences among situations. Therefore, we took a different approach: we considered ecosystem service bundles to identify trade-offs and synergistic combinations of multiple services and their spatial distribution. Self-organizing maps (SOMs) can flexibly identify bundles and effectively manage multiple services [21]. We, therefore, used the SOM approach to identify ecosystem service bundles and, after comparing their trade-offs and synergies, selected the bundles with the most significant synergies as ecological sources.

Ecological resistance surfaces represent the degree to which species are hindered from migrating between different ecological patches [22]. Currently, there are two relatively easy ways to construct resistance surfaces. First, habitat quality (HQ) is used directly as a resistance surface (i.e., high HQ values represent low resistance values); second, direct assignment of values can be performed based on land use data [23]. However, scholars increasingly tend to construct integrated resistance surfaces that introduce some ecological indicators, such as elevation, population density, and nighttime lighting data [24]. In contrast to previous methods, the integrated resistance surface approach can account for more factors in the ecological environment, thereby making the construction process more scientific. To take advantage of this, we referred to previous studies to construct an integrated resistance surface [25]. Moreover, we introduced a human disturbance index (HDI) [26] to measure the degree of human impact on the ecosystem and used it to rationalize the resistance surface.

Ecological corridors are narrow areas that connect independent ecological patches [27]. They are extracted mainly by relying on commonly used connectivity models, such as the least-cumulative-resistance model, gravity model, and circuit theory [28,29]. The least-cumulative-resistance model assumes that paths that create weaker obstacles are more likely to be taken but does not account for driving forces; a path may be easily navigated, but if a species is not forced to take that path, it may not migrate. The gravity model solves this problem by accounting for the strength of the attraction between a new site and a species that is capable of migrating but fails to account for the ease of travel along each path. The circuit theory takes the physical concepts of electrical resistance and current flow and gives them an ecological meaning. It uses the property of random wandering of an electric charge, considers the resistance surface as a conductive surface along which the charge can migrate, calculates the effective resistance value based on the resistance imposed by the surface, and identifies channels along which services and species can flow [4]. This theory has been applied in areas such as wildlife corridor design, disease transmission, and landscape-scale gene flow [30]. Thus, we chose circuit theory to identify ecological corridors.

China’s Guanzhong–Tianshui Economic Zone (GTEZ) is an important economic strategy area, so it is necessary to explore the relationship between ecosystem services and to construct an ecological security pattern to support ecological conservation and development. Our main objectives were to (1) quantify five ecosystem services, net primary productivity (NPP), water yield (WY), soil conservation (SC), food production (FP), and habitat quality (HQ); (2) explore the trade-offs and synergies among these ecosystem services; (3) identify the optimal ecosystem service bundles as ecological sources; (4) construct an integrated resistance surface; and (5) extract ecological corridors to construct the final ecological security pattern.

2. Materials and Methods

2.1. Overview of the Study Area

The GTEZ (104°34′ E to 110°48′ E, 33°21′ N to 35°51′ N) is situated in central Shaanxi Province and southeastern Gansu Province. Its administrative scope encompasses Xi’an, Baoji, Tongchuan, Weinan, Shangluo, Xianyang, and Tianshui (Figure 1). The region is bordered by the Loess Plateau in the north, the Guanzhong Plain in the center, and the Qinling Mountains in the south [31]. Its terrain resembles a trough that is oriented from west to east, with the elevation ranging from 265 m at the center of the trough to 3754 m in the surrounding mountains. There is a typical continental monsoon climate, with the mean monthly temperatures ranging from −5 °C in January to 30 °C in July; the annual precipitation averages 617 mm [32]. The soil in the region has good moisture retention and is suitable for crops, which are mainly wheat and corn [33]. As of 2022, the total population of the GTEZ has reached 3.07 × 10⁷ people, and the GDP (gross domestic product) per capita is 68,300 yuan, an increase of 25,100 yuan compared to 43,200 yuan in 2015, showing significant economic growth [34].

The GTEZ has a profound historical heritage and substantial development potential and is, therefore, highly significant for promoting economic development along the route of China’s Belt and Road Initiative, which is intended to promote trade with Central Asia, the Mediterranean, and Europe [35]. Hence, it is necessary to evaluate the region’s ecosystems to support plans that will promote both regional ecological security and high-quality development.

