Just the once will not hurt: DNA suggests species lumping over two oceans in deep-sea snails (Cryptogemma) Paul Zaharias, Yuri Kantor, Alexander Fedosov, Francesco Criscione, Anders Hallan, Yasunori Kano, Jérémie Bardin, Nicolas Puillandre

To cite this version:

Paul Zaharias, Yuri Kantor, Alexander Fedosov, Francesco Criscione, Anders Hallan, et al.. Just the once will not hurt: DNA suggests species lumping over two oceans in deep-sea snails (Cryptogemma). Zoological Journal of the Linnean Society, Linnean Society of London, In press, ￿10.1093/zoolin- nean/zlaa010￿. ￿hal-02559713￿

HAL Id: hal-02559713 https://hal.archives-ouvertes.fr/hal-02559713 Submitted on 4 May 2020

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Zoological Journal of the Linnean Society

Just the once will not hurt: DNA suggests species lumping over two oceans in deep-sea snails

Journal: Zoological Journal of the Linnean Society

Manuscript ID ZOJ-08-2019-3813.R1

Manuscript Type: Original Article

Forspecies Review delineation < Taxonomy, Only species redescription < Taxonomy, Keywords: Gastropoda < Taxa

The practice of species delimitation using molecular data commonly leads to revealing species complexes and increasing the number of delimited species. In few cases, however, DNA-based taxonomy can lead to lumping previously described species. Here, we delimit species in the genus Cryptogemma (Gastropoda, Conoidea, Turridae), a group of deep- sea snails with wide geographic distribution, by primarily using the mitochondrial COI gene. Three approaches of species delimitation (ABGD, mPTP and GMYC) were applied to define species partitions. All approaches resulted in eight species. According to previous taxonomic Abstract: studies and shell morphology, 23 available names potentially apply to the eight Cryptogemma species that were recognized herein. Shell morphometrics, radular characters, as well as geographic and bathymetric distributions were used to link type specimens to these delimited species. In all, 23 of these available names are here attributed to seven species, resulting in 16 synonymizations, and one species is described as new: C. powelli sp. nov. We discuss the possible causes behind such an apparent ‘over-description’ of species within Cryptogemma, which is here shown to constitute a rare case of DNA- based species lumping in the hyper-diversified superfamily Conoidea.

Just the once will not hurt: DNA suggests species lumping over two oceans in deep-sea snails

Paul Zaharias1*, Yuri I. Kantor2, Alexander E. Fedosov2, Francesco Criscione3, Anders

Hallan3, Yasunori Kano4, Jérémie Bardin5, Nicolas Puillandre1.

1Institut Systématique Evolution Biodiversité (ISYEB), Muséum National d’Histoire

Naturelle, CNRS, Sorbonne Université, EPHE, Université des Antilles, 43 rue Cuvier,

CP 26, 75005 Paris, France.

2A.N. Severtsov Institute of Ecology and Evolution, Russian Academy of Sciences,

Leninski prospect 33, 119071Moscow, Russian Federation

3Australian Museum Research Institute, Australian Museum, Sydney, Australia

4Atmosphere and Ocean Research Institute, The University of Tokyo, 5-1-5

Kashiwanoha, Kashiwa, Chiba 277-8564

5Centre de Recherche en Paléontologie – Paris (CR2P-UMR 7207), Sorbonne

Université‐CNRS‐MNHN, Site Pierre et Marie Curie, 4 place Jussieu, Paris Cedex 05,

France.

*Corresponding author: Paul Zaharias; Institut Systématique Evolution Biodiversité

(ISYEB), Muséum National d’Histoire Naturelle, CNRS, Sorbonne Université, EPHE,

Université des Antilles, 43 rue Cuvier, CP 26, 75005 Paris, France; +33 1 40 79 37 43; [email protected]

Page 1 of 65 Zoological Journal of the Linnean Society

1 2 3 4 1 Introduction 5 6 7 2 The advent of integrative taxonomy (Dayrat, 2005; Will et al., 2005), mainly driven by 8 9 3 the molecular revolution, has led to a great remodeling of the practice of species 10 11 4 delimitation, extending the use of standardized approaches in the field of taxonomy (e.g. 12 13 14 5 Lefébure et al., 2006). Several resulting studies, often primarily based on molecular 15 16 6 data, have revealed numerous species complexes, thus increasing the number of 17 18 7 delimited species within such groups, commonly referred to as “cryptic species” 19 20 8 (Bickford et al., 2007; Fišer et al, 2017). The difficulty in assigning names and/or 21 For Review Only 22 23 9 describe these newly delimited species can constitute an impediment, especially for 24 25 10 non-taxonomists. Indeed, when new species are discovered, nearly half of them remain 26 27 11 undescribed, at least not described in the publication where they are revealed (Pante et 28 29 30 12 al., 2014). Thus, the splitting effect of integrative taxonomy is not accompanied by 31 32 13 adequate taxonomic effort, resulting in a plethora of undescribed species. It has been 33 34 14 suggested that such lack of appropriate taxonomic procedure can have consequences in 35 36 37 15 various fields, such as ecology and conservation (e.g. Iglésias et al., 2010). 38 39 16 Conversely, there have been cases where integrative taxonomy, by incorporating 40 41 17 molecular data, has led to the lumping of species that have been originally described 42 43 18 using morphological and anatomical characters only (e. g. Chan et al., 2018; Dejaco et 44 45 46 19 al., 2016). Such instances may occur when, contrary to the occurrence of “cryptic 47 48 20 species”, intra-specific morphological variability is equal to or greater than inter- 49 50 21 specific variability (e.g. Puillandre et al., 2010). These phenomena may have biological 51 52 53 22 causes (e.g. sexual dimorphism, allometry during ontogeny, phenotypic plasticity), or in 54 55 23 other cases simply be artefacts of a particular taxonomic practice (some taxonomists 56 57 24 tend to split more than others). Unfortunately, these cases are difficult to investigate for 58 59 60 1 Zoological Journal of the Linnean Society Page 2 of 65

1 2 3 4 25 taxa where morphological characterization is not formalized, genomic resources are 5 6 7 26 poor, and/or known specimens are rare and deposited in different institutions. 8 9 27 Nevertheless, investigating whether or not different forms correspond to different 10 11 28 species should rely on a combination of modern species delimitation approaches and 12 13 14 29 solid taxonomic expertise in order to clarify both boundaries and taxonomic status of a 15 16 30 given taxon. 17 18 31 The Conoidea (Neogastropoda) is an extremely diversified lineage of marine 19 20 32 gastropods, traditionally divided into Terebridae, Conidae and Turridae (Bouchet et al., 21 For Review Only 22 23 33 2011). Recent molecular phylogenies have significantly redefined boundaries within the 24 25 34 superfamily (Puillandre et al., 2011; Abdelkrim et al., 2018a), leading to a redefinition 26 27 35 of the turrids. The family Turridae H. Adams & A. Adams, 1853 (1838) is now a 28 29 30 36 monophyletic group that encompasses a few morphologically well-defined 31 32 37 monophyletic genera, such as Gemmuloborsonia, Polystira, Lophiotoma or the 33 34 38 “Xenuroturris/Iotyrris complex”, that have been revised in recent years (Puillandre et 35 36 37 39 al, 2010; Todd & Rawlings, 2014; Puillandre et al., 2017; Abdelkrim et al., 2018b). In 38 39 40 some cases, the integrative taxonomy approach has uncovered “cryptic species” that 40 41 41 were described in the same studies in which they were revealed. However, a 42 43 42 considerable proportion of turrid species are still left in the paraphyletic genus 44 45 46 43 Gemmula, for which it was estimated that less than half the number of species has been 47 48 44 described so far (Puillandre et al., 2012). Puillandre et al. (2012) also called for the 49 50 45 complete revision of Gemmula, and suggested that the high number of species and the 51 52 53 46 suspected global morphological stasis of the group are likely explanations as to why this 54 55 47 group remains unrevised until now. Moreover, revising the Gemmula group would be 56 57 48 tantamount to revising the entire Turridae, given that independent “Gemmula-like” 58 59 60 2 Page 3 of 65 Zoological Journal of the Linnean Society

