Spatial Fleming-Viot models with selection and mutation (Q371428)

A system of interacting (spatially structured) Fleming-Viot diffusions is considered which describes the stochastic dynamics of a multi-genotype population. The geographic space is modelled by a set of colonies (dems) labelled with the countable hierarchical group \(\Omega_N\) (the direct sum of a countable number of copies of the cyclic group of order \(N\)). \(\Omega_N\) mimics a two-dimensional spatial greed as \(N\to\infty\). The population model takes into account selection, mutations and migration effects. The authors focus mainly on the asymptotic (as \(N\to\infty\) and time \(t\to\infty\)) description of different phases of punctuated equilibrium population dynamics in which the transition to a new quasi-equilibrium on a higher level of fitness proceeds first by the occurrence of rare mutants in a few colonies which form a growing droplet of fitter colonies (droplet formation phase). This droplet has negligible intensity for a long time. Then the superior types invade the remaining colonies (emergence phase) and dominate all the sites (fixation phase). A multiple time-space scale analysis is developed to make precise the notion of quasi-equilibrium as equilibrium of a limiting non-linear McKean-Vlasov dynamic at different phases. Separated time scales in different phases are connected by random entrance laws from time \(-\infty\). The main tools to carry out the analysis are various Markov processes dualities. New forms of duality are introduced which allow to calculate the probability that a finite sample of the population has specific genealogical distances and types. In an enriched version of the model, historical processes are considered which record the family relations between individuals and the path of descent through space.