2.2. Data Sources and Processing

Our dataset included data in the following categories:

(1)

Digital Elevation Model data. We obtained the data from the SRTM DEM UTM 90 m resolution digital elevation data product provided by China’s Geospatial Data Cloud (https://www.gscloud.cn/) (accessed on 12 July 2024).

(2)

Land use data. We obtained the data from a publication by Yang and Huang (2023) [36] (https://zenodo.org/records/8176941) (accessed on 12 July 2024). The data had an annual temporal resolution and a spatial resolution of 30 m. The land use types were classified into seven primary categories: cropland, forest, shrub, grassland, water, barren land, and construction land.

(3)

Meteorological data. We obtained temperature, precipitation, and potential evapotranspiration data from the China 1 km resolution monthly average temperature, monthly precipitation, and monthly potential evapotranspiration datasets provided by the National Tibetan Plateau Data Center (http://data.tpdc.ac.cn/) (accessed on 2 October 2023). We obtained solar radiation data from version 3.0 of the China Meteorological Data Network’s Ground Climate Daily Data Set (http://data.cma.cn/) (accessed on 6 October 2023) and converted the data into daily solar radiation values using the Python 2.7 programming language. We spatially interpolated the results to a resolution of 30 m with the ANUSPLIN 4.4 software to generate a grid of solar radiation on a monthly scale.

(4)

Normalized-difference vegetation index (NDVI) data. We obtained this data from the MODIS MOD13Q1 product available on the Google Earth Engine platform (https://earthengine.google.com/) (accessed on 20 October 2023), with a temporal resolution of 16 days and a spatial resolution of 250 m. We derived monthly and annual NDVI values using the maximum synthesis method implemented in the MATLAB R2022a software.

(5)

Soil data. We obtained this data from version 1.1 of the Harmonized World Soil Database (https://www.fao.org/soils-portal/data-hub/soil-maps-and-databases/harmonized-world-soil-database-v20/en/) (accessed on 24 October 2023), characterized by a spatial resolution of 1 km, and it was extracted using ArcGIS 10.7 to obtain the soil texture, soil bulk density, and rooting depth data.

(6)

Population spatial distribution. We obtained this data from the LandScan platform (https://landscan.ornl.gov/) (accessed on 24 July 2024), with a resolution of 1 km, to provide a gridded representation of the population distribution.

(7)

Food production data. We obtained this data from Gotohui (https://www.gotohui.com/) (accessed on 15 July 2024). It provides cereal production figures for Xi’an, Baoji, Tongchuan, Weinan, Shangluo, Xianyang, and Tianshui.

All spatial data were converted to the CGCS2000_3_Degree_GK_CM_108E projection after preprocessing and resampled to a spatial resolution of 100 m. We performed the processing using the projection and resampling functions provided by ArcGIS. The time period was from 2000 to 2022, except for the digital elevation model data and soil data.

2.3. Research Methods

The research methodology presented in this paper encompasses the following steps (Figure 2), which are described in more detail in the following subsections: (1) We evaluated five ecosystem services (NPP, WY, SC, FP, and HQ) using the Carnegie–Ames–Stanford Approach (CASA) and InVEST models; (2) We explored the trade-offs and synergies among ecosystem services using correlation analysis and GWR; (3) We identified ecosystem service bundles using the SOM model and used them to select ecological sources; (4) We selected four natural factors and four social factors as influencing variables, determined each factor’s weight using the analytic hierarchy process (AHP), and then developed an integrated resistance surface; (5) We extracted ecological corridors, pinch points, and barriers using circuit theory to develop an ecological security pattern.

2.3.1. Quantifying Ecosystem Services

The GTEZ confronts significant ecological challenges, including forest and grassland degradation, water scarcity, soil erosion, and obstacles to achieving food self-sufficiency. Consequently, we selected NPP, WY, SC, FP, and HQ as essential ecosystem service indicators for a comprehensive assessment of the relationships among services and identification of ecological sources (

Table 1. Calculation methods for ecosystem service. NPP, net primary production; WY, water yield; SC, soil conservation; FP, food production; HQ, habitat quality.

