Abstract
LTR and non-LTR retrotransposons exhibit distinct patterns of abundance within the Drosophila melanogaster genome, yet the causes of these differences remain unknown. Here we investigate whether genomic differences between LTR and non-LTR retrotransposons reflect systematic differences in their insertion history. We find that for 17 LTR and 10 non-LTR retrotransposon families that evolve under a pseudogene-like mode of evolution, most elements from LTR families have integrated in the very recent past since colonization of non-African habitats ( approximately 16,000 years ago), whereas elements from non-LTR families have been accumulating in overlapping waves since the divergence of D. melanogaster from its sister species, Drosophila simulans ( approximately 5.4 Mya). LTR elements are significantly younger than non-LTR elements, individually and by family, in regions of high and low recombination, and in genic and intergenic regions. We show that analysis of transposable element (TE) nesting provides a method to calculate transposition rates from genome sequences, which we estimate to be one to two orders of magnitude lower than those that are based on mutation accumulation studies. Recent LTR integration provides a nonequilibrium alternative for the low population frequency of LTR elements in this species, a pattern that is classically interpreted as evidence for selection against the transpositional increase of TEs. Our results call for a new class of population genetic models that incorporate TE copy number, allele frequency, and the age of insertions to provide more powerful and robust inferences about the forces that control the evolution of TEs in natural populations.