Five new cryptic freshwater gastropod species from New Caledonia (Caenogastropoda, Truncatelloidea, Tateidae)

Abstract During the course of a project aiming at the reconstruction of the colonization of the South Pacific islands by tateid gastropods based on molecular data we discovered five new species on New Caledonia belonging to the genera Hemistomia and Leiorhagium, respectively. We describe these species based on morphological, anatomical and genetic data. All five species are morphologically cryptic as they closely resemble or are even indistinguishable from known species stressing the importance of a comprehensive taxonomic approach integrating several methods. As a consequence of their small and fragmented geographic ranges and the rapidly progressing anthropogenic land cover changes on New Caledonia, all five species qualify as critically endangered according to the criteria of the IUCN.


Introduction
New Caledonia is famous for being a biodiversity hotspot harboring a high number of endemic species (Myers et al. 2000) including a radiation of small freshwater gastropods belonging to the family Tateidae. This radiation is probably of Oligocene origin and comprises more than 50 species in seven genera Bouchet 1998, Zielske andHaase 2015). Many of these species are extreme narrow-range endemics known from only few or single sites (Haase and Bouchet 1998), a pattern typical for Truncatelloidea in freshwaters worldwide (e.g. Giusti and Pezzoli 1980, Radoman 1983, Haase 1996, 2008, Ponder and Colgan 2002, Liu and Hershler 2005, Hershler et al. 2011, Delicado and Ramos 2012. In the frame of a project aiming at the reconstruction of the colonization of the South Pacific islands by tateids based on molecular data Haase 2014a, b, 2015, Zielske, Ponder andHaase in preparation) we visited New Caledonia in May 2012 in order to collect suitable material for sequencing. During this expedition we found five new species of the genera Hemistomia Crosse, 1872 and Leiorhagium Haase & Bouchet, 1998, respectively (Figs 1, 2), which we describe herein based on morphological, anatomical and genetic data. All five species qualify as morphologically cryptic as they closely resemble or are even indistinguishable from known species (see Pfenninger and Schwenk 2007). The discovery of new cryptic species was predicted by Haase and Bouchet (1998), whose revision was based solely on morphology and anatomy. In general, cryptic species are common among different spring snail families of Truncatelloidea (e.g., Liu et al. 2003;Haase et al. 2007;Delicado and Ramos 2012;Collado et al. 2013).

Material and methods
Snails were fixed in 70% ethanol in the field, transferred to propylene glycol for shipping by courier, and returned to ethanol, this time 96%, after arrival in our lab. For measurements, up to 20 snails per sample were photographed under a Zeiss SteREO Discovery.V20 dissecting microscope with a Zeiss Axio Cam MR3. Five dimensions -shell height, shell width, aperture height, aperture width, body whorl width -were measured using the program AxioVision 40 V4.8. (Zeiss) and whorls counted to the nearest eighth (Kerney and Cameron 1979). Up to six shells were dissolved in diluted hydrochloric acid for dissections. Anatomies were photographed as well. These digital images served as template for drawings made on a graphical tablet. For scanning electron microscopy up to three shells, radulae and opercula were cleaned in 5% sodium hypochlorite. The cephalopodia of up to two males were dried using hexamethyldisilazane (Nation 1983). After sputter coating with gold objects were investigated in a Zeiss EVO LS10 Scanning Microscope.
Morphometric analyses of shell measurements including canonical variates analyses (CVA) maximizing the differentiation of groups in multivariate space, multivariate analyses of variance (MANOVA), assignment tests, and Hotelling's T 2 -tests were conducted in PAST 2.12 (Hammer et al. 2001). Sequential Bonferroni-correction was applied to multiple tests. These analyses also included samples of known, similar species the new ones could be mistaken for (Table 1). The selection of species used in comparisons was based on the phylogenetic analysis.
Phylogenetic analyses were based on a selection of sequences generated by Zielske and Haase (2015), who analyzed fragments of the mitochondrial genes cytochrome oxidase subunit I (COI) and 16S rRNA as well as the nuclear internal transcribed spacer 2 (ITS2). For lab protocols see Haase (2014a, 2015). We restricted the analysis to 3 specimens per species at most and used Kanakyella gentilsiana, Crosseana crosseana, and C. melanosoma as outgroups (Table 1). The alignment of 16S rRNA and ITS2 was generated using secondary structure information using RNAsalsa 0.8.1 (Stocsits et al. 2009) (for details see Zielske and Haase 2015) and checked for ambiguous and randomly similar sites in Aliscore 2.0 ). We defined seven partitions. PartitionFinder 1.1.0 (Lanfear et al. 2012) identified the following scheme and substitution models as optimal among all possible combinations of separate and merged partitions: COI 1 st positions (TrNef+I), COI 2 nd positions (F81), COI 3 rd positions (TVM+I+Γ), 16S rRNA loops (TrN+ Γ), ITS2 loops (TrNef+I+Γ), joint stems of 16S rRNA and ITS2 (K80+I). With these settings, tree reconstructions were conducted in a maximum likelihood (ML) frame-  also Table 1). Table 1. Locality data of all samples and GenBank accession numbers of specimens represented in phylogeny (see also Fig. 1). The last three species represent the outgroup. Specimens are only distinguished in two cases where more than 1 sequence per sample was used. For museum catalog numbers of NeCa-sample voucher material see Zielske and Haase (2015). Paratypes of species described by Haase and Bouchet (1998) (Zwickl 2006) with 500 replicates. Robustness was assessed by bootstrapping with 200 replicates. Type and non-type material is deposited at the Museum National d'Histoire Naturelle in Paris (MNHN) and at the Naturhistorisches Museum Wien (NHMW).