Ecosystem Service	Calculation
NPP	NPP(x,t) = APAR(x,t) × ε(x,t)
	NPP is the net primary productivity of vegetation at time t for pixel x (gC/m2); APAR is the absorbed photosynthetically active radiation (MJ/m2); and ε is the actual light energy utilization (gC/MJ).
WY	WY(x) = [1 − AET(x)/P(x)] × P(x)
	WY is the annual water yield of pixel x (mm); AET is the annual actual evapotranspiration (mm); and P(x) is the annual precipitation (mm).
SC	RKLS = R × K × LS USLE = R × K × LS × C × M SC = RKLS − USLE
	SC is the soil conservation amount (t/hm2); RKLS is the potential soil erosion amount (t/hm2); USLE is the actual soil erosion amount under ecological management measures (t/hm2); and R, K, LS, C, and M indicate the rainfall erosion factor, the soil erodibility, slope length factor, vegetation cover factor, and management factor and account for soil and water conservation.
FP	${FP}_x={FP}_{\mathrm{t}}\times {\displaystyle \frac{{VCI}_x}{\sum _{i=1}^n{VCI}_i}}$
	${VCI}_x=\left({\displaystyle \frac{{NDVI}_x-{NDVI}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{{NDVI}_{\mathrm{m}\mathrm{a}\mathrm{x}}-{NDVI}_{\mathrm{m}\mathrm{i}\mathrm{n}}}}\right)$
	FP is the food production of pixel x (t/hm2); FPt is the total food production in the study area; n is the total number of cropland pixels in the study area; NDVIx is the normalized-difference vegetation index for cropland pixel x; and NDVImax and NDVImin are the annual NDVI maxima and minima, respectively, for cropland.
HQ	${HQ}_{xj}=H_j\left[1-\left({\displaystyle \frac{D_{xj}^Z}{D_{xj}^Z+k^z}}\right)\right]$
	HQxj is the habitat quality of pixel x in land use type j; H is the habitat suitability for that land use type; Dxj is the level of stress on pixel x in that land use type; k is the half-saturation factor, which is usually taken as half the maximum value of Dxj; and Z is a normalized constant that is usually taken as 2.5.

).

(1)

NPP denotes the organic carbon sequestered by green plants per unit time and unit area through photosynthesis after deducting the plant’s respiratory consumption of photosynthate [37]. We used the CASA model to estimate NPP.

(2)

WY denotes the capacity of an ecosystem to produce and retain water, which is influenced by both natural conditions and anthropogenic land cover changes [38]. We used the water yield module of the InVEST model to assess WY. We validated the WY against the total water resources of the Wei River Basin (Water Resources Bulletin of the Shaanxi Provincial Water Resources Department) to ensure that the results are reliable.

(3)

SC denotes the role of soil within an ecosystem in controlling erosion and intercepting sediment flows, which are influenced by topography, vegetation, and various ecosystem components [39]. We used the soil conservation module of the InVEST model to assess SC.

(4)

FP denotes the production of crops such as cereals, pulses, and potatoes provided by the ecosystem [40]. Previous research has demonstrated a strong linear relationship between NDVI and FP [41]. Consequently, we drew upon the findings of Liu et al. (2022) [42] to calculate FP based on the NDVI.

(5)

HQ denotes the capacity of an ecosystem in terms of its ability to provide suitable living and developmental conditions for individuals, populations, and communities, thereby reflecting biodiversity [43]. We used the habitat quality module of the InVEST model to assess HQ.

2.3.2. Quantification of Trade-Offs and Synergies Between Ecosystem Services

We characterized the trade-offs and synergies between pairs of ecosystem services by computing Pearson’s correlation coefficient (r). When 0 < r < 1, the pairs exhibit synergy; when −1 < r < 0, the pairs exhibit a trade-off. We calculated the r values for each pair of ecosystem services in 2000, 2010, and 2022 using the corrplot package for version 4.4.1 of the R software and identified a total of 10 pairs.

