Diversity and taxonomy of Vietnamese Pollicaria ( Gastropoda , Pupinidae )

Species in the genus Pollicaria (Gastropoda: Pupinidae) are conspicuous members of the southeast Asian land snail fauna. Originally erected in 1856, both the genus and its constituent species have been reorganized multiple times with the most recent treatment published in 2013. Collections of Vietnamese Pollicaria during 2014 and 2015 raised questions of the utility of currently used diagnostic shell characters and identification keys in species identification. An examination of the authors’ collections, combined with museum specimens, suggested that at least three species of Pollicaria occur or have historically occurred in Vietnam. It is suggested that P. rochebruni is a senior synonym of P. crossei and treat both taxa as conspecific. A second species, P. mouhoti, was believed to only occur in Cambodia, Laos, and Thailand. A possible third species, based only on previous karyotypic work, is discussed. Our data further suggest that shell features such as color and size lack consistent utility in species-level identifications in Pollicaria.


Introduction
The genus Pollicaria was erected by Gould (1856) to include large pupinid snails of southeastern Asia possessing the following characters: a thick ovate shell, distorted and flattened in the adult; a circular aperture; and an internal shoulder or groove inside the palatal edge of the peristome (Pain 1974).The taxonomy and systematics of Pollicaria have remained in flux for many years.At the same time he erected the genus, Gould (1856) described Cyclostoma pollex Gould, 1856 and made it the type species of Pollicaria.Benson (1859) erected the genus Hybocystis with Megalomastoma gravidum Benson, 1856 from Burma as the type species, recognizing that C. pollex was synonymous with the previously described M. gravidum.Haines described C. myersii Haines, 1858 from Siam, and three more species were described in Hybocystis: H. mouhoti Pfeiffer, 1862 from the Lao Mountains of Cambodia; and H. elephas de Morgan, 1885 and H. jousseaumi de Morgan, P. mouhoti as a separate species.They found karyotype differences between all four species and found two distinct karyotypes in Vietnamese P. gravida; their "big brown shell" and "small orange shell" samples differed in the centromere position of one chromosome pair (Kongim et al. 2010:127).However, they did not karyotype any topotypic P. gravida from Burma.Most recently, Kongim et al. (2013) reassessed Pollicaria using a combination of shell, radula, and reproductive characters and recognized six separate species: P. gravida, P. myersii, P. mouhoti, P. elephas, P. rochebruni, and P. crossei.They correctly recognized that not all Pollicaria possess the shouldering inside the peristome, and assigned the different karyotypes of Vietnamese P. gravida (Kongim et al. 2010) to P. rochebruni ("big brown shell") and P. crossei ("small orange shell").The two species were diagnosed in their key based on shell size and color; P. rochebruni is large (>40 mm) with a brownish shell, while P. crossei is small (<35 mm) with a bright orange shell (Kongim et al. 2013).Both species are thus separate from the small yellowshelled P. gravida from Burma.Kongim et al. (2013) also provided revised distributions for all species and figured type and additional material.
While identifying Vietnamese land snails collected in 2015 by PMH and DVT, we found Kongim et al. (2013) to be problematic in identifying specimens of P. rochebruni and P. crossei.We noted specifically that smaller (30-35 mm) reddish-brown shells and larger (35-40 mm) orange shells were left unidentifiable by the key, and that species-specific characters of the key conflicted with the comparative characters listed by Kongim et al. (2013:28).Of additional interest were records of putative P. mouhoti from Lào Cai, Vietnam (Solem 1966), a species that Kongim et al. (2013) had restricted to Cambodia, Laos, and Thailand, and internet references to additional species in northeast Vietnam and southern China.In an effort to resolve the taxonomic and distribution issues presented by Kongim et al. (2013), we undertook a morphological and molecular examination of Vietnamese Pollicaria specimens, along with original descriptions and other literature relevant to the genus, to address two research questions.First, are P. rochebruni and P. crossei separate valid species?Second, beyond those specimens listed in Solem (1966), is there evidence of additional Pollicaria species in Vietnam?