1 2 3 4 49 lineages are distributed all over the turrid tree (Puillandre et al. 2012). Finally, while a 5 6 7 50 high number of undescribed species was estimated in Gemmula, it is not clear whether 8 9 51 all the independent “Gemmula-like” lineages equally include undescribed species, or if 10 11 52 some lineages will require splitting of previously described species, while others will 12 13 14 53 imply lumping of previously described species, or a combination of both splitting and 15 16 54 lumping. Hence, we propose to partition the taxonomic revision by focusing on smaller 17 18 55 monophyletic groups that would be revised one by one, in order to progressively resolve 19 20 56 the taxonomic predicament of Gemmula as it presently stands. The first group we 21 For Review Only 22 23 57 identified is a clade of deep-water species, hereafter named “Cryptogemma”, that has 24 25 58 already been used as a case-study to illustrate congruent species hypotheses in the 26 27 59 integrative taxonomy framework designed in Puillandre et al. (2012, Figure 5), 28 29 30 60 comprising five species of Turridae attributed to three genera. We completed the dataset 31 32 61 from Puillandre et al. (2012) with recent field sampling and material from museums, 33 34 62 and applied an integrative taxonomy approach primarily based on the analysis of 35 36 37 63 barcode fragment of COI with additional data on the geography and bathymetry 38 39 64 included a posteriori to help attributing existing names to the molecular species 40 41 65 hypotheses. Furthermore, we significantly improved resolution of morphological studies 42 43 66 by carrying out a formalized analysis of the morphology of teleoconch and protoconch, 44 45 46 67 and by examining radulae in all sequenced species. The sequencing of a few type 47 48 68 specimens and the formalized morphological analysis allowed us to attribute 49 50 69 confidently type specimens to molecular species. Our results enabled a comprehensive 51 52 53 70 revision of Cryptogemma: rather than finding the usual splitting effect of molecular- 54 55 71 based integrative taxonomy, our study led to an unexpected number of new synonymies 56 57 72 and only one new species description. 58 59 60 3 Zoological Journal of the Linnean Society Page 4 of 65

1 2 3 4 73 5 6 7 74 Material and Methods 8 9 75 The integrative taxonomy pipeline followed in this study comprises three main steps. 10 11 76 First, most samples were sequenced at least for the CO1 gene fragment, without 12 13 14 77 necessary being preliminarily identified morphologically to the species level. Second, 15 16 78 automated species delimitation methods were used to delimit molecular operational 17 18 79 taxonomic units (MOTUs), resulting in primary species hypotheses. Finally, MOTUs 19 20 80 were assessed using alternative gene fragments, morphology, geography and 21 For Review Only 22 23 81 bathymetry, resulting in final species hypotheses. 24 25 82 Sampling 26 27 83 As a part of our background activity, most neogastropods collected during field 28 29 30 84 expeditions organized by the Muséum national d’Histoire naturelle (MNHN) are 31 32 85 processed for DNA extraction and sequencing of the Barcode fragment of the COI gene. 33 34 86 Among the processed material, all specimens for whom the barcode sequences clustered 35 36 37 87 into the Cryptogemma clade were included in this study. These samples were collected 38 39 88 during field expeditions in the Indo-Pacific and West Atlantic: EBISCO and 40 41 89 KANADEEP in the Chesterfield Islands, CONCALIS, EXBODI, KANACONO, 42 43 90 TERRASSES and NORFOLK 2 in New-Caledonia, MIRIKY and ATIMO VATAE in 44 45 46 91 Madagascar, MAINBAZA in Mozambique, BIOMAGLO in Mayotte, BIOPAPUA, 47 48 92 KAVIENG, MADEEP and PAPUA NIUGINI in Papua New Guinea, AURORA 2007 49 50 93 and PANGLAO 2004 in the Philippines, SALOMON 2 and SALOMONBOA 3 in the 51 52 53 94 Salomon Islands, TARASOC in the Société-Tuamotu islands, TAIWAN 2013, 54 55 95 NANHAI 2014, DONGHA 2014 and ZHONGSHA 2015 in Taiwan, GUYANE 2014 in 56 57 96 French Guyana. Additional samples were collected during Japanese expeditions: T/V 58 59 60 4 Page 5 of 65 Zoological Journal of the Linnean Society

1 2 3 4 97 “Nagasaki-maru” cruise N275 and R/V “Tansei-maru” cruise KT-12-32 in Japan. We 5 6 7 98 completed this sampling with specimens from the Australian museum (AMS), 8 9 99 Zoological Museum of Moscow State University (ZMMU) and the National Museum of 10 11 100 Natural History (USNM) that could potentially belong to this clade, based on their 12 13 14 101 morphology and the available corresponding sequence in GenBank. Twelve specimens 15 16 102 from the AMS were collected during the IN2017_V03 - Sampling the Abyss cruse 17 18 103 (2017). A paratype of Ptychosyrinx lordhoweensis Kantor & Sysoev, 1991 (R/V 19 20 104 “Dmitry Mendeleev”, 16th cruise, st. 1245, N° Lc14353; Kantor & Sysoev, 1991) from 21 For Review Only 22 23 105 the ZMMU was used for DNA extraction. Samples from USNM were also selected for 24 25 106 molecular analyses: three paralectotypes of Gemmula benthima Dall, 1908 (USS 26 27 107 Albatross; Stations 2807, 3360, 3365, USNM N° 96485, 123087, 123092; Kabat, 1996) 28 29 30 108 and one specimen of Ptychosyrinx carynae (Haas, 1949), USNM 832922. For USNM 31 32 109 samples, G. benthima specimens were preserved dry, while the P. carynae specimen 33 34 110 was probably fixed in formalin before conservation in ethanol. 35 36 37 111 MNHN specimens processed before 2012 as well as all AMS specimens chosen for 38 39 112 molecular analysis were anaesthetized using the isotonic solution of MgCl2 prior to 40 41 113 fixation in 96% ethanol. MNHN specimens processed after 2012 were microwaved and 42 43 114 fixed in 96% ethanol (Galindo et al., 2014). In some cases, shells were subsequently 44 45 46 115 drilled to extract the retracted animals. Japanese specimens were boiled in 70–90°C 47 48 116 water for 0.1–0.5 min before preservation in 99% ethanol. ZMMU specimens were 49 50 117 probably fixed and preserved in 75–80% ethanol. 51 52 53 118 54 55 119 DNA sequencing 56 57 58 59 60 5 Zoological Journal of the Linnean Society Page 6 of 65

