However choosy you are about who you mate with, it is unlikely that any mistake will have an appreciable impact on the course of human evolution. In the same situation, a female Drosophila pseudoobscura has a more onerous decision to take. Because this species coexists in some regions with the related Drosophila persimilis, and mating between the two gives rise to sterile males, the female's decision makes the difference between propagating the line and an evolutionary dead end. The genetics of reproductive isolation between the two species is well known; now, Daniel Ortiz-Barrientos and colleagues reveal how this behavioural discrimination is strengthened by natural selection. Their high-resolution genetic analysis is the first to examine the nitty gritty of the final stages of speciation, so filling one of several experimental gaps in the study of adaptation.

D. persimilis and D. pseudoobscura mate in the wild but, because of the sterility of the male offspring, females of both species (but not males) select mates primarily from their own species. How is this behaviour reinforced by natural selection, such that the discriminatory mating behaviour survives in the face of gene flow? D. pseudoobscura females sampled from regions in which the two fly species overlap (in 'sympatry') are much choosier than those that do not live alongside D. persimilis (in 'allopatry'), and the authors show that this 'reinforced mating discrimination' trait is inherited from the female in a dominant manner. Backcrossing the progeny of a sympatric × allopatric cross into an allopatric background identified strong candidate regions for the trait on the X and fourth chromosomes, which were mapped in more detail to three QTL intervals by using 70 microsatellite markers. One of the candidate regions contains as few as five genes, and both the D. pseudoobscura genome sequence and information from Drosophila melanogaster mutants indicate the involvement of the olfactory response.

As with many studies of adaptation, theories abound over experiments. An exciting aspect of the work was to find that the genetic architecture of reinforcement traits within species differs from the basal level of mating discrimination seen between species, which is what some theories would predict.