Tempo and Mode of Transposable Element Activity in Drosophila

Abstract

The evolutionary dynamics of transposable element (TE) insertions have been of continued interest since TE activity has important implications for genome evolution and adaptation. Here, we infer the transposition dynamics of TEs by comparing their abundance in natural D. melanogaster and D. simulans populations. Sequencing pools of more than 550 South African flies to at least 320-fold coverage, we determined the genome wide TE insertion frequencies in both species. We suggest that the predominance of low frequency insertions in the two species (>80% of the insertions have a frequency <0.2) is probably due to a high activity of more than 58 families in both species. We provide evidence for 50% of the TE families having temporally heterogenous transposition rates with different TE families being affected in the two species. While in D. melanogaster retrotransposons were more active, DNA transposons showed higher activity levels in D. simulans. Moreover, we suggest that LTR insertions are mostly of recent origin in both species, while DNA and non-LTR insertions are older and more frequently vertically transmitted since the split of D. melanogaster and D. simulans. We propose that the high TE activity is of recent origin in both species and a consequence of the demographic history, with habitat expansion triggering a period of rapid evolution.

Fig 1. Frequency distributions of TE insertions in D. melanogaster (black) and D. simulans (grey); Only TE insertions for which the population frequencies could be estimated are shown (not overlapping, minimum physical coverage of 30); D. melanogaster: 16,901 insertions; D. simulans: 12,716 insertions.

Fig 2. Distribution of TE insertions in a natural population of D. melanogaster (dm) and of D. simulans (ds).

The TE distribution (outer graph) is compared to the recombination rate (middle graph) and the nucleotide polymorphism (Θ_π, yellow inner graph). TE abundance and recombination rate are shown for windows of 500kb, whereas the nucleotide diversity is shown for windows of 100kb. For overlapping TE insertions (white) no estimates of population frequencies could be obtained. The relationship between the reference genomes is shown in the inside. Note, the inversion on chromosome 3R [47] and the missing pericentromeric regions in the assembly of D. simulans. The maximum nucleotide diversity of the plot is 0.018 and the maximum number of TE insertions 400.

Fig 3. Abundance of different TE families in natural D. melanogaster and D. simulans populations; Significant differences in TE copy numbers from expectations under drift are indicated for the species with a higher number of insertions, assuming equal population sizes in both species (yellow), or a N _e ratio of 1.519 (pink).

Those cases for which both models agree are shown in white. Families with at least one fixed insertion common to both species are highlighted in grey and families with documented HT between D. simulans and D. melanogaster [46] are marked with an arrow. p-value after Bonferroni correction: ** < 0.001; * < 0.01; + < 0.05; Only TE families having in total more than 10 insertions are shown. Foldback (FB) is grouped with TIRs solely for graphic reason.

Fig 4. Procedure for estimating the significance (p) of the difference in TE copy numbers between D. simulans (Dsim) and D. melanogaster (Dmel) from expectations under drift using an equilibrium model.

A.) Simulated equilibrium copy numbers of TE insertions for transposition rates (v) ranging from 0 to 0.003 and two different populations sizes (N = 6,583 black dots; N = 10,000 blue dots). More than 10.000 independent simulations were performed for each population size. For every TE family (e.g. roo and I-element) the maximum likelihood transposition rate (v _ml) is identified, assuming either an about 1.519 times smaller population size in D. melanogaster than in D. simulans (N _Dmel < N _Dsim) or equal population sizes in both species (N _Dmel = N _Dsim). B.) A normal distribution is fitted to the equilibrium copy numbers in a small window around v _ml and p can be estimated from the two tailed area obtained by intersecting the normal distributions with the observed copy numbers in the two species (bottom bars). For details see material and methods.