Construction of Green Ecological Network in Qingdao (Shandong, China) Based on the Combination of Morphological Spatial Pattern Analysis and Biodiversity Conservation Function Assessment

Abstract

Building urban green ecological network systems and increasing urban and rural landscape connectivity are effective ways to improve urban biodiversity and landscape sustainability. The ecological sources in the main urban area of Qingdao City (Shandong, China) were identified based on morphological spatial pattern analysis (MSPA) combined with a biodiversity conservation function assessment, with the ecological corridors established and the ecological network structure optimized. The results showed that (1) the study area lacked high-quality patches with strong landscape connectivity; (2) the potential green ecological network of the study area was composed of 38 ecological sources, 703 ecological corridors, and 284 ecological nodes, effectively connecting urban and suburban green spaces; (3) after optimization, the green ecological network contained a total of 223 important corridors and 61 key nodes, with significantly increased network connectivity; (4) the optimal ecological corridor width in Qingdao was determined to be 30 m. Our study provided important guidance for the construction of ecological security patterns and scientific evidence to support urban green space planning and sustainable development in Qingdao.

1. Introduction

When planning urban green space systems, it is fundamentally important to improve the urban biodiversity and landscape sustainability of urban green spaces if the spaces are limited in number. The accurate identification of effective environmental strategies used to introduce the abundant biological resources of the suburbs into the city has become a popular area of research [1]. In recent decades, a series of environmental impairments have been caused by the continuous expansion of urban construction land, i.e., the ecosystem service value of landscape patches has decreased; the fragmentation of urban green patches has increased; the landscape connectivity between green spaces has decreased; ecological processes, such as energy flow, matter cycle, and information flow, have weakened; and the communication between humans and nature has gradually faded away [2]. Some researchers have considered biological rhythms to be a crucial factor affecting urban biodiversity [3]. However, core patches and ecological corridors are considered to have the strongest positive effects on biodiversity in high-density urban areas [4]. Therefore, it is crucial to construct a reasonable urban green space network system based on both morphological spatial pattern analysis (MSPA) and a biodiversity conservation function (BCF) assessment to improve biological flow efficiency and enhance the biodiversity and sustainability of urban green spaces.

The construction of green ecological networks will effectively connect fragmented sources through ecological corridors to promote the ecological process among patches, increase urban biodiversity, and improve the sustainability of regional ecosystems [5,6]. Numerous studies have investigated the construction and optimization of urban ecological networks, and the fundamental paradigm “identification of ecological sources → construction of resistance surface → extraction of ecological corridor” has been well established [7]. In particular, during the identification of ecological sources, most studies have directly designated patches with high habitat quality, e.g., nature reserves and scenic spots, as ecological sources, and alternatively identified ecological patches based on indicators, e.g., the importance of ecosystem services and ecological sensitivity [8]. However, most of these studies have largely ignored the consideration of the integrity of patches in a landscape’s space. This problem could be adequately solved by applying MSPA, which has been widely used in environmental investigations of urban planning and development. MSPA can not only identify and classify ecological patches based on spatial morphology, but also reveal the structural connections among these patches [9]. Furthermore, the MSPA approach has shown the advantages of small data requirements and the convenient visual outcomes of this analysis. Therefore, as an important method of connecting ecological sources and to ensure species diffusion, ecological corridors are commonly simulated via various methods, e.g., graph theory, the minimum cumulative resistance (MCR) model, and circuit theory. In particular, the MCR model is used to simulate the process of species diffusion overcoming resistance to quantify the potential movement pattern of organisms and to improve the scientific significance of ecological network construction, and is thus widely applied to determine ecological security patterns in territorial spatial planning [10].

