On the taxonomic position of Phaenomenella Fraussen & Hadorn, 2006 (Neogastropoda, Buccinoidea) with description of two new species

ABSTRACT This contribution provides novel information on the anatomy, radula and phylogeny of several species of Phaenomenella Fraussen & Hadorn, 2006, a genus of Buccinoidea Rafinesque, 1815 with unclear affinities. Molecular phylogenetic analysis based on sequences of mitochondrial COI and nuclear 28S rRNA genes of different representatives of Buccinoidea revealed close relationships of Phaenomenella with Siphonalia A. Adams, 1863 both taxa forming a clade with maximal support. The anatomy of two species of the latter genus was examined for the first time for comparative purposes. The subfamily Siphonaliinae Finlay, 1928 was erected for several Recent and fossil genera of Southern Hemisphere Buccinidae Rafinesque, 1815, and is still recognized by current taxonomists (Bouchet et al. 2017). Species of all Recent genera of Siphonaliinae were included in our analysis and the monophyly of the subfamily Siphonaliinae in its original scope is rejected. Molecular and morphological data revealed two still unnamed species of Phaenomenella from the lower bathyal zone of the South China Sea. These species, Phaenomenella nicoi n. sp. and P. samadiae n. sp. are described in the present study.


INtROduCtION
The genus Phaenomenella Fraussen & Hadorn, 2006, with the type species Manaria inflata Shikama, 1971, was establi shed for three species of "Buccinidae" Rafinesque, 1815 from taiwan and South China Sea (Fraussen & Hadorn 2006). Several additional species were described since and presently the genus includes nine species, all except one from off South East Asia (Fraussen 2008;Fraussen & Stahlschmidt 2012;Fraussen & Stahlschmidt 2013). The anatomy of the genus has never been examined, but the radula was illustrated for two species, Phaenomenella inflata (Shikama, 1971) and Phaenomenella angusta Fraussen & Hadorn, 2006. The radula is of general buccinid appearance, which gives no clues of the relationships of Phaenomenella to other buccinid genera. The position of the genus within Buccinidae was not addressed in previous publications, probably due to still unesolved in trafamiliar classification of the family. The intrageneric shell variability of Phaenomenella is high that is hampering provid ing reliable generic diagnosis.
Relationships of Siphonalia with other Buccinidae are not clear. Its isolated position was recognized by Finlay (1928), who proposed a new subfamily Siphonaliinae in the newly established family Buccinulidae. Having been proposed with out diagnosis or discussion, the subfamily originally included several Recent and fossil genera, which were later synonymized with Penion Fischer, 1884 (= †Austrosipho Cossmann, 1906, Verconella Iredale, 1914, Berylsma Iredale, 1924, Aeneator (= †Ellicea Finlay in Marwick 1928, †Pittella Marwick, 1928, as well as Glaphyrina Finlay, 1926(presently attrib uted to Fasciolariidae Gray, 1853Couto et al. 2016) and †Pomahakia Finlay, 1927. All genera except the type one are confined to southern hemisphere, mostly to AustralianNew Zealand region. The validity and scope of the subfamily have not been revised, although it is recognized in the current tax onomy (Bouchet et al. 2017). A recent molecular phylogeny of some southern hemisphere Buccinulinae based on whole mitochondrial genome and nuclear ribosomal sequence data (Vaux et al. 2017) revealed that Recent genera included by Finlay into Siphonaliinae do not constitute a monophyletic group. But Siphonalia itself was not included in the analysis and therefore its relationships remained unresolved.
In the course of expeditions organized by the Muséum na tional d'Histoire naturelle, Paris (MNHN) to the South China Sea several specimens of different species of Phaenomenella and one species of Siphonalia were collected and preserved for dNA sequencing. The examination of this material revealed two still unnamed species. The purpose of the present paper is to provide formal description of the revealed new species, and to clarify relationships of Phaenomenella based both on anatomy and on molecular data of extended dataset of Buc cinoidea, including groups that are conchologically similar to Phaenomenella.

