An integrative test of the dead-end hypothesis of selfing evolution in Triticeae (Poaceae).

Self-fertilization is hypothesized to be an evolutionary dead end because reversion to outcrossing can rarely happen, and selfing lineages are thought to rapidly become extinct because of limited potential for adaptation and/or accumulation of deleterious mutations. We tested these two assumptions by combining morphological characters and molecular-evolution analyses in a tribe of hermaphroditic grasses (Triticeae). First, we determined the mating system of the 19 studied species. Then, we sequenced 27 protein-coding loci and compared base composition and substitution patterns between selfers and outcrossers. We found that the evolution of the mating system is best described by a model including outcrossing-to-selfing transitions only. At the molecular level, we showed that regions of low recombination exhibit signatures of relaxed selection. However, we did not detect any evidence of accumulation of nonsynonymous substitutions in selfers compared to outcrossers. Additionally, we tested for the potential deleterious effects of GC-biased gene conversion in outcrossing species. We found that recombination and not the mating system affected substitution patterns and base composition. We suggest that, in Triticeae, although recombination patterns have remained stable, selfing lineages are of recent origin and inbreeding may have persisted for insufficient time for differences between the two mating systems to evolve.