3 RESULTS

3.1 Key ecological corridors assessment

Although the UAHB ESN had about the same number of ecological sources in 1995 and 2015, it had 27 more key ecological corridors in 2015 than in 1995. This means that the connection between ecological nodes was strengthened in 2015. The total length of the key ecological corridors decreased slightly from 2289.25km in 1995 to 2238.50km in 2015 (see also Figure 3).
In the past 20 years (Table 1), the length of the key ecological corridors in Ningbo, Shaoxing, and Huzhou had increased from 1995 to 2015. The length of the key ecological corridors in Jiaxing remained zero, meaning that there were only isolated sources and/or thegeneral ecological corridors in this area (Zhou, Lin, Ma, Qi, & Yan, 2020). Hangzhou had still the longest key corridors among the five cities and had the best network connectivity. However, the length of the key ecological corridors in Hangzhou had decreased from 1995 to 2015, which was mainly due to urban expansion and the increase of ecological sources area.

3.2 Ecological nodes assessment

3.2.1 Degree and degree distribution

The number of ecological nodes in the network increased from 85 nodes in 1995 to 96 nodes in 2015, which means that the number of habitats for migrating species in the UAHB had increased. The average degree of the network increased from 3.14 in 1995 to 3.36 in 2015, indicates that the network connectivity had slightly improved in the past 20 years.
Figure 4 shows the degree distributions of the UAHB’s ESN in 1995 and 2015. In 1995, the maximum value of degree was 8, the minimum value of degree was 0, and the number of nodes with degrees 8, 7, and 6 only was accounted for 1%. In 2015, the maximum degree was 9, the minimum degree was 0, and there was one node with degree 9, accounting for 1%, and there was no node with degrees 8 and 7.
The proportion of the nodes with a degree greater than and equal to 4 was 29.4% in 1995 and 36.0% in 2015, indicating that the number of the nodes with a large degree increased significantly. However, the number of nodes with degree 3 accounted for the largest proportion in both years (48.2% in 1995 and 47.0% in 2015). The number of nodes with lower degrees (0, 1, and 2 degrees) decreased from 22.4% in 1995 to 17.0% in 2015. Overall, the connection between ecological nodes in 2015 was better than that in 1995.
Table 2 shows that the average degree (k’ ) and the maximum degree (kmax ) of the nodes in Ningbo, Shaoxing, and Hangzhou increased from 1995 to 2015, indicating that the network structure connectivity of these three cities increased. But, in Huzhou, all three indicators (k’ , kmax , andkmin ) decreased from 1995 to 2015, signifying decreasing network connectivity in this city in the past 20 years. Because Jiaxing had only isolated ecological sources in the network, the average degree, maximum degree, and minimum degree of the nodes were 0 in both 1995 and 2015.

3.2.2 Clustering coefficient

Table 3 presents the changes in clustering coefficients in 1995 and 2015. There were 30 and 31 nodes (accounting for 37.97% and 34.07% of the total number of nodes) with a clustering coefficient of 0 (no clustering) in 1995 and 2015, respectively. There were two nodes with a clustering coefficient of 1 (obvious clustering) in both 1995 and 2015. The nodes with a clustering coefficient greater than and equal to 0.4 accounted for 6.33% in 1995 and 11.00% in 2015, while the nodes with a clustering coefficient less than 0.4 accounted for 93.67% in 1995 and 89.01% in 2015, indicating that the cluster characteristics for most nodes were not obvious. The results show that the ESN in 2015 had obvious non-uniformity and the network structure was unstable, but compared with 1995, the structural stability of the ESN was slightly improved.
The average clustering coefficient of the ESN was 0.20 in 1995 and 0.22 in 2015. Although the average value in 2015 was slightly higher, the small world characteristic of the network was not obvious.

3.2.3 The core of node

Figure 5 shows that the maximum number of the core of nodes in the UAHB’s ESN was 3 in 1995 and 2015, and the nodes with a core of 2 accounted for the highest proportion (80-86%). The percentage of the nodes with a core of 1 did not change from 1995 to 2015. The percentage of the nodes with a core of 2 increased from 80% to 86%, whereas the percentage of the nodes with a core of 3 decreased from 16% to 10%.
In terms of spatial pattern, Figure 5 shows that the nodes with a core of 3 were mainly concentrated in the southwest of the UAHB, while the nodes with a core of 2 were mainly distributed in the middle and southeast of the study area. The number of nodes with a core of 2 increased significantly from 1995 to 2015. The nodes with the decreasing core (from 3 to 2) were mainly distributed at the edge of the network, so we should pay more attention to protecting the nodes at the edges of the network in the future. Overall, the network connectivity had improved from 1995 to 2015.

3.2.4 Node comprehensive importance assessment

As shown in Figure 6, the node comprehensive importance was classified into five levels (higher, high, general, low, and lower). There were 10 nodes with high and higher importance levels in both 1995 and 2015. In 1995, more than half of the nodes were at the low level of importance, and 20 nodes were at the general level of importance. In 2015, the number of nodes with the general importance level increased significantly, while the number of nodes with the low importance level decreased evidently. However, the number of nodes with a lower importance level increased slightly.
Figure 6 also shows that the nodes with the high and higher importance levels were mainly distributed in the southwest of the study area, and a small number of nodes were in the south of the study area. Many corridors were passing through these nodes, and the land use in the areas was mainly forest, and the terrain was mainly mountainous and hilly. Therefore, these nodes played an important ecological function in the UAHB’s ESN, they were the key to maintaining the stability of the ESN, and the protection of these nodes should be the focus of the construction of regional ecological security. Moreover, the nodes with the general importance level were mainly distributed in Hangzhou and Shaoxing, while the nodes with the lower importance were mostly dispersed in Ningbo and Huzhou. The protection of the ecological nodes in these cities should be strengthened.

3.3 Network connectivity assessment

Table 4 shows that the index α of the UAHB’s ESN was higher in 2015 than in 1995, indicating that the material circulation in the ESN was smoother in 2015 than in 1995. The two indexes β and γwere also higher in 2015 than in 1995, which indicates that the network connection was complex. Although the network connectivity of the ESN had improved, there were still isolated nodes in the network.
Although the Cost Ratio declined from 0.94 in 1995 to 0.93 in 2015, the network cost was still relatively high, mainly because the ecological corridors (edges) constructed on the large scale of urban agglomeration stretches across more counties and cities, leading to higher network connection complexity. The results show that it is especially important for the regional ESN to give priority to the construction of ecological corridors with higher importance levels.

3.4 Network robustness assessment

As discussed above, the connectivity robustness (R 1) and the vulnerability robustness (R 2) represent the connectivity and the operation efficiency of the ESN, respectively. As shown in Figure 7(a,b,c,d), the initial values of R 1 of the ESN in 1995 and 2015 were both 1.0 and those of R2 in 1995 and 2015 were 0.23. This indicates that the ESN in 1995 and 2015 had the same network connectivity and operation efficiency.
By and large, with more nodes being deleted from the network,R 1 and R 2 decreased under the two disturbance scenarios (NHD and HD). The number of deleted nodes and connected components appeared to be the inverted V-shaped relationship in 1995 and 2015 (Figure 7(e,f)).