Ecological Indicators 138 (2022) 108794

Contents lists available at ScienceDirect

Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind

Connecting urban area with rural hinterland: A stepwise ecological security

network construction approach in the urban–rural fringe
Chen Liang a, b, Jian Zeng b, d, Ruo-Chen Zhang c, *, Qian-Wen Wang b
a
School of Architecture and Art Design, Hebei University of Technology, Tianjin 300130, PR China
b
School of Architecture, Tianjin University, Tianjin 300072, PR China
c
School of Civil and Transportation Engineering, Hebei University of Technology, Tianjin 300401, PR China
d
Resilient City Council, Chinese Society for Urban Studies, Beijing 100835, PR China

A R T I C L E I N F O A B S T R A C T

Keywords: Rapid and disordered urbanization has created a fragmented landscape, the urban–rural fringe (URF), which has
Ecological security network led to the degradation of ecological connections between urban areas and rural hinterlands. The ecological se­
Urban-rural fringe curity network (ESN) offers significant advantages for restoring interrupted urban–rural ecological processes.
Landscape connectivity
However, classic ESN construction approaches are ineffective in the URF when applied to a highly complex
Ecological restoration
Ecological corridor
urbanization landscape. In this paper, a new approach for a stepwise ESN construction in the URF is proposed,
Urban planning which integrates the basic ESN and the connecting ESN to create an optimized ESN. Taking Xiamen as a case
study, the ecological elements of the URF were extracted and reclassified for ESN construction. The scarcity of
transitional patches and complexity of land cover were responsible for the lack of urban–rural connecting cor­
ridors. The final optimized ESN consisted of 19 source nodes with 7 rural structural corridors, 22 urban–rural
connecting corridors, and 8 urban structural corridors. Moreover, the structural performance and spatial
coverage of the ESN were both significantly improved after optimization. The optimal buffer widths of rural
structural, urban–rural connecting, and urban structural corridors were 600 m, 180 m, and 120 m, respectively.
This paper provides an operational framework of ESN construction adapted for the URF as well as management
guidance for policymakers. It can support decisions for regulating urban growth and outlining ecological pro­
tection areas, thus contributing to sustainable urban development.

1. Introduction has become increasingly urgent.

As cities grow, the interdependence and complementarity of urban
In recent decades, rapid and disordered urbanization has induced areas and rural hinterlands become stronger. Kroll et al. (2012) high­
fundamental changes in land use and landscape patterns worldwide lighted that urban regions are essential locations for ecosystem service
(Kalnay and Cai, 2003; Cui et al., 2020). These changes cause serious demands, and the ecological network embedded in the URF is the carrier
environmental degradation in urban and surrounding areas (Peng et al., that maintains diverse urban–rural ecological processes (Peng et al.,
2018a). Located in the transition zone between urban and rural areas 2018b; Yu et al., 2017). Nevertheless, the paradigm of modern urban
(Pryor, 1968), the urban–rural fringe (URF) is not just the frontier of planning has been deeply influenced by von Thünen’s (1875) mono­
urban expansion but also frequently an area of ecological sensitivity. centric model of higher land values clustering at the center of urban
Thus, the dynamic and uneven socioeconomic activities in the URF have areas. This has led to an expansion mode of circular sprawl on the urban
led to intense contradictions between urban construction and ecological periphery (Deng et al., 2020; Allen, 2016), where ecological land cover
protection (Scott et al., 2013). However, changing lifestyles, infra­ is replaced by construction land. Furthermore, as the URF landscape
structure development, increased mobility, and a growing population becomes severely fragmented (Brinkley, 2018), the remaining ecological
are still driving disordered low-density development in the URF (Shaw islands (isolated patches) are no longer able to support the ecological
et al., 2020), which inevitably exacerbates environmental degradation. processes between the urban area and the rural hinterland. Therefore,
As a result, the optimization of ecological security patterns in the URF the ecological risks of urban heat islands, storm floods, and biodiversity

* Corresponding author.
E-mail address: [email protected] (R.-C. Zhang).

https://doi.org/10.1016/j.ecolind.2022.108794
Received 29 December 2021; Received in revised form 5 March 2022; Accepted 18 March 2022
Available online 23 March 2022
1470-160X/© 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
C. Liang et al. Ecological Indicators 138 (2022) 108794

