Spatial network structure and metapopulation persistence

Background/Question/Methods In the last few years there has been an intense research exploring the spatial dimension of population and community persistence. In metapopulation theory, space is introduced as discrete patches of suitable habitat connected by dispersal. Here we use an extension of Levins’ metapopulation model with a contrasting spatial network structure. Specifically, we consider four types of networks on the basis of their degree distribution: regular, random, exponential, and scale-free. While the former represents the most homogeneous network type as illustrated by cellular automata, the latter represents the most heterogeneous network type. With this range of networks we analyze (i) how regional abundance, and (ii) its rate of decline with habitat loss, depend on the structure of the spatial network.

Results/Conclusions For low extinction-to-colonization ratios, regular networks exhibit the highest relative abundances, while scale-free networks exhibit the lowest abundances. Surprisingly, the pattern is the opposite for high extinction-to-colonization ratios. This means that species with a high dispersal ability are more abundant in continuous habitats, whereas species with limited colonization or high local extinction rates have advantage in patchy, heterogeneous habitats. Therefore, the optimum design for a network of protected areas critically depends on the species’ life history. Next, we carry out random node deletion simulations for each kind of spatial network and extinction-to-colonization ratios. Our results show that the effect of habitat destruction differs among network topologies. Regular networks are the most fragile, while random networks are the most robust in the face of habitat loss.