Molecular phylogeny and systematics of Australian and East Timorese Stenothyridae (Caenogastropoda: Truncatelloidea)

The Australian and East Timorese species belonging to the truncatelloid family Stenothyridae are revised using molecular data and morphological characters from the shell, operculum, radula and external and reproductive anatomy. The Australian species Stenothyra australis is redescribed and two previously recognised subspecies are shown to be synonyms. The New Guinean species Stenothyra paludicola van Benthem Jutting, 1963 is redescribed and recorded from the Torres Strait region of northern Australia, and two new subspecies of S. paludicola are described from the Northern Territory and East Timor; S. paludicola topendensis n. subsp. and S. paludicola timorensis n. subsp. respectively. Stenothyra gelasinosa n. sp. is described from Australia, comprising three allopatric subspecies; S. gelasinosa gelasinosa n. sp. and n. subsp. from the eastern seaboard, S. gelasinosa phrixa n. subsp. from northern Australia and S. gelasinosa apiosa n. subsp. from the Pilbara region of Western Australia. Stenothyra frustillum is considered a nomen dubium. Molecular phylogenetic analysis of these taxa and other Asian stenothyrids supports these systematic decisions and provides a preliminary interpretation of relationships within Stenothyridae.


Introduction
The Stenothyridae are a family of small-sized gastropods found in intertidal and shallow-water aquatic habitats in Asia and Australia. While many species are found in brackish estuaries, a large portion of stenothyrid diversity occurs in freshwater environments such as lakes or rivers, and a few taxa are found in unusual habitats such as hot springs (Kosuge 1969). Several major Asian river systems are sites of particularly high stenothyrid diversity (see Brandt 1968Brandt , 1970Brandt , 1974Davis et al. 1986Davis et al. , 1988, and a few of those taxa have been described in detail. Less in known about estuarine stenothyrids, and many taxa named in the early literature are poorly defined and effectively ignored or overlooked in recent studies of estuarine gastropods. Uncertainty over the identity of many stenothyrid species results in speculative identifications in faunal checklists (e.g. Tan and Woo 2010). Attempts to catalogue estuarine gastropod diversity often result in the listing of undescribed stenothyrid taxa (e.g. Walthew 1995;Robba et al. 2003).
A preliminary survey of the literature recovered approximately 100 extant species-level names, belonging to approximately six genus-level groups. Many of these names may be synonyms and, given the lack of assessment of most of these taxa, it is not possible to produce a list of valid taxa. A previous estimate of stenothyrid diversity recognised approximately 60 species globally (Strong et al. 2008). Many fossil taxa have been named in several of the Recent genera and an additional six extinct genera have been proposed (Youlou 1978;Esu and Girotti 2010). Almost all Recent species are placed in the genus Stenothyra Benson, 1856, in which the monotypic subgenus Incolaestuarium Kuroda, 1962 is sometimes recognised. A small number of taxa belong to the Asian genus Gangetia Ancey, 1891, and the recently described genus Farsithyra Glöer & Pešić, 2009 contains a single Iranian species. The lack of generic structure and high proportion of taxa assigned to Stenothyra are probably a consequence of highly conserved shell morphology in the group. Stenothyra is easily recognised by its small, circular aperture and dorsoventrally compressed shell, but the characters used to distinguish species such as whorl proportions, size and shell sculpture recur in many species and are also known to be variable within single taxa (Davis et al. 1986).
Australia's stenothyrid fauna was initially sparsely recorded by C. Hedley and T. Iredale. A single taxon, Stenothyra australis Hedley, 1901 was described from Bowen, central Queensland, and subsequently placed in the genus Obesitena Iredale, 1943 with two additional subspecies, Obesitena australis perdives Iredale, 1943 Williams et al. 2003 from Darwin, Northern Territory and Obesitena australis wildiana Iredale, 1943 from Nudgee, near Brisbane, southern Queensland. Since those original descriptions, no new information on the systematics of Australian Stenothyridae has been published. All three available names appear to belong to estuarine species, based on their coastal localities. Nothing is known of the anatomy, phylogenetic relationships, behaviour, ecology, geographic distributions or higher systematics of any Australian stenothyrid. Three species of Stenothyra were described by van Benthem Jutting (1963) from the southern coastline of New Guinea, but those species have not been subsequently reported. This study aims to document the diversity of Australian stenothyrids using specimens presently held in museum collections as well as material collected from various regions of tropical Australia during this project. Species discrimination will be informed by shell morphology, observations of the living animal, microscopic anatomical details and molecular data. The detailed morphologybased treatment of a Chinese stenothyrid species complex by Davis et al. (1986Davis et al. ( , 1988) provides a good guide for the morphological treatment of Australian taxa. This study also uses molecular data to reconstruct the phylogenetic relationships of the examined Australian species of Stenothyridae and a small selection of Asian stenothyrids. A comprehensive global molecular phylogeny and systematic revision of Stenothyridae is long overdue, but unfortunately beyond the scope of this work.

