Introduction

In recent years, with the rapid advancement of urbanization and industrialization, the expansion of construction land has been accelerating, resulting in the degradation of ecological environment, loss of species habitat, and serious landscape fragmentation in certain regions1. In this context, the rapidly expanding cities are confronted with various issues, such as the decline of habitat quality, the decrease of biodiversity and the threat of ecological security2. As the ecological governance technology advances, a large number of studies have proposed to relieve the contradiction between urban sprawl and habitat destruction by means of the construction of urban ecological network3. The construction of ecological network can not only promote the circulation of ecological materials and energy in the city, but also is of great significance for urban ecological spatial planning and the realization of regional sustainable development4.

Since the 1970s, the study of ecological network has attracted the attention of foreign scholars, who have made repeated attempts on the approaches and the model construction centering on ecological network from the macroscopic angle5,6. Currently, the structure of ecological network has become more mature, and the research on ecological network construction has begun to take shape: ecological source area identification, combined ecological resistance surface construction and potential ecological corridor extraction7. For the identification of ecological source areas, most existing studies take ecosystem services, ecological security, landscape connectivity, the importance of ecological function and other factors as the basis for evaluation, or directly select appropriate ecological patches based on patch area and attribute8,9. The combined resistance surface is mostly obtained based on the resistance assignment or the establishment of relevant evaluation systems. Some scholars have capitalized on nighttime light data, topographic potential index and geological hazard sensitivity to modify the combined resistance surface, in order to make the evaluation results more objective10,11. Minimum cumulative resistance (MCR) and circuit theory models are extensively utilized to extract potential ecological corridors, and important ecological corridors are identified based on gravity models12,Overview of the study area

Located at the source of the Pearl River, the third largest river in China, and the middle and upper reaches of the **sha River, Qu**g City undertakes the construction of three major ecological security barriers in the Pearl River, Yangtze River basins and the western plateau. It is a typical plateau mountain forest city, the second largest economy and second largest city in Yunnan Province. Qilin District (25°08′ ~ 25°36′N, 103°10′ ~ 104°13′E), the headquarter of Qu**g Municipal Committee and municipal government, is the main urban area of Qu**g City. Situated in the east of Yunnan Province, the middle of the east Yunnan plateau and the upper reaches of Nanpan River, it is with an elevation of 1881 m and a total area of 1552.83km2 (Fig. 1). It has a north subtropical monsoon climate, with an average temperature of 16.4℃, annual rainfall of 802.3 mm, and 98% of the days with good air quality throughout the year. In 2023, Qilin District was named “Forest City of the Province” (https://lcj.yn.gov.cn/html/2023/zuixindongtai_0721/69389.html) by the Forestry and Grassland Bureau of Yunnan Province. There are 4 county-level nature reserves in the district, with a forest coverage rate of 48.8%.

Figure 1
figure 1

The study area.

Full size image

Data sources and research methods

Data sources and processing

The data used in this study mainly included administrative division, DEM, slope, land use, NDVI etc. The administrative vector data was derived from BIGEMAP map downloader (http://www.bigemap.com/), and DEM digital elevation data from geospatial data cloud (https://www.gscloud.cn, resolution 30 m × 30 m). Based on DEM and ArcGIS10.7, the slope data was processed through inlaying, mask extraction and projection, and finally obtained by means of 3D Analyst tool, for instance, raster surface. On the basis of the Landsat 8OLI_TIRS remote sensing image (https://www.usgs.gov, resolution 30 m × 30 m, cloudcover 1.13%) in March 2021, the land use data was extracted by the interpretation of ENVI5.3 supervised classification, in combination with visual observation. Combined with the field investigation, the data interpreted was revised and proofread repeatedly. Finally, land use types were generated based on the ArcGIS10.7, including 6 types of woodland, water body, grassland, cultivated land, construction land and other land according to the current national classification standard15, and the confusion matrix was used to verify the accuracy. The overall accuracy reached 86.30% and the Kappa coefficient was 0.8, which met the needs of this research. The NDVI data was based on Landsat8OLI_TIRS remote sensing image (https://www.gscloud.cn, resolution 30 m × 30 m, cloudcover 1.13) in March 2021 and calculated by ENVI5.3.

