Fig. 1: Experimental approach.

We evolved 15 S. pombe populations at 30 °C in lidded YPD 96-well plates for 10,000 generations, collecting and freezing samples approximately every 70 generations in 27% glycerol at −80 °C and sequenced them every 1,000 generations. These populations were part of a larger experimental evolution project, which explains why they occupy two rows of the 96-well plates, and were examined regularly for contamination (Methods). Because the experimental conditions yielded a doubling rate of ten generations per day, we performed a 1 in 1,024 serial dilution every day (1/2¹⁰), which can be done by two consecutive 1:32 dilutions with 96-well plates. This yielded ~100 time points or samples from which we sequenced DNA to identify variants. We also sequenced RNA at the initial (WT) and final time points to quantify gene expression changes. To find targets of selection at the genomic levels, we identified multi-hit genes, that is, genes that were recurrently hit by fixed mutations (frequency ≥75%) that are not synonymous (indels, missense or nonsense) in several populations. At the transcriptomic level, we identified genes that were expressed differentially across several evolved populations compared with WT or with populations not exhibiting a change in a phenotype of interest, for example, sensitivity to OS. These data allowed us to generate hypotheses about adaptation in the ten populations surviving, that is, G1–G5, G9, H9 and H11 and 12, which we tested experimentally. For instance, in the figure, G1 (blue) and G2 (red) both adapt to HS at the cost of becoming more sensitive to OS. Figure created in BioRender; Arnaud, N. https://biorender.com/8pi06vc (2026).