Materials and methods
Material examined in this study was mainly sourced from the collections of the Australian Museum, supplemented by additional collections in tropical Queensland (September 2011) and northern Western Australia (June 2012) as part of a larger study of Australian mangrove microgastropods (see Golding in press). Representative specimens of most species were collected alive and photographed crawling in a petri dish of salt water, using a microscope-mounted camera. Records of pigmentation and foot/tentacle morphology are included in descriptions where known. The animals were preserved for either molecular (95-100% EtOH) or morphological (10% saltwater formalin or 70% EtOH) examination.
Dissections were performed under a stereomicroscope and illustrated using a camera lucida. Radulae were removed from dissected buccal masses by dissolution overnight in a warmed solution of sodium hydroxide. Shells, radulae and opercula were cleaned in a sonic water bath, mounted on stubs and thinly coated with gold for examination by scanning electron microscopy (performed by Sue Lindsay, Microscopy and Microanalysis Laboratory, Australian Museum) using a Zeiss Evo LS-15 machine. Shell measurements were made using a calibrated camera lucida, with whorl counts rounded to the nearest 1/4 whorl.
For molecular analysis, either entire animals or samples of foot tissue were processed using a DNeasy kit (Qiagen, Inc., Hilden, Germany) and QiaCube ® robot to extract genomic DNA. Three gene fragments were amplified by polymerase chain reaction (PCR); COI and 16S from the mitochondrial genome and 28S from the nuclear genome. PCRs were performed in 25 μl volumes containing 1 × PCR buffer, 200 mM each dNTP, 2.0 mM MgCl 2 , 0.5 mM forward and reverse primers, 1.25 units Taq polymerase, and approximately 50 ng DNA. Amplification followed a standard protocol with 35 cycles of 94 • C for 1 min, primer-specific annealing conditions (Table 1) and 72 • C for 1 min. Post-PCR products were purified using ExoSAP-IT proteinase solution (GE Healthcare, USA) and sequenced in both directions by Macrogen Inc. (Seoul, Korea).
Raw sequences were checked for misreads against their electropherogram and compiled using BioEdit v7.0.9 (Hall 1999). All sequences were deposited in GenBank with accession numbers provided in Table 2 (KC439688-KC439930). Ribosomal sequences were aligned using the online MAFFT v.6 server (Katoh et al. 2002) with the E-INS-i option (Katoh et al. 2005) implement for the 16S dataset and Q-INS-i (Katoh and Toh 2008) for the 28S dataset. COI sequences were translated using the invertebrate mitochondrial code and unambiguously aligned by amino acid sequence. Datasets were compiled using Mesquite v2.75 (Maddison and Maddison 2011) and uncorrected pairwise distances computed within and between taxa using MEGA5 (Tamura et al. 2011).
Australian species of Stenothyridae are represented in the molecular phylogenetic analyses by 4-16 specimens Table 2. Identity, location/source, registration and GenBank accession numbers for specimens included in the molecular analyses. All sampling locations are in Australia unless otherwise stated. Sequences generated during this project are in bold and all other sequences were produced by Criscione and Ponder (2012). Taxa described in this manuscript and the accompanying manuscript (Iravadiidae) are named accordingly.  from many populations across a broad geographic range (Table 2). Non-Australian taxa are sparsely sampled and mostly represented by specimens from a single locality. Species from East Timor, Hong Kong, Thailand, Malaysia and the Philippines were included in this study but represent only a small fraction of global stenothyrid diversity. All of the foreign material was collected in mangrove or brackish-water habitats, except for material from the Philippines that was collected from a freshwater lake. Non-Australian taxa were provisionally identified where possible using the available literature (Brandt 1974, Davis et al. 1986). Some non-Australian specimens included in the molecular analysis could not be identified and may belong to undescribed taxa, but are not further treated here. Sequences from additional, non-stenothyrid, taxa were also examined. Sequences of 16S and 28S gene fragments were recently obtained by Criscione and Ponder (2012) for a phylogenetic analysis of Rissooidea and Truncatelloidea, and some sequences from their study are included here. Truncatelloidea currently comprises approximately 23 families, including Stenothyridae (Criscione and Ponder 2012). During the present study additional COI sequences were generated from the original DNA extracts deposited in the Australian Museum collections (Criscione and Ponder 2012). Further sequences from two species of Calopiidae and several Iravadiidae generated as part of a larger project exploring Australian truncatelloid diversity were also included in the datasets.
Two datasets were constructed in order to test the position and monophyly of Stenothyridae in Truncatelloidea and also to examine the internal relationships and molecular diversity of the sampled representatives of Stenothyridae. Phylogenetic hypotheses were explored using maximum likelihood (ML) and Bayesian Inference (BI) methods. ML analyses were performed using RAxML v7.3.2 (Stamatakis 2006) implemented in raxmlGUI v1.2 (Silvestro and Michalak 2011) with 1000 'thorough' bootstrap repetitions. Datasets were partitioned by gene and the GTRGAMMA model was implemented. BI analyses were performed using MrBayes v3.1.2 (Ronquist and Huelsenbeck 2003). Bayesian posterior probability support was estimated by running four Markov chains (10 million generations each, with trees sampled each thousand generations). The first 25% of trees were conservatively rejected as burn-in, and stationarity was confirmed by examination of the log likelihood plot using Tracer (Rambaut and Drummond 2007). A summary consensus tree with support indices was generated by MrBayes. Datasets were partitioned by gene, with the GTR + G + I model of sequence evolution selected for both datasets by MEGA5. Trees were visualised using FigTree v1.3.1 and rooted using the outgroup. Nodal support was considered high for Bayesian probabilities >95% and bootstraps >80%, and moderate for Bayesian probabilities 90-95% and bootstraps 70-80%. Nodes with lower support values were not considered significant.
Systematic descriptions are provided for all Australian taxa identified during this study. In addition, a new subspecies from East Timor is also described. A comprehensive description has been provided for Stenothyra australis; for the remaining taxa, only characters useful for discriminating species and subspecies are included in the descriptions. Unless otherwise stated, locations given for type materials examined are in Australia. Materials examined have been summarised here, but full details are provided in a supplementary file.

