Effects of community and ecosystem properties on an alternative state: A simulation model

Background/Question/Methods

It should no longer be a surprise when communities and ecosystems suddenly change in dramatic ways.  Coral reefs, temperate forests, grasslands, the open ocean, and many other habitats experience large shifts between discrete states.  In the northeast United States, many freshwater lakes and ponds are found either in a state dominated by submerged aquatic vegetation (SAV) or floating plants (FP), suggesting the existence of alternative states.  Field surveys show that the FP state typically only exists in water bodies less than 5 ha and can be composed of a variety of species. My goal was to develop a spatially- and temporally-explicit simulation model to determine the effects of water body size and shape and species composition on the threshold between states.  Existing models of this system are typically not spatially-explicit (i.e., cannot consider most water body properties), combine species into functional groups, or are extremely complex, requiring numerous parameter estimates. I developed a simple, yet biologically realistic, model of floating plant dynamics on a two-dimensional grid, including processes such as density-dependent growth, senescence, overwintering, and wind disturbance. Parameters were estimated from lab experiments and published literature and stochastic simulations were performed in R. 

Results/Conclusions

Over a range of nutrient levels, simulations typically ended in either a FP- or a SAV-dominated state, depending solely on the initial conditions. This result suggests the existence of alternative states. In general, ecosystem properties (e.g., water body size and wind disturbance) had a greater effect than community composition on the tipping point between states. Larger water bodies were less susceptible to a FP state, but only if wind had a prevailing direction. Water body shape did not have a large effect on the results. These results are generally supported by data from field surveys in the northeast US. Further work is necessary on the process of wind disturbance because simulations were highly sensitive to the manner in which this process was modelled. This model demonstrates how a simple, yet realistic, and spatially-explicit model can uncover the mechanisms that determine the threshold between alternative states.  This model is useful because experimental manipulations of whole communities and ecosystems are typically implausible.  It can also be used to develop management strategies (e.g., harvest or pesticide application) to reverse an undesirable alternative state.