1 2 3 4 120 DNA from MNHN samples was extracted using the Epmotion 5075 robot (Eppendorf), 5 6 7 121 following the manufacturers’ recommendations. DNA from AMS samples was 8 9 122 extracted from small pieces of foot muscle by use of a Bioline Isolate II Genomic DNA 10 11 123 extraction kit for animal tissue, following the standard procedure of the manual. DNA 12 13 14 124 from ZMMU and USNM samples was extracted using the EZNA Mollusc DNA Kit 15 16 125 (Omega Bio-Tek Inc.), following the manufacturers’ recommendations. For most 17 18 126 samples, the barcode fragment (658 bp) of the mitochondrial COI gene was amplified 19 20 127 using the universal primers LCO1490/HCO2198 (Folmer et al., 1994). PCR reactions 21 For Review Only 22 23 128 were performed using a previously well-established protocol (Puillandre et al., 2017). 24 25 129 For the ZMMU and USNM samples, the barcode fragment was amplified in two 26 27 130 fragments, one with the LCO1490 and a newly designed primer (conoCOIintR: 28 29 30 131 GCNCATGCHGGNGGNTCWGTW) and the other with another newly designed 31 32 132 primer (conoCOIintF: TCWTCAGCTGCNGTWGAAAGNGG) and the HCO2198 33 34 133 primer. Both fragments are overlapping over ~50 nucleotides. The PCR mix was the 35 36 37 134 same as for the LCO1490/HCO2198, except that BSA (10 mg/mL) was used instead of 38 39 135 DMSO. The amplification procedure was also similar, except that the annealing 40 41 136 temperature was 47°C, with 40 cycles instead of 35. A fragment of the mitochondrial 42 43 137 12S rRNA gene was amplified using the universal primers 12S1/12S3 (Simon, Franke & 44 45 46 138 Martin, 1991) and PCR reactions performed using a previously established protocol 47 48 139 (Puillandre et al., 2011). PCR products were purified and sequenced by the Eurofins 49 50 140 sequencing facility. 28S primers used at AMS were C1’/D2 (Dayrat et al., 2001; Jovelin 51 52 53 141 & Justine, 2001). Amplification of 28S consisted of an initial denaturation step at 94 °C 54 55 142 for 4 min, followed by 30 cycles of denaturation at 94 °C for 30 s, with annealing set to 56 57 58 59 60 6 Page 7 of 65 Zoological Journal of the Linnean Society

1 2 3 4 143 57°C for 45 s, followed by extension at 72 °C for 1 min. The final extension was at 72 5 6 7 144 °C for 5 min. 8 9 145 10 11 146 Species delimitation 12 13 14 147 COI sequences were aligned manually using MEGA X (Kumar et al., 2018); 12S and 15 16 148 28S sequences were aligned using Muscle (Edgar, 2004) as implemented in MEGA X 17 18 149 and then checked by eye. Phylogenetic reconstruction methods were applied to the COI 19 20 150 alignment, plus a concatenation of the COI, 12S and 28S alignments with one specimen 21 For Review Only 22 23 151 per species (see Results section).ABGD, GMYC and mPTP methods were applied to 24 25 152 the CO1 dataset. All three methods are exploratory and aim at finding clusters of 26 27 153 specimens corresponding to species but differ strongly in several aspects. ABGD is 28 29 30 154 based on genetic distances only and automatically detects the gap (the so-called 31 32 155 “barcode gap”) between intra and interspecific distances, then used to delimit species 33 34 156 hypotheses. On the contrary, GMYC and mPTP rely on phylogenetic trees. GMYC uses 35 36 37 157 an ultrametric tree and looks for a transition between speciation and coalescent nodes. 38 39 158 mPTP also delimit species by identifying the transition between the inter and 40 41 159 intraspecific parts of the tree, but requires only a phylogeny with branch length being 42 43 160 proportional to the number of substitutions, and relies on the mutation rate instead of the 44 45 46 161 branching rate. 47 48 162 For ABGD, the web version (http://wwwabi.snv.jussieu.fr/public/abgd/) and the default 49 50 163 parameters were used, with a Kimura (K80) TS/TV model implemented. Maximum 51 52 53 164 likelihoods trees were constructed for COI, and the concatenated (COI + 12S + 28S) 54 55 165 alignments using IQ-TREE v1.6.3 (Nguyen et al. 2014), with 100 bootstraps. For all 56 57 166 alignments, COI was consistently divided into three partitions, corresponding to the 58 59 60 7 Zoological Journal of the Linnean Society Page 8 of 65

1 2 3 4 167 three codon positions, and 12S and 28S considered each as a single partition. The best 5 6 7 168 model for each partition was evaluated using ModelFinder as implemented in IQ-TREE 8 9 169 (Kalyaanamoorthy et al., 2017). The mPTP approach of species delimitation (Kapli et 10 11 170 al., 2017) was performed on the COI maximum likelihood tree, using both ML 12 13 14 171 delimitation with default parameters and MCMC method with a Markov chain of 15 16 172 200,000,000 generations, a sampling frequency each 2,000 generations and a burnin set 17 18 173 at 25%. Bayesian trees were reconstructed using BEAST v2.5.0 (Bouckaert et al., 19 20 174 2019), running 200,000,000 generations on the COI dataset with a sampling frequency 21 For Review Only 22 23 175 every 4,000 generations and a burnin set at 10%. The COI dataset was partitioned by 24 25 176 codon position for model evaluation, and prior values were set as in Puillandre et al. 26 27 177 (2017). The resulting trees were combined using TreeAnnotator as implemented in 28 29 30 178 BEAST2, with node heights corresponding to mean heights of all trees. The “single 31 32 179 threshold” method of GMYC was applied using the resulting tree. Finally, a non- 33 34 180 ultrametric Bayesian tree was reconstructed for the COI, and concatenated datasets 35 36 37 181 using MrBayes v3.2.6 (Ronquist & Huelsenbeck, 2003), each of the two runs consisting 38 39 182 of 8 Markov chains and 20,000,000 generations and a sampling frequency each 2,000. 40 41 183 The analysis was performed on the Cipres Science Gateway 42 43 184 (http://www.phylo.org/portal2). The consensus tree was calculated with burnin set at 44 45 46 185 20%. 47 48 186 Four species of Turridae were used as outgroups: Turris babylonia (Linnaeus, 1758), 49 50 187 Gemmula kieneri (Doumet, 1840), Unedogemmula unedo (Kiener, 1839) and 51 52 53 188 Turridrupa sp. The MolD program (Fedosov et al., 2019) was used to identify unique 54 55 189 combinations of diagnostic sites for each delimited species. 56 57 190 58 59 60 8 Page 9 of 65 Zoological Journal of the Linnean Society

