PS 33-52
Dispersal network structure predicts metacommunity dynamics
Dispersal among communities has important dynamical consequences, particularly in regards to species persistence and community stability. Changes in behavior at the metacommunity scale are understood to depend on not only the rate of dispersal but also the structure of dispersal connections between communities. The objective of this work is to capture general, predictive relationships between dynamics and structural properties in such systems. To this end, we examined the dynamics of a two species, spatially-explicit Rosenzweig-Macarthur model with a tendency toward limit cycles simulated over randomly generated dispersal networks. Dispersal networks were both connected (all patches could be reached by any other patch) and regular (all patches had an equal number of connections) and initialized with randomly distributed starting densities of varying deviation from equilibrium. Dynamical behavior was described by the length of transient dynamics, variability during transience, and degree of synchrony at equilibrium. The structure of the dispersal network was described using a range of graph theoretic indices, but analyses highlighted the importance of the transitivity (ratio of closed triplets to connected triples), path length (# intermediary connections needed to link patches), and symmetry (# of self isomorphisms).
Results/Conclusions
When starting densities are close to equilibrium, the majority of metacommunities tend quickly toward total synchrony, cancelling the effects of dispersal and causing each patch to behave as it would in isolation. However, a minority exhibit either significantly longer transient dynamics or short transients with asynchronous equilibria; the occurrence of these states increases with the variability of starting densities. Metacommunities prone to these dynamics are distinct structurally, exhibiting higher transitivity and mean path length. Notably, varying the position of starting densities within the metacommunity without changing the values could produce distinctly different dynamics from the same spatial network. These results suggest network behavior can be predicted partially from simple structural properties, though dependent on the initial conditions of the system. Increased transient length and asynchrony among patches have both been highlighted as important mechanisms supporting coexistence and diversity in metacommunities, and as such it is worth investigating the properties which promote and predict them further.