All-inclusive descriptions of new freshwater snail taxa of the hyperdiverse family Tateidae (Gastropoda, Caenogastropoda) from the South Island of New Zealand

Four new species and one new subspecies of tateid freshwater gastropods are described from the north of the South Island of New Zealand, Catapyrgus jami sp. nov., Opacuincola lisannea sp. nov., O. gretathunbergae sp. nov., O. mete kahurangi ssp. nov. and Obtusopyrgus farri sp. nov. The species are integratively defined based on a combination of shell morphological, anatomical and mitochondrial DNA data. Morphological and anatomical data were generated by morphometrics, scanning electron microscopy, as well as micro-computed tomography. The genetic data were basis of phylogenetic analyses and incorporated into the diagnoses. The new taxa occur in springs or spring-like habitats, i.e., shallow, slowflowing sections of small streams except for O. mete kahurangi subsp. nov., which was collected from rough rocks in a river, where the snails sat in small depressions. None of the species exceeded 2.75 mm in length. Opacuincola gretathunbergae sp. nov. and Obtusopyrgus farri sp. nov. are pigmented and true crenobionts, while C. jami sp. nov. and the sympatric Opacuincola lisannea sp. nov. have eyes of reduced size and lack epidermal pigment, hence, probably dwell in the transitional zone of epigean and groundwaters.

a principal component analysis (PCA) and univariate tests were conducted in PAST ver. 4.01 (Hammer et al. 2001).

Micro-computed tomography
The anatomy of one female and one male of each new (sub)species was studied using micro-computed tomography at the Imaging Centre of the University of Greifswald. Prior to scanning, shells were dissolved in 0.5M EDTA (pH = 7.5) for two days, post-fi xed in paraformaldehyde for another two days, then rinsed in distilled water for 10 minutes and fi nally transferred to a 0.3% solution of phosphotungstic acid in 96% ethanol for further two days to enhance the contrast of tissues. Mounted in a plastic pipette tip fi lled with 99% ethanol and sealed with hot glue, snails were scanned with an Xradia Micro XCT-200 μCT (Carl Zeiss X-ray Microscopy Inc.) at a voltage of 40 kV, a power of 8 W, and ten times magnifi cation. The resulting image stacks were analyzed and the reproductive organs reconstructed in three dimensions (3D) using AMIRA ver. 6.0.1 (FEI, Visualization Science Group).

Scanning electron microscopy and dissections
Shells, radulae, opercula and penes were investigated by scanning electron microscopy. Two shells of each species were cleaned in ca 2.5% sodium hypochlorite. For the remaining characters, shells had to be dissolved in 1M hydrochloric acid. For radulae and opercula the soft bodies of up to three specimens were dissolved in sodium hypochlorite. If available, penes of two males were dissected free and dried in hexamethyldisilazane (Nation 1983) after dehydration in ethanol and transfer to 100% acetone. Specimens were mounted with carbon tabs, coated with palladium/platinum with a Fisons Polaron SC7640 sputter coater and then photographed in a Zeiss EVO LS10 SEM, again at the Imaging Centre of the University of Greifswald. In the specimens dissected for SEM, we also studied the remaining anatomy including the genitalia. The total numbers of investigated genitalia, including in most cases one μCT-scanned specimen, are given in the species descriptions. Only data of mature individuals are reported.