The MSPA method can identify habitat patches that play a crucial role in landscape connectivity in the study area based on the high pixels of satellite images. Indeed, most of the current studies on ecological network construction based on MSPA have identified the ecological sources in combination with landscape connectivity assessments. For example, critical areas such as nature reserves and forests were selected as prospects for MSPA and then large core patches were selected as ecological sources [11]; the MSPA method was used to identify the ecological sources based on patch size and landscape connectivity [12], and the ecological sources were identified based on the importance and ecological sensitivity of ecosystem services combined with the results of MSPA [7]. It is noted that although patches with high habitat quality and connectivity were generated, these studies ignored the importance of biodiversity conservation functions in ecological sources for the survival and migration of organisms among ecological networks. Meanwhile, the viability of urban wildlife populations is closely related to the quality of urban open spaces [13]. Therefore, in our study, we combined the BCF assessment with MSPA to identify the critical ecological sources for promoting biological migration and building a green ecological network to ultimately improve urban biodiversity in Qingdao City, Shandong, China.

Qingdao is one of the most livable cities in eastern China. The natural environment in the city is well known for its unique picturesque scenes throughout the entire urban area, harboring a variety of rare animals and plants. However, as a high-density area, numerous human disturbances, such as continuous expansion of urbanization and industrial production activities, have led to dramatic changes in urban green spaces [14], increased habitat fragmentation, and serious threats to biodiversity in urban parks [15]. In recent years, Qingdao has gradually recognized the importance of the construction of an ecological civilization, and the area of urban green space has been significantly increased. According to the statistical yearbook of Qingdao [16,17], the green coverage rate of built-up areas has been increased from 39.44% in 2016 to 43% in 2022. However, an effective urban green space system has not been established in the city and the lack of ecological corridors is not conducive to animal migration and diffusion. Therefore, the aims of our study were to improve the biodiversity and landscape sustainability of urban green spaces and to build an efficient urban green space network system, ultimately providing scientific evidence to support the construction of urban biodiversity protection and sustainable developmental strategies during the planning of urban green spaces.

2. Materials and Methods

2.1. Study Area

The study area was located in the urban areas of Qingdao, covering a total of 5 districts, i.e., Shinan, Shibei, Licang, Laoshan, and Chengyang, mainly between 35°35′–37°09′ N and 119°30′–121°00′ E (Figure 1). The total study area was about 827.70 km² with the highest altitude of 1080 m, showing evident maritime characteristics, with an annual precipitation of 600–800 mm [18]. Although the green space coverage rate of the study area was high (about 36%), the green spaces were extremely fragmented in urban areas, with low biodiversity and poor ecological functions. In particular, the Laoshan District, located in the east of the study area, was rich in geological and geomorphic variations with high biodiversity, representing a remarkable ecological environment and a significantly important biological source in Qingdao [19].

2.2. Data Collection and Preprocessing

The satellite Landsat 8 OLI_TIRS image data (spatial resolution 30 m) and elevation data presented in the digital elevation model (DEM) were obtained from Geospatial Data Cloud (https://www.gscloud.cn/, accessed on 7 June 2022). Data of net primary productivity (NPP) were retrieved from the National Earth System Science Data Center (http://www.geodata.cn/, accessed on 9 November 2022). Data of rainfall and air temperature were collected from the Resources and Environmental Science Data Registration and Publication System (https://www.resdc.cn/, accessed on 25 August 2022).

ENVI 5.3 was used to divide the satellite image data of the study area into five categories: farm land, green land, water body, construction land, and unused land. Based on the highly accurate images taken during the same period, a land use and land cover (LULC) classification map (Figure 2) was obtained with a grid size of 30 m × 30 m, with the overall accuracy of the final interpretation being 96% and with a Kappa coefficient of 0.9.

2.3. Identification of Ecological Sources

First, ENVI 5.3 software was used to splice and cut the satellite image data. All green spaces in the study area were obtained via supervised classification and visual interpretation. Then, the MSPA was performed to divide all green patches into seven types of landscape, including core, island, loop, bridge, perforation, edge, and branch, based on spatial arrangement characteristics [20]. The ecological significance of each landscape type is shown in

Table 1. MSPA landscape types and their ecological significance.

