Granopupa in Iran, monophyly, and the fossil Granariinae (Gastropoda, Pulmonata, Chondrinidae)

Abstract Indisputable Chondrinidae, Granariinae species, characterized by shell shape and apertural dentition, are known from Eocene deposits to the Recent. The generic classification of the extant species is based on conchological, anatomical and molecular data that are available now for most of the known species, including ‘Granaria’ persica as a representative of the once problematic group of so-called eastern Granaria species. According to molecular and anatomical characters, these eastern species have to be classified with Granopupa granum in Granopupa. Graniberia gen. n. is introduced for Granaria braunii on the basis of molecular and conchological data. For the pre-Pleistocene species, two generic names are equally well available now, viz. Granopupa and Granaria. Shell characters only do not enable a decision here. For the sake of nomenclatorial stability we propose to use Granaria for these species. Because both molecular and anatomical data most likely will never be known for the fossils, it will remain unclear whether the combined extant and extinct Granaria species form a monophyletic group.


Introduction
The genus-group taxa of the Chondrinidae Steenberg, 1925, are currently characterized by conchological, anatomical, and molecular characters (Gittenberger 1973, Kokshoorn and Gittenberger 2008, 2010).These data were not available for all the species, however, so that not all could be classified accordingly.
The extant Granopupa granum (Draparnaud, 1801), and Granaria Held, 1837, species, and all the fossil chondrinids known from Eocene to Pliocene, have similar shell shapes and, what is more distinctive, the same characteristic arrangement of the apertural teeth, i.e. the palatalis inferior is more prominently developed than the other palatals.The extant so-called Granaria species show a disjunct distribution, with a western group of four species occurring in Europe and an eastern group of three species in the Arabian peninsula and Iran.The western species, viz.G. frumentum, G. variabilis, G. stabilei and G. braunii, are relatively well-known, whereas the eastern group, viz.G. lapidaria, G. persica and G. arabica, was known from shells only.
The shells of Granopupa granum, measuring 3.1-6.0× 1.4-1.8mm, are smaller than those of the European Granaria species, measuring 5.6-9.3 × 2.1-2.6 mm in the smallest, i.e.G. stabilei, and 6.7-19.0× 2.6-4.5 in the largest species, i.e.G. variabilis (see Gittenberger 1973).The generic classification of the species from the Middle East, with shells measuring 4.9-5.5 × 2.1-2.2mm (see Gittenberger 1973), was questionable.They could be either relatively small Granaria, or large Granopupa, because an obvious gap in sizes does not exist.The use of the generic name Granaria for the extant species occurring in the disjunct eastern part of the alleged range of that genus, was based on tradition and on the lack of a clearly preferential alternative.
For the chondrinids from before the Pleistocene (Höltke and Rasser 2013) neither molecular nor anatomical data are likely to ever be known, so that their classification has to be based on shell morphology only.
Recently, one of us (U.B.) collected together with empty shells a live specimen of 'Granaria' persica.As a consequence, both anatomical and molecular data are available for that species now.Therefore, a revised classification of this species is proposed in this article.Additionally, the generic classification of the remaining, so-called Granaria species is dealt with.

Material and methods
A single live animal of 'Granaria' persica was collected by Ulrich Bößneck in Iran, province of Hormozgan, Banooband, Genu Mtn, at 740-1000 m altitude (Fig. 1).The specimen was transferred into ethanol 70%.During dissection, the proximal part of the genital tract was isolated and coloured with organic cochineal dye, subsequently hardened in ethanol 97%, then cleared in Euparal essence and eventually fixed in Euparal as a genital slide.Serial sections were not made, so that the structure of the lumen can only be described by studying the transparent parts of the genital slide with a  (Gittenberger, 1973).Iran, province of Hormozgan, Kuh-e Bakhun, large valley, 980 m alt., 27°55'11"N 56°41'24"E, washed ashore; U. & K. Bößneck & A. Saboori leg.Shell height 5.2 mm (RMNH 336351).Scale bar 1 mm.Photograph by DSJG.regular microscope.The buccal mass was partly dissolved in KOH to isolate the radula, which was cleaned, mounted on a stub and photographed with a SEM.
The material that is used here is housed in Naturalis Biodiversity Center, Leiden, The Netherlands.
DNA was extracted with a DNeasy blood & tissue kit (Qiagen).COI and 16S were amplified using the procedure described in Kokshoorn and Gittenberger (2008).Products were sequenced in both directions (using the same primers) at BaseClear (The Netherlands) and edited in SEQUENCHER 5 (Genecodes Corp.).Sequences for 'Granaria' persica have been deposited in GenBank (accession numbers KT948999 and KT949000 for COI and 16S, respectively).Datamatrices with relevant reference sequences from GenBank were constructed in Geneious PRO 7.0.6.Because no COI sequence is available for Granaria variabilis it was coded as missing data.Both datasets were aligned with MAFFT v.7.017 (Katoh and Standley 2013) using default settings.Conserved regions of the 16S alignment were selected with GBLOCKS v. 0.91b (Castresana 2000).PARTITIONFINDER (Lanfear et al. 2012) was used to check for the best partitioning scheme (COI codon positions and 16S were considered as potential partitions) and substitution models.The translated amino acid sequene of COI was added as a fifth partition for a Bayesian phylogeny reconstruction.None of the suggested partitions could be combined and GTR+G, GTR+G, HKY+I+G, GTR+I+G and aa mixed were specified for COI codon position 1, 2, 3, 16S and the COI amino acid partition, respectively.A phylogenetic analysis was carried out with MRBAYES (Ronquist and Huelsenbeck 2003) v.3.2.3 (10 M generations, 2 runs, 4 chains) hosted on the CIPRES science gateway (Miller et al. 2010).Trees were sampled every 1000 generations, the first 2500 trees were discarded as burnin (relburnin = yes, burninfrac = 0.25).To compare and further explore the COI and 16S datasets, both were analysed separately as well (see Supplementary information).Except for the omission of Granaria variabilis (for which no COI data are available) the MrBayes analysis (partitioning and selected models) for COI was identical to that for the concatenated dataset.For 16S the complete sequences (no characters omitted) were utilized.The selected model again was GTR+I+G.

