Revisited – the species of Tweeting vineyard snails, genus Cantareus Risso, 1826 (Stylommatophora, Helicidae, Helicinae, Otalini)

Abstract The generic allocation of Helix subaperta is clarified by using genetic data and morphological traits of the genital organs; its position within the hitherto monotypic genus Cantareus is corroborated. Further analysis of several specimens of Cantareus apertus from Algeria and Italy revealed that this taxon is composed of two species, C. apertus from Italy, and C. koraegaelius from Algeria. The morphological traits of the genital organs of all three species are discussed, and the definition of the genus Cantareus is amended. All three species confined to Cantareus are re-described, and the syntype specimen of H. aperta is illustrated.


Introduction
The hitherto monotypic genus Cantareus is currently placed in the helicoid tribe Otalini G. Pfeffer, 1930 (http://www.molluscabase.org/aphia.php?p=taxdetails&id=994951) (Neiber and Hausdorf 2015;Razkin et al. 2015). This clade embraces 12 genera of helicoid snails, among them Cornu Born, 1778 and Erctella Monterosato, 1894. Both genera contain species possessing shells that resemble each other to some extent, and they exhibit a similar morphology of their genital organs. The Sicilian genus Erctella was previously reviewed and re-described by Colomba et al. (2011), and resulted in the resurrection of three valid, narrowly endemic species of the group on north-western Sicily.
The tribe originates from the Maghrebinian radiation centre (Korábek et al. 2019), and contains a considerable number of species, many of them only randomly known. One of these problematic species is the enigmatic Helix subaperta Ancey, 1893, which is endemic for a relatively small mountain ridge in the Kabylie in eastern Algeria, the Djudjura Mountains. It has a shell that shows character states typical for both aforementioned genera: it resembles Cornu in its colouration but shows no malleate pattern, and Erctella in the shell form and the considerably developed ribs on the surface of the adult shell. Thus, the starting point for this paper was to clarify the correct taxonomic position of this species. However, adding supplementary specimens, and using genetic data available from other studies revealed that there is another and completely overlooked species living in northern Africa, which turns out to be a member of Cantareus, the Tweeting vineyard snail.

Taxon sampling
The specimens for this study were collected by the authors of the study, particularly by the senior author. Missing sequences for Erctella and Italian Cornu were added for the same specimens used by Colomba et al. (2011;. Freshly sampled animals were preserved in 80 % EtOH. The analysed specimens were removed from their shells, the genital organs were isolated and fixed on a wax-bed. The situs as well as the details of the interior lumina were photographed. Tissue samples were taken from those speci-  GQ402444 Colomba et al. (2011 mens and sequenced, the shells were photographed (in case they were not destroyed when extracting the animal). All shell photos were taken using a Leica M205 C microscope with the Leica DFC425 camera and the IMS Client (Imagic Bildverarbeitungs AG, Glattbrugg, Switzerland).

Phylogenetic analysis
DNA was extracted from a piece of foot muscle tissue using Qiagen Blood and Tissue Kit (Qiagen cat nr. 69506) and the QIAcube extraction robot (Protocol 430, DNeasy Blood Tissue and Rodent tails Standard). Our phylogenetic hypotheses were reconstructed using five phylogenetic markers (mitochondrial COI (657 base pairs (bp)), 16S (374 bp) and nuclear 28S (528 bp), H3 (304 bp) and ITS2 (909 bp)), resulting in a length of 2772 bp (see Table 1). The PCR included the following admixture: 2 µL template, 12.5 µL GoTaq (Promega) polymerase, 8.5 µL of nuclease-free water and 1 µL of both forward and reverse primer (10 µmol) respectively. In cases where the PCR signal was judged too weak, the reaction was repeated using 3 µL template DNA, 3 µL of the previous PCR product and 5.5 µL of nuclease-free water. The amount of GoTaq and primers stayed the same. The PCR was conducted using the following protocols: For COI, the admixture was first heated up to 95 °C for 1 minute (min), followed by 30 cycles of 30 seconds (s) at 95 °C, 30s at 52 °C and 30s at 72 °C, finishing with 3 min at 72 °C. For 16S, the protocol started with 2:30 min at 90 °C, followed by ten cycles of 30s at 92 °C, 30s at 44 °C and 40s at 72 °C, followed again by 30s at 92 °C, 40s at 48 °C and 40s at 48 °C. The protocol for 28S started with 1 min at 96 °C, then went into 35 cycles of 30s at 94 °C, 30s at 50 °C and 1 min at 72 °C, finishing with 10 min at 72 °C. The ITS2 protocol started with 1 min at 96 °C, followed by 35 cycles of 30s at 94 °C, 30s at 44 °C and 1 min at 72 °C, ending with 10 min at 72 °C. For H3 the admixture was first heated up to 95 °C for 3 min, followed by 40 cycles of 45s at 94 °C, 45s at 50 °C and 2 min at 72 °C, finishing with 10 min at 72 °C. The protocols for COI and H3 could be used for both markers. The PCR products were sequenced at the LGC Genomics GmbH (Berlin, Germany) and at Eurofins Genomics (Ebersberg, Germany) using their respective standard protocol. In total, 48 helicid specimens were used, chiefly from the genera Cantareus, Cornu, and Erctella. Five specimens, belonging to Helix pomatia, Massylaea vermiculata, and Massylaea constantina were used as outgroup. Sequences received from LGC and Eurofins were imported into the Geneious 5.4.7 software (Kearse et al. 2012). The forward and reverse sequences for each gene and individual were combined and edited. For each marker, sequences were aligned in Geneious using the MAFFT multiple sequence alignment plugin version 1.3.6 (based on MAFFT v7.308; Katoh et al. 2002, Katoh andStandley 2013), letting the program choose the appropriate algorithm. The sequence length of each alignment was standardized to the length mentioned above. The alignments were concatenated using the "Concatenate sequences or alignments" function in Geneious.