Materials and methods
Specimens collected by the authors were vouchered at the Carnegie Museum of Natural History (CM), Philadelphia, Pennsylvania.Collection data for Vietnamese Pollicaria came from the following museums: Chulalongkorn University, Museum of Zoology, Bangkok, Thailand (CUMZ); Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts (MCZ); Florida Museum of Natural History, University of Florida, Gainesville, Florida (FLMNH); Museum National d'Histoire Naturelle, Paris, France (MNHN), Royal Bel-gian Institute of Natural Sciences (RBINS), and National Museum of Natural History, Smithsonian Institute, Washington D.C. (USNM).Specimens from MNHN, CUMZ, and RBINS were not made available, so we treated the identifications and locality data given in Kongim et al. (2013) as correct.All Vietnamese Pollicaria holdings from FLMNH, MCZ, and USNM were borrowed, identified by the authors based on Kongim et al. (2013), latitude and longitude coordinates estimated, and mapped with DIVA-GIS (Hijmans et al. 2008).
We also searched the Internet for shell dealers selling Pollicaria specimens from Vietnam and surrounding areas.While we lacked the funding to purchase specimens outright, we examined shell images and locality information from the dealer websites.Requests to formally include images and additional information in this manuscript went unanswered.We accepted that this information was not equivalent to that of vouchered specimens, but could be important in determining the diversity and distribution (Turney et al. 2015) of Pollicaria species.
Live Pollicaria specimens were collected in Vietnam by PMH, DVT, and local farmers in 2015, preserved in ethanol, and DNA was successfully sequenced from two of them.One individual collected south of Hoàng Liên National Park had a reddish-brown shell we identified as P. rochebruni; the other was collected from Cúc Phương National Park and had an orange shell that was consistent with P. crossei sensu Kongim et al. (2013).Using foot tissue we followed a standard CTAB and phenol/chloroform DNA extraction method (Saghai-Maroof et al. 1984), then amplified one mitochondrial (16S) and two nuclear (18S and 28S) gene fragments by PCR.Amplification conditions were the same for all fragments and included an initial denaturing step (94ºC for 120 s), 30 cycles (94ºC for 60 s, 50ºC for 60 s, 72ºC for 120 s), and a final extension step (72ºC for 10 minutes).We employed the following primer pairs: 16Sar and 16Sbr (Palumbi et al. 1991), 18S1F and 18S4R (Giribet et al. 1996), and 28SD1F and 28SD6R (Park and Ó Foighil 2000).Sequences were generated with the Sanger method on an ABI 3730xl by Genewiz (South Plainsfield, New Jersey, USA).Contigs were assembled in Geneious (BioMatters Ltd.) and combined with Cyclophoroidea data taken from GenBank (Table 1).Each gene fragment was aligned separately using MUSCLE (Edgar 2004) and the appropriate substitution model selected by Bayesian information criterion in IQTREE (Nguyen et al. 2015).The models chosen were HKY+G4 (Hasegawa et al. 1985) for 16S, JC+I+G4 (Jukes and Cantor 1969) for 18S, and TN+I+G4 (Tamura and Nei 1993) for 28S.The combined three gene dataset was analyzed under maximum likelihood in IQTREE allowing each gene partition to be optimized under its own model.Branch support was estimated using 10,000 ultra-fast bootstrap replicates (Minh et al. 2013) in IQTREE.

