It has been proposed that the insertion time of a long terminal repeat (LTR) retrotransposon can be estimated by the divergence between the two LTRs at the both ends because their sequences were identical at the insertion event. This method is based on the assumption that the two LTRs accumulate point mutations independently; therefore, the divergence reflects the time since the insertion event. However, if gene conversion occurs between LTRs, the nucleotide divergence will be much smaller than expected with the assumption of the independent accumulation of point mutations. To examine this assumption, we investigated the extent of gene conversion between LTRs by applying a comparative genomic approach to primates (humans and rhesus macaques) and rodents (mice and rats). We found that gene conversion plays a significant role in the molecular evolution of LTRs in primates and rodents, but the extent is quite different. In rodents, most LTRs are subject to extensive gene conversion that reduces the divergence, so that the divergence-based method results in a serious underestimation of the insertion time. In primates, this effect is limited to a small proportion of LTRs. The most likely explanation of the difference involves the minimum length of the interacting sequence (minimal efficient processing segment [MEPS]) for interlocus gene conversion. An empirical estimate of MEPS in human is 300-500 bp, which exceeds the length of most of the analyzed LTRs. In contrast, MEPS for mice should be much smaller. Thus, MEPS can be an important factor to determine the susceptibility of LTRs to gene conversion, although there are many other factors involved. It is concluded that the divergence method to estimate the insertion time should be applied with special caution because at least some LTRs undergo gene conversion.

A population genetic simulation framework is developed to understand the behavior and molecular evolution of DNA sequences of transposable elements. Our model incorporates random transposition and excision of transposable element (TE) copies, two modes of selection against TEs, and degeneration of transpositional activity by point mutations. We first investigated the relationships between the behavior of the copy number of TEs and these parameters. Our results show that when selection is weak, the genome can maintain a relatively large number of TEs, but most of them are less active. In contrast, with strong selection, the genome can maintain only a limited number of TEs but the proportion of active copies is large. In such a case, there could be substantial fluctuations of the copy number over generations. We also explored how DNA sequences of TEs evolve through the simulations. In general, active copies form clusters around the original sequence, while less active copies have long branches specific to themselves, exhibiting a star-shaped phylogeny. It is demonstrated that the phylogeny of TE sequences could be informative to understand the dynamics of TE evolution.

In order to understand how DNA sequences of transposable elements (TEs) evolve, extensive simulations were carried out. We first used our previous model, in which the copy number of TEs is mainly controlled by selection against ectopic recombination. It was found that along a simulation run, the shape of phylogeny changes quite much, from monophyletic trees to dimorphic trees with two clusters. Our results demonstrated that the change of the phase is usually slow from a monomorphic phase to a dimorphic phase, accompanied with a growth of an internal branch by accumulation of variation between two types. Then, the phase immediately changes back to a monomorphic phase when one group gets extinct. Under this condition, monomorphic and dimorphic phases arise repeatedly, and it is very difficult to maintain two or more different types of TEs for a long time. Then, how a new subfamily can evolve? To solve this, we developed a new model, in which ectopic recombination is restricted between two types under some condition, for example, accumulation of mutations between them. Under this model, because selection works on the copy number of each types separately, two types can be maintained for a long time. As expected, our simulations demonstrated that a new type arises and persists quite stably, and that it will be recognized as a new subfamily followed by further accumulation of mutations. It is indicated that how ectopic recombination is regulated in a genome is an important factor for the evolution of a new subfamily.

This paper focuses on terminological issues related to the translation of Darwin’s concept of “natural selection” in Japanese. We analyze the historical fate of the different phrases used as translations, from the first attempts in the late 1870s until recent times. Our first finding is that the first part of the Japanese translations never changed during the period considered: “natural” was constantly rendered by “shizen”. By contrast, the Japanese terms for “selection” have dramatically changed over time. We identify some major breaks in the history of Japanese translations for “natural selection”. From the end of the 1870s to the early 1880s, several translations were suggested in books and periodicals: “shizen kanbatsu”, “shizen tōta”, “tensen”. Katō Hiroyuki adopted “shizen tōta” in 1882 and he undeniably played an important role in spreading this phrase as the standard translation for “natural selection”. The most common Japanese translation of the Origin during the first half of the 20th century (by Oka Asajirō in 1905) also used “shizen tōta”. Adramatic shift occurred after WWII, from “tōta” to “sentaku”. While a linear interpretation could suggest a move from a “bad” translation to a better one, a closer analysis leads to more challenging insights. Especially we stress the role of the kanji restriction policy, which specified which kanji should be taught in schools and thus should be used in textbooks: “tōta” was not included in the list, which may have led to the good fortune of “sentaku” in the 1950–1960s. We think the hypothesis of the influence of Chinese translations is not a plausible one. As to conceptual differences between “shizen tōta” and “sentaku”, they remain unconvincing as both terms could be interpreted as a positive or negative process: there is no clear reason to prefer one term over the other from the strict point of view of their meanings or etymology. Then, turning to the way terms are used, we compare translations of natural selection with translations of artificial or sexual selection. First we turn to the field of thremmatology (breeders): there, “tōta” (sometimes spelled in hiragana instead of kanji) often bore the meaning of culling; since 1917, breeders often used “sentaku” as a translation for “selection”. However, quite surprisingly, breeders used two different terms for selection as a practice (“senbatsu”), and “selection” as in “natural selection” (“shizen sentaku”). Finally, we compare possible translations for “sexual selection” and “matechoice”: here again, there are some good reasons to favour “tōta” over “sentaku” to avoid lexical confusion.