Type of Landscape	Ecological Significance
Core	Habitat patches with larger areas, which provide larger habitats for species, are generally used as ecological source areas in the ecological network.
Island	Little isolated patches that are not connected to each other; the possibility of internal material and energy exchange and transfer is lower.
Loop	The ecological corridor connecting the same core areas serves as a channel for energy exchange within the core areas.
Bridge	The ribbon area connecting the core areas plays a corridor role in the ecological network and is an important landscape type for enhancing landscape connectivity.
Perforation	Marginal zones of inner core patches with edge effects.
Edge	The external boundary of the core areas protects the internal core areas from external interference and ensures normal energy flow and material exchange in the core areas.
Branch	A linear landscape with only one end connected to bridge, loop, edge, or perforation.

. The core patches were generally large habitat patches and were considered to be ecological source patches [21].

The patch connectivity index (PC) was used to identify the most important habitat patches and evaluate the fragmentation of the landscape and the connectivity of patches [22]. The patch importance value (dI) was calculated to evaluate the contribution of a single patch to the landscape connectivity index [23]. In this study, both PC and dI were calculated to evaluate the landscape connectivity of core patches using conefor 2.6 software, with the distance threshold set to 2500 m and the connectivity probability set to 0.5, as previously reported [24,25]. The PC and dI were calculated using the following equations, respectively:

$$PC=\frac{{\sum }_{i=1}^n{\sum }_{j=1}^n{a}_i{a}_j{p_{ij}}^{*}}{{A_L}^2}$$

(1)

$$dI=\frac{PC-{PC}_{remove}}{PC}\times 100$$

(2)

In Equation (1), n is the number of patches; a_i and a_j are the areas of patches i and j, respectively; A_L is the total landscape area; and p_ij^* is the maximum probability of species diffusion in patches i and j. The PC value is [0, 1], with the greater value representing a higher connectivity of the landscape. In Equation (2), dI is the connectivity index value of a certain landscape and PC_remove is the connectivity index after the patch was removed. The greater value of dI indicates the greater importance of landscape elements.

The values of NPP, precipitation, temperature, and altitude indicate the capability of habitat patches to maintain biodiversity [26]. Therefore, the BCF index was calculated to select high-quality ecological patches using the following equations:

$${NPP}_{mean}=\frac{{NPP}_{2010S}}{10}$$

(3)

$$P_{mean}=\frac{P_{2010s}}{10}$$

(4)

$$T_{mean}=\frac{T_{2010s}}{10}$$

(5)

$$BCF={NPP}_{mean}\times P_{mean}\times T_{mean}\times \left(1-A\right)$$

(6)

where NPP_mean is the average annual vegetation NPP, P_mean is the average annual precipitation, T_mean is the average annual temperature, A is the altitude, and 2010s represents the sum of the annual data between 2010 and 2020.

The results of landscape connectivity evaluation and BCF evaluation of the core patches were normalized and overlayed, respectively, to obtain the total patch comprehensive evaluation value. The patches with high comprehensive evaluation values were selected as the ecological sources for the subsequent analyses.

2.4. Construction of Ecological Corridors

Species migration between patches is often hindered by different obstacles. A smaller landscape resistance indicates that fewer obstacles are encountered in the process of species migration and the higher efficiency of ecological flow transmission. Therefore, determination of the landscape resistance value is a pivotal step in extracting ecological corridors. The landscape resistance value is generally determined by the type of LULC, elevation, slope, distance to water body, road, and residential site [27,28]. In this study, ArcGIS was used to overlay the landscape resistance factors to obtain the landscape resistance value in the study area (

Table 2. Characteristics of landscape resistance factors in this study.

Landscape Resistance Factor	Content	Classification of Landscape Resistance	Value	Weight
Land use type	Land use and land cover	Green land	1	0.2
		Water body	2
		Farm land	3
		Unused land	4
		Construction land	5
Natural topographic factor	Altitude (m)	0–200	1	0.15
		200–400	2
		400–600	3
		600–800	4
		>800	5
	Slope (°)	0–5	1	0.15
		5–10	2
		10–15	3
		15–20	4
		>20	5
	Distance to water body (m)	0–200	1	0.1
		200–400	2
		400–600	3
		600–800	4
		>800	5
Human activity factor	Distance to road (m)	0–100	1	0.2
		100–200	2
		200–300	3
		300–400	4
		>400	5
	Distance to residential area (m)	0–200	1	0.2
		200–400	2
		400–600	3
		600–800	4
		>800	5

), with each resistance factor weighted using the expert scoring method [29].

