Locating at the northeastern Zhejiang Province and south bank of Hangzhou Bay, Ningbo city (120°55'-122°16'E, 28°51'-30°33'N) is an economic center at the south wing of the Yangtze River Delta, having a jurisdiction over six districts, two counties and two county-level cities, with a total area of about 9543 km² (Figure 1). Topographically, it is high in the southwest and low in the northeast, with a plain area accounting for about 40.30%. There are many low mountains and hills, numerous sea islands and circuitous coastlines, and the cities are mainly seated on the northern plains. It belongs to a subtropical monsoon climate with sufficient heat resources and abundant rainfall, and the dominant vegetations are subtropical evergreen forests, which are mainly distributed in hilly and mountainous areas. Under the topographic influence, relatively fragmented geomorphic units and continuously expanded urban built-up areas have generated important effects on the spatial distribution of ecological sources and corridors; particularly, valley ecological degradation imposes a stress on the connectivity of ecological corridors.

Two scenes of Landsat 8 OLI/TIRS C1 Level-1 imagery, acquired on August 16, 2020, were screened out to perform preprocessing such as radiometric calibration, atmospheric correction, mosaicking and clipping. Land use types were verified and corrected via Google Earth through visual interpretation, and land use types were reclassified into farmland, green land (forest and grassland), water body, built-up land and bare land. The ecological protection red line in Zhejiang Province was vectorized. All data (Table 1) were represented under a unified coordinate system at a spatial resolution of 30 m.

The remote sensing ecological index (RSEI), which integrates four ecological-spectral indices: normalized difference vegetation index (NDVI), wetness (Wet), land surface temperature (LST) and normalized difference built-up and bare soil index (NDBSI), can be rapidly and conveniently used to evaluate regional ecological quality without needing manual parameter setting (Xu, 2013a; 2013b), and it has been extensively applied to objectively evaluating an eco-environmental quality. Definitions of the four ecological-spectral indices with Landsat 8 data are described as follows:

NDVI:

Wet:

The NDBSI is affected by both buildings and bare land, so it is synthesized from index- based built-up index (IBI) and soil index (SI).

Whereρ_2-7 represents the 2nd to 7th bands’ reflectance of Landsat 8, respectively.

The retrieval of land surface temperature (LST) was implemented via Landsat 8 TIRS based on a radiative transfer equation-based method (Yu et al., 2014). The TIR radiance value L_λ includes three parts: the energy of radiance of ground object transmitted to the sensor via the atmosphere, the energy reflected by the atmosphere after the downwelling radiation to the ground surface, and upwelling path radiance L↑ of atmosphere.

where ε denotes the surface emissivity; T_S is the true surface temperature (K); B(T_S) is the thermal radiance of black body and τ stands for the atmospheric transmittance in a TIR band. The radiance B(T_S) of black body with temperature T in the TIR band is calculated as follows:

LST can be acquired using the function of Planck formula:

For TIRS10, K₁ = 774.89 W/(m²·µm·sr), K₂ = 1321.08 K.

After the four indices were calculated, they were fused by conducting a principal component analysis (PCA) and the dimension reduction was implemented by taking the first principal component image. Before conducting a PCA analysis, according to inversion results and actual situation of the study area, the confidence interval for each index was taken within the standard of 2% from its normalized value. The normalized result of the first principal component was taken as RSEI, and then the RSEI was divided into five levels by an equal interval of 0.2.

According to the topological relation between elements and structure, a morphological spatial pattern analysis was used to divide grid data into seven types (zones) of ecological landscapes (core, islet, perforation, edge, bridge, loop and branch) based on mathematical morphological principles including corrosion, expansion, open operation and close operation, and their (the seven ecological landscapes) specific ecological significance was clearly presented (An et al., 2021). The core zone is a complete patch and can serve as a habitat. By referring to relevant references (Xu, 2013a; Xu,2013b) and practical regional status, the areas with RSEI > 0.8 were used as the foreground, and the core zones with an area of > 10 km² were chosen as ecological sources.