MAtERIAL ANd MEtHOdS
The material was collected mostly in the research cruises in the South China Sea (dONGSHA 2014, ZHONGSHA 2015 and in the Philippines (AuRORA 2007) that were organized respectively by the National taiwan university and the Mu ZOOSYSTEMA • 2020 • 42 (3) séum national d'Histoire naturelle, Paris (MNHN) with the Philippines Bureau of Fisheries and Aquatic Resources (BFAR).
Specimens collected were processed with a microwave oven (Galindo et al. 2014): the living molluscs in small volumes of sea water were exposed to microwaves for 1030 s, depending on specimen size. Bodies were immediately removed from shells and dropped in 96% ethanol. Specimens processed in this way are suitable for further anatomical studies after soaking them in 70% ethanol. Specimens are registered in the MNHN col lection and specimens and sequences are deposited in BOLd (Barcode of Life datasystem) and GenBank (table 1). due to technical reasons, the sequenced specimens of Siphonalia spadicea (Reeve, 1847) were not available for anatomical examination and the material (not suitable for sequencing) on two other species of Siphonalia was used for anatomical study.
Dna extraction and pcr total dNA was extracted from the piece of foot using either the dNeasy96 tissue kit or Investigator Kit (Qiagen), fol lowing the manufacturer's recommendations. The barcode fragment of the Cytochrome Oxidase I (COI) gene (658 bp) and a 28S rRNA fragment were amplified using the universal primers LCO1490 and HCO2198 (Folmer et al. 1994) and C1 and d2 (Jovelin & Justine 2001), respectively. PCRs were performed in 20 µl final volume containing approximately 3 ng template dNA, 1.5 mM MgCl2, 0.26 mM of each nucleotide, 0.3 µl of each primer, 5% dMSO and 0.75 µl of either taq Polymerase (Qbiogene) or BioHYtaq dNA polymerase (dialat). The PCR profile for the COI started with 5 min at 95°C fol lowed by 40 cycles with the denaturation at 95°C (35 sec), annealing at 50°C (35 s) and elongation at 72°C (1 min), with final elongation phase at 72°C (10 min). Similar PCR profiles were set for 28S (annealing at 56°C). COI and 28S genes were sequenced in both directions to confirm accuracy of each se quence. The sequencing was performed by Eurofins or in the SIEE RAS molecular facility on an ABI 3500 Genetic analyser.

Morphology and radula
Radulae were cleaned using diluted bleach (NaOCl), airdried, coated with gold and examined by scanning electron microscope teScan tS5130MM in the Institute of Ecology and Evolution of Russian Academy of Sciences, Moscow (IEE RAS). Anatomy was examined on manual dissections. phylogenetic analysis COI and 28S sequences were aligned using ClustalW im plemented in BioEdit v. 7.0.9.0 (Hall 1999); the accuracy of each alignment was checked by eye and if needed modified. COI and 28S sequences of additional 32 buccinoidean taxa, mainly from the datasets of Kantor et al. (2013) andVaux et al. (2017) were accessed from GenBank.
Three datasets were analyzed, a singlegene COI dataset (57 taxa) with three codon positions coded as three indepen dent partitions, a nuclear 28S (45 taxa) as single partition, and a concatenated COI-28S dataset (45 taxa), where four partitions corresponded to three codon positions of COI and to the 28S fragment respectively. The single gene datasets were mainly used to evaluate primary species hypotheses (PSHs), proposed based on the shell morphology, whereas the analysis of concatenated dataset allowed us to estimate relationships of Siphonalia and remaining genera originally included in Siphonaliinae. The lineage comprising buccinids from the biogenic substrates (Kantor et al. 2013) was used to root the tree based on the topology of Buccinidae tree recovered by Galindo et al. (2016). In the RAxML analyses (Stamatakis 2006) robustness of nodes was assessed using the Thorough Bootstrapping algorithm (Felsenstein 1985) with 1000 itera tions. The Bayesian inference analyses (BI) were performed using MrBayes (Huelsenbeck & Ronquist 2001), running two parallel analyses, consisting each of six Markov chains of 20 000 000 generations with default parameters. Para meters of the substitution model were estimated during the analysis (six substitution categories, a gammadistributed rate variation across sites approximated in four discrete categories and a proportion of invariable sites). The trees from the first 5 000 000 generations (25% from total number of genera tions) were discarded as burnin prior to the calculation of consensus trees. Convergence of each analysis was evaluated using tracer 1.4.1 (Rambaut et al. 2014) to check that all ESS values exceeded 200 (with default burning). All analyses were performed on the Cipres Science Gateway (http://www. phylo.org/portal2), using MrBayes on XSEdE (3.2.6) and RAxMLHPC2 on XSEdE (8.2.10) (Miller et al. 2010