GWR serves as an extension of traditional linear regression models. It can modify the parameters of a traditional model by accounting for the geographic coordinates for locations where data were obtained, thereby integrating the spatial characteristics of geographic data into the regression parameters [44]. Therefore, we chose GWR to determine the spatial relationships between ecological services. We input two services into the “GWR Model” [45] package of the R 4.4.1 software as independent and dependent variables to calculate the outcomes. Because there is only one continuous independent variable in the calculation, there is no multicollinearity problem. Next, we reclassified the regression coefficients as positive if they represented a synergy and negative if they represented a trade-off:

$$y_i={\mathsf{\beta }}_0\left(u_i,v_i\right)+\sum _{j=1}^k{\mathsf{\beta }}_{bwj}(u_i,v_i)x_{ij}+{\mathsf{\epsilon }}_i$$

(1)

where y_i is the dependent variable; x_ij is the independent variable; k is the number of independent variables; (μ_i, ν_i) is the spatial location of point i; ε_i is the random error; β₀ (μ_i, ν_i) denotes the intercept at point i; and β_bwj (μ_i, ν_i) denotes the regression coefficient.

2.3.3. Construction of Ecological Security Patterns

Identification of Ecological Sources

Ecological sources are key ecological patches that maintain ecosystem integrity [19]. We used the “kohonen” package in the R software to perform the SOM to identify ecosystem service bundles. The SOM integrates principal-components analysis with K-means clustering to preserve the topology of the input space through the adjacency function and to add spatial information to the analysis for classification processing [46]. So, in bundles, ecosystem services that depend on each other will increase or decrease simultaneously, or one will increase and the other decrease; this helps to reveal the trade-offs and synergies among multiple services [8,47]. At this point, we can use the identified bundles to select combinations with significant synergies between services as ecological sources.

Constructing an Integrated Resistance Surface

Ecological resistance surfaces refer to the ability of a landscape or ecosystem characteristic to obstruct the migration of species. Topography (e.g., elevation, slope) affects the distribution and stability of ecosystems; climatic conditions determine ecosystem type and function; land use change directly alters the structure and connectivity of ecosystems (A32); and socioeconomics exerts pressure on ecosystems through the intensity of human activity (e.g., human disturbance, road networks, population distribution).

Therefore, we chose four natural factors (elevation, slope, vegetation cover, and distance from rivers) and four social factors (land use type, a human disturbance index, the distance from roads, and population) to construct an integrated resistance surface. We used the Yaahp 12.7 software to perform the analytic hierarchy process and determine the assigned value of factors. The consistency ratio was 0.0866, which is <0.10; thus, the results were acceptable. Subsequently, we used ArcGIS to assign resistance coefficient values (ranging from 1 to 5) to the nodes in the resistance surfaces, with larger resistance coefficients indicating higher resistance for species during migration (

Table 2. Resistance factors and the weight evaluation index system. HDI, human disturbance index.

Resistance Factors		Score					Weight
		1	2	3	4	5
Natural factors	Elevation (m)	<600	600–1200	1200–1800	1800–2400	>2400	0.0503
	Slope (°)	<10	10–15	15–20	20–30	>30	0.0503
	Vegetation cover	>0.85	0.75–0.85	0.65–0.75	0.50–0.65	<0.50	0.1458
	Distance from river (km)	0–10	10–20	20–40	40–60	>60	0.0750
Social factors	Land use type	forests, shrubs	grassland, water	cropland	barren	construction land	0.3452
	HDI	<0.2	0.2–0.4	0.4–0.6	0.6–0.8	>0.8	0.2123
	Distance from roads (km)	>20	10–20	2–10	1–2	0–1	0.0349
	Population (×103)	<0.5	0.5–2	2–6	6–10	>10	0.0861

).

Furthermore, we developed a human disturbance index (HDI) as one of the social factors [26]. It is primarily a measure of the extent of human utilization of land and is calculated as follows:

$$HDI=\sum _{i=1}^m\left({\displaystyle \frac{A_i}{A}}\right)\times C_i$$

(2)

where m is the number of land use types; A_i is the area of category i of land use (km²); A is the total area of the study area (km²); and C_i is the coefficient of anthropogenic interference for land use category i. Referring to previous studies, we set C_i for cropland, forest, shrub, grassland, water, barren land, and construction land to 0.55, 0.10, 0.20, 0.23, 0.115, 0.14, and 0.95, respectively, with larger values indicating greater human disturbance [48,49].

