Allelic frequency change
We sequenced between 53 and 74 million reads per pool (BioProject XXXX). After quality trimming and mapping, this sequencing effort resulted in an average coverage of 31.9x (SC mtDNA) and 41.1x (SD mtDNA) for the initial F2 hybrids, 33.9x, 43.1x and 35.4x for the lines evolving under the SC mitochondria, and 33.5x, 33.5x and 34.5x for the lines evolving under the SD mitochondria. After reciprocal mapping to the reference genomes of both parental populations and filtering for fixed SNPs, our approach allowed us to estimate read counts for 1,658,000 ancestry informative SNPs, corresponding to 3,316 genomic windows, and covering the 12 chromosomes of T. californicus .
The allelic frequencies in adult F2 were generally close to 50% (Figure S2), varying between 38 and 64% in the F2s with the SC mitochondria, and between 34 and 61% in the F2s with the SD mitochondria. This variation was relatively stable along each of the 12 chromosomes, consistent with chromosome-wide linkage disequilibrium. In contrast, allelic frequencies in hybrid lines after experimental evolution showed a strong deviation from 50% (Figure S3). This deviation was most pronounced in the lines evolving under the SC mitochondrial background, where the frequency of the SD allele ranged between 5 and 90%. The lines evolving under the SD mitochondrial background showed a more modest allele frequency change, ranging from 20 and 76% for the SD allele. Contrary to F2s, changes in allelic frequency were variable along each of the 12 chromosomes, showing that the effective population recombination rate was enough to reduce linkage disequilibria during the nine months of experimental evolution.
The SC lines showed the most extreme P-values of rejecting the null hypothesis of no allelic frequency change, with genomic windows reaching the threshold of -log10 (P-value) =1.73 in all 12 chromosomes. Larger genomic regions were identified in 8 chromosomes, 7 of which favored the nuclear allele matching the mitochondrial background (i.e. SC allele). The only region in which the opposite allele was favored (i.e. SD allele in chromosome 11) did not show a similar pattern in lines evolving under the SD mitochondria. The SD lines showed less extreme P-values supporting allelic frequency change, with genomic windows above the threshold located in 6 chromosomes. Larger genomic regions were identified in 5 chromosomes, 3 of which favored the nuclear allele matching the mitochondrial background (i.e. SD allele). Notably, the chromosomal regions in which the opposite parental allele was favored (i.e. SC allele in chromosomes 7 and 8), show a similar pattern in the SC lines. Genomic regions responding to selection varied between 62 Kbp – 6.2 Mbp for the SC lines and 512 Kbp – 3.2 Mbp for the SD lines.
From the 3,316 genomic windows analyzed here, 60.5% were not skewed in both lines and occurred in all 12 chromosomes, consistent with a prevalent role of genetic drift. About 6.5% of the windows shows a consistent skew towards the SC allele, mostly in chromosome 7 and in part of chromosome 8, consistent with uniform selection favoring SC alleles. The SD alleles were never consistently favored in both mitochondrial backgrounds. The remaining 33% of the genome show skews dependent on the mitochondrial background and are therefore consistent with divergent selection.
Of the 909 genomic windows that are skewed only in the SC mitochondrial background, 88.6% show a skew favoring the matching SC nuclear allele, in chromosomes 2, 3, 8,10 and 12. The remaining 11.4%, all in chromosome 11, show a skew towards the mismatched SD nuclear allele. Of the 187 genomic windows that are skewed only in the SD mitochondrial background, 87% show a skew towards the matching SD nuclear allele, in chromosomes 4, 6, and 9. The remaining 13% show a skew towards the mismatching SC nuclear allele, but notably all these windows are adjacent to windows likely under uniform selection favoring the same allele in chromosome 7.