Hybridization and recurrent evolution of left–right reversal in the land snail genus Schileykula (Orculidae, Pulmonata)

Abstract The land snail genus Schileykula Gittenberger, 1983 is distributed in arid limestone areas from western Turkey to north‐western Iran. It comprises eight species, which display high variation in shell size and morphology. The cylindrical shells are 5–12 mm in height and the last shell whorls bear several inner lamellae and plicae. Two taxa differ in their chirality having sinistral shells, while all the others are dextrals such as the vast majority of orculids. The aim of this study was to establish a molecular genetic phylogeny of Schileykula and to test whether it conforms to the current morphology‐based classification. Furthermore, we were interested in the phylogenetic position of the two sinistral forms in order to assess whether one or two reversals happened in the evolution of the genus. Nine out of ten species, including all four subspecies of Schileykula trapezensis and three of six subspecies of Schileykula scyphus, were investigated. A section of the mitochondrial cytochrome c oxidase subunit I gene was analyzed in 54 specimens of Schileykula and from a subsample, partial sequences of the mitochondrial genes for the 12S rRNA and the 16S rRNA, and a section of the nuclear H4/H3 histone gene cluster were obtained. The phylogenetic trees based on the mitochondrial sequences feature high support values for most nodes, and the species appear well differentiated from each other. The two chiral forms evolved independently and are not sister lineages. However, some groupings disagree with the present morphology‐based classification and taxonomical conclusions are drawn. Schileykula trapezensis is polyphyletic in the molecular genetic trees; therefore, three of its subspecies are elevated to species level: Schileykula acampsis Hausdorf, 1996 comb. nov., Schileykula neuberti Hausdorf, 1996 comb. nov., and Schileykula contraria Neubert, 1993 comb. nov. Furthermore, Schileykula sigma is grouped within S. scyphus in the mitochondrial and nuclear trees and consequently treated as a subspecies of the latter (Schileykula scyphus sigma Hausdorf, 1996 comb. nov.). Schileykula nordsiecki, whose shell morphology is indistinguishable from that of the neighboring Schileykula scyphus lycaonica, but who differs in its genital anatomy, was confirmed to represent a distinct lineage. The phylogenies produced by the mitochondrial and nuclear data sets are to some extent conflicting. The patterns differ concerning the grouping of some specimens, suggesting at least two independent hybridization events involving S. contraria, S. scyphus and S. trapezensis. The results exemplify the importance of integrating both mitochondrial and nuclear sequence data in order to complement morphology‐based taxonomy, and they provide further evidence for hybridization across distantly related lineages in land snails.