Extracting Ecological Corridors and Identifying Pinch Points and Barriers

We used the Linkage Mapper toolbox and Circuitscape 4.0.5 software to simulate the characteristics of random current travel and the magnitude of current intensity based on circuit theory to determine the direction of ecological flow and the locations of pinch points and barriers in ecological corridors and then identify multiple migration paths of species between ecological sources. We designated pinch points as key protection areas and ecological barriers as the priority areas for restoration.

3. Results

3.1. Spatial and Temporal Distribution of Ecosystem Services

All five ecosystem services within the GTEZ exhibited spatial heterogeneity (Figure 3). NPP was 870.36 ± 58.45 gC/m² (the multi-year average ± the standard deviation), which increased significantly from 2000 to 2022, and its highest values (NPP > 1200 gC/m²) were mainly in the Qinling Mountains, whereas its lowest values (NPP < 300 gC/m²) were mainly in the urban areas of Xi’an and Baoji as well as in the northwestern part of Tianshui.

WY was 94.19 ± 55.29 mm, with a significant increase from 2000 to 2022. Its highest values (WY > 300 mm) were concentrated in the urban areas of Xi’an, Tianshui, and Baoji. These areas have high surface hardness (low water absorption), and precipitation does not easily infiltrate the soil and can produce streams directly. The lowest values (WY < 50 mm) were mainly situated in the Weibei Mountains. The WY was only 66.55 mm for forest, which is due to the high vegetative cover of forest, which traps a lot of water.

SC was 38.06 ± 10.19 kt/hm² but did not change significantly from 2000 to 2022. Its highest values (SC > 100.00 kt/hm²) were in the Qinling Mountains. The lowest values (SC < 16.00 kt/hm²) were in the Guanzhong Plain, which is dominated by cropland and is at increased risk of soil erosion.

FP was 1.13 ± 0.07 t/hm² but did not change significantly from 2000 to 2022. The Guanzhong Plain had high FP values (FP > 3.20 t/hm²). Tianshui also had a high FP value, averaging 1.07 t/hm² for all years.

HQ was 0.38 ± 0.01 but did not change significantly from 2000 to 2022; this indicates poor ecological conditions in the study area. Its highest values (HQ > 0.8) were in the Qinling Mountains and the Weibei Mountains. The lowest values (HQ < 0.35) were in the Guanzhong Plain, where human impacts are high.

3.2. Trade-Offs and Synergies Between Ecosystem Services

The relationships among the five ecosystem services differed both spatially and temporally (

Table 3. Values for trade-offs and synergy between ecosystem services. All correlations were statistically significant at p < 0.001. NPP, net primary production; WY, water yield; SC, soil conservation; FP, food production; HQ, habitat quality.

	NPP × WY	NPP × SC	NPP × FP	NPP × HQ	WY × SC	WY × FP	WY × HQ	SC × FP	SC × HQ	FP × HQ
2000	0.08	0.60	–0.53	0.45	0.29	–0.15	0.19	–0.70	0.44	–0.58
2010	–0.11	0.62	–0.53	0.47	0.06	–0.01	0.04	–0.69	0.45	–0.53
2022	–0.50	0.53	–0.38	0.35	–0.33	0.43	–0.22	–0.65	0.47	–0.44

and Figure 4). During the study period, the NPP × WY value changed from positive (0.08) to negative (–0.50), indicating a shift from synergies to trade-offs. The synergy zones were distributed in the Qinling Mountains, the Weibei Mountains, and Tianshui, and the area increased from 15.91 × 10³ to 20.70 × 10³ km². The NPP × SC, NPP × HQ, and SC × HQ values were always positive, indicating that they were synergies. Their synergy zones were widely distributed, with the area of NPP × SC increasing, the area of NPP × HQ increasing and then decreasing, and the area of SC × HQ decreasing. The NPP × FP, SC × FP, and FP × HQ values were always negative, indicating that they were trade-offs. Their trade-off area decreased throughout the study period, and the synergy zones were mainly distributed in the Guanzhong Plain. The WY × SC and WY × HQ values changed from 0.29 to –0.33 and 0.19 to –0.22, respectively, indicating a shift from synergies to trade-offs. The WY × SC synergy zones were mainly distributed in the Qinling Mountains and Guanzhong Plain, with the area decreasing from 23.32 × 10³ to 19.53 × 10³ km²; the WY × HQ synergy zones were distributed in the Qinling Mountains and Tianshui, with the area decreasing from 32.87 × 10³ to 23.70 × 10³ km². The WY × FP value changed from –0.15 to 0.43, indicating a shift from trade-off to synergy. All regions except the Guanzhong Plain were synergies, and the area increased from 52.10 × 10³ to 56.39 × 10³ km².

