Natural evolving populations experience constantly fluctuating selection strength, which also creates a fluctuating tradeoff between exploration and exploitation. Range expansion, for example, creates semi-persistent spatially-distributed differences in selection strength, particularly among the pioneering agents along the leading edge of each range expansion. The pioneers experience reduced selection strength and in turn experience greater potential for exploration, while selection on the remainder of the population ensures that prior discoveries are not lost.
Here we describe a method to augment pre-existing selection algorithms inspired by the exploration-boosting properties of range expansion events. The key insight is that for productive exploration on deceptive landscapes, mutations must be able to accumulate and persist in some, but not all, lineages. We create artificially drifting lineages of “super explorers” and show that they can be used to improve the performance of another selection algorithm.