Systematic descriptions
Diagnoses and descriptions of Hemistomia and Leiorhagium and data used in our comparisons with the new species were provided by Haase and Bouchet (1998). Locality data include site number, district capital, site, coordinates, and date of collection. Shell measurements are given in Table 2 and not repeated in the descriptions. Type locality. NeCa 12, Bouloupari: Ouaméni-valley, small stream on W-side of road in secondary forest, 21°49'46.9"S, 165°56'42.9"E, 22 May 2012.

Genus
Etymology. The new species is dedicated to the senior author's daughter on the occasion of her 'quinceañera', the 15 th birthday.
Diagnosis. H. andreae sp. n. is very similar to H. cockerelli and H. nyo. It differs from both in a clearer separation of the opercular pegs and a much more delicate penis. The protoconch of the new species has more whorls than H. nyo and the palatal denticle is further behind the outer lip.
External features. Epidermis without pigment, eyes black.
Remarks. This is Hemistomia sp. n. 1 of Zielske and Haase (2015). Both H. andreae sp. n. and H. cockerelli do have the columellar fold in the body whorl assumed to be unique in H. nyo by Haase and Bouchet (1998). H. andreae sp. n. is only known from the type locality.

Genus Leiorhagium Haase & Bouchet, 1998
Type species. Leiorhagium orokau Haase & Bouchet, 1998    Etymology. Adioincola is composed of the name of the area of Adio and the Latin noun incola meaning inhabitant, and thus refers to the type locality of the new species.
Diagnosis. L. adioincola sp. n. is very similar to L. kavuneva and L. clandestinum sp. n. The former pair differs in penial shape, slender vs. basally broad with long terminal filament. L. adioincola sp. n. tends to have fewer radular denticles than L. kavuneva. Genetically, these species differed on average at 9.65% of the positions of COI. Due to the lack of anatomical data, both new species can only be distinguished genetically. Their COI sequences differed on average by 9.5% (p-distance).
Remarks. This is Leiorhagium sp. n. 4 of Zielske and Haase (2015). L. adioincola sp. n. occurs in the area between the villages of Nétéa and Goipin including the Massif d'Adio. Etymology. The new species is named after the Aremu valley, where it has been discovered.
Diagnosis. L. aremuum sp. n. is most similar to L. ajie, which is, however, larger and slightly more slender, lacks the palatal denticle, and has a more massive penis. The prolonged capsule gland is unique among New Caledonian tateids. The COI sequences had a p-distance of 9.4%.
Male genitalia. Proximal end of lobate testis 1 whorl below apex, comprising c. 0.75 whorls, covering proximal end of stomach; vesicula seminalis arising from distal third of testis; penis very long and slender (N=2) (Fig. 8D). Remarks. This is Leiorhagium sp. n. 3 of Zielske and Haase (2015). L. aremuum sp. n. is only known from the type locality. Etymology. The Latin adjective clandestinus means clandestine and refers to the new species' external identity with L. kavuneva.
Diagnosis. L. clandestinum sp. n. is most similar to L. adioincola sp. n. and L. kavuneva. For the distinction from L. adioincola sp. n. see above. Due to the lack of anatomical data, L. clandestinum sp. n. and L. kavuneva can only be distinguished based on 7.6% average sequence divergence of COI (p-distance).

External features. Epidermis without pigment, eyes black.
Remarks. This is Leiorhagium sp. n. 2 of Zielske and Haase (2015). L. clandestinum sp. n. is only known from the type locality.
Etymology. The new species is named after the village of Nétéa, which is closely proximal to our collecting locality.
Diagnosis. L. neteae sp. n. is very similar to L. douii and L. montfaouense. In L. neteae sp. n. the palatal denticle is slightly larger and 1/8 whorl further behind the outher lip. The operculum has only a single denticle compared to 2-3 in L. douii and L. montfaouense. The distal loop of the renal oviduct of the new species forms a 270° loop counter-clockwise, whereas in the other two species this part of the oviduct is bent 180° clockwise. The penis of L. neteae sp. n. is long and slender in contrast to the other species, where it has a broad base and a very long filament.
Remarks. This is Leiorhagium sp. n. 5 of Zielske and Haase (2015). L. neteae sp. n. is only known from the type locality.

Morphometry
The CVA plot (Fig. 9) comparing species of Hemistomia shows the high variability of H. cockerelli. The associated MANOVA was highly significant (Wilk's λ = 0.062, DF 1 = 35, DF 2 = 490.4, F = 13.16, p = < 0.001). Many pairwise comparisons of populations were significant as well (Table 3). H. nyo and H. andreae sp. n. fell within the variation of H. cockerelli. According to the CVA, they were not more different from each other than from populations of H. cockerelli. Assignment and jacknifed assignment tests allocated 80 (62.5%) and 67 (52.3%) of a total of 128 shells to their original sample indicating the considerable overlap of shapes.

