Selection inference in a complex genomic landscape: the impact of polymorphic inversions

Understanding the genetic basis of adaptation is a key objective in evolutionary biology. Although advances in genomic selection scans have greatly improved our ability to detect signatures of adaptation, distinguishing true signals from false positives remains challenging. This task is particularly difficult in regions of reduced recombination, such as polymorphic inversions. In this study, we examine the genome-wide landscape of selection in the European spruce bark beetle ( Ips typographus ), Europe’s most destructive forest pest, which has one of the most complex inversion-associated recombination landscapes known. Using simulation-based analyses and whole-genome resequencing data from 312 individuals across 23 populations, we applied two complementary selection scan methods (nSL and Λ) to assess how inversions influence the detection of adaptive signals. Simulations revealed that the two selection scan methods differ in their susceptibility to false-positive signals within inversions and that partitioning data by inversion genotype (i.e., homozygote classes) can substantially reduce these errors. Consistent with the results of our simulations, our empirical data showed that inversions are highly enriched for selection signals when all inversion genotypes are analyzed but are depleted when only homozygotes are considered. This demonstrates that focusing on homozygote genotypes allows for a more precise identification of putative selection targets and haplotype-specific signals, as it overcomes the confounding effects of suppressed recombination in heterokaryotypic individuals. In contrast, the detection of selection in collinear regions was largely unaffected by the presence of inversions. Our findings highlight that inversions may play an important role in shaping adaptation, underscoring the need to account for species-specific genomic architecture when interpreting signals from selection scans.