Results
In 'Granaria' persica, as in the other chondrinid species, the male part of the genital tract forms a loop because the proximal part of the vas deferens is fixed to the genital atrium (Gittenberger 1973).A prominent flagellum, as is present in both the Granaria and the Solatopupa species, is lacking.The male loop can be subdivided in five parts, which differ in the structure of the lumen and slightly in width; the transitional sites are more or less clearly distinguishable by irregularities in width or curvature of the duct.The segments are described from proximal (starting from the body wall) to distal.
The most proximal segment of the loop, i.e. the penis (Fig. 2: 1), has a muscular wall with very fine, transverse striae and a short ridge in the distal third of the lumen.The adjoining epiphallic part (Fig. 2: 2) has a thin wall and a regular transverse structure, maybe with small papillae in the lumen.It is followed by the narrowest part of the loop (Fig. 2: 3), without any regular, luminal structure.The next part (Fig. 2: 4) is  (Gittenberger, 1973).Iran, province of Hormozgan, Genu Mt., Banooband, 740-1000 m alt., 27°22'01"N 56°09'45"E, dry rocky limestone slope with little vegetation; U. Bößneck leg.Genital tract.Abbreviations: b, bursa copulatrix; pd, pendunculus; r, retractor muscle; v, vagina; vd, vas deferens.The figures 1-5 refer to the segments of the male loop that are mentioned in the text.Scale bar 1 mm.
clearly broader again; the lumen has relatively large papillae, which gradually pass into a more transverse arrangement.The most distal part of the loop (Fig. 2: 5) has a thick wall with very fine transverse and longitudinal striae, and could be considered a part of the vas deferens; the longitudinal striae can be followed over some distance also more proximally.
The radula of 'Granaria' persica has rows of teeth with a tricuspid central tooth, accompanied by adjoining bicuspid teeth, and teeth with more cusps, towards the margin of the radular ribbon.In the specimen that could be studied, the central tooth shows some individual irregularities.It is accompanied by 6 bicuspid teeth; from tooth 7 on, the side cusp is split into two, and more marginally in more, irregular, smaller cusps (Fig. 3).The separate molecular analyses did not result in entirely congruent results (see Supplementary Information), so that the summarizing cladogram that is presented here (Fig. 4) has uncertainties in it.The phylogenetic relationships of all the genera remains unresolved.All reconstructions indicate, however, that 'Granaria' persica is most closely related to Granopupa granum.The position of G. braunii is unclear, but none of the reconstructions assigns that species to a clade with G. frumentum or G. stabilei.
For data regarding the European representatives of the chondrinid species, we refer to Gittenberger (1973) and Kokshoorn andGittenberger (2008, 2010).

