Environmental variability is often expressed by probability density functions of distinct variants, hereafter called states, of the environment (e.g., open vs closed canopy in forests) or distributions of quantitative climatic descriptors (e.g., the North Atlantic Oscillation Index). The extent to which environmental states/values exhibit Generalized Extreme Value (GEV) distributions rather than normal or poisson distributions is under explored in ecology. Sequence, often neglected, can impact population ecology as much as frequency. We model temporal variation by drawing environments at random at each time step from the appropriate distribution, with sampling protocol dependent upon whether current environment does or does not influence the probability of the environments at the next time step. For three examples, we focus first on determining whether the historical record of environmental variation is well described by GEV distributions and second on whether the parameters of the distribution depend on the current state of the environment. We then compare stochastic population growth rate and environment-specific elasticity of it under different four scenarios (2 × 2): using the same distribution parameters at each time step vs using different distribution parameters depending upon the current environmental state crossed with using two different types of distributions: e.g., GEV vs Normal.
Results/Conclusions
We find that GEV distributions are relevant to understanding population dynamics of several species and that sequence is as important as frequency in elucidating population dynamics and its sensitivity to extreme events.