3.3. Distinguishing Ecological Security Pattern

3.3.1. Ecological Source Selection Based on Ecosystem Service Bundles

We identified the optimal ecological service bundles through the SOM, and the spatial and temporal distributions of the various types of service bundles are presented in Figure 5. Overall, the bundles in 2000 and 2010 presented a similar spatial structure. The Qinling Mountains were dominated by bundles I, II, and III, and the Guanzhong Plains were dominated by bundle VI. In 2022, the bundles were more heterogeneous, and the area of bundle IV expanded into the Guanzhong Plains.

Based on the types of ecosystem service bundles in each year, we selected the bundles with significant synergy as the ecological sources; these were type I and II bundles in 2000 and 2022, respectively, and the type II and III bundles in 2010. In addition, to account for the more fragmented ecological patches, we selected the ecological patches with an area larger than 50 km² as the final ecological sources (Figure 6a). We selected 47 ecological patches with a total area of 30.47 × 10³ km², accounting for 38.1% of the study area. They were mainly distributed in the Qinling Mountains and the Weibei Mountains, with more than 90% of the entire source area forested and with good ecological quality.

3.3.2. Resistance Surface Construction

We chose eight factors to establish the integrated resistance surface (Figure 6b). The Qinling Mountains had high elevations and steep slopes, so their resistance values were relatively high. The resistance values in the Guanzhong Plains, which have low vegetation cover, were also notably higher than those in other areas. Resistance values were also relatively high in Xi’an, Tianshui, and Baoji, where the HDI and population densities were high. The greater the distance (proximity) from a river or road, the higher the assigned resistance value. The resistance values were generally relatively low in areas with favorable natural conditions (such as the Qinling Mountains and the Weibei Mountains) and relatively high in areas with intense human activities (such as the Guanzhong Plains and western Tianshui).

3.3.3. Ecological Security Pattern Construction

Based on the ecological sources and resistance surfaces defined in Section 3.3.1 and Section 3.3.2, we described the ecological security patterns of GTEZ using circuit theory and the Linkage Mapper tool. These patterns consisted of three elements: ecological corridors, pinch points, and barriers (Figure 7). We found 58 ecological corridors with an average length of 21.22 ± 27.88 km and a total length of 1231.02 km, with the three longest ones (a length > 78.82 km) all situated in Weinan. We categorized the corridors into three levels based on the current value; the higher the level, the more important it was.

4. Discussion

4.1. Characterization of Ecosystem Service Interactions

Many ecosystem services exist, and elucidating the intricate trade-offs and synergies among these services can improve the understanding of the detrimental effects of altering these services in an ecosystem, thereby supporting efforts to improve protection of the ecological environment. Our results showed high regional heterogeneity and temporal variation in the links among ecosystem services. Overall, NPP × WY, WY × SC, and WY × HQ shifted from synergies to trade-offs; NPP × SC, NPP × HQ, and SC × HQ were always synergies; NPP × FP, SC × FP, and FP × HQ were always trade-offs; and WY × FP shifted from trade-offs to synergies. This result is consistent with previous studies in which supply services (FP) and other services (WY and HQ) were trade-offs [50,51], and regulating services (NPP) and other services were synergies (WY and SC) [7,52]. However, this differs from the findings of Niu et al. (2021) [53], who investigated the synergies between WY and FP. This discrepancy may result from differences in local climate, regional topography, vegetation species, and the nature and intensity of human activity in the two regions.