Circuitscape (McRae and Shah, 2009) simulates and predicts the movement, gene flow and genetic differentiation between animal and plant populations in heterogeneous landscapes based on the circuit theory. The resistance value corresponding to each grid is assigned according to an influence mode and resistance value of resistance sources, and the current value between nodes is iteratively calculated through the pairing, all-to-one and one-to-all patterns. A migration probability of species between ecological sources is positively correlated with a cumulative current value between nodes. In this study, the ecological sources were taken as the nodes, and the terrain gradient, distance from water body, distance from trunk line, population density, degree of landscape fragmentation and land use were chosen as the resistance sources from the aspects of natural endowment, human development and land use types. Next, the type, distance and area differences were planned as a whole. Hierarchical buffer zones were set for both road network and water body, so their resistance values might be set. The Shannon diversity index of land use data was calculated as a fragmentation degree via the software of Fragstats using a moving window method, and natural breaks were used to classify and assign resistance values. The resistance value was set with referring to relevant research (Liu et al., 2018; Dai et al., 2021). A weight of each resistance source was determined using an analytic hierarchy process method (with a consistency ratio of 0.068), and the resistance surface was constructed (Table 2).

By calculating the cost-weighted distance between ecological sources with the nearest distance in pixel, Linkage Mapper (Carroll et al., 2012) was used to create the cost surface of weighted distance and to calculate the least-cost path (LCP). The ratio of cost-weighted distance (CWD) of least-cost path to its length (CWD: LCPL) was chosen to display the path cost. The greater the CWD: LCPL value, the greater the resistance, the worse it was for the species migration between the ecological sources. A global ecological corridor is the LCP between the ecological sources integrated from all ecological sources, and a potential ecological corridor is the LCP between every two specific ecological sources. The potential ecological corridor may pass through an intermediate ecological source. The global ecological corridor is taken as a key ecological corridor and all the potential ecological corridors are not included in the ecological security patterns.

The importance degree of one ecological source in an ecological security pattern can be described by the source centrality. Regarding the ecological sources as nodes in the current, Centrality Mapper (Carroll et al., 2012) was used to take the minimum CWD between any two ecological sources as the resistance, and to generate a cumulative current diagram through the iterative operation similar to the pairing mode in Circuitscape. A greater current value represents the better connectivity of this ecological source with a higher importance in the ecological security pattern. The calculation process relies upon the LCP and minimum CWD.

An identification of pinch points considers that a pinch point is a high-current-density area in an ecological corridor, an area that may be passed by species migration, and a key point that connects between the ecological sources, with a great significance to the connectivity of ecological corridors, thus the pinch point’s protection should be prioritized in the ecological protection. By invocting the pairing calculation mode of Circuitscape, Pinchpoint Mapper inputs the constructed resistance surface, sets the cut-off distance of CWD as 3000, and generates a pinch point diagram through an iterative calculation.

An identification of barrier points considers that Barrier Mapper (McRae et al., 2012) was used to set a circular moving search window with a designated radius D. Based on the least-cost distance LCD₀, if the pixel value within the window is reduced by one unit, then the LCD₀ value declines to LCD₁ after the barrier points are removed, and the improvement score is IS = (LCD₀-LCD₁)/D. The greater the improvement score, the greater the recovery degree of regional connectivity after the barriers are removed. In this study, the search radius was set as 60 m to identify barrier points.

Ecological sources and ecological corridors are areas that should be preferentially protected in a regional ecological security pattern. The regional and hierarchical protection of these areas should be implemented in order to improve the protective effect, reduce the protection cost and maximize the protection benefit. The ecological pinch point is the area with high current value in the ecological corridor, indicating that species are most likely to pass there. Ecological barrier points are areas in an ecological corridor that are not easy for species to pass through. Therefore, they can be expressed as areas in an urgent need of restoration. Current values of both ecological pinch points maintaining a connectivity of ecological corridors and ecological barrier points being urgently restored were superposed after the processes. According to the natural breaking point method, they were divided into prioritized restoration zones, prioritized protection zones and key conservation zones in a descending order of their importance degrees, and the ecological sources that were not included in those zones were specifically posed and set as general conservation zones.