phylogenetic analysis
Phylogenetic analysis of both the the COI and 28S datasets recovered a wellsupported clade comprising sequences of Siphonalia and Phaenomenella (Figs 1; 2). The K2P pairwise distances distribution for Phaenomenella and Siphonalia revealed two modes: below 1.1% and above 7.2% that we interpret as corresponding to the intra and interspecific comparisons respectively (table 2). The seven revealed clusters also corre sponded to the wellsupported monophyletic or single specimen lineages on the COIbased tree but relationships among them are mostly poorly resolved. Specimens of only six cluster recog nized based on the COI were present in the 28S data set, and they formed five reciprocally monophyletic lineages (Fig. 2). Thus the 28Sbased clusters mostly corresponded to those re vealed with COI, except for the unresolved relationship of the specimen MNHNIM201350260. Based on the analysis of singlegene datasets, we conclude that the analyzed specimens of Phaenomenella and Siphonalia comprise seven MOtus. two of the MOtu can be identified as Phaenomenella inflata Shikama, 1971 ( Fig. 1 (1)) and Phaenomenella insulapratasensis (Okutani & Lan, 1994) ( Fig. 1 (5)). One MOtu was represented by a single specimen only in the COI data set ( Fig. 1 (3), MNHNIM201350260). It has strong re semblance to Phaenomenella callophorella (Fraussen, 2004), described off taiwan based on a single specimen from similar depths (500900 m vs 795822 m in our specimen) (Fraus sen 2004), although it differs in having higher spire and less inflated whorls. Since we are not able to identify the variability of this species we attribute our specimen to P. callophorella with some reservation. two specimens ( Fig. 1 (2), MNHN IM201350012 and MNHNIM201341068) are subadults (with shell length less than 15 mm) and have some resem blance to P. thachi Fraussen & Stahlschmidt, 2012. Although they may represent a different species, the limited material available to us prevents us presently from further taxonomic consideration of the species. two MOtus were represented by several wellpreserved adult specimens and cannot be at tributed to any existing species of Phaenomenella. They are described herein as new species, Phaenomenella samadiae n. sp. and P. nicoi n. sp. Finally one MOtu can be unambiguously identified as Siphonalia spadicea (Reeve, 1847).
The analysis of the COI-28S data set (Fig. 3) recovered a wellsupported PhaenomenellaSiphonalia cluster consistent with the one in the COI tree, except for P. inflata, which was not represented in the concatenated data set. Whereas Phaenomenella cf. thachi, P. cf. calloporella and P. nicoi n. sp. formed a wellsupported subclade (BI posterior probability = 0.97), P. samadiae n. sp. showed weakly supported affin ity to Siphonalia spadicea (BI posterior probability = 0.75). This result questions monophyly of Phaenomenella in rela tion to Siphonalia. In the absence of the data on Siphonalia cassidariaeformis, the type species of the genus, and bearing in mind the conchological differences between the genera, we accept the conservative approach and do not synonymize Phaenomanella with Siphonalia.
distribution. -Presently the species was recorded only in the South China Sea at 12051389 m.