Precipitation is a limiting factor for vegetation growth and for the recharge of bodies of water and is an erosion factor for soils [54,55]. Therefore, we compared precipitation data from 2000 to 2022 in our study area and found that Tianshui, the Weibei Mountains, and the Qinling Mountains regions had a large increase in precipitation. This probably explains the NPP × WY and WY × HQ synergies [56]. Also, a longer growth season for vegetation may have resulted from recent climate change [57], thereby increasing the NPP of vegetation where water was not limited. Ecological improvements have led to the expansion of forest areas, and their favorable hydrothermal conditions and rich biomass provide greater capacity for carbon sequestration, soil conservation, and habitat quality; this resulted in synergies between NPP, SC, and HQ [58]. This agrees with findings on the Loess Plateau and in its sub-regions [59,60].

It is noteworthy that the Guanzhong Plains are the main food production base in Shaanxi Province and its core area for the construction of urban agglomerations, so the interior of the plains is heavily impacted by human activity. NPP × WY and WY × HQ were trade-offs in the Guanzhong Plains, which, due to the high water demand for human domestic production in this zone, reduced surface runoff and thereby affected the relationships among ecosystem services [61]. In addition, it has been shown that anthropogenic disturbances alter the potential trade-offs between ecosystem services and show actual synergies and that the increase in synergy will become more noticeable as the intensities of anthropogenic activities increase [62]. This could be the cause of NPP × FP, SC × FP, and FP × HQ being synergies in the Guanzhong Plains. In contrast, the shallow root systems and low biomass of crops can make the soil less porous and less able to retain water, leading to a WY × FP trade-off in the Guanzhong Plains [63].

4.2. Recommendations for Optimization and Management of Spatial Ecological Patterns

We took the ecological sources as the basis for management and combined the ecological corridors, the resistance surface, and the land use types for zoning, thus building an ecological security pattern with six zones and one belt (Figure 8). The belt stretches along the southern edge of the Weibei Mountains through the eastern and western parts of the study area. It is an important pathway for the transfer of ecological information and species migration flow and is flanked by ecological patches of varying density. The six zones are the Qinling Mountains Nature Ecological Reserve, the Weibei Mountains Nature Ecological Reserve, the Key Urban Construction Zone, the Guanzhong Agro-ecological Restoration Zone, the Village Improvement Zone, and the Tianshui Ecological Protection and Control Zone. In the rest of this section, we will discuss our recommendations for each of these zones.

The Qinling Mountains Nature Ecological Reserve has bountiful ecological resources and now represents a relatively complete nature protection system [64] (Figure 8a). For instance, with Qinling National Park, which is the reserve’s main body, the protected area is divided into core areas (elevation > 2000 m), key areas (1500 m < elevation < 2000 m), general areas (the remaining area), and construction control zones based on the characteristics of vertical differentiation of the Qinling mountains and considering the consistency of ecological functions. Implementing area-wide protection and zoning control can be based on these core areas. We should also pay attention to the relationship between NPP, WY, SC, and HQ in the region so that the fragmented synergistic area can be made more complete and the sustainable development of the region can be further ensured.

The Weibei Mountains Nature Ecological Reserve is characterized by fragmented ecological sources, complicated ecological corridors, and a fragile ecological environment (Figure 8b). In recent years, China’s implementation of many ecological conservation projects has increased the areas of forest and grassland in the reserve [12], but the quality of their habitats remains unsatisfactory (HQ averaged 0.40). Thus, we offer the following suggestions: First, the quantity and quality of ecological sources should be improved; second, the pinch point areas at the edge of the ecological source should be considered priority conservation areas; and finally, in conjunction with the level of ecological corridors, priority should be given to improving the first level of corridors [65].

The Key Urban Construction Zone is an area of concentrated human activity and high ecological resistance values (3.42 ± 0.23) (Figure 8c). The excessive development of built-up land in cities has led to extremely poor ecological quality (HQ = 0). Therefore, the construction of garden cities and forest cities must be fully implemented. Controlling the quantity and quality of construction land and promoting a recycling and low-carbon way of production and living would also improve the urban environment. Further, neighboring cities should also build good ecological pinch points in their urban agglomerations and increase green space in order to change the trade-offs between services [66].

The Guanzhong Agro-ecological Restoration Zone contains a dense concentration of ecological barriers (Figure 8d), so efforts should focus on restoration and the construction of a complete ecological corridor. Because the zone is a major hub for food production and faces environmental challenges, such as soil erosion and declining soil quality, we recommend that the soil environment of the zone’s agricultural land be managed to improve its quality. Further expansion of the SC × FP synergistic area is based on ensuring food production.

