Empirical tests of metacommunity theory using an isolation gradient

All metacommunity theories incorporate dispersal as a key process influencing community composition. If communities are examined across a habitat-patch isolation gradient, different metacommunity theories predict contrasting patterns of community divergence and responses of rare species. We used bird and reptile data collected in three years from 168 habitat patches in a fragmented agricultural landscape, and bird data gathered from 63 remnants embedded in a pine plantation over two years to examine predictions arising from six metacommunity theories: A, neutral; B, species sorting; C, species sorting at low isolation but neutral at high isolation; D, neutral at low isolation but species sorting at high isolation; E, mass effects; and F, patch dynamics. We identified three classes of predictions arising from these theories that we could test using community survey data: (1) trajectories of rare species across an isolation gradient; (2) the influence of geographic distance, environmental parameters, and patch isolation on pair-wise comparisons of community divergence; and (3) the influence of isolation and environmental parameters on the divergence of communities from a regional species pool. Some analyses indicated moderately common support for neutral and species-sorting concepts, often acting simultaneously. Opposite responses by different groups of rare species to the isolation gradient showed that neutral and patch-dynamic processes may act on different components of the same community. Different analyses provided support for different theories, and so helped us to avoid being misled by a single analysis. However, generally we found very little consistent support for any of the metacommunity theories. Our analyses suggested that different metacommunity mechanisms act ephemerally, sometimes simultaneously and on different subsets of the fauna in different regions. The complexity of responses means that metacommunity ideas cannot yet be used predictively in a management context. We encourage further testing of the multiple predictions that arise from current metacommunity concepts, using additional data sets. However, if metacommunity processes act simultaneously and ephemerally, the predictions of any one metacommunity theory may be hard to detect at the community level. In that case, examining the ecology of multiple species in the same landscape may be needed to characterize metacommunity processes in nature.