Gastrocopta (Mollusca, Gastropoda, Pupillidae) in the Pilbara region of Western Australia

Abstract Six species of Gastrocopta have been identified from the Pilbara region, Western Australia, by means of comparative analyses of shell and mtDNA variation. Three of these species, Gastrocopta hedleyi, Gastrocopta larapinta and Gastrocopta servilis, have been recorded in the Pilbara for the first time. Gastrocopta sp. CW1 is probably new to science and might be endemic to the region. By contrast, Gastrocopta hedleyi, Gastrocopta larapinta and Gastrocopta mussoni are shown to be widespread.


Introduction
Gastrocopta Wollaston, 1878 is the most speciose pupillid genus in Australia with twelve recorded species (Pokryszko 1996). Its members are found throughout most of Australia except for the humid south-west and south-east corners of the continent (Solem 1991;Pokryszko 1996;Stanisic et al. 2010). The Australian taxa have most recently been revised based on comparative shell morphology by Solem (1986Solem ( , 1989 and Pokryszko (1996). Both works disagree on some details, mainly the morphological separation of Gastrocopta larapinta and Gastrocopta mussoni and the taxonomic distinctness within the size-variable Gastrocopta margaretae complex. Molecular studies that might help to resolve the taxonomic discrepancies have remained unavailable.
Previous works have focussed mainly on the northern, eastern and southern parts of coastal Australia and to a lesser degree on the mid-west and central parts of Australia (Pilsbry 1917;Iredale 1939;Solem 1986Solem , 1989Solem , 1991Slack-Smith 1993;Pokryszko 1996; while the fauna in Western Australian has remained poorly documented. In Western Australia most pupillid specimens have been collected along main roads of the more coastal areas and along major inland roads, but the interior of Western Australia has so far been widely neglected. Being of small size (maximum dimension less than 6mm) and cryptic in nature, pupillids are often ignored when documenting land snail diversity (Nekola 2009). The lack of specimens from inland areas of Western Australia has made it difficult to determine the relationships between west coastal specimens and those from central and eastern Australia. Pilsbry (1917) in his world monograph on the subfamily Gastrocoptinae had little Australian material, except of a few types and vouchers received from Tate (from Central Australia) and Hedley (mostly from Eastern Australia). Solem (1986) was the first author to revise the Australian fauna more comprehensively. He examined the Pupillidae from the south and mid-west coasts of Western Australia and later (Solem 1989) the non-camaenid families (including Pupillidae) from the Kimberley, Northern Territory and Red Centre regions. A second major revision by Pokryszko (1996) extended the area of review in Western Australia only slightly, because only a small amount of additional material from the Western Australian Museum was studied (just 9 lots) and probably because the collection was little expanded since Solem (1986) examined the collections.
Since Pokryszko's (1996) revision, the Western Australian Museum collection of Gastrocopta in the Pilbara region has greatly expanded. Most of this collecting has been associated with expanding mineral operations in the region and improved vehicle access to remote areas. A Western Australian Museum fieldtrip during August 2009 visited the eastern Pilbara area, collected macro-and micro-non-marine molluscs and significantly increased the pupillid collection in that region. This paper (1) presents new data on Gastrocopta in the Pilbara, establishing new records and range extensions; (2) tests the taxonomic significance of morphological characters commonly used for the identification and delimitation of species by using a mitochondrial phylogeny; (3) provides comparative remarks on shell morphology of Gastrocopta species; (4) indicates systematic issues that require clarification by further studies. For detailed comparative analyses of shell characters we refer to Pokryszko (1996) and Solem (1986Solem ( , 1989.

Methods
All Gastrocopta material from the Pilbara in the malacological collections of the Western Australian Museum was examined. Additional specimens from the private collection of Mr Vince Kessner and from the collection of the Field Museum of Natural History, Chicago were also included. In total 545 Gastrocopta lots were studied with distributional maps being plotted by use of the online vector map software available at www.planiglobe.com.