The Village Improvement Zone has a wide distribution of cropland and a high population density (Figure 8e). It is, therefore, important to harmonize the contradictions between ecological protection and socioeconomic development in the region. The focus should, therefore, be on efforts to strengthen the ecological infrastructure, such as improving green space quality and promoting ecological restoration projects [16]. Simultaneously, multi-functional ecological corridors should be established to facilitate ecological flows in more natural landscapes, mitigate the overexploitation of arable land, and form multi-functional agricultural landscapes.

The Tianshui Ecological Protection and Control Zone is an important water conservation area in the upper reaches of the Wei River (Figure 8f). It has sparse ecological sources and corridors but dense pinch points, indicating the need to enhance forest resource cultivation in the district to promote structurally intact and functionally stable forest ecosystems [67]. The zone also faces ecological problems of soil erosion and poor vegetation cover. Therefore, restoration should focus on building soil and water conservation projects with a view to controlling soil erosion by water at the source.

4.3. Research Gaps and Future Plans

In this paper, we explored the relationships among ecosystem services by quantifying the ecosystem services of the GTEZ. To do so, we identified the ecological security patterns for the zone and proposed recommendations for ecological protection and restoration. However, our study has the following shortcomings:

(1)

There are many types of ecosystem services. Although we studied regulating, supplying, and supporting services, we did not account for cultural services. The associated human leisure and recreation services should be incorporated into the assessment system in future research.

(2)

Our analysis relied on correlation and regression analyses to explore the relationships among ecosystem services. However, these methods are relatively simplistic and do not delve into the influencing mechanisms that underlie the relationships. Thus, our analytical methods should be enhanced using methods such as structural equation modeling and random forest analysis in the future to provide insights into the mechanisms by which the external environment influences the relationships among ecosystem services.

(3)

When we defined the ecological security patterns, our focus on the identification of the ecological sources did not involve testing the desirable or required widths of the ecological corridors. In the future, more attention should be paid to objectively defining the width of the ecological corridors and to exploring evaluation methods for this analysis so that the patterns can be optimized according to the feedback results.

5. Conclusions

In this paper, we delved into the spatial and temporal variation in ecosystem services within the GTEZ, defined the trade-offs and synergies among the services, and established an ecological security pattern for the zone. Our main conclusions are the following:

(1)

NPP, WY, SC, and FP generally increased during the study period. In contrast, habitat quality HQ decreased. The high-value areas for NPP, SC, and HQ were mainly distributed in the Qinling Mountains, whereas the low-value areas were mainly distributed in the Guanzhong Plains. The opposite was true for FP.

(2)

NPP × WY, WY × SC, and WY × HQ shifted from synergies to trade-offs; NPP × SC, NPP × HQ, and SC × HQ were always synergies; NPP × FP, SC × FP, and FP × HQ were always trade-offs; and WY × FP shifted from trade-offs to synergies.

(3)

The ecosystem service bundles in different time periods had similar spatial structures. We selected the bundles with significant synergies among NPP, SC, and HQ as the ecological source. We found 47 patches that together accounted for 38.1% of the study area, and they were mainly located in the Qinling Mountains and the Weibei Mountains.

(4)

In the ecological security pattern we constructed, we identified 58 ecological corridors that spread throughout the study area in a net-like structure. The pinch points had forest and grassland as the main land use types. The barriers had cropland and construction land as the main land use types.

(5)

We defined an ecological security pattern with six zones and one belt that spanned the entire study area from west to east.

Based on these analyses, we suggest that the ecological policy of the district revolves around ecological protection and restoration, optimization of ecological corridors, green urban development, sustainable use of cropland, and soil erosion control, aiming to build an efficient and stable ecological security pattern. In the future, our research direction will focus on (1) incorporating the cultural service category into the system for assessing ecosystem services; (2) adopting more complex methods to deeply explore the relationship between services and the influence mechanisms behind them; and (3) placing emphasis on the pre-determination of the width of ecological corridors and the assessment methodology in the ecological security pattern construction.