Population genomics of transposable elements in Drosophila melanogaster

Abstract

Transposable elements (TEs) are the primary contributors to the genome bulk in many organisms and are major players in genome evolution. A clear and thorough understanding of the population dynamics of TEs is therefore essential for full comprehension of the eukaryotic genome evolution and function. Although TEs in Drosophila melanogaster have received much attention, population dynamics of most TE families in this species remains entirely unexplored. It is not clear whether the same population processes can account for the population behaviors of all TEs in Drosophila or whether, as has been suggested previously, different orders behave according to very different rules. In this work, we analyzed population frequencies for a large number of individual TEs (755 TEs) in five North American and one sub-Saharan African D. melanogaster populations (75 strains in total). These TEs have been annotated in the reference D. melanogaster euchromatic genome and have been sampled from all three major orders (non-LTR, LTR, and TIR) and from all families with more than 20 TE copies (55 families in total). We find strong evidence that TEs in Drosophila across all orders and families are subject to purifying selection at the level of ectopic recombination. We showed that strength of this selection varies predictably with recombination rate, length of individual TEs, and copy number and length of other TEs in the same family. Importantly, these rules do not appear to vary across orders. Finally, we built a statistical model that considered only individual TE-level (such as the TE length) and family-level properties (such as the copy number) and were able to explain more than 40% of the variation in TE frequencies in D. melanogaster.

FIG. 1.

TE frequency distribution by family and order.

FIG. 2.

Frequency distribution of all the TEs, LTR, non-LTR, and TIR by local recombination rate where the TE is inserted (A), TE length (B), and polymorphic copy number (C).

FIG. 3.

Comparison of the predicted TE frequencies according to models A, B, C, and E′ and the observed TE frequencies. For each observed frequency class, we plot the observed versus the predicted frequencies according to that model for each of the 637 TEs and the boxplot summarizing the frequencies distribution. We also plot the observed versus the predicted frequencies for the 13 adaptive TEs described in González et al. (2008).