The RSEI was divided into 1-5 levels (Figure 3). The 1st level zones occupied an area of 383.12 km², accounting for 4.01% of the total study area. These zones were mainly distributed in the core area of built-up land, with low NDVI and Wet and high NDBSI and LST values, so the overall habitat quality was poor. The 2nd level zones covered an area of about 1403.84 km² (14.71%), and the main land use types were built-up land and farmland with poor habitat quality. The 3rd level zones occupied an area of 2092.03 km² (21.92%), mainly being wetland mudflats, river systems and some farmlands nearby the coast with medium habitat quality. The 4th level zones occupied an area of about 2583.74 km² (27.08%), mainly including grass land and marginal areas of forest land with high habitat quality. The 5th level zones covered an area of 3080.03 km² (32.28%), mainly distributed in mountainous areas; the main land use type was forest land; and the habitat quality was high. The quality of habitats at different levels showed a hierarchical structure. Overall, the zones affected by human activities to a minor extent enjoyed relatively high habitat quality, while the habitat quality was poor in areas where human activities were concentrated like cities and rural settlements as well as industrial and mining areas, in which the zones with RSEI of 0.8-1.0 were determined. By the morphological spatial pattern analysis (MSPA), a total of 42 ecological sources with a core area larger than 10 km² were selected, covering a total area of 1325.11 km². By referring to the ecological protection red line chart of Zhejiang Province, the zones outside the red line were taken as the ecological sources, and those included in it but not completely overlapped by the red line were considered as ecological protection of the province. Given the definition of the ecological protection red line, a total of 47 ecological sources were generated, with a total area of about 1540.15 km², accounting for 16.14% of the total study area (Figure 4a).

The ecological sources distributed along mountain ranges, as a whole, presenting three ecological source accumulation zones from the northeast to the southwest. According to the ecological protection red line, the southern coast of Hangzhou Bay was added to form a “one patch and three belts” distribution pattern (Figure 4a). From the county-level administrative division, the ecological sources in Ninghai county, Yuyao city and Fenghua district had large areas, being 388.76 km², 242.97 km² and 233.63 km², respectively. The ecological sources in Ninghai, Fenghua and Yinzhou district occupied large areas in the administrative region, with proportions of 22.72%, 18.65% and 18.52%, respectively, and the ecological sources in Zhenhai and Jiangbei districts occupied small proportions, being 12.45 km² and 5.20 km², respectively. The areas of Zhenhai and Jiangbei were relatively small, and they had a low and flat terrain, which was favorable for urban expansion, and thus the built-up land occupied a high proportion. There were a few high-level habitat patches in the two districts. The ratios of ecological source area in each district/county were ranked as Ninghai > Fenghua > Yinzhou > Haishu > Beilun > Yuyao > Xiangshan > Cixi > Zhenhai > Jiangbei.

The resistance surface constructed by combining natural and human factors intuitively reflected the disturbance suffered by the species during their migration between ecological sources (Figure 4b). Among various factors, the human development factor had a relatively high weight, and the contiguous artificial surfaces and fragmentary and sporadic artificial surfaces were the primary causes affecting the connectivity between ecological sources. In addition, influenced by geographical location and topographic factors, the special coastal landform in coastal areas aggravated the progress of small-scale “geographical isolation” of sea islands. The resistance values for the resistance surface in Ningbo city ranged from 8.39 to 60.30; the high-resistance values were mainly concentrated in urban built-up areas; and the overall resistance value was high in Zhenhai, Jiangbei, and Cixi. The high-resistance areas in the northern region presented a flaky distribution; the low mountains and hills were represented by Siming and Wulei mountains that became the low-value areas of northern resistance surface; the southern high-resistance areas centralized in intermontane valleys and coastal areas; and the values were gradually reduced outward from cities and settlements.