Results
A review of the original descriptions of both P. rochebruni and P. crossei suggested that Mabille (1887a) and Dautzenberg and d'Hamonville (1887) described shells of the same species from Tonkin (Figure 1).Authors of both species provided nearly identical data for the size and shape of the shell, aperture, apex, and sutures, and the number of whorls (Table 2).Mabille (1887a) did not provide information on the color of P. rochebruni, but did note the interior shoulder of the peristome.Dautzenberg and d'Hamonville (1887a) gave the color of P. crossei as reddish-brown, did not mention the shouldered peristome, but did note the triangular breathing device.The illustration of P. rochebruni in Mabille (1887b) clearly shows the breathing triangle, though the shouldered peristome is not immediately apparent in the illustrated P. crossei shell (Dautzenberg and d'Hamonville 1887).No mention is made in either description of any small orange shells.A total of 87 museum specimens (Figure 2) was examined and an additional 44 specimens were collected by PMH and DVT.Many shells were weathered and slightly pitted, and most were light tan to medium brown in col-or.Of the 131 total specimens, 124 possessed a shoulder inside the peristome and seven did not.Using the key to Pollicaria in Kongim et al. (2013:25) we were able to identify 12 specimens to species level: three P. rochebruni, two P. crossei, and seven P. mouhoti.Seventeen specimens had a peristome with the interior shoulder, but were between 35 and 40 mm in height and excluded by the key.The remaining 119 specimens also had an interiorly shouldered peristome, but had shell size and coloration combinations different than those given in the key.Using the table of comparative characters given later in Kongim et al. (2013:28), we identified 43 P. rochebruni, two P. crossei, and seven P. mouhoti from the same set of 131 specimens.The remaining shells did not fit any combination of characters given in the table.The seven putative P. mouhoti specimens more closely resembled the specimens of P. myersii figured in Kongim et al. (2013:27) than the lectotype of P. mouhoti figured therein.
Pollicaria specimens are advertised for sale at five websites: eBay.com,yhshells.com,femorale.com,conchology.be, and topseashells.com.Taken together, these specimens suggest that the known distribution of Pollicaria species should be expanded to include Guangxi autonomous region and Yunnan province in southern China.These two regions are adjacent to the northern borders of Vietnam, Laos, and Myanmar.Shells from these two regions were identified by their sellers as P. crossei, P. gravida, and P. rochebruni.All but two Chinese specimens possessed the interiorly shouldered perisotome; the other two possessed the flared lip and no shoulder seen in P. mouhoti and P. myersii.Shell sizes varied up to 45 mm in length, and shell colors varied from light tan through orange, reddish brown, dark brown, and dark purple.Vietnamese Pollicaria specimens varied similarly in color and size to the Chinese samples, and all possessed the interior shoulder in the peristome.These specimens were labelled as P. crossei and P. rochebruni.
DNA sequences from our putative P. rochebruni and P. crossei specimens (Figure 3) were identical for all three genes.Alignment lengths for each gene were 566 positions for 16S, 430 for 18S, and 838 for 28S.Our phylogenetic analysis produced a single tree (log-likelihood -17304.7519)that recovered a monophyletic, well-supported Pupinidae sister to Cochlostomatidae.Pollicaria was resolved as sister to a clade of all other pupinids (Figure 4).

