关键词: 卷羽鹈鹕, 迁徙连通性, 节点移除, 关键停歇地, 水体比例

Abstract

Aims: Long-distance migration by waterbirds depends on a network of wetlands providing successive stopover sites. A site's importance reflects not only its use by birds, but also its position within migration routes and local water conditions during stopover. However, the network consequences of node loss and concurrent surface water conditions are rarely evaluated together. Using multi-year satellite-tracking data from four Dalmatian Pelicans (Pelecanus crispus), we compared the composition and connections of en-route node networks between spring and autumn migration, quantified the effects of individual node loss on network connectivity, and examined whether sites with large node-removal effects also had high surface water proportions during stopover. 

Methods: Separate en-route node networks were constructed for spring and autumn migration. Node strength and directed betweenness were calculated, and articulation points were identified to characterize the direct connections and structural roles of individual nodes. Each node was removed individually from the original network, and the resulting relative loss of pairwise reachability was used to quantify its node-removal effect. Nodes were then removed sequentially in descending order of their single-node removal effects, and the resulting connectivity loss was compared with that under random removal. Sentinel-2 imagery was used to estimate the surface water proportion within a 10-km buffer during the 61-day period centered on the date when an individual first arrived at a node. For each season, the ten nodes with the largest removal effects were selected to examine whether high network impact corresponded to high surface water availability during stopover. A value of 5% was used as an operational threshold for summarizing surface water proportions. 

Results: The median distance between consecutive nodes was 457.9 km in spring, approximately 1.98 times the autumn median of 231.6 km. Among the 11 sites used in both seasons, the spring subnetwork contained only six connections and comprised six connected components, whereas the autumn subnetwork contained 11 connections and comprised three connected components. Thus, the same sites were connected differently between seasons, indicating that spring and autumn migrations did not simply follow the same route in opposite directions. Using the original network as the baseline, removing each en-route node separately caused different losses of pairwise reachability. N038 had the largest single-node removal effect in both seasons, and its removal reduced pairwise reachability by 56.2% in spring and 48.1% in autumn. When approximately 15% of the nodes were removed sequentially in descending order of single-node removal effect, only 12.7% and 19.6% of the original pairwise reachability remained in the spring and autumn networks, respectively; both values were lower than those obtained under random removal. A large node-removal effect did not necessarily coincide with a high surface water proportion during stopover. Of the 20 seasonal nodes ranked among the ten highest-impact nodes in each season, ten had median surface water proportions of at least 5%, whereas the other ten had values below 5%; at several nodes, the proportion was below 1%. 

Conclusion: Node-removal effects and surface water proportions during stopover represent distinct and non-substitutable dimensions of migration-wetland importance: the former reflects a site's role in maintaining connections within the observed network, whereas the latter describes concurrent surface water coverage. When setting conservation priorities, sites that lack recorded alternative connections and have large removal effects should be treated as candidates for priority investigation, with their network position evaluated together with water conditions, water quality, food resources, and human disturbance. For infrequently used articulation points that maintain connections between migration branches, surveys should identify potential alternative wetlands and assess detour risks. For high-impact sites with low surface water proportions, high-resolution imagery and field surveys should verify small water bodies and their actual habitat quality. Such integrated assessments can support migration-route continuity and overall flyway function.

Key words: Pelecanus crispus, migration connectivity, node removal, key stopover sites, surface water proportion