Speciation accompanying the left–right reversal of the entire ontogeny (chiral speciation) is unique to snails and can be visualized by the coiling direction of their shells4 (Fig. 1a). The chirality (or 'handedness' — occurring as 'sinister' and 'dexter' forms) of snails is determined by the maternal nuclear genotype at a single locus5. Because of the physical difficulty of two-way copulation between snails that have opposite coils (Fig. 1b), frequency-dependent selection occurs to eliminate the chiral minority3,6,7,8 and antagonize chiral speciation. Offspring from each mother all develop with an identical chiral phenotype, regardless of their genotype. This peculiar mode of inheritance could be advantageous in enabling the dominant reversal allele to proliferate in small, isolated populations2,9. Once the mirror-image variant exceeds 50%, selection fixes the reversal. Reproductive isolation by handedness would then be complete.

Figure 1: Mirror-image species and copulation in snails.
figure 1

a, Shells of Euhadra: left, E. quaesita (sinistral, or left-handed); middle, E. aomoriensis (dextral, or right-handed); right, E. senckenbergiana (dextral). Scale bar, 10 mm. The remarkable similarity of the shells reflects the chiral speciation of E. aomoriensis from E. quaesita, although E. aomoriensis has traditionally been considered to be a subspecies of E. senckenbergiana (see text). b, Successful two-way copulation between dexters of E. congenita. c, Genital mismatch between a sinistral variant (top) and an ordinary dexter (bottom) of Bradybaena similaris. Arrows indicate genital openings that cannot be joined.

In Euhadra, 4 of 20 species are sinistral. We discovered three notable processes of chiral evolution, based on analysis of mitochondrial DNA phylogeny (see supplementary information). First, the sinistral taxa are all derived from a single sinistral ancestor. Second, reversal to a dextral species from the sinistral ancestor has occurred in at least three independent lineages. Third, all of the haplotypes of a dexter E. aomoriensis are included within a sinister E. quaesita.

This pattern cannot be explained by introgression of mitochondrial DNA between E. quaesita and other dextral species, because opposite handedness prevents hybridization. Both mitochondrial-DNA phylogeny and a similar shell morphology (Fig. 1a) indicate that E. aomoriensis is derived from E. queasita. E. aomoriensis is not merely a dextral morph of E. queasita, as the two differ in shell sculpture although they are partly sympatric. We conclude that E. aomoriensis has speciated from E. quaesita by virtue of its chirality. The two species mutually failed in interchiral copulation despite frequent attempts, as did the chiral variants of Bradybaena similaris3 (see supplementary information). Chiral reversal is therefore an acute pre-mating mechanism of isolation.

Reversal to a dextral species has occurred multiple times after only sinistral evolution, so chiral speciation must have been easier to the dexter than to the sinister. We tested chiral inheritance in the family Bradybaenidae, which includes Euhadra, using a single dextral adult Bradybaena similaris collected with an exceptional sinistral variant3. The dexter, which presumably had been inseminated by multiple conspecific dexters, produced sinistral offspring.

According to the single-locus, delayed-inheritance model with dextral dominance, the dextral mother would have been a recessive homozygote and the sinistral offspring (F1) would have been heterozygotes. Seven pairs of F1 sinisters produced 131 dexters (F2), in which only 30 produced sinistral offspring and others did not. The 30 dexters must therefore have been recessive homozygotes. The ratio of 30 to 101 is consistent with a prediction of allelic segregation at a single locus, with dextral dominance in F22 = 3.01, P = 0.62). The repeated twofold reversals to dextral species must therefore reflect the greater chance of fixation for the dominant allele, as predicted by theory9,10.

The polyphyly of E. aomoriensis in mitochondrial-DNA phylogeny could have resulted from parallel evolution of E. aomoriensis, or from introgression or ancestral polymorphism of mitochondrial DNA. As the parallel evolution of three dextral lineages has occurred, multiple peripheral populations of E. quaesita may also have fixed the dextral allele independently. The polyphyly of ancestral species is expected where prompt speciation recurs in peripheral isolates11. Incipient single-gene speciation without genome-wide changes must allow E. aomoriensis to interbreed with polyphyletic descendants. E. aomoriensis may therefore be a unique example of a polyphyletic animal species.

Introgression is impossible unless the reversal allele has persisted in either species under stringent selection. Any introgression must therefore have been limited to a short time before chiral fixation. Ancestral polymorphism would not easily survive chiral speciation in small, isolated populations.

Left–right reversal in reciprocally mating snails can accomplish pre-mating isolation at a single locus, overcoming the majority rule of handedness. Our results indicate that single-gene speciation is possible, at least in hermaphroditic snails, contrary to the traditionally held view2,12,13.