Ecological corridors were of important channels connecting ecological sources and realizing a material migration, energy transfer and information exchange. The ecological corridors and ecological sources jointly constitute a regional basic ecological security pattern. The least-cost path (LCP) was extracted via Linkage Mapper. A total of 95 ecological corridors were generated among the 47 ecological sources in Ningbo, with a total length of about 708.67 km, among which there were 83 key ecological corridors and 12 potential ecological corridors. The total length of key ecological corridors was 471.19 km (an average of 5.68 km with a maximum of 38.25 km and a minimum of 0.08 km). The key ecological corridors were densely concentrated in the central-west: No.12 ecological source connected the most key ecological corridors (8), and the ecological source of No.42 did not connect with any other ecological sources owing to the administrative boundary and the island landform. The total length of potential ecological corridors was about 237.47 km (an average of 19.79 km with a maximum of 46.69 km and a minimum of 0.93 km). As shown in Figure 5a, the width of ecological corridor denoted the CWD: LCPL value: the narrower the corridor, the lower the CWD:LCPL ratio, namely, the smaller the resistance value. The average CWD:LCDL value of key ecological corridors was 19.74, but that of key ecological corridors between Nos. 4 and 5 ecological sources was the minimum, about 12.03. It could be clearly seen from Figure 5a that the CWD:LCPL values of ecological corridors among ecological sources in southern Yuyao, western Haishu and northwestern Fenghua were low; the more the ecological corridors were densely distributed, the more the favorable overall connectivity, and the lower the species migration cost among these areas.

The cumulative current values of 47 ecological sources ranged from 0 to 453.55. As shown in Figure 5a, the darker the ecological source, the greater the cumulative current value, and the stronger the centrality. The cumulative current value of No.27 ecological source was the maximum, which was the core area of ecological sources and contributed the most to the connectivity of regional ecological sources. The centrality of No.42 ecological source was 0, becoming an ecological isolated island. As shown in Figure 5b, the current value of No.42 ecological source was 0, and the current value of other islands was also 0 under the influences of administrative boundary and coastal landform.

The cumulative current diagram was analyzed by combining the centrality of ecological sources and resistance of ecological corridors in the study area under the one-to-one pattern. The key areas of ecological security patterns in Ningbo were located in the western and southern areas, including southern Yuyao, western Haishu, Fenghua, northern Ninghai and central-east Xiangshan. The areas with high-centrality ecological sources were located in mountainous areas, where the density of ecological corridors was gradually reduced outward from the central-west. The resistance values of ecological corridors passing through the flat areas were large; the distribution density of ecological corridors in coastal areas of Xiangshan was relatively low under the topographic influences of bay and island; but the resistance values of ecological corridors were relatively low; and the cumulative current value was large. The southern coastal zones became narrow and small in human settlements, with intense human development activities, weak connectivity, and low current value due to circuitous lines of bay, thus becoming an ecological marginal area with a small quantity of mudflats and salt marshes. As an important type of ecological land, wetlands are of great importance to the regional ecological security. Affected by the distribution of built-up land and farmland in the northern part, there were a few ecological sources with a great ecological resistance, a low density of ecological corridors, a weak connectivity between ecological sources and a low cumulative current value.

It could be observed from Figure 6 that the prioritized restoration zones are concentrated at the junction between the narrow part inside ecological sources and ecological corridors and they played a decisive role in the integrity of ecological sources and connectivity of ecological corridors. Their ecological quality had a direct impact on the construction of the whole regional ecological network and regional ecological security. Given possible barriers, these zones should be restored as priorities as they were collectively distributed in Ninghai and Fenghua, with an area of 7.30 km² and 5.59 km², respectively. Ecological corridors dominated the prioritized protection zones, constituting a basic network framework of ecological security patterns, which could serve as a channel and bridge for the material migration and energy transfer between ecological sources. Some ecological corridors in Ningbo city should be protected as priorities due to their important location and large resistance, and the resistance of ecological corridors should be mitigated by taking ecological improvement measures while maintaining the current status. Key conservation zones were the core zones with high centrality of ecological sources and some areas with small resistance values along ecological corridors. These zones, which covered a large area, dominated the whole regional ecological quality and was a basis for the regional ecological security, so they should be given first priority among priorities, needing a continuous improvement of habitat quality. General conservation zones were mainly distributed in the edge zones of key conservation zones, with relatively low importance in the ecological network and a low resistance value, too. However, they usually located in the border areas between natural and artificial surfaces, with a fragile ecological environment; they could be easily degraded due to the stress imposed by human activities. Though there was not a high importance degree in the whole ecological network and was a weak effect on maintaining the regional ecological connectivity, they should still be protected and restored. Influenced by topographic factors in southern Xiangshan, both centrality and connectivity of ecological sources were low, so they were classified as general conservation zones.

