A note on population size inspired by the extinction of mammoths

Abstract

This study performs simulations inspired by the reported genome meltdown of a small population of woolly mammoths prior to their extinction. These simulations test the interaction of population size, mutational diameter, and fitness change on two types of fitness landscapes. The first landscape studies a population initialized at a global optimum to assess fitness loss, while the second uses an open-ended function with no global optimum to assess the degree of adaptive radiation possible with different population sizes. Both an age structured non-elitist evolutionary algorithm and a evolution-strategy like biased random walk are used. The simulations demonstrate that small populations are substantially worse at retaining fitness when initialized in a global optimum but also have a substantially greater potential for adaptive radiation and discovery of new niches.