Title: Extinctions of coupled populations and rare-event dynamics under non-gaussian noise
 
Abstract: 
Population dynamics is a cornerstone of biomathematics with applications to genetics, epidemic spreading, and ecology.

In many cases, the population can reach absorbing states such as genetic fixation or population extinction, which are inevitable in the presence of fluctuations.
Estimating the mean time to reach these states is of great importance in evolution and ecology.
Here we look at a canonical setting, of a natural population residing in two locations (patches) coupled by migration, and where the environmental conditions fluctuate. Chances of extinction in the simpler problem of a single patch have been derived long ago and are a fundamental result in the field. For multiple coupled patches, the problem has thus far remained open.
We find that, unlike rare events in many other systems, here the histories leading to a rare extinction event are not dominated by a single path (the instanton). We develop the appropriate framework, which turns out to be a hybrid of the standard saddle-point method, and the Donsker-Varadhan formalism which treats rare events of atypical averages over a long time.

An unexpected prediction of this framework is that a sink patch (where individuals die more than are born) can nonetheless reduce the probability of extinction, even if it normally lowers the population’s size and growth rate.

Getting here