Molecular phylogenies
Phylogenetic reconstruction by BI and ML methods produced consensus trees with identical topologies, with the exception of several weakly supported clades composed of closely-related populations from each subspecies. Only the BI summary tree is shown here, labelled with both Bayesian posterior probabilities (BPP) and bootstrap support values generated by ML analysis (Figs 1, 2). The phylogenetic analysis of Truncatelloidea, including single representatives of most Stenothyra species examined here, found strong support for the monophyly of Stenothyridae (BPP and ML bootstrap values of 100%) and conflicting support (BPP = 100, ML bootstrap value not significant) for a sister relationship with a clade composed of Assimineidae, Truncatellidae, Falsicingulidae and Pomatioposidae (Fig. 1). Together, this group of families corresponds to clade F in the phylogeny produced by Criscione and Ponder (2012). Relationships between Stenothyridae and other truncatelloids were partially resolved by the phylogenetic analysis ( Fig. 1), with strong support for a clade composed of S. australis, S. gelasinosa n. sp. and S. cf. divalis, and for a sister relationship between S. paludicola and S. cf. glabra.
The phylogenetic analysis of Stenothyridae with only two truncatelloid outgroups also provided strong support for the integrity of all species and subspecies of Stenothyra that were represented by more than one individual (Fig. 2). Stenothyra australis, S. gelasinosa n. sp. and S. cf. divalis formed a well-supported clade, with S. australis basal (Fig. 2). A relationship between S. monilifera and an unidentified species from the Malaysian Peninsula (S. sp. 'Johor') was supported by BPP (98%) but not by ML bootstrap values (<80%). Likewise, a clade comprising S. paludicola and S. cf. glabra was moderately supported by BPP values (92%), but not by ML bootstrap values Figure 2. Summary tree from Bayesian analysis of concatenated COI (including 3rd codon position bases), 16S and 28S sequences (10 million generations, trees sampled every 1000 generations). Support indices are BI posterior probability (above nodes, >90%) and ML bootstraps (below nodes, >70%); asterisks indicate a support value of 100%. Taxa described in this manuscript are named accordingly and labelled in blue, with locality data given for each sample. Shell images are all to the same scale.
(<80%). Relationships between these clades and other branches of the tree were not well-supported in either BI or ML analyses (Fig. 2).     Iredale's (1943) subspecies of Stenothyra australis are synonymous with the typical species as comparison of the type specimens shows only small differences that are within the variation of the species, and sequence data from specimens collected at all three type localities supports the existence of a single species-group taxon. The holotype of the Papua New Guinean species S. nebularum has similar spiral sculpture on the upper whorls and is only slightly smaller than some specimens (including one of the paratypes) of S. australis. It is possible that S. nebularum is a junior synonym of S. australis. However, only photographs of the type specimens have been examined and the size difference may prove to be significant when further New Guinean material is examined. Stenothyra australis is generally larger than the other Australian species, and has a distinctively tall, conical spire with linear spiral ridges.