The MCR model was used to compare the cumulative resistance of grid pixels between patches to obtain low-resistance paths for biological migration and diffusion [30]. Based on the landscape resistance values obtained, the Cost Path tool of ArcGIS was used to calculate the raster pixel resistance between two patches and to obtain the minimum cumulative resistance path, i.e., an optimal ecological corridor. The MCR model was established as the following equation:

$$MCR=f_{\mathrm{m}\mathrm{i}\mathrm{n}}{\sum }_{j=n}^{i=m}\left(D_{ij}\times R_i\right),$$

(7)

where MCR represents the minimum cumulative cost of species from one ecological source to another patch in space, D_ij represents the diffusion distance of species from patches i to j, and R_i is the landscape resistance value of patch i.

2.5. Ecological Network Evaluation and Optimization

Green ecological networks generally contain different structures due to various factors such as the length, location, and connection mode of ecological corridors. Network closure index (α), network connectivity index (β), and network connectivity rate index (γ) are usually calculated to evaluate the complexity of green ecological network structures [31] using the following equations:

$$\alpha =\frac{\left(l-\upsilon +1\right)}{\left(2\upsilon -5\right)}$$

(8)

$$\beta =\frac{l}{\upsilon }$$

(9)

$$\gamma =\frac{l}{3\left(\upsilon -2\right)}$$

(10)

where l is the number of ecological corridors and v is the number of ecological nodes. In this study, the intersections of the maximum- and minimum-cost paths were identified as ecological nodes [32]. The value of α is used to describe the loop level of an ecological network, i.e., a larger α value suggests more loops of the ecological network and higher mobility of matter and energy. The index β is defined as the average number of connections of each node and was used to measure the access to the ecological network, and a larger β value indicated a more complex ecological network structure. The index γ represents the interconnection of each node in the ecological network, i.e., a larger γ value implicates higher interconnection.

Based on the ecological network evaluation, the relationships between patches were quantitatively analyzed according to the interaction matrix between patches generated by the gravity model, which was calculated using the following equation:

$$G_{ab}=\frac{{L_{max}}^2\times \mathrm{l}\mathrm{n}{S}_a\times \mathrm{l}\mathrm{n}{S}_b}{L_{ab}^2\times P_a\times P_b}$$

(11)

where G_ab (gravity value) is the relative significance between patches a and b; S_a and S_b are the areas of patches a and b, respectively; P_a and P_b are the average landscape resistance values of patch a and patch b; L_ab is the cumulative landscape resistance value of the corridor between patches a and b; and L_max is the maximum cumulative landscape resistance value of all corridors.

An ecological corridor generally contains several ecological nodes, and the connectivity of an ecological corridor can be judged by the density of its ecological nodes. Therefore, in this study, the most important ecological corridors were selected based on the interaction matrix, and kernel density (N) analysis was used to analyze the density of the most important ecological corridors. In this study, the Ns of ecological corridors were divided into five groups, i.e., high kernel density (N ≥ 723), relatively high density (468 ≤ N < 723), medium density (275 ≤ N < 468), relatively low density (109 ≤ N < 275), and low density (N < 109). All the ecological nodes in the region with high density were identified as key nodes. Finally, the green ecological network composed of important ecological corridors and key ecological nodes was identified as the optimized ecological network.

2.6. Determination of Ecological Corridor Width

When the width of an ecological corridor reaches the fundamental requirements, organisms can efficiently pass through the corridor and populations can rapidly increase [33]. In general, a corridor width between 30 and 60 m can meet the requirements of most animal migrations, while a width of 60–100 m is suitable for bird migration [24]. The existence of high-quality patches can improve the efficiency of ecological processes [34]. However, due to the expansion of urban construction, it may be a challenge to increase the width of the ecological corridors [35]. Therefore, 30, 60, 120, 300, and 600 m widths were used to identify the optimal ecological corridor width based on the proportions of different types of LULC.