Species identifications were based on shell characters, with particular emphasise on the size, shape and quantity of apertural barriers. Specimens were photographed and measured using a Leica MZ16A microscope with Leica DFC500 camera. DNA was extracted from entire specimens taken from their shell by use of a QIAGEN DNA extraction kit for animal tissue following the standard procedure of the manual. Fragments of the mitochondrial 16S rRNA (16S) and of the COI genes were amplified by PCR using the primer pairs: 16Sar and 16Sbr (Palumbi et al. 1991), and L1490 and H2198 (Folmer et al. 1994), respectively. Reactions were performed under standard conditions with an annealing step of 60 s at 55 °C for 16S and at 50 °C for COI. Both strands of purified PCR fragments were cycle sequenced by use of the PCR primers. Electropherograms were manually corrected for misreads, if necessary, and forward and reverse strands were merged into one sequence file using CodonCode Aligner v. 3.6.1 (CodonCode Corporation, Dedham, MA). Sequences have been deposited in GenBank (CO1: KC143966-KC143993, 16S: KC143994-KC144020). Sequence alignments were generated using MUSCLE as implemented in MEGA5 (Tamura et al. 2011). Uncorrected pair-wise genetic distances were calculated using MEGA5 under the option 'pair-wise deletion of gaps'. Prior to the model-based phylogenetic analyses, the best-fit model of nucleotide substitution was identified for each gene fragment using the model proposal function of MEGA5. To infer phylogenetic relationships, we performed Maximum Likelihood (ML) analyses using MEGA5 with Nearest-Neighbor-Interchange (NNI) as heuristic method and automatic generation of the initial tree. Two-hundred ML bootstrap replicates were performed to assess the topology support.
Abbreviations used for depositories of material are: FMNH, Field Museum of Natural History, Chicago, United States; VK, Vince Kessner Private Collection, Adelaide River, Australia; WAM, Western Australian Museum, Perth, Australia. For shell aperture barrier terminology we followed Pokryszko (1996), reproduced here in Fig. 1.

taxonomic part
Six species of Gastrocopta were recorded from the Pilbara region (Table 1). Four species are endemic to Australia, one species is introduced and one species requires further investigation (Gastrocopta sp. CW1). Another species, G. bannertonensis was only collected from the inner mid-west region of Western Australia and was not discussed in this paper. Pilsbry, 1917 http://species-id.net/wiki/Gastrocopta_hedleyi Fig. 2B Gastrocopta hedleyi Pilsbry 1917hedleyi Pilsbry [in 1916hedleyi Pilsbry -1918: 166-167, pl. 27, figs 1-4;Solem 1991: 250;Pokryszko 1996Pokryszko : 1104Stanisic 1998: fig. 17   Distribution. This species has previously been recorded from northern New South Wales and from scattered localities in northern Queensland (Cape York Peninsula), central Australia (Glen Helen area) and northern Western Australia (King Leopold Ranges) (Pokryszko 1996). In addition, it is now recorded from the Hamersley Ranges, the Burrup Peninsula and a few isolated sites from approximately 100 km SSE of Port Hedland in the Pilbara region ( Figure 3).

Gastrocopta hedleyi
Comparative morphology. G. hedleyi shells are slightly smaller (shorter) than those of other Gastrocopta species (excluding G. sp. CW1) recorded from the Pilbara. They typically have (1) a large, usually rounded (sometimes acute) columellar tooth that is drooping at the anterior end (2) a high, strongly convergent upper palatal tooth (3) a long, high, strongly twisted parietoangular tooth that usually comes in close proximity to the upper palatal tooth (4) a prominent infraparietal tooth that is sometimes prolonged as thin ridge on parietal wall (5) often a strong basal tooth (6) very occasionally with a weak interpalatal tooth.
Some G. hedleyi shells (particularly more elongate specimens) can be difficult to separate from the ovate form of G. mussoni but (1) are smaller (slender) when sympatric (2) have a less rounded body whorl (3) have a more strongly sigmoid lower palatal tooth (4) have a larger upper palatal tooth that is usually strongly convergent with the lower palatal (5) have a longer, more strongly twisted parietoangular tooth (6) have a larger, more rounded columellar tooth that is usually drooping at the anterior end .
The cylindrical form of G. mussoni is also very similar to G. hedleyi but (1) has a lower, shorter and less twisted parietoangular tooth (usually at 45 o angle in apertural view) (2) has a shorter and less sigmoid (usually straight) lower palatal tooth (3) generally lacks an infraparietal tooth (4) has a smaller upper palatal tooth (occasionally slightly convergent with lower palatal) (5) has a more acutely angled, slanted columellar tooth, rarely drooping at the anterior end.
Remarks. There is considerable variation in the shell size and barrier length of specimens identified as G. hedleyi during this study. Many specimens grouped as G. hedleyi from the eastern Hamersley Range (eg. Wonmunna, Kalgan Pool) have reduced barriers and often a lower parietoangular tooth (nearing 45 o angle in apertural view) but a large series shows a progression to shells that typically possess a large, strongly convergent upper palatal tooth and a strongly twisted parietoangular tooth. Solem (1989) mentioned that G. hedleyi was somewhat similar to G. pilbarana, although in that case he was actually referring to the ovate form of G. mussoni (see section on G. mussoni).
