Print PagePrediction of spatial spread must take into account considerable uncertainties that are generated by stochastic and nonlinear processes. We derive mechanistic stochastic models at the population and landscape levels by scaling up from stochastic processes at the level of individuals. Randomness of invasive spread is difficult to study in the field since any invasion in nature is just one realization of the stochastic process. Instead, we have been studying variance in spatial spread in highly replicated experimental microcosms using the red flour beetle, Tribolium castaneum.
Results/Conclusions
By fitting the stochastic models to data from this model system we show that the accuracy and uncertainty of predictions for invasive spread depend critically on factors contributing to stochasticity. For example, even a carefully parameterized demographic model with comprehensive sources of stochasticity was unable to explain the observed variance between replicated microcosms until stochastic founder effects were also included. Similarly, microcosms with spatially heterogeneous environments paradoxically had lower variance in spread rates than homogeneous environments. These results show that stochastic biological processes can influence uncertainty in surprising ways.