loss continue to grow in urban areas due to an insufficient supply of 2. Study area and data source
ecosystem services (Aronson et al., 2017; Liang et al., 2021). In
response, urban green belt design took off with the London plan of 1944 Xiamen is located on the southeast coast of Fujian Province, China
(Abercrombie, 1944) and has continued to spread in mega-city planning. (117◦ 53′ E-118◦ 26′ E, 24◦ 23′ N-24◦ 54′ N). The city consists of mainland
Although green belts have proven to be effective in controlling urban and islands and is rich in subtropical marine and terrestrial animals and
expansion and improving ecosystem services (Yokohari et al., 2000; plant resources. Xiamen is a typical high-density city in eastern China,
Casado-Arzuaga et al., 2013), they are still regarded as another ring in with a population of 4.29 million people and a gross domestic product
the urban area, which does not break the circle pattern, and thus cannot (GDP) of 599.50 billion yuan in an urban area of 1699 km2. Since the
solve the problem of urban–rural ecological isolation. 1980s, Xiamen experienced rapid urbanization, which has greatly
The ecological security network (ESN) is a spatial pattern of critical impacted the landscape pattern, especially in the URF. Liang and Zeng
landscape elements that enables the achievement of regional ecological (2021) reported that the Xiamen URF continues to expand and move
sustainability by integrating landscape patterns with ecological pro­ from the central island toward the outer region. Since the ESN will not
cesses (Peng et al., 2018b; Li et al., 2020). The ESN can effectively cross the sea, the study was conducted solely on the mainland region of
promote the regional circulation of material and energy and maintain Xiamen (Fig. 1). The case study for Xiamen is based on the following
the structural and functional stability of the ecosystem (Dai et al., 2021). considerations: (1) Xiamen is a coastal city with a sensitive environment
Based on the patch-corridor-matrix theory proposed by Forman (1995), threatened by disordered artificial exploitation in recent years; (2)
it is well known that the ESN construction process identifies and regu­ Xiamen is a high-density city with little land suitable for construction.
lates strategic nodes and corridors. Hence, for the fragmented landscape Thus, the relationship between urban and rural areas is tense at the URF.
of the URF, the ESN has a notable advantage in restoring interrupted Three types of data are required for this study. (1) LULC data are the
ecological processes through the integration of existing ecological ele­ basis of ESN identification. Gong et al. (2019) produced the Finer Res­
ments. To extract the spatial configuration of the ESN, the minimum olution Observation and Monitoring of Global Land Cover (FROM-GLC)
cumulative resistance (MCR) model is widely used to simulate the product at a 10 m resolution in 2017 (https://data.ess.tsinghua.edu.
linkages and interactions between ecological nodes by calculating the cn/). Before application, it was reclassified as cropland, forest, grass­
resistance surface (Shan et al., 2019; Luo et al., 2020); empirical studies land, water, construction land, and bare land. (2) Nighttime light data
are mostly conducted at early stages in wild areas with natural land were generated by the Luojia-1A satellite image at 130 m resolution
cover (Knaapen et al., 1992). Recently, it has been noted that the MCR (https://59.175.109.173:8888/index.html). (3) Vector data of the
model has been applied to ESN identification in heavily urbanized areas transport network and administrative boundary at a scale of 1:1000000
(Luo et al., 2020; Wang et al., 2021). However, in most cases, the built- were obtained from the National Catalog Service for Geographic Infor­
up area is isolated from the ESN system for two main reasons: (1) the mation of China (https://www.webmap.cn/).
MCR model is not sensitive to the complex resistance surface of the
urban region (Su et al., 2016), and therefore, some simulated corridors 3. Material and methods
are not feasible for planning practice; (2) the ecological patches in the
urban region are small and discontinuous (Liu and Yang, 2015), and The ESN in the URF addresses the landscape fragmentation and
hence, they are often not selected as an ecological source due to their ecological process interference caused by disordered urbanization by
low ecological importance or connectivity. Therefore, this classic optimizing the spatial connection from urban areas to rural hinterlands.
approach cannot be directly applied to the URF without adaptation. Due to the vast difference between socioeconomic systems and natural
Constructing a directional and connecting ESN in a transitional zone ecosystems, the components of the ESN show diverse functions in urban
between urban–rural areas remains challenging. Additionally, economic and rural districts. Hence, a spatial classification system was proposed
booms and population growth have meant that the implementation of (Fig. 2), which was the premise of ESN identification in the URF. In this
the ESN planning faces intense pressure from urban expansion in the study, the ESs were further divided into three categories: urban, tran­
URF (Yu et al., 2017). For example, although greening network planning sitional, and rural. Urban ecological sources (UESs) refer to patches with
started in Xiamen, a mega-city in southeast China, in the 2000s, Xiamen large-scale artificial development activities around, which are primarily
failed to maintain its ecological structure over the following decade
(Fig. S1 in the supplementary material). Despite the application of the
Basic Ecological Line in the planning, most of the reserved ecological
space in the URF was occupied (e.g., forest, grassland, and water) due to
the incompatibility between corridor width management and urban
expansion demand. Hence, it is necessary to identify the compositional
characteristics of land use and land cover (LULC) within different
corridor widths and then support the planning decision of the ESN.
Furthermore, given the high heterogeneity and dynamics of the URF
(Liang and Zeng, 2021), the elements of the ESN need differentiated
management strategies rather than being static and uniform as in pre­
vious work.
In this paper, a new approach to stepwise ESN construction in the
URF is proposed for restoring the ecological connections between urban
areas and rural hinterlands. This paper aims to (1) extract and classify
the ESN elements in the URF in accordance with location and artificial
disturbance; (2) integrate the three identification steps of the basic ESN,
connecting ESN, and optimized ESN to establish directional urban–rural
ecological linkages; and (3) explore the structural characteristics of the
ESN and search for the optimal buffer width for a variety of ecological
corridors. The developed approach can equip policymakers with oper­
ational ESN analysis and identification toolboxes tailored to urbanized
areas, as well as management guidance, thus contributing to a sustain­
able landscape pattern of high connectivity in the URF. Fig. 1. The study area of Xiamen.

2
C. Liang et al. Ecological Indicators 138 (2022) 108794

Fig. 2. The spatial system for classifying ESN elements in the URF.

the ecological space reserved within the built-up area. Urban ecological identification and integration, including two substeps. The first is to
sources are generally isolated, thus limiting ecological benefits. Rural identify basic and connecting ecological corridors. The second is to
ecological sources (RESs) refer to natural patches that have not been optimize ESN based on ecological flow simulation. (3) Evaluation of ESN
artificially disturbed. These patches are perfectly structured and provide for planning decision.
basic ecosystem services to rural regions. Transitional ecological sources
(TESs) refer to ecological patches in a state of transition from rural type
to urban type. Additionally, corridors are also classified based on the 3.1. Extraction and classification of ecological sources
ecological sources connected at both ends (a detailed description is
given in Table S1). Among the three types of corridors, urban–rural 3.1.1. Extraction of ecological sources
connecting corridors and urban structural corridors are the focus of this The morphological spatial pattern analysis (MSPA) method and
study. connectivity index analysis were introduced to extract ecological sour­
Based on the spatial classification system, three modules are ces. As an image processing technique based on mathematical and
included in the approach to identify the ESN in the URF (Fig. 3): (1) morphological principles (Vogt et al., 2009), the MSPA effectively
Extraction and classification of ecological sources. (2) Stepwise ESN identifies landscape types and ecologically significant structures. Here,
the LULC raster map was reclassified into foreground, which included

Fig. 3. The framework of stepwise ESN construction.

3
C. Liang et al. Ecological Indicators 138 (2022) 108794

forestland, grassland, and wetland, and background, which included Table 1

other LULC classes. The calculation was performed using Guidos The evaluation index system of ecological resistance.
Toolbox 3.0 software. Among the landscape types of the MSPA result, Index system Resistance value Weight
core patches larger than 20 hm2 were selected for further connectivity
Factor Level
analysis. To rank the importance of core patches for ecological source
selection, the overall connectivity index (IIC, Eqs. (1)), possible con­ LULC type Cropland 1 0.29
Forest 20
nectivity index (PC, Eqs. (2)), and importance value of the patch (dI, Eqs. Grassland 30
(3)) need to be introduced (Baranyi et al., 2011; Saura and Pascual- Water 50
Hortal, 2007). Among the three indices, dI addresses the importance Bare land 70
of each patch based on its contribution to the overall connectivity of the Construction land 100
MSPA landscape type Core 5 0.24
landscape (IIC and PC). The connectivity analysis was performed using
Bridge 10
Conefor 2.6 software. Loop 20
∑n ∑n ai ∙aj Branch 30
Islet 50
(1)
i=1 j=1 1+nlij
IIC =
A2L Edge 60
Perforation 70
∑n ∑n p*ij ∙ai ∙aj Background 100
Elevation < 200 m 1 0.12
(2)
i=1 j=1 1+nlij
PC = 200 m ~ 400 m 20
A2L
400 m ~ 600 m 60
600 m ~ 800 m 80
I − Iremove
dI = × 100% (3) > 800 m 100
I Slope < 8◦ 1 0.15
8◦ ~ 15◦ 20
where n represents the total number of ecological patches; ai is the area 15◦ ~ 25◦ 60
of patch i; nlij is the number of links in the shortest path (topological 25◦ ~ 35◦ 80
> 35◦ 100
distance) between patch i and patch j; p*ij indicates the maximum prob­
Vegetation coverage < 20% 1 0.20
ability of species spreading between patches i and j; AL is the total 20% ~ 40% 20
landscape area; I is the value of IIC or PC, and Iremove is the value of IIC or 40% ~ 60% 60
PC after the patch is removed. 60% ~ 80% 80
> 80% 100