Typical G. larapinta shells with a small interpalatal tooth (or very occasionally no interpalatal tooth) appear considerably more variable in apertural barrier structure (particularly in high calcareous soils), making their separation from the cylindrical and elongate-ovate forms of G. mussoni difficult. As such the following separation is tentative. G. larapinta shells are typically (1) slightly to moderately larger (obese) (2) have a slightly smaller, more rounded columellar tooth (3) usually a much shorter parietoangular tooth that is positioned lower in apertural view (4) generally possesses an infraparietal tooth (5) often a slightly lower, less convergent upper palatal tooth (6) usually a less sigmoid lower palatal tooth.
Remarks. The separation of G. larapinta (small or no interpalatal tooth) with the cylindrical and elongate-ovate forms of G. mussoni has proved extremely difficult and a more detailed molecular study is required to resolve this issue. Pokryszko (1996) separated these G. larapinta specimens from cylindrical G. mussoni based on shell size (smaller) and columellar tooth angle (less acute). However, from the small genetic data available and from examination of many shells, G. larapinta shells were slightly, to moderately more obese.
Some of those near west coast specimens (Cy Creek) included as cylindrical G. mussoni contained mixed lots of G. larapinta and G. mussoni. Interestingly, Poykryszko had identified a few of these larger Cy Creek specimens as G. larapinta, but included those records as G. mussoni in her publication. Pokryszko (1996) also noted in a large lot of G. mussoni from central Australia (FMNH 201570) that some of the ovate G. mussoni were quite large (many having an interpalatal tooth) but we consider most of those to be G. larapinta with a small or no interpalatal tooth (see G. mussoni section).
It is possible Pokryszko (1996) was alluding to a slender form of G. larapinta when separating cylindrical G. mussoni and G. larapinta (no interpalatal tooth) but this does not reflect accurately in her identifications. During this study specimens from Kalgan Pool (WAM S58005) were unexpectedly grouped within the G. larapinta clade. These specimens, although slightly more slender, have proven difficult to separate from cylindrical G. mussoni specimens identified by Pokryszko (1996) as they (1) have a high, long, lamellate parietoangular tooth (2) have a separated angular tooth (3) have a slightly rounded to angled columellar tooth and (4) lack an infraparietal tooth. They may prove to be a subspecies of G. larapinta and in this sense, Pokryszkos' (1996) separation of cylindrical G. mussoni and G. larapinta (no interpalatal tooth) was correct.
As there is doubt surrounding the distinguishing morphological characters of cylindrical G. mussoni and G. larapinta shells with a small or no interpalatal tooth, the above separation is tentative and a more detailed genetic investigation is required.
Comparative morphology. Shells of G. margaretae are easily distinguished from other Gastrocopta species in the Pilbara by the presence of (1) a moderately to strongly folded columellar tooth (2) a generally large and transverse basal tooth (3) a high and long lower palatal tooth (4) an upper palatal tooth that is moderately to strongly convergent with the lower palatal (5) a weak to strong infraparietal tooth present.
Remarks. Solem (1986) maintained the separation of the west coast species G. wallabyensis from the south coast G. margaretae based on size (smaller) and length of apertural barriers (longer). He also described a new species, G. pilbarana, from the west coast but his separation of it from G. wallabyensis was vague. Solem later (1989) maintained the separation of the central Australian G. tatei from the above species but remarked it was somewhat similar to the west coast G. wallabyensis. Pokryszko (1996) later disagreed, synonymising all species with G. margaretae.
The few specimens sequenced from the south coast of Western Australia (WAM S32048, WAM S32052) could represent genetic isolation by distance or perhaps a different species from those on the west coast (WAM S42834) but more molecular data are required. The southern specimens are (1) much larger with reduced apertural barriers (2) more strongly rounded whorls (conical) and (3) consistently lack an infraparietal tooth. Specimens resembling the smaller west coast form (ie. long apertural barriers and weak to strong infraparietal tooth) have also been recorded from the south west area of Western Australia (Whisson, pers. comm.) where it is often sympatric with Gastrocopta bannertonensis (Gabriel, 1930). It is not known whether there is a continuous distribution between the two areas. Until more detailed molecular work is undertaken we have maintained Pokryszko's (1996) systematic positions.
