Print PageMaintaining connectivity among habitat patches is critical for biodiversity in fragmented landscapes. To guide the management and design of fragmented landscapes, ecological research must resolve the question of how altering habitat connectivity affects the dynamics and synchrony of populations. We examined the role of network structure on experimental metapopulations of springtails (Folsomia candida). Three types of ten-patch habitat networks were constructed as rings with varying degrees of structural randomness: 1) regular lattice networks of degree four (each patch connected to four nearest neighbours); 2) networks with three randomized connections (removing connections between three pairs of patches in the lattice network and rewiring them to connect different pairs); 3) completely randomized networks in which all connections have been assigned at random. Nodes in the networks were plastic vials filled with plaster of Paris and activated charcoal, and connections among nodes allowed for movement through plastic tubing of varying lengths. Replicate networks were run for 180 days with each node receiving 90 feeding events (1 mg of granulated dry yeast) subject to an autocorrelated first-order autoregressive feeding schedule. Networks were initially populated with ten adult individuals in a single node. Population density in each node was digitally recorded every 48 hours producing a large spatio-temporal data set of densities in each network.
Results/Conclusions
Network structure constrained the speed and pattern of spatial spreading through the network. Randomizing network structure delayed the rate of colonization because random connections were longer than nearest-neighbor connections and corresponded to exponentially lower dispersal probabilities. Randomizing network structure also changed the pattern of colonizations. Regular lattice networks were colonized in an even dispersal outwards from the source patch around the ring while randomized networks were characterized by jumps across the ring in a less synchronized fashion. These differences in colonization dynamics among network structures resulted in differences in metapopulation growth. Completely random networks reached carrying capacity twice as slowly as regular lattice networks. Preliminary results indicate that heterogeneity in node centrality (measured as the geometric distance from that node to all other nodes in the network) present only in the randomized networks desynchronized arrival times and initial population growth in nodes. We provide empirical evidence that heterogeneity and randomness in connections among different patches in ecological networks can have large effects on population dynamics. This network approach provides a foundation for longer term goals of creating ecosystem networks that restore functional connectivity for biodiversity in fragmented landscapes.