3.1.2. Classification of ecological sources in the URF

Artificial development is an important feature that distinguishes 3.2.2. Ecological corridor identification and selection
between urban and rural areas. Core patches extracted by MSPA, based The least-cost path (LCP) method was used to identify the corridors
on this characteristic, were further classified into three categories: among the source patches, which were highly operable and feasible.
urban, transitional, and rural. Using ArcGIS 10.2, a 500 m buffer zone Based on the generated ecological sources and the resistance surface, the
was created for each core patch and was classified by quantifying the LCPs between each ecological source pair were calculated using the GIS
proportion of construction land within the buffer zone. The Jenks nat­ distance model. The following formula modified by Yu (1996) was
ural breaks classification method was applied to search for the classi­ applied:
fying boundary. For Xiamen, the boundaries between UESs-TESs, and
TESs-RESs were 40% and 15%. Among the three types of corridors in ∑
i=m
MCR = fmin (Dij × Ri ) (5)
the classification system, urban–rural connecting corridors and urban j=n
structural corridors were the focus of this study.
where Dij represents the spatial distance of a species from source point j
to landscape patch i; and Ri is the ecological resistance of spatial unit i to
3.2. Stepwise ecological security network identification and integration a species migration.
In view of the actual characteristics of the URF, the initially gener­
3.2.1. Construction of the resistance surface in the URF ated corridors needed further screening. The intensity of interaction
In the application of the MCR model, the resistance surface charac­ between the source patches was used to characterize the importance of
terizes the effect of landscape heterogeneity on the flow of ecological potential ecological corridors, where the gravity model served as a
processes (Peng et al., 2018b). The resistance value is not only related to reasonable indicator for its quantitative evaluation (Yang et al., 2020a).
the distance of ecological flow but also associated with land cover, The corridors were classified into three levels of importance according
topographic factors, and artificial disturbance (Dai et al., 2021; Yu et al., to the gravity value. Additionally, continuity and homogeneity are
2017). Hence, the index system of resistance was constructed from five widely known to be the basic attributes of ecological corridors (Luo
aspects: LULC, MSPA landscape type, elevation, slope, and vegetation et al., 2020). However, the frequent and intensive land development in
coverage (Table 1). The indicators were weighted using the analytic the URF has negatively impacted a variety of ecological processes. The
hierarchy process (AHP) method (details are given in Text S1). Addi­ breaking section caused by artificial disturbance will interfere with the
tionally, nighttime light was found to have a significant linear rela­ ecological flow (e.g., the migration of a species) along the corridor
tionship with the distribution of population and GDP (Zhao et al., 2019) (Bailey et al., 2002). Once the breaking length exceeds a certain
and enables the precise correction of the resistance surface (Eqs. (4)). threshold, the corridor will lose its structural function for the entire ESN.
The resistance surface of Xiamen is given in Fig. S2. In this study, the corridor breaking rate (CBR) was proposed to test the
TLI i integrity of each ecological corridor. Considering ecological processes
R* = × R0 (4) such as species migration and the hydrological cycle, 10% was deter­
TLI n
mined as the screening threshold for CBR. The formula was as follows:
where R* represents the corrected resistance value; TLIi is the DN value ∑n ∑k
of nighttime light of grid i; TLIn is the average DN value of nighttime Bi =
α=1 Liα + β=1 Liβ
(6)
light of LULC type n; and R0 is the resistance value calculated by the Li
index system.

4
C. Liang et al. Ecological Indicators 138 (2022) 108794

where Bi is the CBR of corridor i; Liα is the length of the αth intersecting Table 2
The evaluation indicators of ecological network structure.
segment of corridor i and the current construction land; Liβ is the length
of the βth intersecting of corridor i and the construction land in urban Indicator Formulation* Description
planning that reflects the influence of future land development on α index l− v+1 α index reflects the recyclable level of the ecological
α =
ecological corridor; n and k represent the numbers of intersecting seg­ 2v − 5 network, and its value range is [0,1]. The larger the
ments with the current construction land and the construction land in value, the more loops of the ecological network, and
the stronger the material circulation and energy
urban planning, respectively; and Li is the total length of corridor i. mobility.
Additionally, to avoid excessive redundancy of the ESN, all indirect β index l β index reflects the number of connections to each
connecting corridors were removed. A detailed explanation is given in
β =
v node. When β < 1, it is a tree structure; when β = 1, it is
Text S2. a single loop structure; when β > 1, it is a complex
structure.
γ index l γ index reflects the connection degree of all nodes in
3.2.3. Stepwise ecological security network integration γ =
3(v − 2) the network, and its value range is [0,1]. The larger the
value, the higher the connectivity of ecological nodes.
3.2.3.1. Step I: construction of the basic ecological security network. The c index
c = 1−
l c index is also called cost ratio. When the cost ratio gets
d lower, it will be easier to modify the ecological
basic ESN was constructed using the classic methodology in which the
network structure artificially.
sources were selected solely based on the importance or centrality of the
core patches. However, due to the fragmentation induced by urbaniza­ * l is the number of corridors; v is the number of nodes; d is the corridor length
tion to the natural landscape (Peng et al., 2018a), the basic ESN is (km).
usually distributed outside the built-up area. In this study, all core
patches were ranked according to their comprehensive importance (the By comparing the changes in the four indicators, the effect of stepwise
mean value of dIIC and dPC) across the network. The core patches with a ESN integration can be verified, and further implications for planning
comprehensive importance value above 1.00 (Cui et al., 2020) were practice can be realized.
selected as the ecological sources for the basic ESN. The LCPs were then
derived using the method given in 2.2.2. 3.3.2. Spatial distribution of the ecological security network in the URF
The majority of ecosystem services show a distance effect (Farina,
3.2.3.2. Step II: construction of the connecting ecological security network. 2006). In general, areas with dense ESN show better landscape con­
The construction of the connecting ESN is a key step for completing the nectivity, and areas close to the ESN more easily benefit from the edge
classic methodology. The elements of UESs, TESs, urban–rural con­ effect (Fernández et al., 2002). In this study, kernel density estimation
necting corridors and urban structural corridors are necessary to break was used to analyze the spatial agglomeration of the ESN (line ele­
through the ecological isolation between the urban area and the rural ments), and the importance level of ecological corridors was entered
hinterland. Therefore, with the exception of the basic ESN, additional into the population field. Moreover, the cost-weighted distance was used
sources need to be identified from the UESs and TESs in the URF. The to determine the minimum cumulative cost path between the target
patch area is an important attribute that supports ecosystem services pixel and its closest network element. Here, the resistance surface of
(Tolessa et al., 2017); thus, it is regarded as one of the dominant in­ 2.2.1 was used for the calculation.
dicators in searching for additional sources for the connecting ESN.
There are four search principles: (1) should belong to TESs or UESs; (2) 3.3.3. Corridor width of the ecological security network in the URF
should be located inside the URF; (3) should be chosen with preference Corridor width has a significant impact on species migration,
for large areas; and (4) should not be close to the sources of the basic biodiversity maintenance, materials flow and energy flow (van Schalk­
ESN. wyk et al., 2020). Establishing multi-ring buffers for potential corridors
allowed for the area of various LULC types to be calculated. The change
3.2.3.3. Step III: network supplement based on ecological flow simulation. in the landscape structure of ecological corridors was analyzed to
Influenced by the intense artificial disturbance of the URF, the actual determine the optimal buffer width for corridor management. Bentrup
ecological processes between the urban area and the rural hinterland are (2008) noted that water quality and biodiversity are the most sensitive
more complex than those in other districts. The methods of step 1 and indicators of corridor width. For water quality, when a buffer zone
step 2, while addressing the ecological connection concerns in the URF, reaches 30 m, a significant removal effect on mild pollutants can occur.
need further supplementation from the system-wide perspective. For biodiversity, 30.5 m is the minimum recommended width for mul­
Derived from physics, circuit theory was applied to model ecological tiple species. Therefore, the lower limit of the buffer distance was set to
flows in heterogeneous landscapes (McRae et al., 2008). The resistance 30 m. With reference to the relevant studies (Yang et al., 2020b), the
surface constructed in 2.2.1 was introduced for simulation, which was upper limit of the buffer distance was set to 300 m.
applied to Circuitscape 4.0.5 software. Based on the results, these sup­
plementary elements were selected according to two principles: (1) 4. Results
nodes and corridors should be located in the high-value area of current
intensity and should not be repeated with the basic corridor and the 4.1. Distribution and classification of core patches in the URF
connecting corridor; and (2) supplementary nodes and corridors should
be taken from the software database, and should pass both the CBR and 4.1.1. Distribution of the core patches
indirect connection test. Finally, the optimized ESN was obtained. The foreground area of the overall landscape pattern was 716.18
km2, representing 50.55% of Xiamen’s land region. Forest area made up
64.50% of the foreground, which was the dominant LULC type. From
3.3. Evaluation of ecological security network for planning decision inland mountainous areas to coastal urban areas, fragmentation of the
habitat landscape has become more evident (Fig. S3), as has a lack of
3.3.1. Structural performance of the ecological security network in the URF connecting ecological corridors in the URF. Among the 58 core patches
Graph theory (Cantwell and Forman, 1993) was applied to define the with an area greater than 20 km2, the comprehensive importance values
ESN structure as an intelligible set of spatial configurations (a finite set were skewed distributions. The patches of high importance were few in
of nodes, a finite set of linkages, and a composition rule) to portray its number, while the patches with comprehensive importance less than
topological quantities, patterns, and relationships. The basic, connect­ 1.00 reached 49. In terms of spatial pattern (Fig. 4a), the important
ing, and optimized ESNs were evaluated with four indicators (Table 2).