3. Results

3.1. Identification of Ecological Sources

Using the MSPA method, the distribution of different landscape types was obtained in the study area (Figure 3). The results showed that the total area of green land in the study area was 432.82 km², with a core area of about 329.73 km², accounting for 76.18% of the total area. The largest core patch, i.e., the most important biological source, was the forest in Laoshan Mountain in the east of Qingdao, which contained abundant vegetation, with favorable growth conditions and rich biodiversity. The distribution of the Laoshan patch was concentrated with other source patches of built-up areas, which could meet the needs of species migration and energy exchange. These results were consistent with those previously reported [36]. There were few core patches observed in the western urban area, showing relatively scattered distribution, poor landscape connectivity, low efficiency of ecological processes between patches, and hindered ecological flow transmission.

The results of MSPA showed that the average core patch area was 0.14 km² (Figure 3). A total of 50 core patches with an area greater than 0.14 km² were selected for further extraction. Then, based on the values of dI and BCF (

Table 3. A total of 38 core patches identified in the study area ranked by the values of dI and BCF.

Rank of Patch	Patch Number	BCF	dI	Normalized Value of Both BCF and dI
1	34	29,241.59	99.57	13.74
2	21	16.14	12.97	0.41
3	10	5.35	11.90	0.34
4	20	17.60	11.24	0.29
5	4	1.82	10.37	0.23
6	12	8.59	9.83	0.19
7	15	5.70	8.95	0.13
8	11	5.04	8.90	0.12
9	5	2.74	8.87	0.12
10	1	0.74	8.80	0.11
11	36	819.63	4.39	0.00
12	13	9.78	6.78	−0.03
13	29	1.25	5.92	−0.09
14	37	668.27	2.20	−0.19
15	6	6.34	4.15	−0.22
16	26	5.40	4.05	−0.22
17	14	8.67	3.85	−0.24
18	3	12.59	3.19	−0.28
19	7	3.13	2.99	−0.30
20	8	10.11	2.93	−0.30
21	22	8.65	2.93	−0.30
22	35	38.48	2.45	−0.33
23	31	120.99	2.07	−0.34
24	23	4.04	2.44	−0.34
25	19	7.24	2.26	−0.35
26	32	100.17	0.83	−0.43
27	2	43.94	0.91	−0.44
28	18	23.05	0.71	−0.46
29	38	221.98	0.01	−0.46
30	33	86.13	0.45	−0.46
31	24	23.99	0.37	−0.48
32	25	54.59	0.26	−0.48
33	30	5.45	0.41	−0.48
34	17	31.10	0.31	−0.48
35	27	31.64	0.31	−0.49
36	16	19.41	0.33	−0.49
37	28	4.44	0.36	−0.49
38	9	24.55	0.28	−0.49
Average		834.22	6.57	1.43

), a total of 38 patches were identified as the ecological sources, each with a normalized total value greater than −0.5 (Figure 4). These ecological sources were mainly distributed in the east and south of the study area, including Laoshan Mountain, Fushan Mountain, Shiwan Forest Park, and other important ecological protection areas. Patch 34 Laoshan was revealed to have the highest dI and BCF, indicating that this patch contained the most important biological sources.

A large range of variations and generally low values were revealed in the ecological patch dI, ranging from 99.57 (patch 34) to 0.01 (patch 38) with an average of 6.57 (Table 3). In addition to patch 34 Laoshan, high dI values (>11) were revealed in patches 21, 10, and 20, which were mainly located in the central and southern parts of the study area, indicating the lack of high-landscape-connectivity patches in the west of the study area.

The results of the BCF assessment revealed that the areas of higher quality were mainly distributed in the eastern and southern regions, while the densely populated and built-up areas within the city, mainly including Fushan Ecological Park and Zhongshan Park, generally lacked high-quality green patches (Figure 5).

