Correlated evolution in experimental populations of yeast

Abstract

Replicate haploid and diploid lineages of the yeast <italic>Saccharomyces cerevisiae</italic>, selected for &sim;5000 generations in a glucose-limited environment, were used to study correlations among life-history traits and fundamental questions about phenotypic evolution. As expected, growth rate in the selective environment rapidly increased for all lineages for the first &sim;2000 generations and more slowly afterwards. All diploid lineages became haploid before &sim;3000 generations of selection. Yield in the selective environment declined linearly during selection of the haploid lineages but not in the diploid lineages, in which only haploidization lowered yield. The trade-off between rate and yield was predicted by a classical model of biochemical pathways. A second indirect effect of selection on glucose was that growth rate increased and yield decreased in a galactose-limited environment, in both haploid and diploid lines. I demonstrate that this trade-off between rate and yield in the galactose environment results from an ecological feedback between these life-history traits in batch cultures rather than a genetic correlation. Next, I isolated 160 clones from the 5000-generation history of lineages in glucose, and selected them on galactose for &sim;100 generations. As a result, growth rate increased and yield declined in galactose for the clones that grew slowly on galactose, while clones that already grew fast on galactose tended not to respond to selection. This suggests that evolution tends to equalize the fitness of lineages. Finally, I compared the direct and correlated responses to selection in both glucose and galactose in a nested reciprocal selection experimental design. This analysis revealed that (1) better adapted clones paid a larger cost of adaptation than poorly adapted ones; (2) direct and correlated effects were not correlated, such that adaptive changes of large effect tended to have disproportionately small correlated effects and (3) reciprocal correlated responses to selection did not evolve to symmetry. The view of adaptation emerging from this work is that correlated effects show predictable pattern within environments but are unpredictable even between the simplest environments after long-term selection.