Topologies were estimated using two different phylogenetic methods: Maximum Likelihood (ML) and Bayesian inference (BI). The five markers were set as partitions in both of these methods, using a distinct model for the third codon in protein-coding genes (COI, H3). The Maximum Likelihood (ML) topology was estimated using the RAxML 7.2.8 (Stamatakis 2014) plugin of Geneious with the GTR gamma Nucleotide model and 1'000 bootstrap replicates.
The Bayesian tree, which was used as a basis for the combined tree ( Fig. 1), was reconstructed with MrBayes 3.2.6 (Huelsenbeck and Ronquist 2001) using the mixed substitution model (which incorporates model testing into the MCMC), invgamma rate variation, a Markov Chain Monte Carlo (MCMC) chain length of 10,000,000 generations, a subsampling frequency of every 4,000 generations, the first 100,000 generations were discarded as burn-in, four heated chains and a chain temperature parameter of 0.2. Calculations were performed on the UBELIx (http://www.id.unibe. ch/hpc), the HPC cluster at the University of Bern.

Molecular taxonomy
The Bayesian and RaxML reconstructions yielded the same topology for all species involved and are shown in Fig. 1.
All three genera treated here in the analysis split in monophyletic lineages, and the nodes on the generic level have high support values. The species H. subaperta turned out to be a member of Cantareus rather than of Cornu, as could be expected by the colour pattern of its shell. The specimens from northern Africa, which had been identified as C. apertus so far, form a well-supported (95/1) lineage separate from all Italian specimens available in the study. For this species, the nominal taxon name Helix aperta var. globulosa Bourguignat, 1863 from Constantine is available. It should be stressed that the specimen MVHN_2013 (Razkin et al. 2015) originates from Djelfa, a city in the southwest of Tizi Ouzou (shell not seen by the present authors). The Italian specimens of C. apertus show some genetic differentiation as they split into two major clades; interestingly, the two specimens from Amantea in Calabria (NMBE 560941) occur each in one of these lineages. The addition of nuclear markers in Erctella consolidated their topology on species level with high support values.