Molecular analyses
DNA was extracted from three specimens per sample with the E.Z.N.A® Mollusc DNA Kit (Omega Bio-Tek Inc.) by crushing the entire snail and following the manufacturer's protocol. We amplifi ed two mitochondrial fragments: 658 bp of the cytochrome c oxidase subunit I gene (COI) using Folmer et al. (1994) primers LCO1490 and H1298, the latter modifi ed at position 12 (A instead of G) by Zielske et al. (2011) and ca 500 bp of the 16S ribosomal RNA gene (16S) with primers 16Sar-L and 16Sbr-H (Palumbi et al. 1991). Polymerase chain reactions (PCR) were performed in a total volume of 11 μL and consisted for COI of 1 μL of DNA solution (~20 ng), 4.10 μL of water, 5 μL of HS MyTaqTM RedMix (Bioline), 0.40 μL of 1% BSA and 0.25 μL of each primer (from a 10 pmol stock solution). For 16S, the mix was similar except that we added 4.60 μL of water and 0.20 μL of each primer. The temperature profi le for COI was 1 min of initial denaturation at 95°C followed by 40 cycles comprising 20 s denaturation at 95°C, 30 s annealing at 48°C, and 1 min extension at 72°C, and a fi nal extension at 72°C for 5 min. For 16S we had a touch-down protocol with 1 min initial denaturation at 95°C, 10 cycles with 20 s denaturation at 95°C, 20 s annealing starting at 60°C and dropping by 1 degree in each cycle to 51°C, and 1 min extension at 72°C, followed by further 25 cycles consisting of 20 s denaturation at 95°C, 20 s of annealing at 51°C, and 1 min extension at 72°C, and the 5 min fi nal extension at 72°C. PCR products were visualised on a 1% agarose gel and purifi ed with an exonuclease I and shrimp alkaline phosphatase mix. Cycle sequencing was conducted using the BigDye™ Terminator ver. 3.1 Cycle Sequencing Kit (Applied Biosystems) with 50% replaced by halfBD (Sigma-Aldrich) and the PCR primers. The cycle sequencing products were cleaned with magnetic beads using the HighPrepTM DTR Dye Terminator Removal Clean Up (MagBio Genomics) and then sequenced on an ABI 3130xl Genetic Analyser (Applied Biosystems).
Sequences were edited in Geneious ver. 10.2.3 (https://www.geneious.com) and BioEdit 7.0.5.3 (Hall 1999) and aligned with the data of Haase (2008) using MAFFT with the default settings (Katoh et al. 2019). The alignment was fi nally trimmed to 644 bp for COI and 482 bp for 16S. An exhaustive search with PartitionFinder ver. 2.1.1 (Lanfear et al. 2017) suggested a total of four partitions, one per COI codon position and one for 16S. The best fi tting substitution models based on the Bayesian information criterion were TRN+I+G, HKY+I and GTR+G for 1 st to 3 rd codon positions, respectively, and TIM+I+G for 16S. Based on this scheme, we conducted a maximum likelihood (ML) phylogenetic analysis using Garli ver. 2.1 (Zwickl 2006) with the optimal tree inferred from 500 replicates. Robustness was assessed with 500 bootstrap replicates summarized in a 50% majority rule consensus tree calculated with PAUP* ver. 4.0b10 (Swofford 2002). The best fi tting models for a Bayesian analysis conducted in MrBayes ver. 3.2.3 (Ronquist et al. 2012) with its more restricted model collection were GTR+I+G (COI, 1 st codon position), HKY+I (2 nd position), TVM+G (3 rd position) and HKY+I+G (16S). MrBayes was run for 2 Mio generations with every 100 th tree sampled, a burnin of 5000 and otherwise default settings. The average standard deviation of split frequencies reached 0.0046, effective sample sizes exceeded 200, and the potential scale reduction factors reached or closely approached one for all parameters indicating convergence of all estimates.
Pairwise uncorrected genetic distances were computed after pairwise deletion of missing data for COI in MEGA X (Kumar et al. 2018). For the four new species, we identifi ed diagnostic molecular characters based on the alignments of COI (Electronic Supplement 1) and 16S (Electronic Supplement 2) using QUIDDICH (Kühn & Haase 2020), a package written in R (R Core Team 2020). In these diagnoses, we compared each new species of Catapyrgus and Obtusopyrgus to their respective only other congeneric species sequenced or known so far, and the two new species of Opacuincola to all other congeneric species. Among the latter new species, one was also compared to its sister species. The other one did not have a single sister species in our analyses.

Phylogeny
The phylogenetic analyses (Fig. 2) recovered the major clades from previous analyses (Haase 2005(Haase , 2008Zielske et al. 2017). Their relationships were slightly different, however, basally unsupported. All genera received high bootstrap support as well as posterior probabilities and all new taxa unambiguously fell into one of three known genera, viz. Catapyrgus, Opacuincola and Obtusopyrgus, respectively. According to these reconstructions, the four new species were well differentiated against respective congeneric species. Only the taxon found along the Fenian track was paraphyletic with respect to Opacuincola mete Haase, 2008. This was one reason why it was classifi ed as a subspecies of the latter (see below).

Shell morphology
Overall shell morphology was compared in a PCA (Fig. 3) based on the fi ve shell measurements (

Systematic descriptions
The summary statistics of the shell parameters of the new taxa are given in Table 1 and not repeated in the descriptions. Anatomical information is based on the specimens prepared for SEM and on the μCT scans, i.e., in total up to three specimens of each sex. Diagnostic molecular characters for the four new species are provided in Table 2.

