Single-node vs. multi-node centrality in landscape graph analysis: key habitat patches and their protection for 20 bird species in NE Spain

Habitat connectivity is a major concern in biodiversity conservation. Network analysis provides efficient tools for assessing landscape connectivity and identifying priority areas for protection. Widespread approaches consist of ranking individual habitat patches by their importance for connectivity. However, depending on the spatial arrangement of patches, and on the complementarity or redundancies between different patches in the network, the group of patches that together best contribute to connectivity may differ from the top individual patches. Here we apply individual ranking (single-node) and group prioritization (multi-node) methods to identify key patches in the habitat network of 20 bird species in Catalonia, Spain. We compare single-node evaluation using the landscape index probability of connectivity (PC) and two multi-node evaluations that focus on two different aspects of connectivity: reachability and fragmentation. We quantify how well the species' habitats and key patches are covered by the Natura 2000 network of protected areas, as given by all three methods. We discuss some species-specific differences between results, and general, multi-species solutions. The key patches for reachability were widely scattered, while patches that best prevented fragmentation were concentrated in core areas. Key patches according to single-node PC were in intermediate position, although more similar to the fragmentation patches. The patches that minimized fragmentation were highlighted as more crucial for low-mobility species, while reachability patches scored higher for long-distance dispersers. For most species, the key patches were not better protected than total habitat. We identify two main areas that concentrate priority patches for all the studied species (including the endangered Neophron percnopterus) and offer suggestions for the potential improvement of the Natura 2000 network. Our work provides a significant step towards the inclusion of multi-node prioritization in landscape network analysis. The concept of multi-node reachability, especially, provides an additional viewpoint to the assessment of connectivity. The multi-patch algorithm we applied optimizes efficiency, overcoming computational limits associated with the high number of combinations that potentially arise in multi-node analysis. We believe that a combination of multi-node evaluations and PC has the potential to increase the realism and applicability of landscape network analysis for biodiversity conservation.