Notes. In Table 3, the most important character states of the shells and the genital organs of the Otalini subgroup according to Razkin et al. (2015: 108, fig. 2) including the genera Cantareus, Cornu, and Rossmaessleria are shown. Within the Otalini, Cantareus, Cornu, and Erctella share the synapomorphy of a globular to slightly conical shell, other genera in the tribe tend to have more flattened shells (character 1). In all other shell traits, there is no apomorphy that discriminates between Cantareus, Cornu, and Erctella, using COI, 16S, H3, 28S, and ITS2. Table 3. Character matrix including the genera Cantareus, Cornu, Erctella and Rossmaessleria. 1. Last whorl of the shell: 0: occupying more than two thirds of the shell height, 1: occupying less than two thirds of the shell height -2. Teleoconch colour patterns: 0: none, 1: up to 5 spiral bands, 2: a reticulate pattern -3. Teleoconch surface: 0: smooth, sometimes with longitudinal riblets and growth lines, 1: granulated, 2: with wrinkles, 3: strongly wrinkled and irregularly reticulated, 4: ribbed -4. Penis form: 0: short, 1: elongate -5. Epiphallus length: 0: as long as penis, 1: at least three times the length of penis -6. Penial flagellum: 0: twice the length of the epiphallus, 1: clearly more than twice the length of the epiphallus -7. Penial lumen: 0: with numerous crests; 1: smooth -8. PP1 0: not shifted laterally, 1: shifted laterally, leaving a small pore as a connection between epiphallus and penis near its base -9. PP2: 0: pp2 reduced to a septum, 1: reduced to a annular pad, 2: pp2 present -10. Diverticulum: 0: as long as vesicle stem + vesicle, 1: slightly longer than vesicle stem + vesicle, 2: much longer (twice and more) than vesicle stem + vesicle, V: length variable -11. Atrial stimulator: 0: small, 1: medium, 2: large. and Erctella on generic level. On the level of the genital organs, the three genera share the synapomophy of the simple pore connecting epiphallus and penial chamber (character 8), while Rossmaessleria shows the plesiomophic state with two functional penial papillae. In Cornu, the flagellum is much longer than in the other genera. The phylogenetic value of this character state is not clear within the Otalini, within the Helicini, it is considered a plesiomorphy (Neubert 2014). Other character states like ratios in the bursa copulatrix complex (character 10). A massive atrial stimulator can be found in Cantareus and Cornu, while in Erctella, it is relatively small (character 11). Large and massive stimulators are found in many taxa of Helicidae, so a reduction of this system is here interpreted as an apomorphic character state.
Remarks. The change of the status of Cantareus from a monotypic to a polytypic genus causes some nomenclatorial problems. The type species of the genus is Helix naticoides Draparnaud, 1801 from France, which so far has been considered a synonym of Helix aperta Born, 1778, with the specimen preserved in the Born collection in the NHMW as the name bearing syntype of aperta (Fig. 2).The origin of Born's specimen is unknown, and there are almost no shell morphological differences to the Algerian lineage, which proved to represent a separate species (Fig. 1). The correct origin of Born's specimen could probably be clarified genetically by applying NGS methods using shell fragments of the syntype specimen, but this is beyond the scope of this paper.
Anticipating a north African origin of the syntype NHMW-MO 14005 by fixing its type locality in Algeria ends up in a chaotic rearrangement of species names in the group. For Europe, the name naticoides would be reactivated with its last use as an accepted species in 1850 (!). The north African species would then be named apertus contradicting 170 years of permanent use. By fixing the use of the name Helix koraegaelia Bourguignat in Locard, 1882, to the north African lineage, this problem is resolved, and the stability or universality of names used in zoology is guaranteed.

Cantareus apertus (Born, 1778)
Description. Shell thick, medium sized if compared to other helicid species, with a depressed spire and a large last whorl occupying more than two thirds of the complete height of the shell; protoconch small, consisting of 1.5 smooth whorls; teleoconch consisting of approximately 4 whorls, separated by a deep, sometimes crenulated suture; basic colour of teleoconch greenish-brownish, often with longitudinal yellow streaks and a few scattered zig-zag markings; surface of teleoconch smooth, but also often covered by low longitudinal riblets; aperture almost perfectly rounded, enormously large, old specimens with an inconspicuous whitish lip; umbilicus always completely closed.
Genital organs: penis short, club-shaped, epiphallus short, of the same length as penis, mrp attaching in the distal third of epiphallus or even closer to penis; flagellum twice the length of the epiphallus; atrial and penial lumen with numerous crests, penial chamber lumen is wrinkled, pp2 a short broad papilla with a central perforation structured by thick annuli to almost completely reduced forming a septum; pp1 a blind papilla, in a central position inside the penial chamber, elongate, sometimes with a broadened tip; epiphallial pore in a lateral position; distal epiphallial lumen with six broad pilasters, the proximal lumen with elongated ridges.
Vagina short, stem of pedunculus thickened and short, diverticulum slightly longer than the vesicle stem + vesicle, longer than the flagellum; glandulae mucosae longer than the dart sac, with a thickened basal part and two subsequent ramifications, tubules thin and weak, less than 10 tubules per stem; atrium with a medium sized stimulator flap.