3.2. Construction of Ecological Corridors

Based on the six types of resistance factors (Figure 6) and the MCR model, the landscape resistance was obtained (Figure 7). The results showed that the overall landscape resistance values in the study area were high, with evident variations observed between the east and west regions of the study area. The eastern Laoshan District was revealed to have the smallest resistance value, while the landscape resistance values in the central urban area with a high density of construction land were larger, which was not conducive to material exchange and energy flow. Based on the MCR model, a total of 703 ecological corridors with a total length of 935.29 km were initially generated among 38 ecological sources, and these ecological corridors constituted a potential green ecological network (Figure 8). The distribution of ecological corridors in the central and southern parts was relatively dense, indicating that it is beneficial to build ecological networks with more green patches inside the central urban area.

3.3. Evaluation and Optimization of the Potential Green Ecological Network

Based on the potential green ecological network (Figure 8), a total of 284 potential ecological nodes were identified (Figure 9), with the values of the network connectivity indices α, β, and γ being 0.75, 2.48, and 0.83, respectively. These results showed that there were numerous closed loops in the potential green ecological network, the material cycle and energy flow were flexible, and the anti-interference ability was strong, which is conducive to further optimization.

As the most critical and vulnerable locations of a potential green ecological network, most of these ecological nodes were distributed in green land, while a small number of nodes were distributed in farm land and construction land, which enhanced the node connection function, ensuring the integrity of the entire potential green ecological network system.

Then, based on the gravity model, the ecological corridors with high gravity values were extracted as important corridors (Figure 10). The results showed that the gravity values of the potential ecological corridors were generally small with large variations, ranging from 59.49 to 887,457.89 with an average of 4913.66 and median of 715.24. The gravity values between each of the three patches 16, 29, and 30 and other patches were relatively high. In particular, patch 16 was surrounded by multiple patches, and the distances between patches were largely the same, which could facilitate effective connections among the ecological sources of various regions. Similarly, both patches 29 and 30 were shown to be important for maintaining the overall landscape connectivity in the northwest area.

There were 223 ecological corridors with gravity values greater than 1500, accounting for about 32% of the total number of potential ecological corridors. Therefore, the ecological corridors with gravity values greater than 1500 were identified as important corridors, and the rest were regarded as general corridors. The most important corridors were concentrated in the central urban and suburban transitional areas, indicating that the interactions between urban and suburban patches were important for maintaining the overall function of a green ecological network system.

Kernel density analysis was performed based on the important corridors to identify a total of 61 key nodes (Figure 11 and Figure 12). Among them, 26 key nodes were located at the edges of the ecological source patches, indicating that the ecological source patches were readily affected by the surrounding environment. These key nodes were concentrated in the northwest and south of the study area, which was similar to the areas of patches 16, 29, and 30, again indicating the importance of these ecological source patches in the green ecological network. Another 24 key nodes were located around the roads in a north–south linear distribution, indicating that the green corridors along the roads could provide appropriate living environmental resources in the process of biological migration. The other 11 key nodes were located along the rivers. Adequate water resources around these key nodes could not only provide appropriate habitats and nutritional supplies, but also introduce certain obstacles to biological migration. Therefore, biological corridors should be built at key nodes to reduce the resistance of creatures crossing rivers.

Based on the 223 important corridors and 61 key nodes, the values of network connectivity indices α, β, and γ of the optimized green ecological network were increased by 87%, 48%, and 52% to 1.39, 3.66, and 1.26, respectively, compared with those of a potential green ecological network without optimization. These results indicated that the structure of the optimized green ecological network was more stable with a stronger ability to resist external interference, and the efficiency of ecological flow was higher, which could significantly promote the migration of creatures between ecological sources.

3.4. Determination of Ecological Corridor Width

In our study, 30, 60, 120, 300, and 600 m widths were used to identify the appropriate width of ecological corridors based on the proportions of different types of LULC (Figure 13). The results showed that as the corridor width increased, the proportion of construction land increased, whereas the proportion of green land gradually declined. At 30 m, forest land accounted for the largest portion, and a largely optimal overall land type was obtained. Meanwhile, human interference was minimal, which could generally ensure the survival and migration of various organisms in the corridors. Therefore, it was determined that 30 m was the appropriate width of ecological corridors in our study area.

At the width of 30 m, analysis of landscape types showed that the core area accounted for 58.61% of the total area of the ecological corridors (

Table 4. Characteristics of landscape types with the ecological corridor width of 30 m.