description (holotype) Shell
Shell broad fusiform with truncated base (Fig. 4AC)  whorls. Ribs nearly orthocline, broadly spaced, 12 on first preserved whorl, 14 on antepenultimate. Aperture broad ovate, white inside, angulated posteriorly, outer lip thick, slightly reflected. Parietal wall and columella with narrow but thick glossy callus with yellowish band along edge.
Shell covered with light olive adhering periostracum, form ing densely spaced low axial lamellae, obsolete on cords, but visible in interspaces.
Operculum spanning most of aperture, with distinctly turned leftwards terminal nucleus and weakly angulate in upper part. Radula (Fig. 5A, B) Examined in holotype and sequenced specimen MNHN IM201361674. Very similar in both specimens; central tooth with rather short rectangular basal part with weakly arcuate anterior margin and three short triangular broad cusps. Lateral teeth tricuspate with weakly curved basal side, attached to membrane. Outermost cusp recurved, medium long, inner cup weakly recurved, about 2/3 of outer cusp length. Intermediate cusp shortest, situated slightly closer to inner cusp.
Anatomy two specimens studied -MNHNIM201361674, male, sequenced paratype; MNHNIM201361617, female, holo type. Soft body partly extracted from the shell. Head rather large, with two thick long tentacles. Eye lobes poorly defined, not pigmented in both examined specimens, eyes obviously absent. Mantle of female (Fig. 6B) approximately square in shape, with long siphon. Ctenidium comprises ¾ of mantle length and in average 1/5 of mantle width; bipectinate sym metric osphradium slightly narrower than ctenidium and ¾ of its length. Capsule gland medium large, covered by thick rectum and terminated by large vagina. In male's mantle, prostate gland welldeveloped, situated parallel and equal in size to rectum. Penis (Fig. 6 C) flattened, terminating in seminal papilla shifted to left side and not surrounded by a circlular fold.

reMarks
The new species demonstrates some variability in shell shape with more inflated shell outline in smaller specimens. The new species is most similar in the shell shape to P. mokenorum Fraussen, 2008 from the Andaman Sea, differing in better pronounced spiral cords. Another somewhat similar species is P. insulapratasensis (Okutani & Lan, 1994), which possesses smaller, more ovoid shell with faster incrementing teleoconch whorls and a longer and more distinct canal.

description (holotype) Shell
Shell fusiform with high spire and attenuated siphonal canal (Fig. 7AC), fragile, white under periostracum. Siphonal canal medium long, slightly inclined to left and crossing coiling axis. Protoconch paucispiral, of about 2 whorls, par tially eroded as well as upper teleoconch whorls, teleoconch whorls 6 ½ in number. [Better preserved protoconch found in specimen MNHNIM201361592 (Fig. 7L): bulbous, of nearly 2 whorls, diameter 2.05 mm, exposed height 2.1 mm. Protoconch surface eroded, sculpture not seen. Boundary with teleoconch marked by weak (about 5) orthocline ribs.] tel eoconch whorls convex, weakly angulated at shoulder. Suture distinct, shallowly impressed. Spiral sculpture of distinct flat on top spiral cords, covering entire shell surface. Number of cords increasing from 8 on first teleoconch whorl to 19 on penultimate whorl, on last whorl 55 cords, of which about 20 on canal. Cords differing slightly in width, with most narrow on subsutural ramp, interspaces between cords about 1/3½ of cord's width. Strong axial ribs present on entire shell, suture to suture on uppermost teleoconch whorls, gradually becom ing obsolete on subsutural ramp and absent on ramp of last and penultimate whorls. On last whorl ribs distinct only on shoulder, producing its angulated appearance. Number of ribs stable, i.e., 1516 per whorl. Aperture ovate, white inside, angulated posteriorly, outer lip thin, evenly convex, concave at transition to canal. Pari etal wall and columella with narrow and thin glossy callus.
Shell covered with very light olive adhering periostracum, forming densely spaced low axial lamellae visible in interspaces between cords.
Operculum partially abraded, when intact (Fig. 7 d) span ning most of aperture with distinctly turned leftwards terminal nucleus and rounded upper part.

Radula (Fig. 5 C-F)
Examined in five specimens, including holotype. Rather similar in all specimens; central tooth with rectangular basal part with weakly arcuate anterior margin and three medium long triangular broad cusps, central one shorter and narrower than lateral ones. Lateral teeth normally tricuspate with weakly curved basal side, attached to membrane. Outermost cusp recurved, medium long, inner cusp weakly recurved, about 2/3 of outer cusp length. Intermediate cusp shortest, situated slightly closer to inner cusp. In one sequenced specimen (MNHNIM201361673, Fig. 5F)

Measurements (holotype)
Shell length 41.3 mm, last whorl length 26.9 mm, aperture length (without canal) 16.0 mm, diameter 16.5 mm. In the largest specimen studied, shell length reaching 43 mm.
Anatomy two specimens examined: MNHNIM201361636, male, and MNHNIM201361637, female, sequenced paratype (similar in both studied specimens). Soft body partly extracted from the shell. Head medium large, with two long tentacles and large black eyes on lobes at bases of tentacles. Mantle similar to that of Phaenomenella samadiae n. sp. Penis flat tened, with seminal papilla situated on its top and surrounded by circle fold of skin (Fig. 8B).

