Positive Feedbacks, Bistability and Nonlinearities Following Deforestation

Forest vegetation can interact with its surrounding environment in ways that enhance conditions favorable for its own existence. Removal of forest vegetation has been shown to alter these conditions in a number of ways, thereby inhibiting the re-establishment of the same community of woody plants. The effect of vegetation on an environmental variable along with vegetation susceptibility to the associated environmental conditions implies a positive feedback: changes in the internal conditions controlling this variable such as deforestation could inhibit the re-establishment of woody vegetation cover that in turn would act to further degrade the conditions necessary for forest regeneration. Sudden changes in the structure and functioning of ecosystems can be attributed to the existence of multiple stable ecosystem states, which are commonly associated with positive feedbacks. Forest ecosystems exhibiting multiple stable states are characterized by one stable state with low vegetation and another stable state with a full vegetation cover. Once a state shift occurs, the system remains locked in a state of low vegetation until there is a change in the environmental conditions controlling the vegetation dynamics. In this dissertation, I use a combination of experimental and modeling techniques to examine a number of feedbacks in which deforestation can lead to a loss of conditions necessary to sustain forest vegetation. These feedbacks were either previously undiscovered or have been poorly studied. They involve processes including: landsliding, permafrost melting, canopy trapping of nutrients, groundwater-salinity dynamics and microbial-phosphorus interactions. For each of these feedbacks, bistability was shown to result following deforestation. Several of the feedbacks also exhibited a noise-induced state where the dynamics never reached either of the two deterministic states but were instead stabilized in an intermediate state in the presence of noise. Management practices were shown to influence the convergence rate of the system to its attractors. Importantly, findings from this dissertation show that deforestation in areas affected by these feedbacks can lead to irreversible state shifts where the forest vegetation cannot recover following forest removal.