Stenothyra gelasinosa n. sp. Remarks
Representative specimens of Stenothyra gelasinosa n. sp. form a monophyletic clade in the molecular analysis, but are arranged in three geographically discrete groups that are also differentiated by several morphological characters. These taxa are here recognised as subspecies of the S. gelasinosa n. sp. The shells of S. gelasinosa n. sp. are similar in shape to Stenothyra saccata van Benthem Jutting, 1963, from Papua New Guinea. However, the holotype of S. saccata is 30% taller than any specimen of S. gelasinosa n. sp. collected in Australia. The holotype of S. saccata was examined during this study from photographs only, and the presence or absence of the spiral pitted sculpture that characterises S. gelasinosa n. sp. could not be confirmed and were not mentioned in the original description. These Australian and New Guinean species are probably closely related, but at this stage there is sufficient evidence from the size of the animals to erect a new species for the Australian taxa. Sequence data strongly supports the monophyly of S. gelasinosa n. sp. and each subspecies reciprocally. The subspecies have low sequence divergence within each taxon, but 4.80-5.41% divergence of the COI gene between subspecies. Stenothyra gelasinosa n. sp. can be distinguished from other Australian taxa by its small size, dome-shaped spire and spiral rows of oblong pits or dimples on the upper whorls. A comprehensive description is provided below for S. gelasinosa gelasinosa n. sp. & n. subsp., with comparative descriptions provided for the other two subspecies.

Etymology
Named for the sculpture of dimples (Greek, dimple = gelasinos) on the upper whorls that distinguishes this species from other Australian stenothyrids.

Description
Shell (Figs 2; 8B, C). Ovate, with rounded to angled inflation of last whorl, dorsoventrally compressed; spire short, dome-shaped; sutures moderately deep, upper whorls slightly convex; nonumbilicate; up to 4 1/4 whorls including protoconch, SL = 2.07-2.26 mm, SD 1 = 1.34-1.53 mm, SD 2 = 1.10-1.17 mm (ratio SD 1 :SD 2 ∼1.24) ( Table 5). Slightly lustrous, pale golden brown; periostracum thin. Sculpture on first 3/4-1 teleoconch whorl of 9-12 spiral rows of oblong pits, terminating irregularly. Aperture smaller than preceding whorl, near circular, posterior margin slightly angulated; outer lip strongly prosocline. Protoconch (Figs 8D, E). Dome-shaped; 1 3/4-2 whorls. Smooth; transitional varix well developed. Operculum (Figs 9A, B). As for S. australis. Radula (n = 3) (Fig. 9C). Central tooth 2 + 1 + 2/3 − 4 + 3 − 4; central cusp large, secondary cusps and basal denticles diminishing outwardly. Lateral teeth 3 − 4 + 1 + 8 − 9. Marginal teeth with subequal cusps; inner marginal teeth with ∼30 cusps on tip and distal half of outer edge; outer marginal teeth with ∼10 cusps on distal third of inner edge. (Figs 10A,  B). Head-foot transparent grey with mottled white pigment; darkest around eyes, on ventral surface of snout, dorsal midline of anterior foot and dorsal surface of foot anterior to operculum; eyes encircled by granules of canary-yellow pigment. Snout with bright patch of canary-yellow pigment on each lateral surface, surrounded by saddle-shaped outline of black pigment (often only faintly visible, usually appearing as two parallel transverse black lines) and overlaid with scattered cream to orange 'freckles' either side of snout midline; lips unpigmented. Cephalic tentacles transparent, patterned with 7-10 narrow rings of white pigment, black pigment restricted to diffuse band at base of each tentacle. Anterior foot with enlarged lateral lobes, posterior margin rounded. Metapodial tentacle present, narrow, contractile, with sparse granules of white pigment; size variable, 30-50% of extended foot length; mobile, usually held parallel to substrate or slightly upturned in life. Mantle visible through transparent shell, mottled black, white and yellow with prominent broken 'T-shaped' pattern of black pigment; interior surface of anterior mantle margin transparent with alternating bands of black and white to yellow pigment, one black band coinciding with pallial tentacle. Pallial tentacle present on right side of anterior mantle margin; small, semi-circular lobe demarcated by ciliary ridge across base of tentacle, with short, cylindrical tentacle at tip of lobe; held against exterior surface of shell in life.