5
C. Liang et al. Ecological Indicators 138 (2022) 108794

Fig. 4. The importance value and classification of the core patches.

ecological patches were mainly distributed in the inland mountainous with high integrity and therefore had the largest average area. The mean
areas, and the comprehensive importance of ecological patches in the area of transitional patches was in the middle, reflecting the trend of
piedmont plain was generally low. According to Xiamen’s URF bound­ gradient change in the ecological patch scale from the urban area to the
ary proposed by Liang and Zeng (2021), the comprehensive importance rural hinterland. In addition, Xiamen’s URF mainly included urban and
value of patches within it was below the average level of the land region. transitional patches, and was adjacent to a large area of natural patches;
This phenomenon indicated that the distribution of ecological elements therefore, this area became the frontier space connecting urban and
in the URF of Xiamen was unbalanced, and the long distance of rural ESNs.
important patches from built-up areas limited the access of residents to
ecosystem services.
4.2. Spatial pattern of urban–rural connecting ecological security network
4.1.2. Classification of the core patches
4.2.1. Basic ecological security network
In the classification of core patches, the proportions of natural,
The top 10 ecological sources, as determined by their comprehensive
transitional and urban patches were 44.72%, 18.34% and 36.95%,
importance, were selected for basic ESN identification. Among these
respectively. The number of ecological patches in the urban area and
ecological sources (Table S2), 60% were RESs, followed by TESs (30%)
rural hinterland are considerable, while the number of transitional
and UESs (10%). After obtaining the potential LCPs, the 18 paths with
patches is insufficient (Fig. 4b). According to the patch-corridor-matrix
gravity values above 1.00 were extracted (Table S3). A large number of
model proposed by Forman (1995), transitional patches mainly play the
simulated corridors near the Tong’an and Xiang’an urban areas were
ecological role of stepping stones, which are essential node elements in
eliminated, mainly due to a lack of support from the transitional patches
the construction of ESNs. In terms of average area, natural, transitional,
that serve as stepping stones. Hence, ecological flow seriously decreases
and urban core patches reached 25.66 km2, 4.73 km2 and 2.54 km2,
in long-distance transmission, reducing the importance of corridors. All
respectively. The urban core patches were located in high-density built-
corridors underwent two tests (Table 3): corridors 7–10 and 5–10 with
up areas with large spatial constraints; therefore, the average area was
CBRs higher than 10% were removed, and two indirect connecting
the smallest. The natural core patches were located in the outer suburbs
corridors 4–7 and 3–7 were visually identified and removed. From the

Table 3
Ecological corridors screening of the basic ESN.
Number Corrridor ID Length (km) Cumulative resistance Gravity value CBR (%) Description

1 7-10 26.82 669842.69 1.01 15.97 Discarded for high CBR

2 5-10 15.66 434380.84 2.31 9.20 (14.20)* Discarded for high CBR
3 8-10 6.01 179216.00 12.22 8.55
4 4-9 9.40 302605.56 4.01 7.80
5 7-9 15.71 463139.03 1.80 5.81
6 2-5 13.96 422879.75 2.20 5.55
7 6-9 11.84 360638.88 2.72 4.41
8 1-9 10.40 305455.66 3.93 4.08
9 2-9 16.60 466524.66 1.68 2.66
10 1-4 13.52 414450.25 2.27 2.53
11 4-6 11.89 318371.63 3.71 2.10
12 3-6 13.24 398481.53 2.28 1.66
13 6-7 5.84 145869.89 18.55 1.39
14 4-7 17.53 449629.16 2.03 1.30 Discarded for indirect connection
15 3-7 18.26 529546.25 1.41 1.25 Discarded for indirect connection
16 1-2 17.86 521152.28 1.43 1.07
17 5-8 10.04 316952.31 3.60 0.40
18 3-4 11.82 385858.94 2.53 0.00

* The CBR calculation result in the brackets were based on intersecting with the construction land in urban planning.

