Evidence for divergence in the X chromosome over time and between sexes
We next compared the samples from the 35 CH males in 2017 to those we sequenced from a CH collection of 38 males in 2001 to identify changes over time between populations (due to elevated FSTbecause of differences in allele frequencies between populations). We find little differentiation between the two timepoints (median FST = 0.0004, 99th percentile = 0.0143), and find no significant enrichments (GO p -value < 0.05) in the upper 97.5th percentile of FST. However, we do find divergence in the genes at the telomere of the X chromosome (Supplementary Figure 8A, Muller element A, 35-40.5Mb, median FST = 0.0029). Looking at actual allele frequency differences between time points, the minor allele frequency increases between 2001 and 2017 at the X telomere while most other chromosomes appear to show little change. Interestingly, there is also evidence of recent selective sweeps in the telomere of X (Supplementary Figure 7A). The minor allele frequency has decreased on average on Muller element B between 2001 and 2017 (Supplementary Figure 9A). This suggests something else may be influencing allele frequency change on Muller B compared to other autosomes.
We also compared the allele frequencies between 2017 male samples versus 2017 female samples. Again, FST is extremely low genome wide (median FST = 0.0004, 99thpercentile = 0.0501), but we again find a peak of FST at the telomere of the X chromosome (Supplementary Figure 8B), which we find when comparing all populations sexes and in a total population male versus female comparison. Again, we find no significant enrichments in the 97.5th percentile for FST, as most of the divergent genes currently have no functional annotation. We also compared the raw allele frequency change of synonymous variants. Strangely, the population minor allele frequency of euchromatic SNPs on the X chromosome are found at higher frequencies in females (Supplementary Figure 9B), while the X telomere SNPs are overrepresented in male samples. These results are consistent when examining each population separately, suggesting sex specific biases in the X chromosome are found in every populations. It is possible that this signal is caused by an ascertainment bias for SNP calling in females, resulting in more accurate SNP calls in one of the sexes in the euchromatin which is not seen in the heterochromatin. Alternatively, the region of the X chromosome with multiple overlapping inversions could be female-biased due to a female driver, resulting in its overrepresentation in females (and an overrepresentation of the alternate variants in males) (Burt and Trivers 2006). Finally, the X chromosome may be adapting to the skewed sex-ratio associated withD. innubila ’s male-killing Wolbachia (Kageyamaet al. 2009; Unckless 2011b).