External morphology and colouration in life
Male reproductive system (Figs 9D, E, H; 11A). Testis, seminal vesicle and vas deferens as for S. australis. Penis cream or white, with sparse yellow flecks; cilia restricted to narrow strip along outer margin. Penial stylet rolled to form a hollow, curved, 'syringe' shape with open slit at tip; embedded in flesh, with presumed sperm duct opening via oval aperture near base. Female reproductive system (Fig. 11B). As for S. australis. Distribution (Fig. 7B) Common in mangrove forests along the eastern coast of Australia. The most southern record is at Woy Woy, Broken Bay, just north of Sydney, although specimens were not detected at this site during the winter months. There are no records of S. gelasinosa gelasinosa between Woy Woy and Brisbane, but they are common around Brisbane and further north at least as far as Cooktown in northern Figure 6. Anatomy of Stenothyra australis Hedley, 1901, from Nudgee, QLD, AMS C.470968. A, Male reproductive system, in natural position; B, proximal male reproductive system unravelled; C, female reproductive system, with pallial oviduct displaced to right; D, distal female reproductive system unravelled. Scale bars = 1 mm. Anatomical labels: asp, accessory sperm pouch; bu, bursa; csd, common sperm duct; od, oviduct; ov, ovary; pe, penis; po, pallial oviduct; pr, prostate; sd, sperm duct; sr, seminal receptacle; sv, seminal vesicle; te, testis; vd, vas deferens; vd L1, L2 and L3, 1st, 2nd and 3rd loops of vas deferens.
Queensland. Specimens collected at Weipa on the west side of Cape York belong to another subspecies of S. gelasinosa n. sp. It appears that S. gelasinosa s.s. does not extend around the tip of the Cape York Peninsula. Stenothyra gelasinosa gelasinosa lives in submerged mangrove leaf litter and can also be found living in estuaries on shallow subtidal Zostera seagrass (for example, at Woy Woy and Morton Bay).

Remarks
Stenothyra gelasinosa gelasinosa is intermediate in size between S. gelasinosa phrixa n. subsp. and S. gelasinosa apiosa n. subsp. Like S. gelasinosa apiosa, its protoconch is smooth and lacks the spiral sculpture of S. gelasinosa phrixa. Sequence divergence of the COI gene within S. gelasinosa gelasinosa is 0.17%, but compared to S. gelasinosa phrixa and S. gelasinosa apiosa the divergence is 5.01% and 5.41% respectively. Each subspecies has a distinct geographic distribution, and thus specimen locality is a good indicator of subspecies identity.

Description
Shell (Figs 2; 8F-I). As for S. gelasinosa gelasinosa n. sp. and n. subsp., except angled inflation of last whorl more pronounced; up to 4 1/2 whorls including protoconch, length 2.10-2.66 mm, diameter of last whorl 1.47-1.72 mm laterally, 1.22-1.53 mm dorsoventrally (ratio SD 1 :SD 2 ∼1.18) ( Table 5). Slightly lustrous, pale golden, sometimes with faint white growth striae. Protoconch (Figs 8H, I). As for S. gelasinosa gelasinosa, except for sculpture of ∼20 delicate, undulating, spiral ridges. Operculum and radula (n = 2). As for S. gelasinosa gelasinosa. External morphology and colouration in life (Fig. 10C), male (Fig. 9F) and female reproductive systems. As for S. gelasinosa gelasinosa. Distribution (Fig. 7B) Found in the mangrove forests of Queensland and the Northern Territory between the western coast of Cape York Peninsula and Darwin. The precise western boundary of the distribution of S. gelasinosa phrixa is not known, but during recent fieldwork the species was not detected around Broome. Specimens from further south, at Port Hedland, belong to a different subspecies of S. gelasinosa(see below).

Remarks
Stenothyra gelasinosa phrixa is the largest of the three subspecies of S. gelasinosa. It has a slightly more pronounced inflation of the last whorl, and occasionally has faint white growth striae. The protoconch has a unique sculpture of spiral wavy lines, unlike the smooth protoconchs of the other subspecies of S. gelasinosa. Sequence divergence within the subspecies is 1.17%, but divergences from S. gelasinosa gelasinosa and S. gelasinosa apiosa n. subsp. are 5.01% and 4.80% respectively (Table 3).

Etymology
This subspecies is named for the characteristic sculpture on its protoconch, which resembles ripples on the surface of water (Greek, phrix = ripple).

Material examined
Holotype. Nickol Bay, Karratha, Western Australia, 20 • 43 24 S 116 • 52 16 E, coll. R. Golding and M. Hill,4 Jul. 2012, on leaf litter and filamentous roots in pools Distribution (Fig. 7B) In mangrove forests along the coast of Western Australia between Port Hedland and Karratha. The extent of the distribution beyond these two locations is unknown, but S. gelasinosa apiosa was not found during recent fieldwork in Broome or Derby.

