EVOLUTIONARY PREDICTIONS FROM INVARIANT PHYSICAL MEASURES OF DYNAMIC PROCESSES

The invariant physical measure of a dynamic: process comprises the information necessary to compute statistical quantities of the system. It allows us to replace time averages over the system's trajectory by integrals over state-space. In particular, it allows us to compute the long-term growth rate of a fluctuating population and hence to make evolutionary predictions. This is explained here using a one-dimensional difference equation. The invariant physical measure reflects three types of selection in this model: K-selection when the resident has a stable equilibrium; r-selection when the resident exhibits complex dynamics; and c-selection, i.e. selection for lower complexity when the resident undergoes large fluctuations. For all three evolutionary scenarios the interaction of the physical measure with the higher moments of the distribution of offspring numbers is crucial. It is also shown how stochastic noise can affect the invariant physical measure and the evolutionary predictions made from it.