6
C. Liang et al. Ecological Indicators 138 (2022) 108794

partially enlarged view (Fig. 5), corridor 7–10 passed through the cur­ corridors, 6–7 and 8–10, connecting the urban area and the rural hin­
rent built-up area of Xiang’an New Town, resulting in the interruption of terland. The connecting network built more ecological corridors from
the east–west ecological connection within Xiang’an District. The the inland mountainous forest toward bay areas, optimizing the
calculated CBR of corridor 5–10 that intersects the current construction urban–rural connectivity. However, there was still a large blank area in
land is 9.8% and fails to meet the exclusion standard. However, the CBR the middle and south of the Xiang’an District and west of Haicang Dis­
of corridor 5–10 that intersects the construction land of urban planning trict, which lacks the coverage of ecological corridors.
is 14.20%, indicating that the corridor is likely to be occupied by urban
development. Corridors 4–7 and 3–7 has typical indirect connection 4.2.3. Optimized ecological security network
characteristics. According to the results of the current simulation (Fig. S5a), most of
Overall, the basic ESN relies on the mountainous forest resources of the simulated ecological flows were located in the URF, confirming the
the hinterland of Xiamen, forming a semi-enclosed spatial pattern. The importance of this region in the urban–rural ecological connection.
connection between the urban area and the rural hinterland is very Moreover, the high intensity of the simulated current indicates that the
weak, especially in Haicang and Xiang’an Districts, mainly for two ecological flow is strong and that the location has a great demand for
reasons: (1) there is a lack of UESs and TESs in the ecological sources space for the ecological corridor. The path with a strong current can be
that can support ecological flow; and (2) the potential ecological divided into two types: (1) river corridors between the coastal bay area
corridor lacks stepping stones, and the long distance is not conducive to and the inland mountainous forest area. Due to the narrow rivers and
the stable operation of ecological processes. streams, their ecological flow intensity was significantly higher than
that of the common land corridors. (2) Ecological corridors near the
4.2.2. Connecting ecological security network coastline. The substantial occupation of ecological resources in coastal
To supplement the basic ESN, 8 ecological patches were selected areas has resulted in a reduction in potential corridors. Hence, the in­
from the URF as additional sources (Table S4), including 2 TESs and 6 tensity of ecological flows between patches was high. Based on the
UESs. The imbalanced distribution of ecological space in the Xiamen simulation, a total of 7 ecological corridors and 1 ecological source were
land weakens the importance of landscape patches in the URF and built- supplemented (Fig. S5b).
up area. Therefore, the landscape connectivity of additional source In the optimized ESN (Fig. 6), there were 7 rural structural corridors,
patches was generally low. After ecological source supplementation, the 22 urban–rural connecting corridors, and 8 urban structural corridors. It
gravity values of the source interaction matrix in the basic ESN increased was confirmed that optimization of the ESN solved the problem of
to different degrees (Table S5). The additional sources filled the hollow insufficient coverage of the ecological corridors. Additionally, the south
areas of the basic ESN and reduced the cost-weighted distance. On the of the Haicang District and the middle of the Xiang’an District showed a
one hand, the efficiency of ecological flow transmission between the very active ecological flow in the simulation. Nevertheless, most of the
ecological sources was enhanced, and on the other hand, the ecological corresponding ecological corridors were interrupted and occupied by
circulation of rural areas was introduced into urban areas through the urban construction, causing irreversible damage to the network. This
ESN. suggests that ESN protection must be carried out as soon as possible
Through the CBR assessment and the elimination of indirect con­ before it is damaged.
necting corridors, the connecting ESN with 30 ecological corridors was
obtained. The calculation details are given in Text S3, Table S6, and
Fig. S4. The corridor density in the urban fringe increased significantly,
especially in Tong’an District. In the basic ESN, there were only two

Fig. 5. Potential LCPs screening of the basic ESN in the URF.

7
C. Liang et al. Ecological Indicators 138 (2022) 108794

Fig. 6. The optimized ESN in the URF of Xiamen.

4.3. Characteristics of the optimized ecological security network in the sidewalks, squares, parks, and other spaces within urban areas to create
URF a network pattern and thus maximize the diversity and stability of
ecosystem services. For high-density ESN areas, the development in­
4.3.1. Structural performance characteristics tensity should be limited in accordance with the corridor distribution to
The stepwise ESN identification and integration was effective in embed the future urban layout into the ESN. As shown in Fig. 7b, the
restoring the urban–rural ecological connection. Based on the change of optimized ESN demonstrates good accessibility, and most areas of the
four indicators (Table 4), the α index represents the recyclable level of URF can be exposed to the short-range ecological benefits of the
the network in which more closed loops were formed due to the sup­ corridor. It was noted that the industrial zone in the piedmont plain in
plement of ecological connections between the urban area and the rural Jimei District had reduced accessibility despite the high-density network
hinterland. The α index of the optimized ESN reached 0.48, which was coverage, which indicates the lack of close ecological corridors around
45.45% higher than that of the basic ESN and was the best indicator of it.
improvement. The β index reflects the number of external connections of
the source nodes, and the values of the three networks were all higher 4.3.3. LULC composition of different corridor widths
than 1.00, reaching the standard of a complex network (Chen et al., The differences between the three types of ecological corridors given
2021). The change rates of the γ index and c index were found to be the in Fig. 2 were explored through the composition of LULC in different
highest in the optimized ESN. Therefore, the optimization of ESNs corridor buffer widths. The overall results showed that (Fig. 8a) forest
should not only pursue the protection of ecological sources but also was first, followed by crop land. Among the 6 LULC types, water and
strengthen the connections in the structure to improve the spatial construction land were the most sensitive to changes in buffer width.
performance. The urban structural corridor passed through the built-up area; there­
fore, the proportion of construction land increased as the corridor
4.3.2. Spatial distribution characteristics expanded (Fig. 8b). When the buffer width was above 300 m, the growth
The spatial distribution of the optimized ESN was evaluated based on rate began to decelerate. Additionally, there were a number of urban
density and accessibility. As shown in Fig. 7a, the high-density area of structural corridors adjacent to tidal flats or rivers, so the proportion of
ESN was distributed in Jimei District and Tong’an District, while the water remained between 40% and 60%. In the urban–rural connecting
network density was the lowest in Haicang District. The urban area had corridors (Fig. 8c), with the exception of bare land, the proportions of
a low ESN density, and it was impractical to open ecological corridors in the other five types of LULC changed significantly. This is because the
the built-up area. Therefore, it is necessary to integrate the greening of urban–rural connecting corridors passed through the URF, so the
composition of the LULC was influenced by its heterogeneity. The con­
Table 4 struction land changed the most: its proportion increased from 8.51% to
Structural performance comparison of ecological network optimization in the 26.94%. Rural structural corridors were primarily located in the hin­
URF of Xiamen. terland, and their matrix composition was relatively mature. Therefore,
Indicator Basic ecological Connecting ecological Optimized ecological with the change in buffer width, the proportion of LULC remained stable
network network network (Fig. 8d). The ranges of variation of construction land and water were
Value Value Change rate Value Change rate
relatively higher than those of other LULC types.
(%) (%)

α index 0.33 0.42 +27.27 0.48 +14.29

β index 1.40 1.67 +19.29 1.89 +13.17
γ index 0.5833 0.625 +7.15 0.7059 +12.94
c index 0.9167 0.9166 − 0.01 0.9108 − 0.63

8
C. Liang et al. Ecological Indicators 138 (2022) 108794

Fig. 7. The density and accessibility of the optimized ESN.

Fig. 8. LULC composition of different widths of various ecological corridors.