Discussion
Our phylogenetic analyses based on DNA sequence data confirmed the suspicion of Haase and Bouchet (1998) that additional cryptic species in this snail fauna will be identified once molecular methods are applied emphasizing the huge morphological variability of certain nominal species. Recent accounts on tateid gastropods from Vanuatu and Fiji (Zielske and Haase 2014a, b) have revealed extensive radiations of morphologically very similar species. However, in contrast to the New Caledonian taxa, the radiations on those archipelagos are comparatively young (Zielske and Haase 2015). Four of the five species described here are hardly distinguishable from known taxa based on measurements despite being genetically well differentiated with even uncorrected distances (see Fregin et al. 2012) of at least 7.4% to their next similar congeners. Whether this means that morphologically similar species occupy similar niches is impossible to tell at this stage because the relationship of shell morphology to habitat has not been investigated among truncatelloidean gastropods except for a few accounts on Potamopyrgus antipodarum (Haase 2003, Holomuzki and Biggs 2006, Kistner and Dybdahl 2013. Although ranges overlap or are contiguous, sibling species have not (yet) been encountered in sympatry, i.e. in the same spring or stream.
The new species provide an additional truncatelloid example stressing the importance of an integrative taxonomic approach combining morphological, anatomical and genetic methods (e.g. Haase et al. 2007, Delicado andRamos 2012). Given the mosaic nature of evolution of these small gastropods with morphologically as well as genetically cryptic species (e.g., Haase et al. 2007, Haase 2008, Zielske et al. 2011, Delicado and Ramos 2012, Liu et al. 2013), we do not adhere to a fairly strict scheme of species identification as advocated elaborately e.g. by Schlick-Steiner et al. (2010). Instead we advocate the approach of Padial and de la Riva (2010) who have a more natural vision of the evolutionary processes potentially involved in speciation. For instance, they acknowledge that the congruence of different character sets, pivotal for taxonomic decisions for Schlick-Steiner et al. (2010), may be plesiomorphic.
Genetic differentiation was an important indicator of species status. Pairwise pdistances > 7.4% are far above any threshold suggested by advocates of barcoding (e.g., Hebert et al. 2003Hebert et al. , 2004Ratnasingham and Hebert 2007). However, again we do not adhere to a strict scheme as there may be no mitochondrial differentiation between good species as well as considerable variation within species of spring snails (e.g. Haase 2008;Zielske et al. 2014a; see also Fregin et al. 2012). That genetic differentiation does reflect species status for the new taxa is also indicated by the comparison of their branch lengths to branch lengths among morphologically well defined species in our phylogenetic analysis.
While conducting our morphometric analyses we appreciated that the measuring methods applied for the material described previously (Haase and Bouchet 1998) and for this account are incompatible. Obviously, using an ocular micrometer fitted to a dissecting microscope produced inaccurate data, although the measurements were quite consistent judging from the fairly low coefficients of variation, which were of a similar order of magnitude as those computed for the present data. Therefore, we had to re-measure the old samples used in our comparisons.
Another methodological problem almost expectedly occurred in the field. All collections made for our previous monograph (Haase and Bouchet 1998) were georeferenced from maps. This proved to be fairly inaccurate when we tried to relocate sites in 2012 guided by GPS. Additional difficulties arose from recent road development and land-use changes. Many villages are now accessible on much broader roads than 20 years ago. Construction has obviously destroyed small road-side springs and seepages and changed the course of streams. Other sites were destroyed by extensive fires affecting entire valleys or hills. Crosseana melanosoma, in our analysis part of the outgroup, used to be common when first collected in 1992. Now we found only a few specimens. It remains to be seen whether there are other (extant) populations in the unexplored hinterland of Boyen. In contrast, H. yalayu, collected in a few seepages on Col d'Amoss in the far Northeast in 1989, is now probably extinct. The entire area has lost its primary vegetation. Today, the fire resistant niaouli (Melaleuca quinquenervia) and shrubland are dominating and streams harbor a very depauperate fauna.
Four of the five new species were found in single sites and the fifth was found at only two sites. Considering the vulnerability of small habitats like springs and the rapid anthropogenic development and changes on New Caledonia just outlined immediately raises concern regarding the chances of long-term survival of these species (see also Haase et al. 2010). Most sites we surveyed were rather easily accessible, close to roads, so that one can assume that there are other populations deeper in the forests or forest remnants. Nevertheless, given that the area of occupancy of each species is certainly less than 10 km 2 , that ranges of spring snails are almost naturally severely fragmented, and the rapidly progressing change of land cover, areas of occupancy as well as habitat, hence the numbers of populations will decline. Therefore, all five species and probably the majority of New Caledonian tateids qualify as critically endangered according to the criteria (CE, B2,a,II-IV) of the International Union for Conservation of Nature (IUCN 2012).