1 2 3 4 191 Shell and radular morphology 5 6 7 192 Radulae were prepared by standard methods (Kantor & Puillandre, 2012) and examined 8 9 193 by scanning electron microscopes JEOL JSM 840A and Tescan VEGA II LSU at the 10 11 194 MNHN. Protoconchs and shells were measured in standard position and the number of 12 13 14 195 protoconch whorls counted according to Bouchet & Kantor (2004). 15 16 196 Variation of shell shapes was analyzed through their outlines. Initially, we made a 17 18 197 selection of shells of sequenced specimens that were not broken, truncated or aberrant. 19 20 198 To complete this dataset, we obtained photographs of 20 name-bearing types 21 For Review Only 22 23 199 (corresponding to 21 species names) that were deemed high-quality, of which 18 were 24 25 200 suitable for morphometric analysis (nearly complete shells). In total, shape acquisition 26 27 201 was performed for 197 sequenced specimens, 18 types and 8 unsequenced specimens of 28 29 30 202 C. periscelida. 31 32 203 Shell shape was acquired from photographs using a Photoshop CC 2019 script. Outlines 33 34 204 were reconstructed with the package Momocs (Bonhomme et al. 2014) from the R 35 36 37 205 software. We used Elliptical Fourier analysis (Giardina and Kuhl, 1977), which presents 38 39 206 many advantages over alternative Fourier analyses (Bonhomme et al. 2014), to 40 41 207 decompose the closed outlines of the shells into periodic functions. The number of 42 43 208 harmonics necessary to describe the shape was selected to reach 99.9% of the 44 45 46 209 cumulative harmonic power, ensuring that the majority of shell shape outlines were 47 48 210 considered. Shapes were also checked by eye to ensure that enough complexity was 49 50 211 captured. The obtained coefficients were normalized to allow the superposition of the 51 52 53 212 first ellipse and make the shapes comparable. We checked that all the reconstructed 54 55 213 outlines were appropriately aligned. A principal component analysis (PCA) was 56 57 214 performed on the harmonics coefficients in order to capture an optimal shape variation 58 59 60 9 Zoological Journal of the Linnean Society Page 10 of 65

1 2 3 4 215 with a minimal number of principal components. Finally, a linear discriminant analysis 5 6 7 216 (LDA) was performed to find the linear combination of principal components that 8 9 217 maximizes difference between specimens from species delimited with DNA. Fourier 10 11 218 analysis does not take size into account; only shapes are compared. As the size of the 12 13 14 219 shells may provide some discriminatory information, we added the length of each 15 16 220 specimen as a variable for the LDA. The appropriate number of principal components to 17 18 221 use in the LDA model was carefully chosen to find optimal equilibrium between 19 20 222 maximizing the differences and overfitting. A leave-one-out cross-validation procedure 21 For Review Only 22 23 223 was used to find the number of PC axes leading to the best percentage of species 24 25 224 predictions (see Figure S1). In order to test if our species attribution to each molecular 26 27 225 species partition was correct, type specimens were not included in the LDA training set 28 29 30 226 but subsequently added in the LDA morphospace. Another reason not to include the 31 32 227 types in the training set was to extract at best the morphological differences between 33 34 228 molecular species without the influences of type specimens morphology. 35 36 37 229 All the material examined for this study is summarized in Table S1. 38 39 230 40 41 231 42 43 232 Results 44 45 46 233 A total of 303 COI, 32 28S and 18 12S sequences of Cryptogemma were used for this 47 48 234 study, of which respectively 255, 8 and 16 were newly obtained. The USNM and 49 50 235 ZMMU samples yielded poor DNA quality, making them difficult to sequence with a 51 52 53 236 classical Sanger sequencing approach. Nevertheless, the 12S was obtained for the 54 55 237 USNM96485 and USNM123087 paralectotypes of Gemmula benthima collected 128 56 57 238 years ago. For the ZMMU paratype, the USNM123092 paralectotype and the 58 59 60 10 Page 11 of 65 Zoological Journal of the Linnean Society

1 2 3 4 239 USNM83292 lot of Ptychosyrinx carynae, the COI was obtained using the 5 6 7 240 conoCOIintR and conoCOIintF newly designed primers. The USNM123092 lot had 8 9 241 previously been sequenced and published by Todd & Rawlings (2014) for the COI and 10 11 242 12S fragments (GenBank accession number KM218745, KM218648), despite the fact 12 13 14 243 that the samples were most probably formalin-fixed. Nevertheless, sequencing an 15 16 244 additional specimen from the same lot was necessary as the lot might include specimens 17 18 245 from different species. 19 20 246 All ABGD, mPTP and GMYC analysis (see Figure S2) on the COI dataset resulted in 21 For Review Only 22 23 247 the same partition with eight species hypotheses. Both COI and 28S proved, in this 24 25 248 clade, to be appropriate markers for species delimitation, confirming the five COI-based 26 27 249 species hypothesis (C. praesignis, C. timorensis, C. tessellata, C. unilineata and C. 28 29 30 250 powelli sp. nov.) in Puillandre et al. (2012). However, new 28S sequences were not 31 32 251 produced here. The 28S sequencing of C. aethiopica and C. periscelida failed, but these 33 34 252 two species are well recognizable morphologically and are not genetically close to a 35 36 37 253 sister species, making it less probable that the COI-based species hypotheses are 38 39 254 incorrect. The case of C. praesignis and C. phymatias could be considered problematic 40 41 255 as they are found to be sister species and are separated bathymetrically. However, the 42 43 256 single sequence of 28S recovered for C. phymatias (C571715) showed a divergence 44 45 46 257 with the available 28S sequences of C. praesignis comparable to the divergence 47 48 258 between other sister species (results not shown herein; see Puillandre et al., 2012; 49 50 259 Puillandre et al., 2017). 51 52 53 260 54 55 261 For these eight species hypotheses, we searched for all names available and potentially 56 57 262 applicable (see also the Taxonomy section). By consulting with the relevant literature 58 59 60 11 Zoological Journal of the Linnean Society Page 12 of 65

1 2 3 4 263 on Turridae (e.g. Powell, 1964; Tucker, 2004) and by comparison of type specimens or 5 6 7 264 their images to shells of sequenced specimens, we found 23 names (see Taxonomy 8 9 265 section) that we applied tentatively to this partition. A total of 18 name-bearing types 10 11 266 corresponding to 19 names were of sufficient quality to be included in the 12 13 14 267 morphometric study. The other name-bearing types were discussed and assigned to 15 16 268 molecular species using only the shell photographs or drawings and extrinsic parameters 17 18 269 such as the bathymetry or geography (see Taxonomy section). 19 20 270 Despite considerable intraspecific morphological heterogeneity, the morphometric 21 For Review Only 22 23 271 analysis yielded instructive results: using shape alone, the leave-one-out cross- 24 25 272 validation procedure successfully predicted correct species attribution of 84% of 26 27 273 sequenced specimens and confirmed placement of 14 types out of 18 a priori attributed 28 29 30 274 to our species hypotheses (see Figure S3). By adding shell size, we increased the score 31 32 275 of the leave-one-out cross-validation procedure to 90.64% of the sequenced specimens, 33 34 276 and 15 out of 18 types (Fig. 1). The three type specimens whose placement was 35 36 37 277 incorrectly a priori attributed were Bathybermudia carynae (only representative of its 38 39 278 species), Pleurotoma praesignis and Ptychosyrinx bisinuata japonica (see also 40 41 279 taxonomic section). The most discriminant axis (Fig. 1, axis 1: 71.92% of trace) 42 43 280 corresponds to the variation range of the siphonal canal length and the width of the 44 45 46 281 shell, while variation along the second axis (15.57% of trace) mainly corresponds to the 47 48 282 intensity of the curvature of the outer aperture lip. 49 50 283 The phylogenetic trees resulting from the COI (Fig. 2) and concatenated-gene analyses 51 52 53 284 (Fig. 3) showed poorly supported interspecific relationships, except for the 54 55 285 Cryptogemma phymatias/C. praesignis sister relationship and the (C. unilineata, C. 56 57 286 timorensis, C. powelli) clade. Thus, the two molecular species PSH 11 and PSH 14 in 58 59 60 12 Page 13 of 65 Zoological Journal of the Linnean Society