Discussion
For genera and species of the extant Granariinae we refer to Kokshoorn and Gittenberger 2010.DNA sequence data from this study do not allow (and were neither intended) to test for the monophyly of Granariinae or Chondrininae.
In the Granaria species of which the genital tract had been studied in some detail (Gittenberger 1973), viz. the type species G. frumentum (Draparnaud, 1801), G. variabilis (Draparnaud, 1801), G. stabilei (E.von Martens, 1865), and G. braunii (Rossmässler, 1842), the male loop is provided with a prominent flagellum.In these species, the lumen of the entire proximal half of the loop, which may be considered homologous with the penis, has some longitudinal ridges only, which continue into the flagellum; before the transition into the vas deferens, the lumen is covered with many fine papillae.The Solatopupa species, which differ in shell shape and the structure of the radula, have the same bauplan of the genital tract (Gittenberger 1973).
The radula of 'Granaria' persica has the bauplan that is considered plesiomorphic because it is known from Granaria, Granopupa, Abida and snails of many other pulmonate genera that are not feeding on algae or lichens and are not obligatory rock-scraping (Gittenberger 1973, Breure andGittenberger 1982).The radulae of both Chondrina and Rupestrella have the apomorphic rock-scraping condition, i.e. a series of virtually identical unicuspid teeth in the central part of the rows of teeth (Breure and Gittenberger 1982).
In 'Granaria' persica, there is no flagellum and, according to the luminal structure, the penis is restricted to the proximal third of the male loop; the segment of the loop before the vas deferens is devoid of small papillae.According to the structure of the genitalia, G. persica and G. granum are sistergroups and, as a consequence, should be considered congeneric.This view is convincingly supported by the molecular phylogenies (Fig. 4, Suppl.material 1, 2; pp.1.0, 1.0, 0.86), which also show Granopupa granum and G. persica as sistergroups.We suggest to classify in Granopupa the three chondrinid species from the eastern part of the range of the family, that were classified in Granaria by Gittenberger (1973), and considered closely related, viz.Granopupa arabica (Dohrn, 1860), Granopupa lapidaria (Hutton, 1849) and Granopupa persica (Gittenberger, 1973).
The Iberian 'Granaria' braunii does belong to neither the otherwise monophyletic group Granaria (Fig. 4, Suppl.material 1; pp.0.71, 0.75) nor to the clade Granaria-Solatopupa (Fig. 4, Suppl.material 2; pp.0.64, 0.98), as defined above.Morphologically 'G.' braunii cannot be distinguished from Granaria and Solatopupa on the basis of the structure of the genital tract, whereas the apertural armature of the shell has the Granaria & Granopupa bauplan with a most prominent palatalis inferior.Apart from that, however, 'G.' braunii is not particularly similar to any of the other chondrinids, and therefore, short of molecular data, its closest relative was considered unknown by Gittenberger (1973: 62).The molecular phylogenetic analyses that could be performed now (Fig. 4, Suppl.material 1, 2) indicate that 'G.' braunii should not be classified with Granaria (i.e. G. frumentum, G. variabilis and G. stabilei).Hence we introduce a monotypic genus for this species.
Remarks.The three extant Granaria species, the fossil taxa that are currently classified with Granaria, and the four known Granopupa species, all differ from Graniberia braunii in the characters mentioned in the diagnosis.
In all Granaria and Granopupa species and in the other Chondrinidae species with an apertural dentition that is not reduced, viz.several Abida and Chondrina species, the infracolumellaris is clearly visible.Therefore, an obsolete infracolumellaris as in Graniberia braunii, is considered the apomorphic character state.For both the location of the palatal lamellae and the prominence of the apertural lip this is also concluded.A similar reasoning is accepted here, with only the marginal note that very few Abida and Chondrina species have more or less clearly developed a thickened apertural border, whereas in only very few Abida species deep palatal folds occur.Two subspecies are currently recognized, viz.Graniberia braunii braunii (Rossmässler, 1842) and G. braunii marcusi (Gittenberger & Ripken, 1993).
Derivatio nominis.Graniberia after the distribution of a genus resembling Granaria in the Iberian peninsula.

Monophyly of the Granaria s.l. species
There are no conchological differences known to distinguish between Granaria and Granopupa.Even the disputable use of a difference in shell size in not tenable anymore.As a consequence, the generic classification of the fossil so-called Granaria species is problematic.Unless an overlooked diagnostic character of the shells will be discovered, it will remain impossible to decide in a particular case for either Granopupa or Granaria on the basis of shell morphology.The generic classification of the fossil chondrinids is problematic anyway, because the diversification of the chondrinid lineages may have taken place an unclear period of time after the Eocene, as is suggested by the fossil record, and by the application of a molecular clock model to the molecular phylogenetic reconstruction (Kokshoorn and Gittenberger 2008).Thus, the oldest 'Granaria' species, which is known from the Eocene, and the taxa from younger deposits [see Höltke and Rasser (2013)], may be ancestral to the species in the combined six genera that are now considered to constitute the Chondrinidae.The genus Granaria, as it is actually accepted in the literature for both extant and extinct species might be polyphyletic.

Figure 4 .
Figure 4. Bayesian phylogeny reconstruction of Chondrinidae based on a concatenated dataset of nucleotide (COI and 16S) and amino acid (COI) sequences.All depicted apertures are after Gittenberger (1973), except F and I which are after Kerney, Cameron and Gittenberger (1980).Aperture D is not of Granopupa persica, but of the closely related Granopupa arabica.Inset J shows a photo of Granaria grossecostata, after Höltke and Rasser (2013: fig.5) (with permission).