Fitness recovery
Both control parental lines went extinct after two generations of experimental evolution. In contrast, all hybrid lines survived the nine months of experimental evolution, with the exception of SC3, which went extinct in the 6th month. Most of the surviving lines showed an increase in population size (6/9 of the SC lines, and 6/7 of the SD lines), with census size showing an increase between 1.1-fold in SD4 to 9-fold in SD6 (Fig. 2A).
Regarding fecundity (Fig. 2B), F2 hybrids had lower fitness than the parental with the respective mitochondrial lineage, confirming F2 hybrid breakdown in this life history trait. Hybrid breakdown was more extreme in the comparison with the SC mitochondria (P-value= 0.005), relative to that with the SD mitochondria (P-value= 0.017), consistent with stronger breakdown in SC mitochondrial background. Of the hybrid lines that survived experimental evolution, 7 SC and 6 SD lines had sufficient virgin females to assess fecundity. Mean fecundity ±1SE of all hybrid lines was within the range observed in the respective F2 hybrids (mean plus ±1SE), consistent with no or little recovery from initial hybrid breakdown.
Regarding survivorship (Fig. 2C), both reciprocal hybrids had lower fitness than parental populations with the respective mitochondrial lineage, yet non-significant (P-values> 0.125), confirming breakdown in this trait. The ±1SE intervals overlapped in SD but not in the SC comparison, again suggesting that F2 hybrid breakdown is stronger in the SC mitochondrial background. A total of 8 SC and 6 SD lines had sufficient gravid females to assess survivorship of nauplii. Of these, 4 SC lines and 4 SD lines had survivorship (mean ±1SE) at or above the fitness of the respective parental, suggesting recovery from the initial hybrid breakdown. These differences were significant for the SC4, SC7, SC8, SD3 and SD7 lines (respective corrected P-values= 0.001, 0.009, 0.017, 0.043, 0.043; Fig. 2C).
Based on fewer replicates measured during the experiment, we observed a large temporal variability on survivorship (Fig. 2D, Fig. S1). The increase of survivorship was more noticeable among SC lines, where F2 breakdown was stronger, and lines stayed for a few generations at the fitness level of their parental population.
Based on these results, for the lines evolving under the SC mitochondria, we chose to sequence SC4, SC8 and SC10 because, from lines with the highest increase of population size, these were the ones with highest survivorship at the end of the experiment. For the lines evolving under the SD mitochondria, we chose to sequence SD3, SD4, and SD6 because they showed some increase in survivorship, without a noticeable decrease in fecundity (contrary to SD7). We note that, although the choice of lines is necessarily subjective, given that this experimental design did not employ direct selection on a specific phenotype, surviving lines showing population growth would be good representatives of recovery from initial F2 breakdown.