Comparing evolvability and pleiotropy across environments

Abstract The environment can play an important role in determining evolutionary outcomes (Reboud and Bell 1997; Stanton et al. 2000; Gresham et al. 2008; Cooper and Lenski 2010; Becks and Agrawal 2012; Bailey et al. 2015). Populations may increase in fitness more after evolution in one environment than in another (Hegreness et al. 2008). This may be because the distance between the ancestral genotype and the top of the locally accessible fitness peak may be greater in one environment or the other. Additionally, the rate at which mutations occur and populations move up the peaks could differ. For this reason, comparing evolvability across environments presents an interesting problem. Here, we show an environment impacts evolvability in two ways, directly by impacting the rate of adaptative evolution, and indirectly by limiting the range of fitness outcomes that are possible. We show that correlated responses are often highly idiosyncratic, due to variation in the range of possible fitness outcomes in the environment and differences in pleiotropic effects across environments but can also be predicted from the ancestral growth rate regardless of the environment in which populations evolve. Interestingly, we also show a negative correlation between increase in an environment X following evolution in environment Y and the increase in environment Y following evolution in environment X. These results highlight the necessity to measure fitness in both environments when comparing the evolvability or repeatability of evolution across environments.