Are local processes really deterministic?

Background/Question/Methods

Local processes (e.g., biotic and biotic-environment interactions) are usually assumed to be deterministic. If so, communities invaded by the same set of species should converge to a single community state. However, this is rarely the case. One explanation is that the order of species colonization (history) matters. To test whether variation in history leads to alternative community states (and, by implication, whether local processes are deterministic if history does not matter), we conducted two experiments. To remove the influence of history, we constructed a regional species pool (RSP) by mixing the contents of 17 natural rock pools. Using the mixture, we created 40 replicated beaker communities, each with the same species structure and placed them outdoors in a protected area. To test for environmental determinism, we redistributed the RSP into the previously emptied 17 pools and added two plastic beakers containing RSP mix. One beaker allowed water exchange via a fine net (5 micron mesh) while the other was isolated (exposed to pool temperature and light only).  No dispersal was allowed into any of the beakers. Experiments ran for 4 months.

Results/Conclusions

In the first treatment alternative states continued to form over time indicating that biological interactions were not deterministic and that stochasticity played an important role in shaping communities under biological control only.  Completely isolated beakers placed in natural pools produced a single state, while beakers that allowed water interchange and the pools themselves showed several states, whose number increased over time. Desiccation and daily fluctuation of environmental parameters is characteristic of this rock pools. This fact, coupled with convergence seen in the completely isolated beakers, suggest that environmental forcing determined the outcome (community structure). Beakers open to water exchange and the natural pools diverged. The overall picture that emerged from the treatments involving natural setting supported some of the expectations but not others, which prevented us from drawing unambiguous conclusions.