Remarks. The anatomy of the genital organs of C. apertus has been investigated by several authors, for example Hesse (1919), Germain (1930) and Giusti et al. (1995). Schileyko (2006Schileyko ( : 1801Schileyko ( , fi.g. 2308 presented also details of the penial lumina. In his picture of the genital organs of an animal collected in the surroundings of Pisa, Italy, he misinterpreted the morphology of the epiphallial papilla (pp1) suggesting that it was a functional papilla as in many other helicid genera (the illustrated shell comes from Arles, France, and thus does not belong to the dissected specimen). Secondly, in his specimen, the penial papilla (pp2) was completely reduced, so only the perpendicular wall forming the basis of pp2 was left. This led to the misapprehension that in the genus Cantareus, this papilla is reduced, and only a "septum" is left in the place of the papilla.  Description. Shell medium sized to large, thin, globose with a broad to relatively acute conical spire; protoconch whitish, large, with a diameter of up to 6 mm and 2.5 smooth whorls; basic shell colour olive yellowish with up to five separate brown spiral bands; teleoconch covered by a dense granulation, sometimes accompanied by very fine, deep spirals; teleoconch usually covered by irregularly arranged riblets of even ribs, usually stronger around the umbilical area; periostracum thick, often preserved on the shell in small patches; in eroded shells, ribs and riblets whitish; aperture large, elongate oval, slightly thickened forming a lip callus, with a parietal callus in fully adult specimens; aperture whitish inside, with the spiral bands shining through the thin shell; peristome sharp; umbilicus closed, periomphalum covered by a thickened calcareous layer.
Genital organs. Penis short, epiphallus reaching at least three times the length of penis, mrp attaching in the distal third of epiphallus; flagellum twice the length of the epiphallus; penial lumen smooth, pp2 a short broad papilla with a central perforation structured by thick annuli, pp1 a blind papilla, the epiphallial pore in a lateral position; distal epiphallial lumen with broad pilasters, the proximal lumen with elongated ridges.
Vagina short, stem of pedunculus thickened and short, diverticulum longer than the vesicle stem + vesicle, longer than the flagellum; glandulae mucosae longer than the dart sac, with a thickened basal part and two subsequent ramifications, tubules thin and weak, less than 10 tubules per stem; atrium dominated by a massive stimulator.   Figures 10-14. Cantareus subapertus. Anatomical details of the genital organs; specimen collected at Ighil Bourmi, leg. H. Bouaziz-Yahiatene, NMBE 550458 10 situs of the genital organs, 46 mm total length 11 partly everted genital atrium with the atrial stimulator 12 distal penial tube with pp2 13 penis and epiphallus completely opened showing both papillae, and the internal structure of the penial chamber and the epiphallus 14 detail of the penial lumen; note: the needle represents the epiphallial canal, with pp2 bent upwards to show the ending of the canal. Abbreviations: ag = albumen gland; as = atrial stimulator; div = diverticulum; ds = dart sac; ep = epiphallus; f = flagellum; gm = glandulae mucosae; hd = hermaphroditic duct; mrp = musculus retractor penis; p = penis; pp1 = penial papilla 1; pp2 = penial papilla 2; spo = spermoviduct. All figures not to scale.

Distribution.
As far as known, this species is restricted to the Djudjura Mts., where it inhabits quite high altitudes. It also occurs in the northern promontory of this mountain ridge.
Remarks. In the description of Helix subaperta, Ancey (1893) mentioned in a footnote that he already shared this species under the name H. mazzuliopsis with his correspondents. Pilsbry's shells (1893) were purchased from a shell dealer (see label in Fig. 8E) bearing the name H. mazzulopsis, which he consequently used! Moreover, Pilsbry remarks: "I have been unable to find any description or mention of this form in the literature"; thus, H. mazzulopsis cannot be considered an emendation or an incorrect subsequent spelling of H. mazzuliopsis. Strictly speaking, he introduced a new name, and corrected his error two years later in the "Index to Helices" (Pilsbry 1895: 120)  Remarks. Cantareus koraegaelius is a species that is almost inseparable from its congener C. apertus. This also explains why Bourguignat recorded this species from the complete distribution area of the latter species (and including the Algerian lineage). All specimens left in Bourguignat's collection originating from the localities mentioned are syntypes of Helix koraegaelia. Thus, the type lots contain two different species. To unambiguously fix the use of this specific name, we herewith select the single specimen MHNG-MOLL 117907 from "Djemaa N'Saharidj" in Algeria as lectotype. This locality in Tizi Ouzou is very close to the places, where the anatomically and genetically well-known specimens (see below) have been recorded. The application of the name H. koraegaelia is herewith restricted to specimens exhibiting the character states as explained in this paper forming a new species.
Description. Shell thin, medium sized, with a relatively elevated spire and a large last whorl occupying more than half of the complete height of the shell; protoconch medium-sized, consisting of 1.5 smooth whorls; teleoconch with approximately four whorls, separated by a deep, crenulated suture; colour of teleoconch brownish, surface of teleoconch covered by low longitudinal riblets, which are more prominent below the suture, disappearing on the last whorl; aperture rounded, very large, with an inconspicuous whitish lip; umbilicus always completely closed.