Discussion
Based on museum, online, and our own data, we believe that at least two, and possibly three, species of Pollicaria occur or have historically occurred in Vietnam.We consider the first, P. rochebruni, to comprise both P. rochebruni sensu stricto and P. crossei pre- Kongin et al. (2013).We treat P. rochebruni and P. crossei as synonyms based on the similarity of their original descriptions, the natural variation seen among shells of both species, and preliminary molecular evidence.The name P. rochebruni (publication date 14 May 1887) has priority over P. crossei (1 July 1887) and should be applied from hereon.We follow Kongim et al. (2013) in keeping P. rochebruni separate from P. gravida, given the lack of specimens of Pollicaria with an internally shouldered peristome from between Burma and Vietnam.This implies that the two species are geographically separate; based on the underlying geology, those regions separated approximately 40 million years ago (Yin 2010).However, since museum records only document presence not absence, future collections may suggest a continuous distribution.Additional karyotype and sequence analysis is needed to confirm or refute the separation between topotypic P. gravida and P. rochebruni.The revised distribution of P. rochebruni now includes Vietnam south to Đà Nẵng, with a possibility of stretching northeast into southern China (Figure 5).
We identified the second Vietnamese Pollicaria species as P. mouhoti, based on Solem's (1966) identifying collections of shells with peristomes lacking an internal shoulder and possessing a flared and partially reflexed lip to that species.We identified seven museum specimens which keyed out to P. mouhoti based on Kongim et al. (2013).Shells of P. mouhoti were collected by Solem (1966) in Vat Son, and we found additional museum specimens from Thái Nguyên, Gia Phu, Muong, and Bắc Kạn.Internet records suggest similar shells can be found in southern China, but it is unknown if they are the same species.The distribution of P. mouhoti in Vietnam appears restricted to isolated areas in the north.
A possible third Vietnamese Pollicaria species is based on karyotype differences that Kongim et al. (2010) uncovered in nominal P. gravida from Vietnam.Kongim et al. (2013) assigned their two P. gravida karyoptyes to P. rochebruni and P. crossei, the former having a large brown shell and the latter a small orange shell.We did not generate chromosome spreads for any individuals, but based on the figures in Kongim et al. (2010) we find the karyotypes between the brown and orange individuals to be equivocal.We also feel the radula differences listed in Kongim et al. (2013) are non-diagnostic characters.Kongim et al. (2010) believed that their small, orange-shelled snail was the result of sympatric speciation due to chromosomal evolution as outlined in King (1993).This hypothesis is consistent with studies of similar patterns of intra-and interspecific karyotype variation seen elsewhere (e.g.Ayala and Coluzzi 2005).Whereas authors have provided possible selective pressures that may accompany the karyotypic divergence (e.g.elevation and temperature), there is insufficient data available to pose any similar hypothesis for this Pollicaria species.Given Our preliminary molecular evidence suggests that shell color and size may not be reliable for consistent species identification in Vietnamese Pollicaria.Sequences from one small, reddish-brown shelled snail (cf.P. rochebruni) and one large, orange-shelled snail (cf.P. crossei) were identical at three genetic loci.Our orange-shelled snail was larger than the nominal P. crossei in Kongim et al. (2013), and was collected alongside similarly-sized individuals with brownish and reddish shells.We feel both of our sequenced specimens represent P. rochebruni, supporting our belief that shell color in Pollicaria species can vary greatly.This contrasts with Kongim et al. (2013), who felt that shell color was diagnostic for many species.
Despite our best efforts and those of Kongim et al. (2013) to untangle the taxonomy of Pollicaria, we recognize the need for more work on the genus.Karyotypes of topotypic P. gravida remain unknown so no comparisons to the Vietnamese and other taxa can be made.Additional field work in southeastern Asia and southern China is needed to determine the true ranges of each species.

Family Pupinidae
Genus Pollicaria Gould, 1856 Type species.Megalostoma gravidum Benson, 1856(=Cyclostoma pollex Gould, 1856), designated by Pain (1974).Discussion.By synonymizing P. rochebruni and P. crossei, we suggest that all Pollicaria specimens from Vietnam that possess an interiorly shouldered peristome are a single species, excluding some smaller (< 35 mm) individuals with bright orange shells.We found live specimens of P. rochebruni at the bases of and in the valleys between karst outcrops in northern Vietnam.Reports from local farmers indicated that snails are more abundant during the spring rains, when native residents eat them as a delicacy.Some Chinese shells found for sale on the Internet are consistent morphologically with P. rochebruni but may represent a separate species.

Figure 4 .
Figure 4. Phylogenetic tree (log likelihood= -17304.7519)based on three ribosomal genes (mt 16S, nuclear 18S and 28S) showing Pollicaria as sister to all other pupinid taxa.Values at branches are bootstrap values based on 10,000 ultrafast replicates.

Figure 5 .
Figure 5. Revised distribution map for North Vietnamese Pollicaria species.Pollicaria rochebruni is present at all sites shown.Black dots indicate the presence of P. mouhoti, while gray dots indicate the presence of the putative new species of Kongim et al. (2013).Coordinates for localities taken from museum data were estimated from historical sources.

Table 2 .
Morphological character comparison taken from the original descriptions of Pollicaria rochebruni and P. crossei.

Table 1 .
GenBank accession numbers for all taxa included in the phylogenetic analysis.