This guest post is by Sandeep Venkataram and Dmitri A Petrov on their paper (with Diamantis Sellis) Venkataram et al. Ploidy and the Predictability of Evolution in Fisher’s Geometric Model
Since Gould’s famous thought-experiment (Gould, 1990) on “replaying the tape of life”, scientists have been interested in the predictability of evolution. Gould wondered whether it is possible to forecast evolution, and determine the path or the final destination of the evolutionary process from a given starting population. It is also possible, however, to ask whether we can retrocast evolution, and reconstruct the true evolutionary trajectory given the final state and possibly the ancestral state. Forward predictability analysis tries to predict the future evolutionary trajectory or future adapted state of an evolving population, while backwards predictability analysis tries to determine the likelihood of the possible alternative adaptive trajectories that lead to the observed adapted state.
Predictability has been empirically studied to a limited extent due to the laborious nature of such studies (e.g. Ferea et al 1999, Weinreich et al 2005 and Tenaillon et al 2012). We overcome these limitations by analyzing simulated adaptive walks under Fisher’s geometric model. To our knowledge, we are the first to study both of these types of predictability in a single system. We compare the predictabilities of haploid and diploid simulations, and find that forward and backward predictability are inversely correlated in this model. We attribute this inverse correlation to the presence of overdominant mutations and balanced polymorphisms in our diploid simulations and the lack of such mutations in the haploids (Sellis et al 2011).
We observe that the presence of balanced polymorphisms in diploids leads to a number of novel dynamics when studying predictability. It greatly increases the phenotypic diversity in diploid adaptive walks, leading to low forward predictability relative to haploids. We also detect mutations which are stably maintained but subsequently lost in diploid adaptive walks, and are thus hidden from sampling at the end of the simulation. We show that these hidden mutations, which also go unobserved in almost all empirical studies, strongly limit the inferences that can be made when analyzing backward predictability. Finally, we observe that when the same set of mutations is introduced into a diploid population in different orders, the final adapted allele is often balanced against different intermediate alleles, resulting in different adapted population states.
Our results show the importance of considering stable polymorphisms when analyzing adaptive trajectories, and detail, for the first time, some of the limitations in conducting such analysis using empirical data. In natural population, stable polymorphisms can be generated in both haploids and diploids by a wide range of mechanisms, including niche construction, frequency dependent selection, balancing selection and spatially and temporally fluctuating selection pressures. Therefore, our results should be relevant for all natural populations, regardless of ploidy.
Ferea, T., Botstein, D., Brown, P. O., & Rosenzweig, R. F. (1999). Systematic changes in gene expression patterns following. Proceedings of the National Academy of Sciences of the United States of America, 96(August), 9721–9726.
Gould, S. J. (1990). Wonderful Life: The Burgess Shale and the Nature of History (p. 352). W. W. Norton & Company.
Sellis, D., Callahan, B., Petrov, D. A., & Messer, P. W. (2011). Heterozygote advantage as a natural consequence of adaptation in diploids. Proceedings of the National Academy of Sciences of the United States of America, 2011, 1–6. doi:10.1073/pnas.1114573108
Tenaillon, O., Rodriguez-Verdugo, a., Gaut, R. L., McDonald, P., Bennett, a. F., Long, a. D., & Gaut, B. S. (2012). The Molecular Diversity of Adaptive Convergence. Science, 335(6067), 457–461. doi:10.1126/science.1212986
Weinreich, D. M., Delaney, N. F., Depristo, M. a, & Hartl, D. L. (2006). Darwinian evolution can follow only very few mutational paths to fitter proteins. Science, 312(5770), 111–4. doi:10.1126/science.1123539