1 2 3 4 287 the study of Puillandre et al. (2012), which were attributed to the genus Ptychosyrinx 5 6 7 288 Thiele, 1925, did not form a monophyletic group. The genera Cryptogemma Dall, 1918 8 9 289 (type species: Gemmula benthima Dall, 1908), Ptychosyrinx Thiele 1925 (type species: 10 11 290 Pleurotoma (Subulata) bisinuata Martens, 1901) and Pinguigemmula McNeil, 1961 12 13 14 291 (type species: Pinguigemmula okinavensis McNeil, 1960), despite being easily 15 16 292 recognizable morphologically, showed a level of molecular divergence among them that 17 18 293 was similar to, or even lower than, the level of divergence found among species in most 19 20 294 other turrid genera, thus suggesting synonymization of the two latter genera with 21 For Review Only 22 23 295 Cryptogemma. 24 25 296 All species showed large, oceanic or transoceanic, sympatric distributions. Despite 26 27 297 having large sympatric distribution, the C. phymatias/C. praesignis sister species are 28 29 30 298 found in different bathymetric zones, ~1400–3000 m and ~300–1400 m respectively, 31 32 299 with the exception of two specimens, one for each species, found at the same station 33 34 300 CP2752 (1378–1436 m) during the AURORA 2007 expedition, off Luzon Island, in 35 36 37 301 Philippines (see Table S1). All species geographic (Fig. 4) and depth (Fig. 5) ranges 38 39 302 were represented using only sequenced specimens, except for C. periscelida. For the 40 41 303 latter, the COI sequencing did not work for the 13 MNHN samples available but 42 43 304 information on geography and bathymetry was included nonetheless. 44 45 46 305 47 48 306 Abbreviations 49 50 307 Museums and repositories 51 52 53 308 AMS: Australian Museum of Sydney, Sydney, Australia. 54 55 309 AWMM: Auckland War Memorial Museum, Auckland, New Zealand 56 57 310 BPBM: Bernice Pauahi Bishop Museum, Honolulu, USA. 58 59 60 13 Zoological Journal of the Linnean Society Page 14 of 65

1 2 3 4 311 FMNH: Field Museum of Natural History, Chicago, USA. 5 6 7 312 MNHN: Muséum national d’Histoire naturelle, Paris, France. 8 9 313 NHMUK: Natural History Museum of United Kingdom, London, UK. 10 11 314 USNM: National Museum of Natural History, Smithsonian Institution, Washington, 12 13 14 315 DC, USA. 15 16 316 ZMB: Museum für Naturkunde, Humboldt-Universität, Berlin, Germany. 17 18 317 ZMMU: Zoological museum of Moscow University, Moscow, Russian Federation. 19 20 318 ZSIC: Zoological Survey of India, Calcutta, India. 21 For Review Only 22 23 319 24 25 320 Others 26 27 321 AL: Shell aperture length. 28 29 30 322 PD: Protoconch diameter. 31 32 323 PL: Protoconch length. 33 34 324 R/V: research vessel. 35 36 37 325 St.: station. 38 39 326 40 41 327 Taxonomy 42 43 328 44 45 46 329 Superfamily Conoidea Fleming, 1822 47 48 330 Family Turridae H. & A. Adams, 1853 (1838) 49 50 331 Genus Cryptogemma Dall, 1918 51 52 53 332 54 55 333 Type species: Gemmula benthima Dall, 1908 (OD) 56 57 334 Ptychosyrinx Thiele, 1925 (type species Pleurotoma bisinuata Martens, 1901 – OD) 58 59 60 14 Page 15 of 65 Zoological Journal of the Linnean Society

1 2 3 4 335 Bathybermudia Haas, 1949 (type species Bathybermudia carynae Haas, 1949 – OD) 5 6 7 336 †Pinguigemmula McNeil, 1960 (type species †Pinguigemmula okinavensis McNeil, 8 9 337 1960 – OD) 10 11 338 12 13 14 339 Included species: Cryptogemma aethiopica (Thiele, 1925); C. periscelida (Dall, 1889); 15 16 340 C. phymatias (Watson, 1886); C. powelli sp. nov.; C. praesignis (Smith, 1895); C. 17 18 341 tessellata (Powell, 1964); C. timorensis (Tesch, 1915); C. unilineata (Powell, 1964). 19 20 342 21 For Review Only 22 23 343 Remarks: According to WoRMS (checked July 2019), Cryptogemma comprises 11 24 25 344 species, and has never been revised. A consultation of available type photos and 26 27 345 published images on Cryptogemma species indicated that all of 11 species except the 28 29 30 346 type species C. benthima and C. phymatias, the senior synonym of C. benthima, should 31 32 347 be excluded from the genus, as the majority of them lack key characters such as the 33 34 348 narrow fusiform shell and the well-marked peripheral anal sinus. Moreover, studied 35 36 37 349 samples from Japan most certainly corresponding to the Cryptogemma corneus as 38 39 350 pictured in Hasegawa (2009: figs 335–338), showed closer affinity to other conoidean 40 41 351 families (e.g. Horaiclavidae) than to the Turridae, based on both, the sequence of the 42 43 352 barcode fragment of COI, and radular morphology (results not shown). The eight 44 45 46 353 species of Cryptogemma listed herein have wide distributions, most of them covering 47 48 354 the whole Indo-Pacific tropical region, and are not found at depths shallower than 49 50 355 200m. 51 52 53 356 Protoconch consisting of 4 to 5.25 whorls. Shell shape of the teleoconch combined with 54 55 357 size is an important factor to distinguish species (see Results section), despite high 56 57 358 intra-specific variability. Widest variability range observed in shell proportions, with 58 59 60 15 Zoological Journal of the Linnean Society Page 16 of 65

1 2 3 4 359 shells with wide last whorl and long siphonal canal being at one extreme of the range, 5 6 7 360 and more elongated shells with short siphonal canal on the other. 8 9 361 10 11 362 Cryptogemma phymatias (Watson, 1886) 12 13 14 363 (Fig. 6) 15 16 364 17 18 365 Pleurotoma phymatias R. B. Watson, 1886. 1920 m, Philippine Islands, 16°42'N, 19 20 366 119°22'E (Expedition H.M.S. Challenger on November 13, 1874, st. 205). 21 For Review Only 22 23 367 Gemmula benthima Dall, 1908. 2323 m, Gulf of Panama. 24 25 368 Gemmula benthina – Dall, 1918: 318 (lapsus calami); Powell, 1964: 279 (lapsus 26 27 369 calami). 28 29 30 370 Pleurotoma truncata Schepman, 1913. 2798 m, Banda Sea, Indonesia, 6°24'S, 31 32 371 124°39'E. 33 34 372 Bathybermudia carynae Haas, 1949. 3109 m, off Bermuda, 32°08.2'N, 64°33'W. 35 36 37 373 38 39 374 Remarks: This species was not present in the sampling of Puillandre et al. (2012) but 40 41 375 was included in the phylogeny of Puillandre et al. (2011) under the name Ptychosyrinx 42 43 376 carynae. 44 45 46 377 Due to the frequent loss of the shell apex, there are very few available protoconchs for 47 48 378 this species. Only the eroded protoconch of the type specimen of Bathybermudia 49 50 379 carynae could be measured, with PD= 1.3 mm and PL > 1.6 mm, consisting of more 51 52 53 380 than 3 whorls. 54 55 381 Radula medium long, about 2.6 mm in length (0.44 of AL), composed of 74 transverse 56 57 382 rows of teeth. Marginal teeth 118–128 µm long (mean 120 µm, n = 5), duplex. Anterior 58 59 60 16 Page 17 of 65 Zoological Journal of the Linnean Society