5. Discussion high urbanization level of Xiamen, the study took 15% of construction
land as the threshold determining the optimal buffer width of the three
5.1. Optimal buffer width of diverse ecological corridors in the URF types of corridors in Fig. 8. The buffer width of the urban structural
corridor was between 120 m and 180 m. Taking into consideration the
Although corridor buffer width is an important determinant of ESN, limited space in urban areas, the lower limit of 120 m was selected as its
it has often been developed around static principles in previous works buffer width. The buffer width of the rural–urban connecting corridor
(van Schalkwyk et al., 2020) without considering intracorridor hetero­ was 180 m. However, after the buffer width of the rural structural
geneity (Luo et al., 2020) and intercorridor differences. The threshold corridor reached 600 m, the proportion of construction land still did not
effect of the ecological corridor width provides an important basis for exceed 15%. According to Bueno et al. (1995) and Zhu et al. (2005), 600
the implementation of planning. In the URF, the artificial disturbance of m is the basic corridor width needed for maintaining a natural, species-
the ESN can be characterized by the adjacent construction land. Yang rich landscape structure, and it was therefore taken as the buffer width
et al. (2018) concluded that 10% of the construction land should be of the rural structural corridor. In comparison, the width of urban
considered the standard for delineating corridor width. In view of the structural corridors was close to similar studies (Yang et al., 2018; Ding

9
C. Liang et al. Ecological Indicators 138 (2022) 108794

et al., 2022), where 100 m is set for the buffer width. However, in the ecological corridors, different optimal buffer widths may coexist (Dong
wild region outside the urban center, 100 m is inadequate for some et al., 2020), which deserves further study.
necessary ecological services, such as hydrological regulation and soil
conservation (Dong et al., 2020). This illustrates the importance of 6. Conclusion
corridor classification before identifying buffer width. Furthermore, the
corridor widths proposed here are recommended values. In practice, the The traditional mode of urbanization brings development benefits to
specific control width should be determined by fully considering the humanity but gradually isolates the urban system from the natural
land ownership and benefit trade-offs. Under pressure from urbaniza­ system. The novelty of this study lies in the adaptation of ESN to the
tion, the urban–rural connecting corridors need stricter protection in the URF, a typical region of transition, heterogeneity, and dynamics, thus
URF. achieving the stepwise identification and systematic integration of key
ecological elements. Despite the excessive fragmentation and degrada­
5.2. Contrast between the basic and connecting ecological security tion of green patches in the URF, the developed approach can make the
networks best use of existing ecological resources based on the MSPA, the MCR
model, and circuit theory. Moreover, it also provides guidance for pol­
The largest difference between ESN construction in the URF and icymakers and stakeholders. Through the Xiamen case study, this
other regions is reflected in the organization of ecological flows. In approach has proven effective in the construction of ESNs in the URF of
general, ESN construction focuses on the integrity and continuity of the highly complex urbanization landscapes. The findings are as follows: (1)
structure to achieve free and smooth ecological flows (Borrett and Xiamen’s important ecological patches are mainly distributed in the
Scharler, 2019). The basic ESN can meet its requirements by outlining mountainous hinterland, far from the built-up area. The URF contains
the pattern of key ecological elements. In this study, the basic ESN was many islet and edge patches but few core patches, which indicates the
located on the rural side of Xiamen’s URF, and it consisted mainly of degradation of its environment. (2) In the land region of Xiamen, the
forest. This result is consistent with other studies (Dong et al., 2020; proportions of rural, transitional, and urban patches are 44.72%,
Peng et al., 2018a) with few ecological corridors in urbanized areas. In 18.34%, and 36.95%, respectively. The scarcity of transitional patches
contrast, the URF’s ESN construction emphasizes the directional guid­ and the complexity of LULC are responsible for the lack of urban–rural
ance of ecological flows and the ecological connections between the connecting corridors. (3) The ESN in URF consists of basic and con­
urban area and the rural hinterland. Therefore, a subjective selection of necting parts. The final optimized ESN has 19 source nodes with 7 rural
ecological sources is inevitably required to shape urban–rural ecological structural corridors, 22 urban–rural connecting corridors and 8 urban
corridors (Jalkanen et al., 2020). It was found that the lack of ecological structural corridors. (4) After optimization, both the structural perfor­
stepping stones limited the urban–rural connectivity in the Xiamen case mance and the spatial coverage of the ESN were significantly improved.
study, where 123 of the 153 potential LCPs were identified as infeasible The optimal buffer widths of rural structural corridors, urban–rural
paths. Hence, the quantitative classification of UESs, TESs, and RESs connecting corridors and urban structural corridors were 600 m, 180 m
helps to identify the different functions of ecological patches, supporting and 120 m, respectively.
the construction of a connecting network. Moreover, it was found that
the rank correlation was lower between the comprehensive importance CRediT authorship contribution statement
and area of source patches in the connecting network (Fig. S6). This
suggests that patch area had a weaker influence on the selection of Chen Liang: Writing – original draft, Writing – review & editing,
supplementary sources, while spatial considerations played a larger role. Conceptualization, Methodology, Software, Data curation. Jian Zeng:
Regardless, the connecting ESN is an extension of the basic ESN, which Formal analysis, Project administration, Validation, Supervision, Fund­
adds to the redundancy of the overall network. It has also proven to be ing acquisition. Ruo-Chen Zhang: Writing – review & editing, Visuali­
effective at promoting the resilience of regional ecosystems (Wang et al., zation, Formal analysis, Software, Validation, Funding acquisition.
2021). Qian-Wen Wang: Investigation, Validation, Software.

5.3. Possible future applications and limitations Declaration of Competing Interest

The stepwise ESN construction approach shows much potential in The authors declare that they have no known competing financial
the identification and restoration of urban–rural ecological connections interests or personal relationships that could have appeared to influence
in the URF. On the one hand, this approach decomposes the complex the work reported in this paper.
ESN construction process into several intelligible and applicable steps.
On the other hand, the results obtained by this approach can fully meet Acknowledgement
the demand of planning practice. It can support decisions for regulating
urban growth and outlining ecological protection areas. Moreover, the This research was supported by the China Postdoctoral Science
social structure and interest relationships in the URF are very complex. Foundation funded project (Grant No. 2020M670615) the Natural Sci­
Hence, policy support is necessary for ESN planning implementation ence Foundation of Hebei Province (Grant No. E2021202088), the Na­
(Luo et al., 2020). For example, diverse stakeholders, such as local ex­ tional Natural Science Foundation of China (Grant No. 52078330), and
perts, communities, and developers, should be included in the decision the National Key Research and Development Program of China (Grant
process. No. 2016YFC0502903).
However, this approach also shows uncertainties and limitations.
First, the MCR model is the core of corridor simulation, while resistance
Appendix A. Supplementary data
surface construction does not take into account all the relevant factors,
such as ecosystem services, ecological risk, and traffic network influence
Supplementary data to this article can be found online at https://doi.
(Xiao et al., 2020; Li et al., 2020; Ding et al., 2022). Second, this
org/10.1016/j.ecolind.2022.108794.
approach does not focus on specific ecological species or processes
(Aronson et al., 2017). Thus, further explorations are needed in future
References
studies to enhance applicability in specific application scenarios.
Finally, the optimal buffer width of ecological corridors lacks a response Abercrombie, P., 1944. Greater London Plan, first ed. His Majesty’s Stationery Office,
to the spatial heterogeneity of the URF. However, in the same type of London.