Comparative morphology. Cylindrical and elongate-ovate forms of G. mussoni can be mistaken for G. larapinta specimens (with a small or absent interpalatal tooth) but (1) are slightly to moderately slender (2) have a higher, usually longer parietoangular tooth (3) have a larger, more strongly slanted and acutely angled columellar tooth, with its posterior edge often forming a prominent wide ridge along the columellar wall (4) quite frequently have an upper palatal tooth that is slightly (occasionally moderately) convergent with the lower palatal (see also earlier section on G. larapinta) . G. mussoni very occasionally possesses a small interpalatal tooth, usually located close to the upper palatal.
The typical ovate form of G. mussoni can be confused with G. hedleyi (particularly those with reduced apertural barriers) but (1) are larger (obese) when sympatric (2) have a less sigmoid lower palatal tooth and (3) have a smaller, less convergent upper palatal tooth (see also earlier section on G. hedleyi).
Remarks. There appears to be two size forms in G. mussoni, the larger ovate form and smaller, slender cylindrical form, and in agreement with Pokryszko (1996) both are confirmed as ecological phenotypes from the CO1 and 16S sequences. Based on specimens identified by Pokryszko (1996) and during this study, there is enormous variation in shell shape, shell size and apertural barrier structure between and including these two forms.
The ovate form of G. mussoni appears to be most common in the Pilbara. Prior to Pokryszko's 1996publication, G. pilbarana Solem, 1986 was described from the Shark Bay area with an isolated record from the Chichester Range (north of Roy Hill). This species was synonymised with G. margaretae (Cox, 1868) by Pokryszko (1996) although the Chichester Range paratype was not included in that study. This Chichester Range record is actually the ovate form of G. mussoni.
Some of those specimens tentatively identified as the elongate-ovate form of G. mussoni during this study (Wonmunna; Cy Creek; Cloud Break, Barrow Island) are (1) much larger (obese) than the usual elongate-ovate form (2) have the parietoanangular tooth lower (45 o ) (3) usually have a supraparietal tooth and (4) quite frequently possess a small interpalatal tooth. These specimens may prove to be the somewhat variable G. larapinta with a small or no interpalatal tooth, but in the absence of a larger series of specimens and more detailed molecular data, we have left them as G. mussoni.
The nature of many cylindrical G. mussoni identified by Poykrosko (1996) and during this study requires more work. It is probable we have lumped the slender form of G. larapinta from Kalgan Pool (no interpalatal tooth) with cylindrical G. mussoni. Fig. 2A Gastrocopta pilbarana Slack-Smith 1993: 91. Distribution. This species is recorded from the Cape Range and from an isolated site on Barrow Island (Figure 3).
Remarks. Solem (1989) identified specimens from the Kimberley and Northern Territory as G. recondita (Tapparone-Canefri, 1883) but in a later review, Pokryszko (1996) regarded that species as extralimital to Australia, describing the Australian representatives as a new sister species, G. stupefasciens.
G. sp. CW1. is very similar to G. stupefaciens and G. recondita but (1) is smaller (2) has longer apertural barriers and (3) has a thick, solid, non-lamellate columellar tooth and is here within regarded as a new species. Some of Solems` G. recondita specimens from limestone outcrops near Katherine (station WA-685) and Lake Argyle (station WA-248) have a similar columellar tooth structure and their relationship to G. sp. CW1 needs further work. Slack-Smith (1993) listed cavernicolus specimens from the Cape Range as G. pilbarana (which was later synonymised with G. margaretae) but those specimens were in fact G. sp. CW1. She suggested that although this population of snails was ameliorated with the limestone caves of the Cape Range, although it was not generally cavernicolus. The accumulation and breakdown of leaf litter within caves combined with calcareous rocks was deemed advantageous for snails. Solem (1991) discussed an affinity with limestone for his G. recondita. The few records of G. sp. CW1 from the limestone dominated Barrow Island and Cape Range show similar requirements.
Distribution. This species has previously been recorded from just north of Broome (Quondong Point) across northern Australia to mid-eastern Queensland and offshore islands (Solem 1989, Shea 2006). In addition, it is now recorded from a single locality within the Karratha town site (Figure 4).
Comparative morphology. The shells of G. servilis are easily distinguished from other Pilbara Gastrocopta by their (1) strongly rounded whorls (2) short, straight columellar tooth which is perpendicular to the mid-columellar wall (3) very long angular tooth which is fused with the parietoangular tooth (4) weak to absent basal tooth and (5) weak to absent upper palatal tooth.
Remarks. G. servilis has been a recent introduction to the residential gardens of Karratha.