1 2 3 4 383 (inner) 1/3 of tooth length solid, very narrow in dorsal view, pointed, in posterior 2/3 5 6 7 384 major and accessory limbs broadly bifurcating, accessory limb with clear constriction at 8 9 385 about half tooth length, slightly shorter than major limb. Central formation with distinct 10 11 386 narrow carinated cusp and lateral inconspicuous flaps with indistinct lateral and anterior 12 13 14 387 margins (Fig. 7B–C). 15 16 388 Although this species can be distinguished from its congeners by the frequent loss of the 17 18 389 first whorls, and from its sister species Cryptogemma praesignis by the slightly larger 19 20 390 diameter of the shell, the bathymetry is its most distinct characteristic. C. praesignis is 21 For Review Only 22 23 391 found between ~300 and ~1400 m depth, while Cryptogemma phymatias is the only 24 25 392 Turridae to our knowledge to be exclusively found below ~1400 m deep. Moreover, it is 26 27 393 the only Turridae so far, and possibly the first reported benthic gastropod, to have a 28 29 30 394 Pacific–Atlantic distribution (Fig. 4; see Discussion). Most studied type specimens have 31 32 395 very eroded shells, but the variability of the last whorls, siphonal twist and aperture 33 34 396 among them is in agreement with the variability in the sequenced specimens 35 36 37 397 Noteworthily, the etymology of this species combines “Cryptogemma” (hidden gems) 38 39 398 and “phymatias” (one who has tubercules), resulting in a confusing combination. 40 41 399 42 43 400 List of COI diagnostic sites (position: character state): [379: T, 493: C, 622: C]. 44 45 46 401 Counting starts right next to the 3’ end of the LCO1490 primer (from 1 to 658). 47 48 402 49 50 403 Distribution: From the Central Indo-Pacific (Banda Sea) to the North Atlantic Ocean 51 52 53 404 (Bermuda Is.) (Fig. 4A), from ~1400 to ~3000 meters depth (Fig. 5). This species is 54 55 405 expected to be found in the Indian Ocean, as is the case with other Indo-Pacific species 56 57 58 59 60 17 Zoological Journal of the Linnean Society Page 18 of 65

1 2 3 4 406 of Cryptogemma, but it has not been documented there, due presumably to a lack of 5 6 7 407 sampling at bathyal depths. 8 9 408 10 11 409 Cryptogemma praesignis (Smith, 1895) 12 13 14 410 (Fig. 8A–K) 15 16 411 17 18 412 Pleurotoma praesignis Smith, 1895. 1234 m, off Colombo, Ceylon. 19 20 413 ? Pleurotoma microscelida Dall, 1895. 642 m, South of Oahu Island Albatross st. 3475 21 For Review Only 22 23 414 Pleurotoma (Subulata) bisinuata Martens, 1901. 1134 m, off East Africa, 1°49'N, 24 25 415 45°29'E. 26 27 416 Pleurotoma rotatilis Martens, 1902. 1134 m, off Mogadishu, Somalia, East Africa, 28 29 30 417 1°49'N, 45°29'E. 31 32 418 Pleurotoma (Surcula) lobata G. B. Sowerby III, 1903. 805 m, off Cape Natal, Durban 33 34 419 and 567 m, off Buffalo River, East London, S. Africa. E. A. Smith, 1906. 35 36 37 420 Ptychosyrinx bisinuata japonica Okutani, 1964. 620 m, Sea of Enshu-nada, Japan, 38 39 421 34°25.7'N, 137°58.5'E. 40 41 422 Ptychosyrinx lordhoweensis Kantor & Sysoev, 1991:205, figs. 1, 2. 1210 m, Lord Howe 42 43 423 Rise, off eastern Australia, 30°24'S, 161°51'E. 44 45 46 424 47 48 425 Remarks: This species corresponds to PSH 11 in Puillandre et al. (2012). 49 50 426 The protoconch exhibits a wide size range, with PD = 1.05–1.38 mm and PL = 1.4–1.93 51 52 53 427 mm, and from 4 to 5.2 whorls. 54 55 428 Radula long, about 3.4 mm in length (0.44 of AL), formed of 72 transverse rows of 56 57 429 teeth. Marginal teeth 142–151 µm long (mean 149 µm, n = 5, or 1.9% of AL), duplex. 58 59 60 18 Page 19 of 65 Zoological Journal of the Linnean Society

1 2 3 4 430 Anterior (inner) 1/3 of tooth length solid, narrow in dorsal view, pointed, in posterior 5 6 7 431 2/3 major and accessory limbs broadly bifurcating, accessory limb with clear 8 9 432 constriction and bent at about half tooth length, shorter than major limb. Central 10 11 433 formation with distinct narrow carinated cusp and lateral inconspicuous flaps with 12 13 14 434 indistinct lateral and anterior margins (Fig. 7A). 15 16 435 This species is morphologically close to Cryptogemma timorensis, although it differs 17 18 436 from that species in its generally smaller size (~15–35mm SL), in having a narrower last 19 20 437 whorl, and the presence of a tertiary apertural notch in mature females. Cryptogemma. 21 For Review Only 22 23 438 praesignis also differs from C. timorensis in having a much longer radula (0.44 vs 0.2 of 24 25 439 AL) and relatively longer marginal teeth (1.9% of AL vs 1.35%). Anatomical 26 27 440 examination of Australian material (C571714 and C571704) shows that mature males 28 29 30 441 (SL 29.4 and 34.9 mm respectively) can obtain an extremely large and muscular penis, 31 32 442 with a width almost equal to the width of the animal itself, and with a very large, long 33 34 443 lateral appendage situated distally. The penial tip is rather blunt, with the opening 35 36 37 444 situated distally. A third, smaller Australian specimen (C571757; SL 25.7 mm), did not 38 39 445 possess a well-developed papilla, suggesting that such a feature may develop with 40 41 446 increasing maturity and therefore not be present in subadults. Furthermore, the shells of 42 43 447 C571714 and C571704 did not possess a tertiary apertural notch. The observation of 44 45 46 448 C571714 and C571704 supplements the findings of Kantor & Sysoev (1991) that the 47 48 449 mature females of this species develop a tertiary apertural notch. The authors 49 50 450 hypothesized that this structure was possibly connected with the process of fertilization. 51 52 53 451 Such a feature has been observed in C. aethiopica specimens as well (Fig. 9G, I), but 54 55 452 not in other Cryptogemma species. 56 57 58 59 60 19 Zoological Journal of the Linnean Society Page 20 of 65