Landscape Type	Area (km2)	Proportion of Ecological Corridor Area (%)
Core	12.67	58.61
Islet	0.64	2.95
Perforation	0.57	2.63
Edge	4.77	22.08
Loop	0.16	0.76
Bridge	1.16	5.38
Branch	1.64	7.58
Total area of green land	21.61	79.36
Total area of ecological corridor	27.23	100

). In the construction of ecological corridors, some of the landscape elements, such as unused land, could be transformed into green land, which could increase the size of the core area and accelerate the ecological process. The edge is the transition area between core and background, which accounted for about 22.08% of ecological corridors, indicating that the ecological corridors were susceptible to external interference. Therefore, it was recommended that the boundary of the edge should be strictly protected in order to reduce human interference.

4. Discussion

4.1. Identification of Ecological Sources

Although numerous studies have applied the MSPA method to construct green ecological networks [37,38], there is a lack of ecological source identification methods based on biodiversity [39]. Previous studies have graded ecological protection or drawn a red line based on the importance of biodiversity, but have ignored the dynamic process of biological flow [40]. Biodiversity plays a vital role in maintaining the proper functions and sustainability of ecosystems [41], and the BCF assessment can indirectly reveal the regional biodiversity status by evaluating the patch habitat quality [42]. Given these concerns and the very dense environment in Qingdao, our study further combined the MSPA method and BCF assessment to optimize the ecological source identification method and then construct a green ecological network. The results of ecological source patches were largely consistent with the current ecological resources in the study area [36], indicating the appropriate use of the comprehensive identification method applied in our study.

A total of 38 ecological sources were selected and largely coexisted with the national parks listed in the Qingdao “13th Five-Year” Urban Construction Plan [43], which provided the theoretical basis for the subsequent urban green space system planning. Compared with the 26 ecological patches identified in the green ecological network in Xiamen [44], more ecological sources were selected in our study, showing higher dI values of ecological sources, probably due to the high fragmentation of green land, small patch area, and uneven distribution in Qingdao. Similarly, 10 ecological sources were selected based on patch area and habitat quality in Beijing [45], showing that although the number of ecological sources was smaller, the spatial distribution of ecological sources was similar between Beijing and Qingdao, wherein they were mainly detected in mountainous areas.

Furthermore, the source patches overlapped with the areas of high-BCF spaces, which were mainly distributed in the eastern nature reserve and the southern green park of the study area. This was largely due to these areas being located near the coastline, which mainly had the land-use type of green land. Other factors, such as higher elevation, greater relief of terrain, mild climate, and more precipitation than the interior part of the peninsula, made these patches with high BCF suitable for the survival and reproduction of creatures, although strong BCF was observed only in a few large parks along the coast. However, the dI results revealed the weak connectivity between these high-quality patches and other patches. Therefore, it is important to increase the areas of green land in order to improve the quality and connectivity among ecological patches [46].

4.2. Distribution of Landscape Resistance

Qingdao is a typical peninsula city with a high-density urban area. Our results showed that the areas with high landscape resistance values were concentrated in roads and construction land, while the old city at the top of the peninsula was the most dense with the largest landscape resistance (Figure 6 and Figure 7). The results were consistent with those previously reported [47,48], showing the spatial distribution of landscape resistance in highly urbanized areas [49]. The large areas of parks such as Zhongshan Park and Fushan Park showed high BCF, but the landscape resistance between these patches was high, making it difficult for the animals in the Laoshan area to migrate to the inner green land of the city. The ecological corridors constructed in this study formed a dense ecological network, which was conducive to introducing the rich ecological resources of Laoshan Mountain into the existing high-quality green space in the city and alleviating the lack of ecological resources in the urban area of Qingdao.