Diagnosis
Catapyrgus jami sp. nov. is most similar to C. matapango, however but is, genetically distinct at 59 alignment positions. Furthermore, it differs from its congeners in the shape of the bursa copulatrix, which is globular rather than kidney-shaped.

Etymology
Catapyrgus jami sp. nov. is named after Jochen A. Modeß, musician and composer who until his retirement from the university in 2019 has signifi cantly shaped the cultural life of the city of Greifswald for over 25 years. The name is based on the initials of the dedicatee.  Table 1. Paratypes (Fig. 4B) NEW ZEALAND • 9 specs; same collection data as for holotype; NMNZ.M.330188.

Remarks
Catapyrgus jami sp. nov. is most similar to C. matapango in terms of shell size and shape (Fig. 3). The latter is the only other species of the genus for which sequence data exist. The divergence of the species was fairly large (Fig. 2) with a COI p-distance of 0.07, and 45 and 14 diagnostic characters in COI and 16S, respectively ( Table 2). The reduced eyes suggest that this species dwells in the transition zone of epigean and ground waters. The new species occurred sympatrically with Op. lisannea sp. nov.

Diagnosis
The new species is a slender-conical Opacuincola most similar to the smaller Op. terraelapsus. Opacuincola takakaensis is larger and more conical. Compared to Op. terraelapsus, the new species has a larger bursa copulatrix reaching much farther behind the albumen gland than in the latter, and the huge penis has no subterminal swelling and has a lobe that points forward rather than to the right. Compared to all other congeners, Op. lisannea sp. nov. had 20 diagnostic DNA positions.

Etymology
Opacuincola lisannea sp. nov. is dedicated to Lisanne Verhaegen, the sister of the fi rst author, on the occasion of her 30 th birthday. She is in part responsible for the debut of the fi rst author as a biologist by proofreading numerous of her applications, including the one for her PhD position, resulting in the discovery of this new species.

Material examined
Holotype ( Description SHELL (Figs 4C-D, 5B-C). Slender-conical to pupiform, about 1.8 times as high as than wide, whitetranslucent with light brown periostracum; protoconch with fi ne pits comprising ca 0.85 whorls (Fig. 6B-C); entire shell with 3.75 to 4.375 whorls, teleoconch with fi ne longitudinal ridges on fi rst 0.25 whorl, then without structure apart from growth lines; umbilicus narrow; aperture orthocline, almost circular, only slightly higher than wide.

Remarks
With respect to all other sequenced congeners, Op. lisannea sp. nov. had 14 diagnostic characters in COI and six in 16S, respectively (Table 2). In the phylogeny, it was a well-supported sister species to fi ve other species of the genus (Fig. 2). Morphologically, the new species is most similar to Op. terraelapsus. The latter is smaller (shell height, ANOVA: df = 2, 50; F= 79.17; p < 0.0001; Tukey's pairwise posthoc tests: p < 0.004 in all three cases), but the species cannot be distinguished in shape (shell height/ shell width, Kruskal-Wallis test: H = 4.409; p = 0.110; pairwise Mann-Whitney U-tests: p > 0.05 in all three cases). This is perfectly refl ected in the PCA, where the species largely overlap only along PC2 (Fig. 3). In February 2016, we again failed to fi nd Op. terraelapsus. It was originally collected in

Diagnosis
The new species is most similar to Op. ngatapuna in terms of shape and epidermal pigmentation. It differs from the latter in 11 diagnostic DNA positions, in being much larger and in penial morphology. The penis and penial lobe of Op. gretathunbergae sp. nov. are considerably more delicate.

Etymology
The dedicatee of this new species is the Swedish teenage climate activist Greta Thunberg. Starting with a single-person school strike and demonstration to save our climate she has sparked the global movement "Fridays for Future" supported primarily by young people and managed to fi nally get momentum in global politics toward action against climate change after warnings of scientists have been largely ignored for more than 30 years. We wish her and the movement the endurance necessary to keep the pressure up!

Material examined
Holotype ( Description SHELL (Figs 4E-F, 5D-E). Blunt-conical to pupiform, about 1.65 times as high as than wide, whitetranslucent with brown periostracum; protoconch almost smooth with fi ne pits comprising ca 0.75 whorl (Fig. 6D); entire shell with 3.5 to 4.25 whorls, teleoconch initially with fi ne longitudinal ridges, then without structure apart from growth lines; umbilicus narrow; aperture orthocline, slightly higher than wide.