Molecular phylogeny
Two mitochondrial gene fragments, COI and 16S, have been analysed. The data sets contained 27 sequences of Western Australian Gastrocopta (five each of G. bannertonensis and G. mussoni, 12 of G. larapinta, two or three, respectively, of G. margaretae, and two of G. hedleyi) as well as 16 Genbank sequences of several American Gastrocopta species that stem from the study of Nekola et al. (2012). Two to three sequences each of Vertigo spp. and Pupilla spp. were used as out-group to root the trees. Maximum Likelihood analyses of the COI and 16S fragments resulted in identical tree topologies (Figs 6-7). All species as delineated by their shell formed monophyletic sequence clusters. The six Australian species formed a monophyletic crown group nested amongst a basal assemblage of American lineages. The species G. hedleyi, G. mussoni and G. larapinta are more closely related with each other as are G. margaretae and G. bannertonensis, which corresponds well with columellar tooth structure i.e. large and ascending versus small and short, respectively. Intraspecific evolutionary divergences were on average 1% (max. 4%) in COI as well as on average 1% (max. 2%) in 16S in all Australian species but G. margaretae (Tables 2-3). In G. margaretae intraspecific genetic distances were found to be significantly higher than in any other Australian species (16% in COI and 5.3% in 16S). Apart from G. margaretae, the intraspecific divergence was about an order of magnitude smaller than the observed interspecific distances of 5-26% (on average 18%) in COI and 2-14% (on average 9%) in 16S. Only in G. margaretae did the amount of intraspecific genetic differentiation overlap with the range of interspecific genetic distances.    Tamura-Nei (1993). The rate variation among sites was modelled with a gamma distribution (shape parameter = 0.4).

Discussion
Based on shell morphology, all Gastrocopta species recorded here (except G. sp. CW1) have relatively large distributional ranges. Although limited, the molecular data supports the shell-based delineation of the six species recognized herein. The molecular data also confirms the large distributional ranges of G. larapinta and G. mussoni by including samples from areas that are about 100 and 550 kilometres apart, respectively. The apparently widespread distribution of Gastrocopta species is probably due to the common ability in which a single Gastrocopta adult can self-fertilize their eggs and establish a new population (Nekola 2009). They also have the ability to be transported large distances, through either their small and light structure (via wind and water) and/or by their nature to mucous seal to objects such as bark, leaves and vertebrates (Slack-Smith 1993, Nekola 2009). Some of the species G. hedleyi and G. sp. CW1 are at the southern limits of their range. It is probable that G. hedleyi has arrived in the Pilbara as a result of dispersal potential whereas G. sp. CW1, found only in the Cape Range in the Pilbara (and an isolated record from Barrow Island) might represent relictual populations from the Miocene. Both Cape Range and Barrow Island contain moist, well sheltered limestone gorges and caves.
Other recorded species represent a range extension from the red centre. These include G. mussoni and G. larapinta which are common in the Pilbara. This is not suprising given their affinity to arid or semi-arid environments, which persist throughout much of the Pilbara. Future collecting will no doubt show a mostly continuous distribution for these species between the Pilbara and the red centre.
The present CO1 and 16S molecular data set, although small (only 27 individuals sequenced) mostly supports the taxonomic revision of Pokryszko (1996) based exclusively on shell morphology (i.e. apertural barriers). However, more detailed molecular work is needed to sort out some systematic issues: (1) the relationship of the west coast and south coast populations of G. margaretae (2) the relationship of G. sp. CW1 to similar specimens in the Kimberley region (3) the morphological separation of G. larapinta and G. mussoni.
The Australian species are less well differentiated by means of evolutionary divergence than the American Gastrocopta species, which are separated from each other by interspecific pair-wise Tamura and Nei (1993) distances of 7.8-28% (on average 20.4%) in COI and 2.3-22% (on average 13.9%) in 16S. Evolutionary divergences of the Australian Gastrocopta species are also lower than the genetic distances found in other Western Australian land snails, such as the Camaenidae. In this group interspecific sequences in COI were usually larger than 6% (e.g., Köhler 2011; Köhler and Johnson 2012) (16S distances were not compared because the analysed gene fragments differed in length).
There appears to be tremendous variation in shell shape and size between and within populations of some Gastrocopta species, and often this is associated with variation in apertural barrier structure. This can make separation of species difficult, particularly G. larapinta, G. hedleyi and G. mussoni which share similar apertural barrier structures. It is advisable to collect a large series of individuals so the wide variation in apertural barrier structures can be seen.

Conclusion
In summary, G. hedleyi, G. larapinta, G. sp. CW1 and G. servilis are recorded from the Pilbara region for the first time. G. servilis has been a recent introduction to the residential gardens of Karratha. G. hedleyi, G. larapinta and G. mussoni were shown to be common across the Pilbara. G. sp. CW1 may represent an undescribed species.