Current records show that most Schileykula taxa are distributed in a patchy, parapatric pattern, and different species usually do not co-occur at the same localities but live in close vicinity. For example, S. maculata has only been recorded from a single site within the range of Schileykula trapezensis (Páll-Gergely & Asami, 2013).
Schileykula nordsiecki, which was found at a single locality between two disjunct populations of S. s. lycaonica, differs from the latter in its genital anatomy, while it is indistinguishable based on shell morphology (Hausdorf, 1996). Schileykula attilae and S. batumensis are known to inhabit opposite sides of the same castle hill (Páll-Gergely, 2010). Schileykula batumensis and S. t. acampsis are the only Schileykula taxa known to occur in sympatry over a broad range, and the presence of populations with intermediate shell forms in the overlapping area was interpreted as an indication for hybridization between the two species (Neubert, 1993). Apart from the latter two taxa, potential hybrid populations were not reported for any other Schileykula taxa so far. However, the comparison of mt and nc sequence data in more recent studies showed that hybridization is not rare in land snails. There is evidence for hybridization between the orculid species Orcula gularis and Orcula pseudodolium (Harl, Páll-Gergely, et al., 2014), and the mixing of genetically and morphologically extremely distinct populations of Orcula dolium . Evidence for introgressive hybridization between morphologically divergent land snails was found in the family Bradybaenidae in the genus Mandarina (e.g., Chiba, 2005) and between the genera Ainohelix and Ezohelix (Morii, Yokoyama, Kawata, Davison, & Chiba, 2015). RADseq data on the genus Pyramidula also provided evidence for minor ancestral gene flow between Pyramidula pusilla (Gittenberger & Bank, 1986) and Pyramidula saxatilis (Hartmann, 1842) (Razkin et al., 2016).
Hybrid populations were investigated also in the genus Albinaria in Crete with SNP genotyping (Lammers et al., 2013). The presence of a hybrid population of the clausiliid species Micropontica caucasica (Schmidt, 1868) and Micropontica circassica (Boettger, 1888) in the Lagonaki plateau in the Caucasus, analyzed with AFLP markers, was interpreted as a case of hybrid speciation (Koch, Neiber, Walther, & Hausdorf, 2016).
A peculiarity of Schileykula is the presence of a sinistral (shell coiled counterclockwise) species and a sinistral subspecies, whereas all others are dextral (shell coiled clockwise). The sinistral subspecies (S. t. contraria) is known to occur between populations of the dextral S. t. trapezensis. Schileykula inversa, which is a sinistral species, lives close to populations of the dextral S. s. erecta, more than 200 km west of S. t. contraria. In gastropods reproducing by internal fertilization, the mating of snails with different chirality can cause genital mismatch (Asami, Cowie, & Ohbayashi, 1998;Gittenberger, 1988;Lipton & Murray, 1979). This mismatch is supposed to result in frequency-dependent selection against the chirality type with lower frequency (Johnson, 1982). Yet, there are examples of chirally dimorphic populations of snail species (subgenus Amphidromus) that exhibit no difficulty of interchiral mating (Nakadera et al., 2010;Schilthuizen et al., 2007;Sutcharit, Asami, & Panha, 2007;Sutcharit & Panha, 2006). The role of chirality reversal in speciation, that is, whether different chiral types cause sexual isolation ("single gene speciation"), has been a matter of debate (Hoso et al., 2010;Richards et al., 2017;Ueshima & Asami, 2003;Yamamichi & Sasaki, 2013). Van events involving S. contraria, S. scyphus and S. trapezensis. The results exemplify the importance of integrating both mitochondrial and nuclear sequence data in order to complement morphology-based taxonomy, and they provide further evidence for hybridization across distantly related lineages in land snails.

K E Y W O R D S
12S, 16S, COI, histone genes, molecular phylogeny, taxonomy Batenburg and Gittenberger (1996) investigated factors influencing the ease of fixation of a change in coiling direction. They found that the population size, number of invaders, and dominance of the mutant chirality gene are of higher importance than maternal effects and mobility. They also emphasize that the shift in coiling direction usually does not prevent mating in high-spired snails, but that it often leads to reproductive isolation in snails with globular shells (Batenburg & Gittenberger, 1996). While dextral species in temperate regions are by far more common (often exceeding 99%), in Turkey, sinistral land snail taxa reach 5.5% (excluding the Clausiliidae which are mostly sinistral; Gittenberger, Hamann, & Asami, 2012).
Five out of the 45 orculid taxa are sinistral: Orculella heterostropha commagenensis (Neubert, 1988), Orculella heterostropha heterostropha (O. Boettger, 1905) Orculella menkhorsti sinistrorsa Hausdorf, 1996, S. inversa, and S. t. contraria. So far, phylogenetic relationships of these taxa have not been investigated with molecular genetic methods. Schileykula t. contraria has been reported only from three close sites around which the dextral S. t. trapezensis occurs, raising the question of whether they are actually separate species.
In the present study, we tested the morphology-based systematics of Schileykula by molecular phylogenetic analyses using DNA sequences of three mitochondrial (mt) genes and one nuclear (nc) sequence region. We included almost all extant species (except for S. robusta) and most subspecies of Schileykula. We also included several specimens of Sphyradium doliolum (Bruguière, 1792) from the monotypic genus Sphyradium Charpentier, 1837, the sister group of Schileykula (Harl et al., 2017). Sphyradium doliolum has a wide distribution from the Pyrenees in the west to northern Iran in the east, partially overlapping with that of Schileykula (Hausdorf, 1996).
Besides the establishment of a DNA-based phylogeny of Schileykula, we addressed the following specific questions: (a) Is the polytypic S. trapezensis, with four morphologically distinct subspecies, monophyletic? (b) Is S. nordsiecki a part of the S. scyphus group or is it a distinct species as the anatomy suggests? (b) Did hybridization occur between nearby occurring Schileykula taxa in north-eastern Turkey, the most speciose region within the range of the genus? (d) Do the two sinistral taxa represent independent lineages? That is, did a single reversal occur in the common ancestor of the two species or were there two independent reversals?