1 2 3 4 453 The general morphology, and more specifically the presence of a tertiary apertural notch 5 6 7 454 in Pleurotoma bisinuata and Ptychosryinx bisinuata japonica, supports the 8 9 455 synonymization of these names with C. praesignis. Because the holotype of P. bisinuata 10 11 456 japonica is a comparatively large specimen (39 mm), the result of the LDA and the 12 13 14 457 leave-one-out cross-validation procedure attributed it to C. timorensis. Possibly for this 15 16 458 same reason, the holotype of P. praesignis (42 mm) also fell into the C. timorensis 17 18 459 range. The result of the LDA analysis and the leave-one-out cross-validation procedure 19 20 460 when considering only shape predicted the two holotypes as C. praesignis, (see Figure 21 For Review Only 22 23 461 S3) thus justifying the attribution of P. praesignis to this molecular species. The 24 25 462 sequencing of a paratype of P. lordhoweensis confirmed its conspecificity with C. 26 27 463 praesignis. The name Pleurotoma lobata, although conchologically very similar to P. 28 29 30 464 bisinuata, has never been synonymized based on the reportedly different morphology of 31 32 465 the radula: the absence of a central tooth (= central formation as understood herein) in 33 34 466 the specimens of P. lobata examined by Barnard (1958), contrasting with the large 35 36 37 467 unicuspid rectangular-based central tooth described in P. bisinuata by Thiele (1929, p. 38 39 468 359). Powell (1964) suggested that Thiele could have mixed his radular preparations, 40 41 469 but in light of the radulae type of Cryptogemma (Fig. 8), all composed of a unicuspid 42 43 470 central formation, it is more likely that either it was Barnard, who mixed up the radula 44 45 46 471 preparation, or that the specimens examined by Barnard are not Cryptogemma 47 48 472 specimens. The holotype of Pleurotoma rotatilis is almost identical conchologically to a 49 50 473 sequenced juvenile from the East African coast (MNHN-IM-2013-62900; Fig. 8J). 51 52 53 474 Some doubts remain as to the status of Pleurotoma microscelida, but the size and 54 55 475 overall last whorl morphology of the type specimen show a stronger resemblance to C. 56 57 476 praesignis than to any other Cryptogemma species. 58 59 60 20 Page 21 of 65 Zoological Journal of the Linnean Society

1 2 3 4 477 5 6 7 478 List of COI diagnostic sites (position: character state): [232: A, 319: T, 407: C, 409: T] 8 9 479 10 11 480 Distribution: This species is known to occur from the West Indian Ocean to the East 12 13 14 481 Pacific Ocean (Fig. 4B), from ~300 to ~1400 m of depth (Fig. 5). 15 16 482 17 18 483 Cryptogemma timorensis (Tesch, 1915) 19 20 484 (Fig. 8L-P) 21 For Review Only 22 23 485 24 25 486 †Pleurotoma timorensis Tesch, 1915. Timor. Pliocene. 26 27 487 †Pleurotoma ktolemandoënsis K. Martin, 1933. Ktolemando. Miocene. 28 29 30 488 Ptychosyrinx timorensis teschi Powell, 1964: 291 (22–853), pl. 223, figs. 5, 6. 759 m, 31 32 489 NW off Sipadan Island, Borneo. 33 34 490 35 36 37 491 Type locality: Pliocene of Timor Island. 38 39 492 40 41 493 Remarks: This species corresponds to PSH 14 of Puillandre et al. (2012). 42 43 494 Protoconch commonly eroded, with PD = 1–1.2 mm and PL = 1.3–1.675 mm and 44 45 46 495 number of whorls varying from 4 to 4.5. 47 48 496 Radula medium short, about 2.5 mm in length (0.2 of AL), formed of 45 transverse 49 50 497 rows of teeth. Marginal teeth 162–172 µm long (mean 168 µm, n = 5, or 1.35% of AL), 51 52 53 498 duplex. Anterior (inner) 1/3 of tooth length solid, medium broad in dorsal view, 54 55 499 lanceolate, in posterior 2/3 major and accessory limbs broadly bifurcating, accessory 56 57 500 limb with clear constriction and bent at about half tooth length, nearly the same length 58 59 60 21 Zoological Journal of the Linnean Society Page 22 of 65

1 2 3 4 501 as major limb. Central formation with short and obtuse cusp and lateral inconspicuous 5 6 7 502 flaps with indistinct lateral and anterior margins. Posterior margin thickened and 8 9 503 uninterrupted along its length (Fig. 7I). 10 11 504 This type of this species, Pleurotoma timorensis, is a fossil specimen from Timor island 12 13 14 505 (Indonesia) described by Tesch, and Martin described another very similar specimen 15 16 506 from Buton island (Indonesia), Pleurotoma ktolemandoënsis, without reference to the 17 18 507 work of Tesch. However, Martin furtherly recognized the affinities between P. 19 20 508 timorensis and P. ktolemandoënsis (Martin, 1935). Finally, Robba et al. (1989) 21 For Review Only 22 23 509 synonymized P. ktolemandoënsis with P. timorensis. Powell (1964) described the 24 25 510 subspecies Ptychosyrinx timorensis teschi based on an extant specimen, which Sysoev 26 27 511 (1996) elevated to species rank. Both authors’ arguments for separating the fossil 28 29 30 512 species from the extant was the ‘much broader’ shell of the extant Ptychosyrinx 31 32 513 timorensis teschi specimen. Examination of the morphological variability of the 33 34 514 sequenced specimens showed that dimensions and shape of fossil specimens is close to 35 36 37 515 the inferred average breadth of the species; some specimens show broader aperture and 38 39 516 a taller last whorl than others, despite having the same number of teleochonch whorls. 40 41 517 The results of the LDA showed that P. timorensis and P. timorensis teschi fall into the 42 43 518 variability of the sequenced specimens. The broader last whorl and the generally larger 44 45 46 519 size (~30-55mm) are characteristic for C. timorensis when compared to C. praesignis. 47 48 520 49 50 521 List of COI diagnostic sites (position: character state): [73: G, 214: G, 334: G, 511: G] 51 52 53 522 54 55 56 57 58 59 60 22 Page 23 of 65 Zoological Journal of the Linnean Society

1 2 3 4 523 Distribution: Found in the West Indian Ocean to the Central Indo-Pacific (Fig. 4C), 5 6 7 524 from ~300 to ~1200 meters depth (Fig. 5). Curiously, this species has not been found in 8 9 525 New-Caledonia, despite considerable sampling effort in this region. 10 11 526 12 13 14 527 Cryptogemma aethiopica (Thiele, 1925) 15 16 528 (Fig. 9) 17 18 529 19 20 530 Pleurotoma aethiopica Thiele, 1925. 638 m, off Somalia, East Africa, 0°27'S, 21 For Review Only 22 23 531 42°47.3'E. (Expedition Deutschen Tiefsee, st. 253). 24 25 532 Pleurotoma fusiformis Thiele, 1925:176(210), pl. 22(34), fig. 24. 614 m, Nias-südkanal, 26 27 533 0°15.2'N, 98°08.8'E. 28 29 30 534 Gemmula thielei Finlay, 1930. (nom. nov. for Pleurotoma fusiformis Thiele, 1925). 31 32 535 †Pleurotoma trincincta Martin, 1935:113, pl. 2, figs. 2, 2a. Buton Island, SE Celebes. 33 34 536 Oligocene. 35 36 37 537 †Pinguigemmula okinavensis McNeil, 1960. Okinawa. Shinzato Tuff Member, Miocene 38 39 538 or Pliocene. 40 41 539 Pinguigemmula luzonica Powell, 1964 1964:278(22–790), pl. 215, figs. 3, 4. 326m, off 42 43 540 Hermana, Menor Island, Luzon Island, Philippines. 44 45 46 541 Pinguigemmula philippinensis Powell, 1964:278(22–790), pl. 215, figs. 5, 6. 512m, off 47 48 542 Santiago, west Luzon Island, Philippines. 49 50 543 51 52 53 544 Remarks: This species was not included in the study by Puillandre et al. (2012). 54 55 545 Protoconch commonly eroded, with PD = 1.05-1.25, PL = 1.375-1.75 and number of 56 57 546 whorls ranging from 4.2 to 5. 58 59 60 23 Zoological Journal of the Linnean Society Page 24 of 65