Research methods

Analysis of landscape patterns

MSPA, proposed by Vogt et al.16, is an image processing method that relies on the principle of mathematical morphology to measure, identify and segment the spatial pattern of raster images. Referring to relevant studies17 and based on land use data, woodland was extracted as foreground data (assigned value 2), and the rest was background data (assigned value 1). Based on the Guidos Toolbox software, eight-neighborhood image thinning analysis was applied to analyze the binary raster 8-bit Tiff data, and 7 non-overlap** landscape elements, including core area, island, pore, and edge area, were obtained, so that structural elements such as core area and corridor number were directly displayed in the results. Among them, the core area was the best choice for species habitat due to its large area, small degree of fragmentation, and complete shape. Consequently, the core area was selected as the alternative source area. According to the Gudios Toolbox (GuidosToolbox_Manual (europa.eu)), the threshold value of the core area was 17/117.

Identification of ecological source areas

Important ecological source areas are areas where material exchange and energy flow are sufficient, and their accurate identification is the key to the subsequent construction of ecological network. This study took patch area size and landscape connectivity index into account in a comprehensive manner to select important source areas18. Two landscape connectivity indices, integral index of connectivity (IIC) and probability index of connectivity (PC), can be used to reflect the connectivity of landscape patches and calculate the importance value of patches (dPC) to landscape connectivity19,20. The calculation formula is as follows:

$$IIC = \frac{{\sum\nolimits_{i = 1}^{n} {\sum\nolimits_{j = 1}^{n} {\left( {\frac{{a_{i} \cdot a_{j} }}{{1 + nl_{ij} }}} \right)} } }}{{A_{{}}^{2} }}$$
(1)
$$PC = \frac{{\sum\nolimits_{i = 1}^{n} {\sum\nolimits_{j = 1}^{n} {a_{i} \cdot a_{j} \cdot p_{ij}^{ * } } } }}{{A_{{}}^{2} }}\left( {0 < PC \leqq 1} \right)$$
(2)
$$dPC = \frac{{PC - PC_{remove} }}{PC} \times 100\%$$
(3)

where: n is the total number of patches, a is the patch area, nlij is the number of connections between patches i and j, and P*ij is the maximum probability of species migration path between patches i and j. A is the total landscape area, and PCremove is the landscape connectivity of remaining patches after randomly removing patch i.

Construction of resistance surface

A resistance surface consists of one or more resistance layers. Resistance factors such as land use type, DEM, slope, distance from water body, distance from road and distance from residential site are often used in relevant researches, and different combinations are available depending on research needs27. The calculation formula is as follows:

$$MCR = f_{{{\text{min}}}} \sum\limits_{j = n}^{i = m} {(D_{ij} \times R_{i} )}$$
(4)

where: f is an unknown increasing function, m is the number of landscape unit i, and n is the number of ecological source j. Dij is the spatial distance from j to i, while Ri is the cmbined resistance coefficient of i to the movement of a certain species.

The gravity model can quantitatively evaluate the intensity of the interaction between patches, and the greater the value of the interaction force, the more important the potential corridor between the two in the regional ecosystem28. The calculation formula is as below:

$$G_{ij} = \frac{{L_{\max }^{2} \ln (S_{i} )\ln (S_{j} )}}{{L_{{{\text{ij}}}}^{2} P_{i} P_{j} }}$$
(5)

where: Gij is the interaction force between patch i and j, Lmax is the maximum resistance value of all corridors, and Si is the area of patch i. Lij is the cumulative resistance value of the corridor between i and j, and Pi is the resistance value of patch i.

Analysis of Network Connectivity

Connectivity can reflect the connectivity degree of ecological network in a quantitative fashion29. In this study, ecological network connectivity analysis approach was utilized to quantitatively describe the closure, complexity and connectivity of the potential ecological network structure constructed, by calculating three indexes of network closure (α), line point rate (β) and network connectivity (γ)30,31. The calculation formula is as follows:

$$\alpha = \frac{L - V + 1}{{2V + 5}}$$
(6)
$$\beta = \frac{L}{V}$$
(7)
$$\gamma = \frac{L}{{3\left( {V - 2} \right)}}$$
(8)

where: L is the number of corridors and V is the number of nodes.

