Robustness of dispersal network structure to patch loss

Network theories have been largely applied to the investigation of spatial ecology, and there is a new trend to use them to explore the responses of complex dispersal networks to patch loss. Combining both network and metapopulation approaches, we used a spatially explicit patch model to compare the robustness of networks with different heterogeneities to both random and selective patch removal. We found that species in more heterogeneous networks can persist at higher extinction-to-colonization ratios. In addition, dispersal networks with higher heterogeneity display stronger robustness to random patch loss, suggesting that previous models based on lattice- or randomly-structured networks might underestimate species extinction thresholds. However, such strong tolerance to random patch removal comes at a high cost in that these networks are extremely vulnerable to selective removal of the most connected patches (so-called keystone patches), as this removal mode directly leads to a rapid decline in the total number of links among patches. We further explored the mechanism underlying these outcomes via network analysis, and found a strong positive correlation between overall metapopulation size and the largest cluster size. Concerning ecological conservation and management, our findings suggest that future efforts should focus on considering species dispersal networks by identifying and preserving the keystone patches, and as such, optimizing the connectivity between existing habitat patches should be an effective strategy to rescue the endangered species.