From the construction results of restoration zones in a county-level administrative division, the areas of subzones were ranked as key conservation zones > prioritized protection zones > general conservation zones > prioritized restoration zones (Table 3). The different ecological remediation zones in each district or county were obviously different in area, and the prioritized restoration zones in Ninghai and Fenghua were large, accounting for 70.58% of the total area of all prioritized restoration zones. Although human activities were intense in Jiangbei, Zhenhai and Beilun, the ecological sources and corridors were less distributed. They are located at the vacancy of an ecological network, and the area of prioritized restoration zones was small. The area of prioritized protection zones in Fenghua and Ninghai was large, and their ecological corridors were staggered in the two areas, where ecological protection should be enhanced to ensure a regional ecological security. Among the key conservation zones, Ninghai and Yuyao occupied large areas, being 261.91 km² and 243.87 km², respectively, and their administrative areas were large with a high proportion of ecological sources. The coastlines were circuitous in Xiangshan and Ninghai, together with a large amount of sea islands, so the general conservation zones occupied large areas (50.44 km² and 37.01 km², respectively). In addition, the connectivity of ecological sources was weak in Xiangshan.

By combining the area analysis of county-level administrative regions with ecological restoration zones, the ecological land area in Ningbo was about 1898.39 km², accounting for 19.89% of the total study area. The ecological land areas in the county-level administrative regions were ranked as Ninghai > Yuyao > Fenghua > Xiangshan > Yinzhou > Cixi > Haishu > Beilun > Zhenhai > Jiangbei, and the total ecological land area in Ninghai, Yuyao and Fenghua accounted for about 58.49% of total ecological land area in the study area, so they were of great importance to the regional ecological security. The ecological land areas in Ninghai, Fenghua, Yuyao and Yinzhou accounted for large proportions (25.80%, 24.56%, 23.49% and 20.43%, respectively) of their respective administrative regions, while those in Jiangbei and Cixi accounted for small proportions (5.41% and 10.27%, respectively) of their corresponding administrative regions. The areas of ecological restoration zones in Haishu, Beilun, Yinzhou, Ninghai, Yuyao and Cixi were ranked as key conservation zones > prioritized protection zones > general conservation zones > prioritized restoration zones, being identical with the total area relations among the four ecological restoration zones. Efforts should be made to strengthen the conservation of ecological sources as well as the ecological adjustment and optimization of edge zones of ecological sources. The areas of the four ecological restoration zones in Jiangbei, Zhenhai and Fenghua were ranked as prioritized protection zones > key conservation zones >general conservation zones > prioritized restoration zones, among which the area of prioritized protection zones was relatively large, and the emphasis should be laid on the removal of ecological barrier points based on the protection of ecological corridors. The areas of the four ecological restoration zones in Xiangshan were ranked as key conservation zones > general conservation zones > prioritized protection zones > prioritized restoration zones, and the protection of edge zones of ecological sources should be reinforced.

An ecological security pattern is a regional ecological line of defense. The ecological restoration zones constructed on the basis of ecological security patterns are helpful to the precise optimization of the ecological security patterns. Compared with the existing forbidden development zones (Li et al., 2021), it is more conducive to maintain connectivity between different areas, and the variation of ecological importance within the protected area is also enhanced.