reMarks
The new species is highly variable in shell shape. Some of the specimens are much more slender (Fig. 7d, I, G) and the axial ribs are either very weak or obsolete. The specimen with no axial ribs was collected at a maximal depth (16341683 m), but there is not clear correlation with depth, since syntopic specimens can have strong or weak axial ribs. Nevertheless the molecular data clearly indicates the conspecifity of "typical" angulated specimens with welldeveloped ribs and smooth ones. The specimens collected at greater depth have the pro toconch and upper teleoconch whorls more eroded or missing.
In general shape the new species has some resemblance to P. mokenorum Fraussen, 2008 from the Andaman Sea, differ ing in better pronounced axial ribs in later teleoconch whorls and more attenuated narrow siphonal canal. Also P. nicoi n. sp. has smaller size (maximal shell length 42 mm versus 55.6 mm in P. mokenorum).

Anatomy
Soft body (no. 1, female, Fig. 10A, B, E, F, no. 2, female, Fig. 10C) with approximately 3 whorls. Head short and broad, with short contracted tentacles. Eyes small, situated at small lobes in the middle of tentacles (Fig. 10B, C, eyes). Foot contracted, with very narrow propodium and large oper culum with terminal nucleus. Penis of spm. no. 3 (Fig. 10d) medium long, flattened, contracted, with small (contracted) rounded seminal papilla at the top, surrounded by circular fold of skin. Mantle with very long siphon in dissected specimens (longer than half mantle width).
digestive system. Proboscis half everted out of rhynchodaeum, thick, contracted (Fig. 10E, F, pr). Proboscis retractors (prr) attached to rhynchodaeum along both sides of anterior oe sophagus (mostly on its right side), connecting rhynchodaeum to lateral walls of body haemocoel. Anterior oesophagus short and wide, dorsoventrally flattened, along ventral side of proboscis (Fig. 10E, F, aoe) into relatively small rounded valve of Leiblein (Fig. 10F, vl), situated immediately anterior to nerve ring (Fig. 10F, nr). Salivary glands medium small (about 0.25 proboscis length) (Fig. 10E, F, sg), with very thin strongly convoluted salivary ducts (Fig. 10F, sd) following along anterior oesophagus. Gland of Leiblein large (Fig. 10E, F, gl), following along thick, round in section posterior oesopha gus (Fig. 10E, F, poe). Stomach spanning about 0.3 whorl (Fig. 10G). Posterior mixing area not large in spm. no. 1, large in spm. no. 2 (Fig. 10G, I, pma). Posterior oesophagus and intestine medium wide in both specimens. Opening of posterior duct of digestive gland (found in spm. no. 2) located near oesophageal opening (Fig. 10I, pdg), opening of anterior duct (found in spms. nos. 1 and 2) located near beginning of intestine. Inner stomach wall between two openings with longitudinal fold (Fig. 10H,  coMpleMent to description Radula Radula rather similar in both specimens (Fig. 11C, d); central tooth with rectangular basal part and weakly arcuate anterior margin and three medium long triangular broad cusps, central one equal in length but slightly narrower than lateral ones. Lateral teeth tricuspate with weakly curved basal side, attached to membrane. Outermost cusp recurved, medium long, inner cusp weakly recurved, about 2/3 of outer cusp length; inner cusp in right longitudinal row of specimen spm. no. 2 partially subdivided (Fig. 11d). Intermediate cusp shortest, situated closer to inner cusp; inner cusp of spm. no. 2 partially sub divided in left longitudinal row.
Anatomy (spm. no. 1, male, Fig. 12) Head very short and broad, tentacles short, contracted, with small eyes at lobes. Foot contracted, propodium moderately wide, operculum oval with terminal nucleus. Penis rather large (Fig. 12C), flattened, with long narrow seminal papilla in deepening at the top. Mantle with medium long siphon.
digestive system. Proboscis partly everted out of rhyn chodaeum, with contracted walls. Multiple proboscis retrac tors attaching mostly along right side of anterior oesophagus (Fig. 12d, E, prr), connecting rhynchodaeum and lateral walls of body haemocoel. Buccal mass slightly shorter than retracted proboscis (Fig. 12F, bm), attaching to its walls by multiple odontophoral retractors (odr). Radula lying in middle of buc cal mass and attached to proboscis walls by median retractor (Fig. 12F, mrr). Salivary glands (Fig. 12d, E, sg) medium large (0.4 proboscis length), oval, with salivary ducts following on both sides of anterior oesophagus. Anterior oesophagus wide, dorsoventrally flattened (Fig. 12E, aoe), valve of Leiblein rounded, medium large. Posterior oesophagus (poe) relatively narrow. Gland of Leiblein large, folded beneath nerve ring (Fig. 12E, gl). Stomach spanning about 0.3 whorl (Fig. 12G). Posterior mixing area not large (Fig. 12G, H, pma). Intestine medium wide. Opening of posterior duct of digestive gland located near oesophageal opening (Fig. 12I, pdg), opening of anterior duct located closer to beginning of intestine. Inner stomach wall between two openings contains longitudinal fold (Fig. 12H, lfl), lined with low oblique folds, rest part of inner and outer stomach wall lined with moderately high transverse folds.