10
C. Liang et al. Ecological Indicators 138 (2022) 108794

Allen, A., 2016. Environmental planning and management of the peri-urban interface: Liang, C., Zhang, R.-C., Zeng, J., Shen, Z.-J., 2021. A land-use decision approach
perspectives on an emerging field. Environ. Urban. 15 (1), 135–148. https://doi.org/ integrating thermal regulation, stormwater management, and economic benefits
10.1177/095624780301500103. based on urbanization stage identification. Sci. Total Environ. 779, 146415. https://
Aronson, M.FJ., Lepczyk, C.A., Evans, K.L., Goddard, M.A., Lerman, S.B., MacIvor, J.S., doi.org/10.1016/j.scitotenv.2021.146415.
Nilon, C.H., Vargo, T., 2017. Biodiversity in the city: key challenges for urban green Liu, T., Yang, X.J., 2015. Monitoring land changes in an urban area using satellite
space management. Front. Ecol. Environ. 15 (4), 189–196. https://doi.org/10.1002/ imagery, GIS and landscape metrics. Appl. Geogr. 56, 42–54. https://doi.org/
fee.2017.15.issue-410.1002/fee.1480. 10.1016/j.apgeog.2014.10.002.
Bailey, S.A., Haines-Young, R.H., Watkins, C., 2002. Species presence in fragmented Luo, Y., Wu, J., Wang, X., Wang, Z., Zhao, Y., 2020. Can policy maintain habitat
landscapes: modelling of species requirements at the national level. Biol. Conserv. connectivity under landscape fragmentation? a case study of Shenzhen, China. Sci.
108 (3), 307–316. https://doi.org/10.1016/S0006-3207(02)00119-2. Total Environ. 715, 136829. https://doi.org/10.1016/j.scitotenv.2020.136829.
Baranyi, G., Saura, S., Podani, J., Jordán, F., 2011. Contribution of habitat patches to McRae, B.H., Dickson, B.G., Keitt, T.H., Shah, V.B., 2008. Using circuit theory to model
network connectivity: redundancy and uniqueness of topological indices. Ecol. Indic. connectivity in ecology, evolution, and conservation. Ecology 89 (10), 2712–2724.
11 (5), 1301–1310. https://doi.org/10.1016/j.ecolind.2011.02.003. https://doi.org/10.1890/07-1861.1.
Bentrup, G., 2008. Conservation Buffers: Design Guidelines for Buffers, Corridors, and Peng, J., Pan, Y.J., Liu, Y.X., Zhao, H.J., Wang, Y.L., 2018a. Linking ecological
Greenways, first ed. Department of Agriculture, Forest Service, Southern Research degradation risk to identify ecological security patterns in a rapidly urbanizing
Station, Asheville, NC. landscape. Habitat Int. 71, 110–124. https://doi.org/10.1016/j.
Borrett, S.R., Scharler, U.M., 2019. Walk partitions of flow in Ecological Network habitatint.2017.11.010.
Analysis: review and synthesis of methods and indicators. 106, 105451. doi: Peng, J., Yang, Y., Liu, Y.X., Hu, Y.N., Du, Y.Y., Meersmans, J., Qiu, S.J., 2018b. Linking
10.1016/j.ecolind.2019.105451. ecosystem services and circuit theory to identify ecological security patterns. Sci.
Brinkley, C., 2018. Fringe benefits: adding rugosity to the urban interface in theory and Total Environ. 644, 781–790. https://doi.org/10.1016/j.scitotenv.2018.06.292.
practice. J. Plan. Lit. 33 (2), 143–154. https://doi.org/10.1177/ Pryor, R.J., 1968. Defining the rural urban fringe. Soc. Forces 47 (2), 202–215.
0885412217726772. Saura, S., Pascual-Hortal, L., 2007. A new habitat availability index to integrate
Bueno, J.A., Tsihrintzis, V.A., Alvarez, L., 1995. South Florida greenways: a conceptual connectivity in landscape conservation planning: comparison with existing indices
framework for the ecological reconnectivity of the region. Landsc. Urban Plan. 33 and application to a case study. Landsc. Urban Plan. 83 (2–3), 91–103. https://doi.
(1–3), 247–266. https://doi.org/10.1016/0169-2046(94)02021-7. org/10.1016/j.landurbplan.2007.03.005.
Cantwell, M.D., Forman, R.T.T., 1993. Landscape graphs: ecological modeling with Scott, A.J., Carter, C., Reed, M.R., Larkham, P., Adams, D., Morton, N., Waters, R.,
graph theory to detect configurations common to diverse landscapes. Landsc. Ecol. 8 Collier, D., Crean, C., Curzon, R., Forster, R., Gibbs, P., Grayson, N., Hardman, M.,
(4), 239–255. https://doi.org/10.1007/BF00125131. Hearle, A., Jarvis, D., Kennet, M., Leach, K., Middleton, M., Schiessel, N., Stonyer, B.,
Casado-Arzuaga, I., Madariaga, I., Onaindia, M., 2013. Perception, demand and user Coles, R., 2013. Disintegrated development at the rural-urban fringe: re-connecting
contribution to ecosystem services in the Bilbao Metropolitan Greenbelt. J. Environ. spatial planning theory and practice. 83, 1-52. doi: 10.1016/j.progress.2012.09.001.
Manage. 129, 33–43. https://doi.org/10.1016/j.jenvman.2013.05.059. Shan, N., Zhou, K.X., Pan, Y., Tang, F.K., 2019. Research advances in design methods of
Chen, N.N., Kang, S.Z., Zhao, Y.H., Zhou, Y.J., Yan, J., Lu, Y.R., 2021. Construction of biodiversity conservation corridors. Acta Ecol. Sin. 39 (2), 411–420. https://doi.org/
ecological network in Qinling Mountains of Shaanxi, China based on MSPA and MCR 10.5846/stxb201802080348.
model.[J]. Chinese. J. Appl. Ecol. 32 (5), 1545–1553. http://www.cjae.net/CN/10 Shaw, B.J., van Vliet, J., Verburg, P.H., 2020. The peri-urbanization of Europe: a
.13287/j.1001-9332.202105.012. systematic review of a multifaceted process. Landsc. Urban Plan. 196, 103733.
Cui, L., Wang, J., Sun, L.u., Lv, C., 2020. Construction and optimization of green space https://doi.org/10.1016/j.landurbplan.2019.103733.
ecological networks in urban fringe areas: a case study with the urban fringe area of Su, Y.X., Chen, X.Z., Liao, J.S., Zhang, H.O., Wang, C.J., Ye, Y.Y., Wang, Y., 2016.
Tongzhou district in Beijing. J. Clean. Prod. 276, 124266. https://doi.org/10.1016/j. Modeling the optimal ecological security pattern for guiding the urban constructed
jclepro.2020.124266. land expansions. Urban For. Urban Green. 19, 35–46. https://doi.org/10.1016/j.
Dai, L.u., Liu, Y., Luo, X., 2021. Integrating the MCR and DOI models to construct an ufug.2016.06.013.
ecological security network for the urban agglomeration around Poyang Lake. China. Tolessa, T., Senbeta, F., Kidane, M., 2017. The impact of land use/land cover change on
Sci. Total Environ. 754, 141868. https://doi.org/10.1016/j.scitotenv.2020.141868. ecosystem services in the central highlands of Ethiopia. Ecosyst. Serv. 23, 47–54.