Genital organs. penis elongate, club-shaped, epiphallus as long as penis, mrp attaching in the distal third of epiphallus or even closer to penis; flagellum twice the length of the epiphallus; penial lumen smooth; pp2 a broad acute conical papilla with a central perforation structured by thick annuli; pp1 a blind papilla with a slightly broadened apex, the epiphallial pore in a lateral position; atrial and penial lumen with numerous strong crests; distal epiphallial lumen with six broad pilasters, the proximal lumen with elongated ridges.
Vagina short, stem of pedunculus thickened and short, diverticulum extremely longer than the vesicle stem + vesicle, and almost three times longer than the flagellum; glandulae mucosae longer than the dart sac, with a thickened basal part and two   Distribution. the two genetically identified specimens originate from Eastern Algeria.
Remarks. The description of the genital organs is based on the specimen NMBE 534199 from Tigzirt; unfortunately, the other specimen from Draa Ben Khedaa was subadult with only partially developed genital organs.
It is almost impossible to define differences in shell morphology between this new species and C. apertus. In the two genetically identified specimens, the protoconch of C. koraegaelius seems to be larger than in C. apertus, and the shell colour is more or less uniformly brown without any yellowish or greenish streaks. However, the morphology of the genital organs is in fact different: the large triangular pp2 is strikingly different to all what is known so far from the Italian C. apertus, where pp2 is very short to almost completely reduced, so that a "septum" is left.
Cantareus species (?) Fig. 9 Helix aspersa: Kobelt 1888, Iconographie (2)  This specimen was collected by Kobelt in the Gorge d'Isser; it lacks the malleation typical for Cornu aspersum, and thus is here considered to rather constitute a species in Cantareus than in Cornu. However, it also lacks the riblets on the teleoconch, but also has the typical granulation on the whorl exactly like in the specimens from the Djudjura Mts. This form might represent another species close to C. subapertus, but preserved specimens from the canyon are needed to decide about its status. This form might be a separate species endemic to the Gorge d'Isser.

Discussion
The main results of this work consist of the allocation of H. subaperta in the genus Cantareus, and the recovery of a third species in Cantareus, i.e., C. koraegaelius. The minute granulation of the teleoconch, which is a new shell morphological trait for Cantareus, can also be found in other Helicidae like for example Helix Linnaeus, 1758 (Neubert 2014), and thus represents a plesiomorphic character state above the species level. Cantareus koraegaelius can almost be considered a cryptic species, because its shell does not deviate in any major trait from the shells of its sibling species, C. apertus (Born, 1778). The separation between C. apertus and C. koraegaelius is mainly based on the clear genetic data, and all traits discussed to separate the shells of the two species have currently to be considered as first impressions. Only the shell morphological and anatomical study of a larger number of specimens from the Algerian clade can corroborate the stability of the characters discussed here. It also has to be proven whether or not C. apertus is also present in Algeria, which might well be possible. The distance between Tizi Ouzou and Constantine is > 200 km as the bird flies, so it can be estimated that C. koraegaelius constitutes a more widespread species than C. subapertus, which in fact seems to be a small-range endemic species restricted to submontane to alpine environments of the Djudjura Moutains.
The data presented here suggest the need for a more careful investigation of the phylogenetic relationships among the populations of C. apertus from Sicily and southern Italy. Recent studies on species with a wide Mediterranean distribution like Rumina decollata (Linnaeus, 1758), Massylaea vermiculata (O. F. Müller, 1774) and Cornu aspersum (O. F. Müller, 1774), have shown a remarkable genetic divergence (Prévot et al. 2013;Bouaziz-Yahiatene et al. 2017;Sherpa et al. 2018). This will also hold true for C. apertus, which is probably introduced by human activities to other Mediterranean areas like southern France and Greece. Neubert and Bank (2006: 105) argue that the transformation of the papilla system represents a synapomophic character state for an "Eobania group" based on the state of knowledge of this time. Walther et al. (2016: 399) remark that this is wrong because Massylaea [= Eobania] is found on a cluster separate to the Rossmaessleria/Cornu/ Cantareus lineage (Razkin et al. 2015: 108, fig. 2). Consequently, this transformation occurred convergently within Otalini. We fully concur with this statement, although we must note that Razkin et al. (2015: 114) also states "...Otalini in the concatenatedgene tree and Helicini in the nuclear rRNA tree were not supported by NJ analysis". The current research on the Otalini will hopefully include the missing genera, and deliver enough data to stabilise the phylogenetic structure of the tribe.