| Taxon sampling
We performed DNA analyses on 56 specimens of 14 Schileykula taxa. Only S. robusta and three subspecies of S. scyphus, S. s. cilicica Hausdorf, 1996, S. s. crassa (Pilsbry, 1922, and S. s. erecta, were not included. For out-group comparison, we included 31 individuals of Sp. doliolum from a variety of places in its wide distribution including Austria, Hungary, Greece, Croatia, Romania, Slovenia, and Turkey. Before they were broken for tissue preparation, shells of all specimens were pictured with a WILD MAKROSKOP M420 and a NIKON DS Camera Control Unit DS-L2 in frontal, lateral, apical, and umbilical view. Remaining parts of specimens are stored in the tis-

| Distribution maps
Distribution maps of all Schileykula taxa were prepared using ArcMap Desktop 10.0 and manually edited in Adobe Photoshop CC v.2015.01 (Adobe Systems). Distribution data originate from Hausdorf (1996), Neubert (1993), and the present authors. Figure 1 shows a rough overview on the known localities of Schileykula taxa as well as the sampling sites of specimens investigated in the present study. A detailed view of distribution ranges in north-eastern Turkey is provided in Figure 2.  annealing temperatures is listed in Table 2. LGC Genomics. If the complete H4/H3 sequence could not be obtained by direct sequencing, due to the large fragment size and/ or the presence of different alleles in the variable IGS region, PCR products were cloned. In case that specimens were heterozygous and the IGS sections differed in more than a single insertion/deletion (indel), up to four clones were sequenced. Before cloning, PCR products were excised from 1% agarose gels and purified using the QIA quick Gel Extraction Kit (QIAGEN), extended by A-endings with the DyNAzyme II DNA Polymerase (Thermo Fisher Scientific) and

| PCR and cloning
then cloned with the TOPO-TA cloning kit (Thermo Fisher Scientific).
Plasmid preparation and sequencing of the cloned fragments were performed at LGC Genomics using M13 universal primers.

| Nucleotide sequence analyses
The raw forward and reverse sequences were manually aligned in BioEdit v.7.1.3 (Hall, 1999) and checked for errors. Prior to the phylogenetic analyses, sequences of the separate data sets were aligned, which was straightforward for the COI because there were no indels.

| Mitochondrial gene trees
In  In addition, we also provide the uncorrected p-distances in Table S1.

| Nuclear gene trees
Concerning the H3/H4 region, several individuals provided more

| Median-joining networks of H4, IGS, and H3
While the BI and ML trees were generated with the complete H4/H3 region, median-joining haplotype networks were calculated separately for the three sections (H4, IGS, H3) in order to visualize potential recombination products between different variants (Figure 4).
In particular, the placement of three clones of S. s. scyphus (5491 c1

| D ISCUSS I ON
We performed phylogenetic analysis on 18 species and subspecies of the land snail genus Schileykula using nc (H4/H3) and mt (COI, 12S,

16S) data sets, with several samples of the monotypic sister genus
Sphyradium as out-group. It is worth mentioning that the mt phylogeny roughly reflects anatomical characters: The taxa grouped in the two main clades in the mt tree each share similar anatomical traits with respect to the size and shape of the penial cecum ( Figure S2).

| Hybridization
The nc phylogeny of Schileykula reveals complex patterns, which indicate at least two hybridization events between distinct taxa.