Remarks
Stenothyra gelasinosa apiosa is the smallest subspecies of S. gelasinosa. It is often a slightly darker shade of brown, and has a slightly taller spire than S. gelasinosa gelasinosa or S. gelasinosa phrixa. Within-species COI sequence divergence is 0.67%, and the divergence from S. gelasinosa gelasinosa and S. gelasinosa phrixa is 5.41% and 4.80% respectively (Table 3).

Etymology
This subspecies is named for its remote distribution on the western coast of Australia (Greek, apiosa = far away).

Remarks
Unlike other species of Stenothyra described here, S. paludicola lives in billabongs, swamps and coastal streams in brackish or mainly freshwater environments rather than coastal mangrove forests. It has a distinct shell shape, with the whorl profile more rounded and inflated (with correspondingly deeper sutures) than other Australasian stenothyrids. Molecular sequence data strongly supports the monophyly of S. paludicola, and the reciprocal monophyly of two subspecies within the taxon. Other distinguishing features of the species are the asymmetrical 'wings' on the interior surface of the operculum, less dorsoventrally compressed shell and small protoconch with punctate sculpture.
Operculum (Figs 13A, B). As for S. australis, except for prominent angulation at posterior apex; semicircular 'wings' on interior surface asymmetrical, with anterior projection ∼30% larger than posterior projection. Radula (n = 2) (Fig. 13C). Central tooth 2 + 1 + 2/3 − 4 + 3 − 4, central cusp large, secondary cusps and basal denticles diminishing outwardly. Lateral teeth 5 − 7 + 1 + 8 − 9. Marginal teeth with subequal cusps; inner marginal teeth with ∼28 cusps on tip and distal half of outer edge; outer marginal teeth with ∼15 cusps on distal third of inner edge. External morphology and colouration in life. The animal has not been observed alive in this study. Specimens preserved in 70% EtOH have two parallel black bands crossing the snout and scattered black flecks on the cephalic tentacles.
Male reproductive system (Figs 13D, E; 14A). Testis, seminal vesicle and vas deferens as for S. australis. Colour and ciliation of penis not observed in life. Penial stylet rolled to form a hollow, curved, 'syringe' shape with minute open slit at tip; embedded in flesh, but without any other obvious aperture for sperm duct.
Female reproductive system (Fig. 14B). As for S. australis. Distribution (Fig. 7C) Known from New Guinea in Papua, Indonesia from the type localities around Merauke and from Saibai, Horn and Prince of Wales Islands in the Torres Strait region of Australia (Fig. 5C). One locality at Utingu on the Cape York Peninsula is known from mainland Australia, but temporary, wet-season brackish-water lagoons are not well-sampled in this area and the species may be more widespread. Found in permanent or seasonal freshwater to brackish pools (billabongs) behind mangroves and in brackish swamps.

Remarks
Specimens in the Australian Museum collections from Torres Strait and Cape York appear to be conspecific with the type material of Stenothyra paludicola from Merauke in New Guinea. The shells are very similar in size and shape, although sculptural details were not visible on the available photograph of the holotype. Unfortunately, the lack of freshly collected material prevents observation of the colouration and behaviour of the animal, and molecular data could not be obtained. There are sufficient morphological grounds for separating the other subspecies of S. paludicola from S. paludicola paludicola, with the shell of the latter subspecies having characteristic spiral bands of minute pitted sculpture.
Protoconch (Figs 12H, I), operculum and radula (n = 3) (Fig. 13F). As for S. paludicola paludicola. External morphology and colouration in life. The animal has not been observed alive in this study. Specimens preserved in 70% EtOH have an unpigmented snout and scattered black flecks on the base of the cephalic tentacles.
Male and female reproductive systems. As for S. paludicola paludicola.
Distribution (Fig. 7C) Found in brackish to freshwater billabongs, rivers and streams in coastal to inland (but still tidally influenced) areas of the Northern Territory, between the western coast of the Gulf of Carpentaria, through Arnhem Land and Kakadu to the Cobourg Peninsula.

Remarks
Stenothyra paludicola topendensis is distinguished from S. paludicola paludicola by its unsculptured, glossy shell with more convex upper whorls. It is more similar to S. paludicola timorensis n. subsp., but has a slightly taller spire. It is also geographically isolated from the other subspecies of S. paludicola. Within-species sequence divergence of the COI gene is 0.43% and divergence from S. paludicola timorensis n. subsp. is 3.00% (Table 3).

Etymology
Named for the distribution of this subspecies along the tropical coastline of the Northern Territory, a region known to Australians as the 'Top End'.

