Population structure in the D. innubila genome is associated with segregating inversions
As mentioned previously, FST is significantly higher on Muller element B compared to all other elements in all populations (Supplementary Figure 2, GLM t-value = 30.02, p -value = 3.567e-56). On Muller element B, regions of elevated FSTare similar in each population (Supplementary Figure 2). Additionally, Muller element B has elevated Tajima’s D compared to all other Muller elements (Supplementary Figure 3), suggesting some form of structured population unique to Muller element B (GLM t-value = 10.402,p -value = 2.579e-25). We attempted to identify if this elevated structure is due to chromosomal inversions, comparing FST of a region to the presence or absence of inversions across windows (using only inversions called by both Delly and Pindel (Ye et al. 2009; Rausch et al. 2012)). We find several inversions across the genome at appreciable frequencies (89 total above 1% frequency), of which, 37 are found on Muller element B (spread evenly across the entire chromosome) and 22 are found at the telomeric end of Muller element A (Figure 3A). The presence of an inversion over a region of Muller element B is associated with higher FST in these regions (Figure 3A, Wilcoxon Rank Sum test W = 740510, p -value = 0.0129), though these inversions are not unique or even at different frequencies in specific populations (FST < 0.22, χ2 test for enrichment in a specific population p -value > 0.361 for all inversions). Genes within 10kbp of an inversion breakpoint have significantly higher FST than outside the inverted regions consistent with findings in other species (Figure 3C, GLM t-value = 7.702, p -value = 1.36e-14) (Machado et al. 2007; Noor et al. 2007), while inside inverted regions show no difference from outside (Figure 3C, GLM t-value = -0.178, p -value = 0.859). However, all regions of Muller element B have higher FST than the other Muller elements (Figure 3C, outside inversions Muller element B vs all other chromosomes: GLM t-value = 7.379, p -value = 1.614e-13), suggesting some chromosome-wide force drives the higher FST and Tajima’s D. Given that calls for large inversions in short read data are often not well supported (Chakraborty et al. 2017) and the apparently complex nature of the Muller element B inversions (Figure 3A), we may not have correctly identified the actual inversions and breakpoints on the chromosome. Despite this, our results do suggest a link between the presence of inversions on Muller element B and elevated differentiation in D. innubila and that this may be associated with local adaptation.
Figure 3: Summary of the inversions detected in theDrosophila innubila populations. A. Location and frequency in the total population of segregating inversions at higher than 1% frequency and greater than 100kbp. B. Tajima’s D andC. FST for genes across Muller element B, grouped by their presence under an inversion, outside of an inversion, near the inversion breakpoints (within 10kbp) or on a different Muller element.