Subst. model (Akaike information criterion)
Alignments for phylogenetic tree inference  (Liao, 1999). The primary driving force for concerted evolution in tandemly repeated multigene families is probably intra-chromosomal homogenization, whereas inter-chromosomal genetic exchange is much rarer (Liao, 1999;Liao, Pavelitz, Kidd, Kidd, & Weiner, 1997;Schlötterer & Tautz, 1994). In two studies on the genus Orcula Held, 1837, the H4/H3 sequence section was analyzed in over a hundred specimens. Approximately 90% of the Orcula specimens were homozygous regarding the H4/H3 loci and only three provided more than two distinct variants Harl, Páll-Gergely, et al., 2014), which were consequently interpreted as alleles of one histone cluster and not as paralogues.
Another possible explanation for some sequence patterns would be ancestral polymorphism in the H4/H3 genes. However, the nc haplotypes of S. t. contraria and S. t. trapezensis, which cluster with S. scyphus, are very similar to each other and also to S. scyphus (Figure 3b).
Considering that these haplotypes were affected throughout time in the same way by mutations as those found in the other specimens, they should also be more distinct (have longer branches in the tree), which is not the case. The morphological similarity of S. t. trapezensis and S. t. contraria and the close vicinity of populations further support the assumption of introgression after secondary contact.
However, genome-wide nuclear sequence data would be required to confirm our assumption and to clarify the true extent of admixture between populations.
Geographically, hybridization between S. trapezensis and S. scyphus might have occurred particularly in the eastern Black Sea Region (Gümüşhane, Bayburt, Trabzon, and parts of Erzurum province), where their distribution ranges partly overlap. Hybridization might have occurred also in the north-easternmost part of Turkey (Erzurum and Artvin provinces), where five Schileykula taxa are distributed in a relatively narrow range (see Figure 2). Neubert (1993) reported hybrid specimens in populations of S. batumensis and F I G U R E 3 Phylogenetic trees reconstructed based on the alignments of the (a) combined mt COI, 12S, 16S nucleotide sequences and (b) the nc H4/ H3 nucleotide sequences. The scale bars indicate the expected number of substitutions per site according to the models of sequence evolution applied.
Black dots indicate nodes with high Bayesian inference posterior probabilities and ML bootstrap values (see figure). Specimens marked by an asterisk are sinistral. Cloned sequences in the nc tree are indicated by a "c" following the ID number S. t. acampsis, which are the only Schileykula taxa known to occur in sympatry over a broad range. We only sampled a few specimens and localities each. Therefore, although the sequence data did not indicate mixing between these two taxa, analyzing a larger sample covering a wider area could reveal hybridization between them and/ or other taxa.   (Hausdorf, 1996). Both the mt and nc phylogenies (Figure 3)  Schileykula sigma has been considered as an independent species because of its peculiar shell characters, in particular, the sigmoid formation of the columellar lamella (Hausdorf, 1996), the genital anatomy, however, is similar to that of S. scyphus (Páll-Gergely, 2011

| Origin of sinistral taxa
The two sinistral species do not form sister groups in the nc and mt trees, suggesting that the change of coiling direction from dextral to sinistral happened two times independently within the genus.
However, since our data indicate that hybridization affected several taxa (including S. t. contraria), it is possible that there was a single origin of sinistrality in this group, which has subsequently been obscured by hybridization events. Generally, Turkish Orculidae exhibit a comparably high proportion of sinistral taxa (Gittenberger et al., 2012). Sinistral snails are more frequent in high-spired taxa, and opposite-coiled specimens have less difficulties in mating than enantiomorphic pairs possessing flat or globular shells (Asami et al., 1998). Whether the changes in chirality might have contributed to the speciation of both S. inversa and S. contraria remains speculative.

| Classical taxonomy versus molecular phylogeny
The present study provides a phylogenetic basis for the systematics of Schileykula for the first time. Our DNA-based analysis allowed us to re-evaluate some pre-existing hypotheses on the systematics within Schileykula. However, this revised systematics should be further tested using a higher sample size and optimally analyzing genome-wide nuclear sequence data. Figure 5 shows selected specimens of the taxa studied.

| CON CLUS ION
The molecular genetic data presented here provide a first phylogenetic hypothesis based on a combination on mt and nc DNA sequences. Our results suggest some taxonomic revisions. Since S. trapezensis is polyphyletic, we propose treating the four subspecies as independent species. Schileykula nordsiecki was confirmed to represent an independent lineage distinct from S. scyphus.
Schileykula sigma was shown to be closely related to S. scyphus and is therefore treated as a subspecies of the latter. The sinistral taxa S. contraria and S. inversa did not emerge as sister lineages in the mt and nc trees, implying that the change of the coiling direction might have happened two times independently. The incongruences between mt and nc trees suggest at least two independent hybridization events involving S. contraria, S. scyphus, and S. trapezensis. However, in order to shed more light on the complex patterns, future studies should include a larger number of samples from more localities as well as they would benefit from the analysis of additional nc markers.

ACK N OWLED G EM ENTS
We are very grateful to Zoltán Fehér, Takashi  F I G U R E 5 Adult shells of orculid taxa (Schileykula and Sphyradium spp.) of which nucleotide sequences were obtained in the present study. Numbers below the photographs are individual identification codes of each specimen that correspond with those in Table 1. The picture of Schileykula aculeata shows an empty shell from the same locality, because only juveniles were available for the molecular genetic analyses

S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section at end of the article.   Hausdorf, 1996 (right) to highlight the differences of relative penial caecum sizes.