The impact of natural selection on short insertion and deletion variation in the great tit genome.

Insertions and deletions (INDELs) remain understudied, despite being the most common form of genetic variation after single nucleotide polymorphisms. This stems partly from the challenge of correctly identifying the ancestral state of an INDEL and thus identifying it as an insertion or a deletion. Erroneously assigned ancestral states can skew the site frequency spectrum, leading to artificial signals of selection. Consequently, the selective pressures acting on INDELs are, at present, poorly resolved. To tackle this issue, we have recently published a maximum likelihood approach to estimate the mutation rate and the distribution of fitness effects (DFE) for insertions and deletions. Our approach estimates and controls for the rate of ancestral state misidentification, overcoming issues plaguing previous INDEL studies. Here we apply the method to INDEL polymorphism data from 10 high coverage (∼44X) European great tit (Parus major) genomes. We demonstrate that coding INDELs are under strong purifying selection with a small proportion making it into the population (∼4%). However, among fixed coding INDELs, ∼71% of insertions and ∼86% of deletions are fixed by positive selection. In non-coding regions we estimate ∼80% of insertions and ∼52% of deletions are effectively neutral, the remainder show signatures of purifying selection. Additionally, we see evidence of linked selection reducing INDEL diversity below background levels, both in proximity to exons and in areas of low recombination.