Deng, Y.u., Qi, W., Fu, B., Wang, K., 2020. Geographical transformations of urban sprawl: https://doi.org/10.1016/j.ecoser.2016.11.010.
exploring the spatial heterogeneity across cities in China 1992–2015. Cities 105, van Schalkwyk, J., Pryke, J.S., Samways, M.J., Gaigher, R., 2020. Corridor width
102415. https://doi.org/10.1016/j.cities.2019.102415. determines strength of edge influence on arthropods in conservation corridors.
Dong, J., Peng, J., Liu, Y., Qiu, S., Han, Y., 2020. Integrating spatial continuous wavelet Lands. Ecol. 35 (7), 1175–1185. https://doi.org/10.1007/s10980-020-01008-6.
transform and kernel density estimation to identify ecological corridors in Vogt, P., Ferrari, J.R., Lookingbill, T.R., Gardner, R., Riitters, K.H., Ostapowicz, K., 2009.
megacities. Landsc. Urban Plan. 199, 103815. https://doi.org/10.1016/j. Mapping functional connectivity. Ecol. Indic. 9 (1), 64–71. https://doi.org/10.1016/
landurbplan.2020.103815. j.ecolind.2008.01.011.
Ding, M., Liu, W., Xiao, L.i., Zhong, F., Lu, N.a., Zhang, J., Zhang, Z., Xu, X., Wang, K., von Thünen, J.H., 1875. Der isolirte Staat in Beziehung auf Landwirthschaft und
2022. Construction and optimization strategy of ecological security pattern in a Nationalökonomie, Vol. 1, second ed. Wiegandt, Hempel & Parey, Berlin.
rapidly urbanizing region: A case study in central-south China. Ecol. Indic. 136, Wang, T., Li, H., Huang, Y., 2021. The complex ecological network’s resilience of the
108604. https://doi.org/10.1016/j.ecolind.2022.108604. Wuhan metropolitan area. Ecol. Indic. 130, 108101. https://doi.org/10.1016/j.
Farina, A., 2006. Principles and Methods in Landscape Ecology: Toward A Science of ecolind.2021.108101.
Landscape, second ed. Springer-Verlag, Dordrecht. Xiao, S.C., Wu, W.J., Guo, J., Ou, M.H., Pueppke, S.G., Ou, W.X., Tao, Y., 2020. An
Fernández, C., Acosta, F.J., Abellá, G., López, F., Dı́az, M., 2002. Complex edge effect evaluation framework for designing ecological security patterns and prioritizing
fields as additive processes in patches of ecological systems. Ecol. Model. 149 (3), ecological corridors: application in Jiangsu Province. China. Landsc. Ecol. 35 (11),
273–283. https://doi.org/10.1016/S0304-3800(01)00464-1. 2517–2534. https://doi.org/10.1007/s10980-020-01113-6.
Forman, R.T.T., 1995. Land Mosaics: The Ecology of Landscapes and Regions, first ed. Yang, M., Liao, Z.Z., Shi, L.Y., 2020a. Multi-scale ecological infrastructure planning of
Cambridge University Press, Cambridge. Xiong’an New Area. Acta Ecol. Sin. 40 (20), 7123–7131. https://doi.org/10.5846/
Gong, P., Liu, H., Zhang, M.N., Li, C.C., Wang, J., Huang, H.B., Clinton, N., Ji, L.Y., Li, W. stxb201912062651.
Y., Bai, Y.Q., Chen, B., Xu, B., Zhu, Z.L., Yuan, C., Suen, H.P., Guo, J., Xu, N., Li, W.J., Yang, J., Zeng, C., Cheng, Y., 2020b. Spatial influence of ecological networks on land use
Zhao, Y.Y., Yang, J., Yu, C.Q., Wang, X., Fu, H.H., Yu, L., Dronova, I., Hui, F.M., intensity. Sci. Total Environ. 717, 137151. https://doi.org/10.1016/j.
Cheng, X., Shi, X.L., Xiao, F.J., Liu, Q.F., Song, L.C., 2019. Stable classification with scitotenv.2020.137151.
limited sample: transferring a 30-m resolution sample set collected in 2015 to Yang, Z.G., Jiang, Z.Y., Guo, C.X., Yang, X.J., Xu, X.J., Li, X., Hu, Z.M., Zhou, H.Y., 2018.
mapping 10-m resolution global land cover in 2017. Sci. Bull. 64 (6), 370–373. Construction of ecological network using morphological spatial pattern analysis and
https://doi.org/10.1016/j.scib.2019.03.002. minimal cumulative resistance models in Guangzhou City, China. Chin. J. Appl. Ecol.
Jalkanen, J., Toivonen, T., Moilanen, A., 2020. Identification of ecological networks for 29 (10), 3367–3376. http://www.cjae.net/CN/10.13287/j.1001-9332.201810.019.
land-use planning with spatial conservation prioritization. Landsc. Ecol. 35 (2), Yokohari, M., Takeuchi, K., Watanabe, T., Yokota, S., 2000. Beyond greenbelts and
353–371. https://doi.org/10.1007/s10980-019-00950-4. zoning: a new planning concept for the environment of Asian mega-cities. Landsc.
Kalnay, E., Cai, M., 2003. Impact of urbanization and land-use change on climate. Nature Urban Plan. 47 (3–4), 159–171. https://doi.org/10.1016/s0169-2046(99)00084-5.
423 (6939), 528–531. https://doi.org/10.1038/nature01675. Yu, K.J., 1996. Security patterns and surface model in landscape ecological planning.
Knaapen, J.P., Scheffer, M., Harms, B., 1992. Estimating habitat isolation in landscape Landsc. Urban Plan. 36 (1), 1–17. https://doi.org/10.1016/S0169-2046(96)00331-
planning. Landsc. Urban Plan. 23 (1), 1–16. https://doi.org/10.1016/0169-2046 3.
(92)90060-D. Yu, Q., Yue, D.P., Wang, J.P., Zhang, Q.B., Li, Y.T., Yu, Y., Chen, J.X., Li, N., 2017. The
Kroll, F., Muller, F., Haase, D., Fohrer, N., 2012. Rural-urban gradient analysis of optimization of urban ecological infrastructure network based on the changes of
ecosystem services supply and demand dynamics. Land Use Pol. 29 (3), 521–535. county landscape patterns: a typical case study of ecological fragile zone located at
https://doi.org/10.1016/j.landusepol.2011.07.008. Deng Kou (Inner Mongolia). J. Clean. Prod. 163, S54–S67. https://doi.org/10.1016/
Li, S., Xiao, W., Zhao, Y., Lv, X., 2020. Incorporating ecological risk index in the multi- j.jclepro.2016.05.014.
process MCRE model to optimize the ecological security pattern in a semi-arid area Zhao, M., Zhou, Y.Y., Li, X.C., Cao, W.T., He, C.Y., Yu, B.L., Li, X., Elvidge, C.D.,
with intensive coal mining: a case study in northern China. J. Clean. Prod. 247, Cheng, W.M., Zhou, C.H., 2019. Applications of satellite remote sensing of nighttime
119143. https://doi.org/10.1016/j.jclepro.2019.119143. light observations: advances, challenges, and perspectives. Remote Sens. 11 (17),
Liang, C., Zeng, J., 2021. Three-dimensional dynamic identification and spatial evolution 1971. https://doi.org/10.3390/rs11171971.
mechanism of urban-rural fringe: a case study of Xiamen. Geogr. Res. 40 (5), Zhu, Q., Yu, K.J., Li, D.H., 2005. The width of ecological corridor in landscape planning.
1404–1420. https://doi.org/10.11821/dlyj020200416. Acta Ecol. Sin. 25 (9), 2406–2412. http://www.ecologica.cn/stxb/article/abstract
/050937.

11