Ecological security includes human survival, human development and environmental resources (Liu and Chang, 2015). The ecological protection red line can reflect the regional ecosystem services (Liu et al., 2015). The habitat suitability, ecological importance and necessary ecological service value denoted by RSEI and the ecological protection red line can meet the needs of human survival, human development and environmental resources. A reasonable selection of conservation and restoration zones is a key to maximize impacts of ecological engineering. Therefore, an efficient ecological regulation can be realized by using more objective ecological environmental evaluation indices, combining with the red line of ecological protection, selecting areas with good habitat quality as ecological sources and constructing an ecological security pattern to perform hierarchical protection and restoration.

In this study, Ningbo city was taken as the study area. The construction of resistance surface was restricted by the administrative boundaries in the study area, while the surface status outside the boundaries was not considered. In reality, ecological processes are not influenced by the administrative boundaries. Due to considering the special coastal landform in the study area, sea islands become isolated ecological islands, and the “geographical isolation” may still exist within a small scale island due to the obstruction from seawater. Hence, it was reasonable that there were no ecological corridors in the ecological sources between islands and mainland in this study. During the extraction of ecological corridors, ecological pinch points and barrier points, the cost-weighted distance method and circuit theory were utilized, and the simulated possible species migration paths between ecological sources were regarded as the ecological corridors. Actually, the species that are most possible to migrate between dispersed ecological sources are animals with flying ability like birds (Liu et al., 2021), and other species may migrate in areas with densely distributed ecological sources. Therefore, the resistance values were set in a simulation process, not focusing on one single species, but instead, on factors that influence migration. The resistance values were used to characterize the difficulty of species migration. There is no one unified standard in the current literatures. Therefore, in this study, the resistance surface was constructed by combining a type and distance with area differences from multiple aspects like natural endowments and influences of human activities, which were more persuasive.

The analysis results regarding the ecological restoration zoning under the background of an ecological security pattern in a city area with a relatively perfect internal structure can only serve as the ecological defense line and provide a reference for maintaining the current regional ecological status in terms of landscape patterns. However, biological processes in different restoration zones should be restored in order to realize the ecological restoration. Of course, influenced by artificial boundaries and a study scale, a transregional study will better facilitate the construction of ecological security pattern (Dong et al., 2021), which remains to be solved in the future. Besides, the field survey and verification of ecological pinch points and barrier points in the future study should be stressed in an effort to better serve the accurate restoration and protection of ecologically vulnerable areas.

The regional habitat quality was evaluated through a remote sensing ecological method. A morphological analysis method was used to identify ecological sources by considering the ecological protection red line. Further, an ecological resistance surface was constructed under a full consideration of natural endowment and human development from differences of multiple aspects including type, area and distance. The ecological corridors, ecological pinch points and barrier points were extracted using the circuit theory; an ecological security network was constructed; and ecological restoration zones were identified and mapped. In this study, several conclusions were derived from the analysis results as follows:

(1) A total of 47 ecological sources were identified in Ningbo city, with a total area of about 1540.15 km², which accounted for 16.14% of the total study area. A total of 83 key ecological corridors were extracted, with a total length of about 471.19 km, and thus an ecological network pattern of Ningbo was formed. The total ecological land area was about 1898.39 km² (19.89% of total study area).

(2) The core areas of ecological network collectively distributed in the western and southern regions, including southern Yuyao, western Haishu, Fenghua, northern Ninghai and central-west Xiangshan. The ecological sources presented a “one patch and three belts” distribution pattern, and the low-resistance ecological corridors were concentrated in southern Yuyao, western Haishu and central-west Fenghua.

(3) The prioritized restoration zones in the ecological restoration zones were concentrated in areas between the narrow part inside ecological sources and ecological corridors. The ecological corridors played a dominant role in the prioritized protection zones; the key conservation zones were mainly the core zones of ecological sources with high centrality and some zones with small resistance values on ecological corridors; and the general conservation zones were mainly distributed in the edge areas of key conservation zones.

(4) The ecological land use area was large in Ninghai, Yuyao and Fenghua. The prioritized restoration zones and prioritized protection zones in Ninghai County and Fenghua District occupied large areas, and thus the emphasis should be laid on ecological restoration and protection in order to improve the ecological environment, keep the connectivity of ecological network and maintain the regional ecological security.