Although the evolution of reproductive isolation has been a focus of intense interest since the birth of the Modern Synthesis, its quantification is often not a trivial exercise. Among animal taxa, most research effort has focused on Drosophila [1, 2], birds [3, 4], amphibians [5, 6], and fish [7, 8]. In gastropod mollusks, the mechanisms of reproductive isolation have been elucidated only in the marine prosobranch snail Littorina [9, 10], opisthobranch sea slugs , certain terrestrial (stylommatophoran) pulmonates [12, 13], and in the freshwater (basommatophoran) pulmonates that will be the focus of the present work.
The barriers to reproduction that may evolve between a pair of populations have customarily been divided into prezygotic and postzygotic components, both of which are typically inherited in a complex and polygenic fashion. Coyne  reported that the shortened copulations observed between Drosophila simulans and D. mauritiana, for example, are controlled by genes on all three major chromosomes, the two autosomes more important than the X chromosome. With regard to postzygotic reproductive isolation, experiments with F2 backcrosses among Drosophila species have consistently demonstrated that hybrid sterility is attributable to genes on every arm of every chromosome . Presgraves  estimated approximately 191 hybrid-lethal incompatibilities between D. melanogaster and D. simulans. Thirteen loci in seven linkage groups were implicated in the origin of postzygotic isolation speciation among lake whitefish by Rogers and Bernatchez .
A positive correlation has often been documented between degree of reproductive isolation and overall genetic divergence [18–20]. This relationship was first demonstrated in the D. willistoni species complex of South America by Ayala et al.  using variation at allozyme-encoding loci, and has since been confirmed in Drosophila generally [1, 2], Lepidoptera , fish , amphibians [6, 24], and birds . Most of these studies have focused on postzygotic reproductive isolation, which tends to evolve more slowly than prezygotic isolation . Although a lag is generally noted between the accumulation of genetic differences and the onset of reproductive barriers in these studies , the temporal relationship between genetic divergence and reproductive isolation seems sufficiently predictable to prompt theoretical exploration of a "speciation clock" [18, 27].
Since the advent of modern molecular methods for phylogenetic reconstruction, increased attention has turned toward the detailed correspondence between gene trees and the reproductive relationships among the populations from which marker genes have been sampled. As might be expected from the complex and polygenic nature of reproductive barriers, however, their correlation with any single gene or small number of genes arbitrarily selected to reconstruct a population phylogeny under the neutral model may be poor . In both the D. melanogaster and D. willistoni species groups, for example, genes sampled from narrowly restricted but reproductively isolated species nest within the branches of gene trees constructed from their more widely-dispersed progenitors, rendering the ancestor species paraphyletic [25, 29–31]. In the group of D. pseudoobscura, none of 16 genes sampled from 10-20 individuals unambiguously returned the known reproductive relationships among three populations, relative to an outgroup . Such discrepancies have led some systematists to endorse "phylogenetic" species concepts based not on reproductive isolation, but rather on the evolutionary history of marker genes .
The purpose of the work we present here is to examine the origin of reproductive isolation in a simultaneous hermaphrodite from a phylogenetic perspective. We focus on freshwater pulmonate snails of the genus Physa, which by virtue of their reproductive plasticity, ease of culture, and availability of genetic markers have become important models for the study of mating systems generally [34–36]. The best-known species is Physa acuta, apparently a North American native introduced to Europe in the early nineteenth century and subsequently spread to six continents . In cultured populations, male fertility develops around 6 weeks post-hatch, and female fertility is added around week 7, with the onset of self-fertilization around week 9, if no partner is supplied [38–41]. However, the reductions in parental fecundity and F1 viability engendered by self-fertilization seem to select strongly for outcrossing [39, 42–46].
North America is also home to approximately ten lesser-known species in the family Physidae, differing from P. acuta by reproductive anatomy, habitat, and minor aspects of shell morphology . Swamps and ditches in the southeastern United States are inhabited by Physa carolinae, bearing a darker and more slender shell than P. acuta, and river margins are inhabited by Physa pomilia, reaching adulthood at a smaller size . Both P. carolinae and P. pomilia have two-part penial sheaths, with a larger muscular portion and a smaller glandular portion, while the penial sheath of P. acuta is not subdivided. Further north, more stable freshwater environments are inhabited by Physa gyrina, distinguished by a two-part penial sheath with approximately equal glandular and muscular portions [49, 50].
In recent years substantial research effort has been directed toward quantifying the reproductive isolation among North American physid populations. The experimental tools available are directly analogous to the "mate-choice" and "no-choice" tests that have become established by common practice with dioecious animals displaying obligate sexual reproduction , slightly modified in their interpretation for hermaphrodites.
Physa mate-choice tests have been modelled closely on the experimental design used with Drosophila in population cages for many years [1, 2, 51]. Such methods have been applied to examine specific relationships in the freshwater pulmonate snail Biomphalaria [52, 53] and in land snails [54, 55], as well as in the dioecious prosobranch snail Littorina [56, 57]. Although Physa are simultaneously hermaphroditic, copulation is unidirectional, with prospective partners typically vying to assume the male role. The complex behaviors documented in pairs of mating physids [58, 59] have been attributed both to sexual conflict (the rejection of males by females ) and to gender conflict (between hermaphroditic animals vying to mate as male ). Thus while the reproductive relationships among pairs of physid populations can be quantified using standard mate-choice statistics, each datum is not a "choice" in any sense, but rather the outcome of a contest. Any consistent approach to scoring that outcome yielding a single observation per individual might be employed.
The use of no-choice tests to evaluate reproductive isolation was also pioneered with fruit flies  and such techniques have been used to explore specific relationships in the freshwater pulmonate snail Bulinus [62, 63] and in the marine prosobranch Lacuna . But the interpretation of such experiments is again somewhat ambiguous in hermaphrodites such as Physa, since any individual snail always has a "choice" to outcross with the single partner provided or to self-fertilize. Thus in addition to comparing the fertility and fecundity of outcrossed pairs to their corresponding incross controls, the hybridity of first-generation progeny must be confirmed using genetic markers, and any offspring of self-fertilization subtracted.
Here we report an expansion of our previous work to include both mate-choice and no-choice tests among all pairs of P. acuta, P. carolinae, P. pomilia and P. gyrina. We add a second population of P. acuta to test for discrepancies due to paraphyly, such as have been the concern of the proponents of phylogenetic species concepts. The results of these 10 possible pairwise comparisons are then cast into a phylogenic framework (a "species tree") using the mtDNA sequence data of Wethington and Lydeard  and Wethington et al. . By this approach we are able to infer the relative rates of evolution for hybrid sterility, hybrid inviability, and sexual incompatibility.