A gzipped archive of some data table associated with this project.

We evolved an outbred diploid sexual population of budding yeast with weekly episodes of recombination in the presence of five different stressors along with YPD for 21 weeks. We culled populations in which asexual cheaters evolved, and pool-sequenced two replicate populations per drug at six timepoints (weeks 1, 5, 9, 13, 16 and 21). Given that the base population used to begin the experiment was derived from a collection of 18 known founder strains, we were able to estimate founder haplotype frequencies throughout the genome from the poolseq data and examine haplotype change over time. For the first 16 weeks of the experiment, evolution is extremely repeatable between replicates within a drug, but different between drugs. For all populations the entire genome shows haplotype frequency change, either due to direct or indirect selection acting on variation in the base population, despite high rates of recombination and un-recombined haplotype block sizes of less than 100kb. Haplotype trajectories are dominated by an early purging of the dominant local haplotype, followed by a single replacement haplotype quickly increasing in frequency and then changing very little. The new equilibrated haplotype frequencies occur long before fixation in most cases, supporting a trait-based model of phenotypic evolution. Throughout the genome a consistent pattern of change is that one of the 18 local haplotypes changes a great deal, with the others changing very little. We finally repeat the analysis of the experiment ignoring founder haplotype information, focusing instead on directly ascertained frequency changes at SNPs. When we only employ directly ascertained SNP frequencies many of the above patterns observed are not easily detected.