Remarks
The sister relationship of Op. gretathunbergae sp. nov. to Op. ngatapuna was fairly well supported (Fig. 2). The average COI p-distance was 0.014 and there were eight type 1 characters in COI and three in 16S (Table 2). Morphologically, the new species is larger. Univariate tests comparing shell dimensions could not be conducted, though, because of the small sample size available for Op. ngatapuna. But the PCA (Fig. 3) and the data in Haase (2008) are clear regarding the size difference. Anatomically, only the male genitalia could be compared because this information is lacking for Op. ngatapuna (Haase 2008). The well-developed eyes indicate that the new species is a true crenobiont.

Etymology
Opacuincola mete kahurangi subsp. nov. is named after Kahurangi National Park, the second largest National Park in New Zealand in the northeast of the South Island, where the type locality is situated.

Material examined
Holotype (

Description
SHELL (Figs 4G-H, 5F-G). Broadly-conical to globular, only about 1.2 times as high as than wide, translucent with brown periostracum; protoconch almost smooth, comprising ca 1 whorl (Fig. 6E); entire shell with 3.125 to 4 whorls, teleoconch without structure apart from growth lines; umbilicus a wider slit; aperture large, wider than high, about half as high as total shell height and more than half as wide as total width, apertural lip slightly sinuated both ad-and abapically.
PCA is unambiguous (Fig. 3). Anatomically, they are practically identical and in the mitochondrial phylogeny not separable (Fig. 2). Both are known each from a single locality, which are ca 90 km apart (as the crow fl ies). Because of the high overall similarity, we distinguish these forms only as subspecies, although they can easily be told apart by their shells. However, we do not know if there exist connecting populations between both sites mediating gene fl ow in a stepping stone-like fashion. Considering similar cases among Tateidae form New Zealand and other places, we might be witnessing species in statu nascendi or already be dealing with genetically incompatible but mitochondrially undifferentiated, young species (e.g., Haase 2005Haase , 2008Zielske & Haase 2014a, 2014b. Nuclear genetic data might be more informative in this ambiguous situation. The new subspecies represents one of the rare cases where a small tateid is not restricted to springs or small streams. Due to the unresolved situation in the mitochondrial phylogeny (Fig. 2), it is not reasonable to list diagnostic alignment positions.

Diagnosis
In the new species, the central tooth of the radula has more cusps on the edge and less on the basis than in the only other known representative of the genus, Ob. alpinus. The bursa copulatrix is smaller and more elongate compared to the larger, more globular one in the latter. As a consequence, the seminal receptacle reaches far behind the bursa in Ob. farri sp. nov., whereas in Ob. alpinus it lies entirely against the bursa. These species differ at fi ve diagnostic alignment positions of type 1.

Etymology
Obtusopyrgus farri sp. nov. is named after Gareth Farr, acclaimed New Zealand percussionist and composer integrating non-European music styles including Maori music into Western classical music resulting in the most fascinating and colorful synthesis of different musical expressions. His alter ego, the drag queen Lilith LaCroix, is also colorful.

Material examined
Holotype ( Description SHELL (Figs 4I-J, 5H-I). Blunt-ovate to pupiform, about 1.7 times as high as than wide, light brown, translucent with brown periostracum; protoconch almost smooth, comprising up to 1 whorl (Fig. 6F); entire shell with 3.625 to 4.25 whorls, teleoconch initially with very fi ne longitudinal ridges, then without structure apart from growth lines; umbilicus narrow; aperture orthocline, as high as wide.

Remarks
The new species is slightly larger than Ob. alpinus (shell height: Mann-Whitney U-test: z = 2.869, p = 0.004), but in terms of shape, they cannot be distinguished (shell height/shell width: Mann-Whitney U-test: z = 0.764, p = 0.445). The genetic and phylogenetic distinction of both species of Obtusopyrgus was only based on 16S as sequencing of COI failed (Fig. 2). There were fi ve diagnostic characters ( Table 2). The well-developed eyes indicate that Ob. farri sp. nov. inhabits epigean waters.