Material examined
Holotype
Distribution (Fig. 7C) Known only from two shallow, tidal streams in East Timor, in the Manufahi and Lautem districts.

Remarks
This subspecies of S. paludicola is geographically isolated on the island of Timor. It has a similar shell to S. paludicola topendensis but is shorter spired and has less strongly convex upper whorls. COI sequence divergence within the subspecies is 0%, and divergence from S. paludicola topendensis n. subsp. is 3.00% (Table 3).

Etymology
Named for the distribution of this species in East Timor.

Remarks
Stenothyra frustillum was described by Benson (1856) from a specimen in the Cuming collection, with the type locality as 'Australia'. Benson (1856) provided a brief description of the species but did not figure the type specimen, and it has not been subsequently figured or described by any author. The shell is small (1.5 mm long by 1 mm wide) and smooth, but otherwise typically stenothyrid. A recent request for the holotype revealed that it is lost from the collections of the Museum of Natural History in London (J. Ablett, pers. comm.). It is possible that this small species was based on a specimen of what is here described as S. gelasinosa. A specimen 1.5 mm long could belong to this taxon, but without type material, precise locality or detailed original description of S. frustillum it is impossible to resolve its identity. Consequently I consider Stenothyra frustillum a nomen dubium.

Discussion
The systematic results of this study have significantly increased the known stenothyrid diversity in the Australasian region. Prior to this study, one Australian species, S. australis (with two subspecies) and three New Guinean species were recognised, with no stenothyrids known to occur in East Timor. Here the two subspecies of S. australis have been synonymised with S. australis and S. nebularum is a probable synonym of that species. Three subspecies of the newly-named species S. gelasinosa are thought to be endemic to Australia and there are three subspecies of S. paludicola (one endemic to northern Australia, one occurring in both Australia and New Guinea, and one subspecies endemic to East Timor). Stenothyra saccata from New Guinea is still recognised as a valid species, but its relationship to the Australian fauna is unresolved. Clarification of the diversity of stenothyrids in this region will greatly improve our capacity to identify these species in future studies of estuarine gastropods. These findings should provide a basis for further species discovery in the region, as this revision is certainly not a complete account of their diversity. In particular, temporary habitats such as seasonal brackish to freshwater pools may be under-represented in surveys of tropical areas due to the difficulties associated with field work during the wet season. These pools and billabongs have been identified as key habitat for S. paludicola, and they may contain further undescribed stenothyrid species in other parts of Australia, New Guinea and South East Asia. This study is the first use of molecular data to examine stenothyrid systematics and phylogeny. Analysis of sequence divergence and phylogenetic relationships demonstrates the utility of this data for species discrimination in Stenothyridae. Comparison of within-taxon and between-taxon uncorrected sequence divergence provided a sound basis for determining specific, subspecific and population-level differences. Less than 1.5% divergence of the COI gene was considered to meet intraspecific variation, while 3-5.5% was evidence of subspecific diversity and >10% was found between species. In this study, values of sequence divergence fell consistently into these three categories, without any intermediate values. Observation of mitochondrial sequence divergence in marine invertebrates has previously shown that widely distributed species have higher levels of within-taxon sequence divergence (genetic structure), relative to species occupying a narrower geographic range (for example, Waters et al. 2005). Contrary to this expectation, S. gelasinosa gelasinosa has the lowest within-taxon COI sequence divergence (0.17%) among the Australian and East Timorese taxa, but is widely distributed along the coastline of eastern Australia. The extent of the distribution is largely due to the discovery of an isolated population in southern New South Wales (north of Sydney), detectable during the summer months but seemingly not in winter. It is possible that gene flow is maintained by the southerly distribution of larvae by the Eastern Australian Current, which extends to Sydney, allowing for population of suitable habitat during the warmer months but resulting in local extinctions in cooler seasons.
The collection of molecular data from multiple specimens sampled from a wide range of locations provided some insights into the phylogeographic patterns within and between species. While some taxa were widespread, others were found to be endemic to a single region. This contrast is most evident in S. australis and S. gelasinosa, both of which are found (often sympatrically) in the same mangrove forest microhabitat. Stenothyra australis is widespread across the northern coastline of Australia, from at least Brisbane to Karratha, with low genetic diversity across its broad range. Stenothyra gelasinosa occupies an almost identical range (extending further south on the east coast), but is separated into three molecularly and morphologically distinct subspecies. The distributions of these subspecies are abutting in one instance (across the tip of Cape York Peninsula) and separated by an unknown distance in another (the Kimberly region) (Fig. 7B), but molecular data confirms that, at least as far as the present material is concerned, they are not sympatric. These contrasting phylogeographic patterns in closely related species cannot be explained with the present data, but may reflect different life history or dispersal strategies.