dISCuSSION
Results of the phylogenetic analysis suggest close affinities of Siphonalia and Phaenomenella that remained unnoticed previously. Fraussen & Hadorn (2006), while describing Phaenomenella, compared it to Manaria and Eosipho, but not to Siphonalia. The shell outline of some Phaenomenella (e.g. Phaenomenella insulapratasensis) is rather similar to Siphonalia: the shell is stout, with strongly convex whorls and a recurved siphonal canal. Species of Phaenomenella though have a much larger (about twice) protoconch in comparison with Siphonalia.
The intrageneric variability of shell shape in Phaenomenella in its current definition is very high (Fraussen & Stahlschmidt 2013) and in its extremes there is no resemblance between the two genera. It should also be born in mind that some of the most diverging species of Phaenomenella were not yet sequenced and may fall into other lineages. Representatives of both genera are also anatomically similar, particularly in the digestive system characters. Both Phaenomenella and Siphonalia have a broad, medium long in the contracted state proboscis, medium large salivary glands and a large gland of Leiblein. It should be mentioned that despite these general similarities, there are no unique morphological characters uniting both genera. The radular morphology is very similar in both genera (radula of one more species, S. marybethi Parth, 1996 was illustrated in Zhang & Zhang 2018), however, as in the case with the body anatomy, it is of rather generalized buccinid appearance; similar radular morphology can be found in many unrelated genera -eg. Our molecular analysis did not recover Phaenomenella as monophyletic. In both COI and combined COI+28S analyses the internal relationships within Phaenomenella -Siphonalia clade are not resolved. Siphonalia spadicea cluster without sig nificant support with P. samadiae n. sp. We have only a single species of Siphonalia in our analyses so it is too preliminary to change the classification on the basis of the incomplete dataset. Therefore we presently retain the validity of Phaenomenella, although it is possible that Phaenomenella and Siphonalia can belong to a single genus. One of the distinctions between the genera is the depth range of known species. Generally, species of Siphonalia dwell at shallower depths -from 10 to 300 m (Okutani 2000), while Phaenomenella is recorded at 1901389 m (Fraussen & Stahlschmidt 2013; herein). The new species are attributed to Phaenomenella based on stronger conchological similarity to other species of the genus rather than to species of Siphonalia. unfortunately the protoconch of P. samadiae n. sp. was decollated in all available specimens, but the protoconch of P. nicoi n. sp. is large globose, similar to other species of Phaenomenella.
The analysis of a broader dataset of Buccinoidea rejected the monophyly of Siphonaliinae in its original scope. None of the Recent genera, originally included by Finlay (1928) in the subfamily, that are Penion, Aeneator and Glaphyrina, are closely related neither to each other, nor to Siphonalia. The system of Buccinidae and Buccinoidea in general is still far from being resolved, with many problematic buccinoidean lineages (see e.g. Couto et al. 2016;Harasewych 2018). There fore the rank of the inferred clade Siphonalia + Phaenomenella can be resolved only after obtaining the robust phylogeny of the entire superfamily Buccinoidea.