Repeatability and Predictability in Experimental Evolution

Independent populations often use the same phenotypic and genetic solutions to adapt to a selective challenge, suggesting that evolution is surprisingly repeatable. This observation has inspired a shift in focus for evolutionary biology towards predictive studies, but progress is impeded by a lack of insight into the causes for repeatability, which prevents tests of forecasting models outside the original biological systems. Experimental evolution with microbes could provide a way to identify the causes of repeated evolution, directly test forecasting ability and develop methodology, but a range of difficulties limits successful prediction. This chapter discusses the limitations on forecasting of experimental evolution, what can and cannot be predicted on different biological levels and why predictions will often fail. Focusing on experimental populations of bacteria, the importance of selection, mutational biases and genotype-to-phenotype maps in determining evolutionary outcomes is discussed, as well as the potential for including these factors in forecasting models. The chapter concludes with a discussion on the desired properties of experimental evolution models suitable for testing forecasting models.