Molecular phylogenetic analysis of Stenothyridae using multiple representatives of each taxon was a useful tool for resolving taxa, but was less informative about the deeper relationships in the group. A few sister-species relationships were well supported (e.g. S. gelasinosa and S. cf. divalis from Hong Kong), but many nodes lacked strong support from the BI and ML analyses. Based on this information, it is difficult to reconstruct the evolutionary history of Stenothyridae. This situation may be improved by the addition of other species, in particular from freshwater habitats. Only one truly freshwater species was included in this study, an unidentified species from Sampaloc Lake on Luzon in the Philippines, which was positioned at the base of the tree. It is premature to speculate on the relationships between freshwater and estuarine species, but future studies may explore this interesting dimension of stenothyrid evolution.
Low sampling density and a lack of support for deeper nodes in the tree also prevent an assessment of genuslevel nomenclature in Stenothyridae. Almost all named stenothyrids have been assigned to Stenothyra, a situation which poorly reflects their global diversity and subtle but distinct variety of shell and anatomical morphologies. Some clades recovered in the molecular analysis are morphologically distinct, such as S. monilifera and an unidentified species from Johor in peninsular Malaysia. These taxa both have larger shells etched all over with distinctive spiral rows of pits. Too little is known of these species to speculate on the need for generic recognition, but it is likely that further genus-level taxa will be required when a comprehensive systematic revision of Stenothyridae is undertaken.
The more broadly sampled phylogenetic analysis exploring the monophyly of Stenothyridae and its relationship with other truncatelloids produced a similar tree to that of Criscione and Ponder (2012). This result is unsurprising since the majority of the non-stenothyrid sequences were first used in that study, but increased sampling in Stenothyridae did not greatly alter the family-level relationships.
Anatomical examination of these species was restricted to external morphology and the details of the reproductive system and radula. The configuration of the reproductive anatomy was of particular significance because of inconsistent reports in the literature. Kosuge (1969) described the anatomy of three Japanese species, S. edogawaensis (Yokoyama, 1927), S. thermaecole Kuroda, 1962and S. japonica Kuroda, 1962. Subsequently, Davis et al. (19861988) produced a detailed anatomical description of the Chinese species S. divalis (Gould, 1859), S. jinghongensis Davis, Guo and Hoagland, 1986and S. hunanensis Möllendorff, 1888. Davis et al. (1986 questioned the many discrepancies between the two anatomical accounts, in particular the lack of pallial prostate and configuration of the sperm storage organs and ducts in the female reproductive system of the Japanese species, as reported by Kosuge (1969).
The configuration of the reproductive system of three Australian taxa is presented here, and was found to be consistent among those taxa. When compared to the previous descriptions, the proximal male reproductive system is almost identical to the arrangement described by Kosuge (1969), with a bulbous, peanut-shaped seminal vesicle unlike the simple coiled portion of the vas deferens described for S. divalis (Table 7). However, a pallial prostate is definitely present, as reported by Davis (1986), but which was found to be absent by Kosuge (1969). The female reproductive system is structurally very similar to that of S. divalis, S. jinghongensis and S. hunanensis, with openings at both the posterior and anterior mantle cavity, single bursa duct connecting the bursa bulb to the common sperm duct, spherical seminal receptacle and an accessory sperm pouch. Kosuge's (1969) description of the female reproductive tract does not have much in common with the species examined here, and may describe the anatomy of a pomatiopsid rather than a stenothyrid due to the absence of an accessory sperm pouch and connection of the sperm duct to the bursa. The anatomy of the Japanese stenothyrids should be revisited to confirm or refute the reproductive ground plan of Stenothyridae proposed by Davis et al. (1986) and confirmed here for the Australian species.
Comparison between the Australian, Timorese and Chinese stenothyrid species for which detailed anatomical descriptions are available reveals several useful characters for separating species and inferring relationships. The seminal vesicle is either an elongate convolution of the vas deferens (Chinese species), or a discrete bulbous structure (Australian/Timorese species). Anterior male and female reproductive structures are also informative, including the shape of the penial stylet (narrow or spatulate), ciliation patterns on the penis, enlargement of the ejaculatory duct (unique to S. jinghongensis), position of the accessory sperm pouch and thickness of the bursa duct (Table 7). With this very limited dataset, it is already possible to identify discrete anatomical characters, or combinations of characters, that distinguish species and groups of species. The addition of further taxa and anatomical characters will permit an examination of evolutionary patterns in Stenothyridae.

Supplementary data
Supplementary file 1. Full list of material examined.