4.3. Characteristics of Ecological Corridors

The MCR model was used to construct a potential green ecological network in the urban areas of Qingdao, which was composed of 703 ecological corridors and 284 ecological nodes. These results were consistent with the Qingdao Park Urban Construction Plan (2021–2035) [50], which identified Laoshan Mountain as the ecological core and proposed the spatial pattern of a “two belts and multiple corridors into a network” model. Comparing the potential green ecological network constructed in our study with the other high-density urban areas, it was found that the number of ecological corridors and ecological nodes in each region were quite different. For example, although the area of the Chengdu–Chongqing region is about 200 times bigger than that of our study area, the ecological network of this region contains only 17 ecological sources, 33 ecological corridors, and 29 ecological nodes [51]. This was probably because the distance between the sources is relatively long in the Chengdu–Chongqing region. Meanwhile, the ecological network of Guangzhou is similar to our study area, containing 35 ecological sources, 59 ecological corridors, and 39 ecological nodes, with a total length of 817.4 km [52]. Compared with the potential green ecological network in our study, the number of ecological sources and the total length of ecological corridors was similar to that in Guangzhou, but the numbers of ecological corridors and ecological nodes were relatively small. This was likely because the ecological corridors and ecological nodes in Guangzhou were extracted based on circuit theory, with the high-flow corridors and important nodes in the ecological process selected, whereas the MCR model was applied in our study to simulate the possible paths of biological migration, which could effectively connect the fragmented ecological patches.

4.4. Optimization of Potential Green Ecological Network

By optimizing the potential green ecological network, the priority areas for protection and construction could be accurately identified. The results showed that the values of the α, β, and γ indices (1.39, 3.66, and 1.26) of the optimized green ecological network in our study were higher than those (0.27, 2.03, and 0.72) of the green ecological network in Fuzhou [53], indicating that the green ecological network optimized by the gravity model and kernel density was more stable with higher ecological process efficiency. However, higher values of the α, β, and γ indices (3.206, 6.412, and 2.422) were detected in the optimized ecological network in the Dianchi Lake region [54], which treated the ecological sources as ecological nodes in the calculation of the network structure.

Previous studies identified an ecological corridor width of 100 m simulated in the Yanqing District of Beijing [55]. In our study, a shorter optimal ecological corridor width of 30 m was determined, probably due to the large proportion of construction land and the small number of forest patches in Qingdao. Therefore, it was recommended that the ecological corridors should be widened to the highest level possible, and the high-quality forest patches should be included in the boundary of the ecological corridors in order to improve the efficiency of biological migration [56,57]. In ecological corridors, the edge area can be easily affected by human activities. Compared with the internal species that require a single environmental resource condition, ecological corridors are more conducive to adapting to the migration of marginal species in a changing environment [58].

4.5. Suggestions for Ecological Construction of Qingdao

In Qingdao, it was recommended that the new urban areas should make full use of unused land and other patches with low ecological benefits by transforming them into green land patches, which will be conducive to the sustainable development of the city. However, the characteristics of land were well solidified in the old town area, making it largely impossible to add a large area of green space. Therefore, it is necessary to improve the quality of existing green space patches and to optimize the community structure. Furthermore, food-producing plants could provide sufficient nutrients for organisms [59]. To summarize, during ecological construction, native plants should be included on the basis of existing vegetation to provide adequate food and habitats for organisms in order to enhance the potential attractiveness of urban green patches to suburban organisms [60].

Furthermore, the vegetation in Qingdao is mainly characterized by deciduous broad-leaved forest, and organisms lack food and habitats in both autumn and winter [61]. Therefore, it is recommended that during the planning of ecological corridors, the proportion of evergreen tree species in the community should be increased in order to reduce the seasonal impact on the quality of ecological corridors [62], to improve the attraction of ecological corridors to organisms, and to increase the possibility and success rate of biological migration in the corridors.

5. Conclusions

In this study, ecological sources were identified based on the combination of MSPA and BCF assessment and the green ecological network in the urban area of Qingdao was constructed based on the MCR model to explore the spatial distribution of habitat quality and ecological resistance. The ecological network was further optimized based on the gravity model and kernel density analysis to clarify the spatial distribution and conservation importance of the green ecological network. Our study indicated that the construction of the green ecological network played an important role in promoting the landscape connectivity of ecological patches between urban and suburban areas and maintaining the urban biodiversity in Qingdao.