Discussion
Due to the high number and proportion of endemic species, New Zealand is regarded as one of 36 biodiversity hotspots (Myers et al. 2000;Mittermeier et al. 2004; see also Kier et al. 2009;Veron et al. 2019) and the number of species discoveries keeps growing. A quick search in the Web of Science with the search terms "New Zealand species" in "Topic" and only the three selected taxonomic journals European Journal of Taxonomy, ZooKeys or Zootaxa in "Publication Name" revealed a total of 143 new species published only in the years 2018 and 2019. With the four new species introduced here, the number of tateid gastropod species, all of them endemic to New Zealand (including P. antipodarum being invasive in many parts of the world, see Introduction), increased to 68. Although we do not know the entire ranges of the new species, they appear to fi t into the pattern of narrow-range endemism typical for the New Zealand representatives of the family (see Introduction). Their discovery is remarkable for two reasons: 1) our fi eldwork did not target species other than P. antipodarum (see Verhaegen et al. 2018b); and 2) we spent hardly two weeks in 2016 in the north of the South Island. Considering this and the fact that all fi ve localities were easily accessible along tracks and roads and only in the case of Op. m. kahurangi subsp. nov. in more than two hours walking distance from the car park, we can predict that still many more species await discovery.
To our best knowledge, this is the fi rst paper applying μCT scanning in truncatelloidean gastropods to reconstruct their genital system. In comparison to dissections, there are some pros and cons to this approach. In contrast to dissections, the original organ situs remains intact and apart from the fact that the shell had to be dissolved in order to optimize the contrast among the tissues, the specimens do not have to be destroyed so that they can be preserved and included in the type series. On the downside, the resolution of thin structures like the vas deferens or the oviduct has often been wanting (e.g., Fig. 9) and although one of us (MH) has a lot of experience with histological serial sectioning and their reconstruction (e.g., Haase & Bouchet 1998), these structures occasionally could not be detected. Similarly, delicate details could not be resolved in particular in cases where an organ or parts of an organ bend onto itself such as the hook-shape of the penial lobe in Op. lisannea sp. nov. (Figs 10B, 11) or the entire penis of C. jami sp. nov. All of this might be possible in dissections depending on the skills of the researcher and the size of the snails. It has to be stated though, that our material has been fi xed only in ethanol. It is thus very likely that the drawbacks of μCT-scanning can be overcome by fi xation in e.g., formalin so that also very small species can be studied in full anatomical detail.
Incorporating DNA character data into taxonomic diagnoses has already been suggested about 20 years ago. However, as of November 2015 Renner (2016) has found only 98 descriptions explicitly including DNA data, although morphologically cryptic taxa can usually only be identifi ed based on other genetic data. One reason for the seeming reluctance to use DNA characters also in diagnoses was possibly the lack of suitable tools for their identifi cation (Kühn & Haase 2020). With the correction of this defi ciency (Hütter et al. 2020;Kühn & Haase 2020) cryptic taxa such as our tateid gastropods will hopefully be formally described at an increasing rate and not stay in taxonomic crypsis (Schlick-Steiner et al. 2007). In contrast to Renner (2016), we do not advocate the replacement of morphological descriptions and diagnoses by DNA-based diagnoses, as we still usually deal with phenotypes and therefore cannot dispense with morphological data in taxonomy. Characters of all qualities should be seen as complementary in the sense of what has been dubbed integrative taxonomy (Dayrat 2005;Padial et al. 2010;Schlick-Steiner et al. 2010).
Our descriptions were based on relatively few individuals and some of the sequenced specimens even shared identical haplotypes. One might therefore question how meaningful DNA-based diagnoses are when data are limited. This touches on the general question of how to deal with rarity in taxonomy. Do we need a minimum number of specimens per population and populations per species to warrant taxonomic recognition? Considering that a large proportion of all species across all taxa are rare (Lim et al. 2012), often because their ranges are very restricted or their habitats diffi cult to access as in many tateids, one can only conclude with these authors that rarity is an inherent property of biodiversity and the neglect of rare taxa would impact severely on conservation biology, ecology and evolutionary biology. In taxonomy, this means that the variation of characters, regardless if morphological or molecular, may be limited at the time of description. A diagnosis may have to be extended or amended once more specimens become available. But again, this holds for morphological characters as well as for molecular characters. Thus, a molecular character identifi ed as diagnostic and fi xed for one state (A, C, G, T, or gap), a type 1 character according to Kühn & Haase (2020), might erode to a type 2 or type 3 character or eventually no longer be diagnostic, adding new information. This is analogous to a case where two species initially considered to differ in size no longer do so after new populations have been discovered.
Diagnoses are relative. With that respect we found it particularly interesting that in Op. gretathunbergae sp. nov., well defi ned from its sister species Op. ngatapuna at eight positions in COI and three in 16S rRNA, only a single diagnostic position or type 1 remained in the comparison across all congeners (Table 2). This indicates a high frequency of homoplastic mutations in COI. Similar observations were made among tateids from New Caledonia (Zielske & Haase 2015).