Results and analysis

Analysis of MSPA landscape patterns

As shown in Fig. 2 and Table 2, there were7 landscape types analyzed based on MSPA, with a total area of 496.46km2, accounting for 31.97% of the whole study area. The core area was 400.58km2, occupying 80.69% of the ecological space area, with the largest proportion, mainly distributed in the west, southeast and northeast of the study area, with large patch area and strong stability. However, the lack of connectivity among the three core patches was not conducive to the material exchange and energy flow among the patches. The proportion of marginal areas and pores was 12.31% and 2.10%, respectively, indicating that there was a serious issue of patch fragmentation in the study area, and it was necessary to strengthen the landscape connectivity between patches in the core area. As the corridor connecting patches in the ecological network, the bridge area was distributed near the core patch, accounting for 1.52%, displaying that there were few structural corridors in the landscape pattern of the study area, not beneficial to species migration, and corridor optimization was called for. As a temporary habitat for the flow of ecological factors, isolated islands scattered in the study area took up only 0.53%. The stability of the ecological network can be enhanced by adding step** stones in the future. The remaining landscape types were loops (0.43%) and branches (2.42%).

Figure 2
figure 2

MSPA landscape analysis.

Full size image
Table 2 MSPA classified statistics.
Full size table

Evaluation of the connectivity of important ecological source areas

Referring to relevant studies

Figure 4
figure 4

Combined resistance surface.

Full size image

Extraction and analysis of important ecological corridors

According to the above analysis, 91 potential ecological corridors were acquired by calculating the minimum path from source patch to target patch based on the cost distance and cost path through Spatial Analysis tool. On the basis of the gravity model, the interaction matrix among 14 important ecological source areas was constructed to quantitatively analyze and evaluate the relative importance of potential corridors (Table 4). The 16 corridors with the greatest interaction force were extracted as important corridors, with the redundant corridors deleted, so as to obtain the important ecological corridors in the study area (Fig. 5).

Table 4 Interaction matrix among important ecological source areas.
Full size table
Figure 5
figure 5

Distribution of important ecological corridors.

Full size image

As shown in Table 4, the interaction between source area 1 and 4 was the strongest, indicating that the resistance of biological migration between source areas was the least and the degree of connectivity was the highest. Taking a prominent position in the ecological network, corridors should be strengthened in terms of control and protection. The distance between source area 3 and 12 was the furthest, and the interaction between the two was the weakest, demonstrating that the landscape resistance of the ecological corridor between the source areas was relatively large and the connectivity was weak, which could not fulfill the requirements for species migration and energy diffusion, undermining the species richness and biodiversity of the study area. Optimization was needed in this regard.

From Fig. 5, it could be seen that important ecological corridors were mainly distributed in parts of the west, northwest, northeast and southeast. The connectivity was strongest in the northeast, majorly located in Wutai Mountain and Langmushan Mountain county-level nature reserves, but the four parts were relatively independent and had poor connectivity with each other. It followed then that the ecological network in the study area was not sound enough to form complete network connection. Ecological nodes could be supplemented on the basis of the extracted corridors to promote the material exchange and biological exchange in the study area.

Analysis of ecological corridor network connectivity

By calculating the ecological network connectivity, it was observed that the α, β and γ indices before the optimization of ecological network in the study area were 2.36, 6.5 and 2.53, respectively. As shown in Fig. 6, according to the importance degree and location distribution of general ecological source areas, this study selected 6 complementary important ecological source areas as step** stone patches, including 1 in the south, 1 in the northwest, 2 in the southeast, and 2 in the middle. Based on the supplementary calculation of the original important ecological source areas, 99 new planned ecological corridors were obtained. After optimization, there were 20 important ecological source areas and 190 ecological corridors in the study area. The α, β and γ indices of the ecological network after optimization were 3.8, 9.5 and 3.5, respectively, which greatly improved the connectivity of the ecological network and optimized the ecological network. In the future planning and construction, more attention could be attached to step** stone patches, and the ecological network system of the study area could be improved by constructing new ecological corridors.

Figure 6
figure 6

Optimized ecological network system of the study area.

Full size image