1 2 3 4 547 Radula long, about 3.4 mm in length (0.37 of AL), composed of 77 transverse rows of 5 6 7 548 teeth. Marginal teeth 150-159 µm long (mean 155 µm, n = 5, or 1.66% of AL), duplex. 8 9 549 Anterior (inner) 0.4 of tooth length solid, narrow in dorsal view, awl-shaped, in 10 11 550 posterior part major and accessory limbs broadly bifurcating, accessory limb with clear 12 13 14 551 constriction and bent at about half tooth length, thin, nearly same length as major limb. 15 16 552 Central formation with long, very narrow, sharp, carinated cusp and lateral flaps with 17 18 553 distinct posterior and antero-lateral margins. Flaps not completely fused with cusp (Fig. 19 20 554 7E). 21 For Review Only 22 23 555 The former Pinguigemmula genus is easily distinguishable from Gemmula in having a 24 25 556 broadly conic spire, a strongly constricted base and a long, straight siphonal canal 26 27 557 (Powell, 1964). Although expressing a wide variety of forms, the molecular analysis 28 29 30 558 resulted in a single species hypothesis. Several species were described based on the 31 32 559 sculptural details such as the number of gemmate cords (1, 2 or none). The sculpture 33 34 560 seems to correlate with geography, with the forms from West Indian Ocean having 35 36 37 561 smooth inter-suture sculpture, while the forms from East Indian-West Pacific usually 38 39 562 possess two or three gemmate cords. Some large specimens of this species possess a 40 41 563 similar ‘tertiary notch’ to that of C. praesignis, suggesting sexual dimorphism also for 42 43 564 this species (Kantor & Sysoev, 1991). 44 45 46 565 47 48 566 List of COI diagnostic sites (position: character state): [115: C, 307: A, 418: G] 49 50 567 51 52 53 568 Distribution: Found in East Africa to Central Indo-Pacific (Fig. 4E), from ~400 to ~850 54 55 569 meters of depth (Fig. 5). 56 57 570 58 59 60 24 Page 25 of 65 Zoological Journal of the Linnean Society

1 2 3 4 571 Cryptogemma tessellata (Powell, 1967) 5 6 7 572 (Fig. 10A-D) 8 9 573 10 11 574 Gemmula tessellata Powell, 1967:439 (22–734a), pl. 315. 183–219m, off Waikiki, 12 13 14 575 Oahu Island, Hawaiian Islands (Collected by Dr. Pat Burgess) 15 16 576 17 18 577 Remarks: This species corresponds to PSH 12 in Puillandre et al. (2012). The 19 20 578 protoconch of the holotype is 4.5 whorls according to Powell (1967) and PD = 1.22 mm 21 For Review Only 22 23 579 and PL = 1.64 mm according to measurements inferred from photographs of the 24 25 580 holotype. Protoconch with high variability, with PD = 1.175-1.3, PL = 1.525-1.975 and 26 27 581 consisting of 4 to 5.5 whorls. 28 29 30 582 Radula medium long, about 1.2 mm in length (0.29 of AL), composed of 51 transverse 31 32 583 rows of teeth. Marginal teeth 86-91 µm long (mean 89 µm, n = 2, or 2.1% of AL), 33 34 584 duplex. Anterior (inner) 1/2 of tooth length solid, medium broad in dorsal view, 35 36 37 585 triangular, accessory limb weak, thin, without constriction, much shorter than major 38 39 586 limb, but nearly of same width (the marginal teeth on Fig. 8F are not fully sclerotized). 40 41 587 Central formation with long, sharp and carinated central cusp, slightly curved in profile, 42 43 588 and with lateral conspicuous flaps with distinct margins. Flaps not completely fused 44 45 46 589 with cusp. Anterior margin of central formation strongly concave (Fig. 7F). 47 48 590 The ‘light form’ of C. tessellata has been misidentified in Puillandre et al. (2012) as 49 50 591 Xenuroturris gemmuloides Powell, 1967 as it shares superficially similar features such 51 52 53 592 as the white-yellowish shell punctuated with regular brown-orange spots, and the small 54 55 593 size (~15-25mm). The ‘brown-orange form’ of C. tessellata, closer to the holotype, is 56 57 594 quite different from the ‘light form’ not only in coloration, but also in having a stouter 58 59 60 25 Zoological Journal of the Linnean Society Page 26 of 65

1 2 3 4 595 outline and a more tuberculate subsutural fold (Powell, 1964). Although readily 5 6 7 596 distinguished from the other Cryptogemma species due to its color pattern and small 8 9 597 size, LDA analysis indicates that this species more closely resembles small adults of C. 10 11 598 praesignis. 12 13 14 599 15 16 600 List of COI diagnostic sites (position: character state): [46: C, 61: A, 316: T] 17 18 601 19 20 602 Distribution: The sequenced specimens were found in New Caledonia only and the 21 For Review Only 22 23 603 holotype is from the Hawaiian Islands (Fig. 4C), from ~200 to ~500 meters depth (Fig. 24 25 604 5). The protoconch characteristics, similar in shape, size and number of whorls to C. 26 27 605 praesignis, imply that the species may have a much broader range than what is currently 28 29 30 606 documented, possibly covering the entire Central Indo-Pacific. 31 32 607 33 34 608 Cryptogemma unilineata (Powell, 1967) 35 36 37 609 (Fig. 10E-H) 38 39 610 40 41 611 Gemmula congener unilineata Powell, 1967:437(22–716a), pl. 313. 366 m, off Waikiki, 42 43 612 Oahu Island, Hawaii (Expedition Pele, June 13 1964). 44 45 46 613 47 48 614 Remarks: This species corresponds to PSH 13 in Puillandre et al. (2012). 49 50 615 Protoconch commonly retained, with PD = 1.075-1.275, PL = 1.375-1.925 and number 51 52 53 616 of whorls ranging from 4.75 to 5.5. 54 55 617 Radula long, about 3 mm in length (0.31 of AL), composed of 99 transverse rows of 56 57 618 teeth. Marginal teeth 121-127 µm long (mean 125 µm, n = 5, or 1.27% of AL), duplex. 58 59 60 26 Page 27 of 65 Zoological Journal of the Linnean Society

1 2 3 4 619 Anterior (inner) 0.4 of tooth length solid, very narrow in dorsal view, awl-shaped, in 5 6 7 620 posterior part major and accessory limbs broadly bifurcating, accessory limb with clear 8 9 621 constriction and bent at about half tooth length, slightly shorter than major limb. Central 10 11 622 formation with long sharp carinated cusp and lateral flaps with distinct posterior and 12 13 14 623 antero-lateral margins. Flaps not completely fused with cusp (Fig. 7G). 15 16 624 This species has been named based on the characteristic carinated brown-orange 17 18 625 subsutural band. The brown-orange subsutural band has been found in C. powelli n. sp. 19 20 626 and in some specimens of C. praesignis, as well as in other Turridae such as Gemmula 21 For Review Only 22 23 627 cosmoi (Sykes, 1930), but it is generally smoother and thinner to that of C. unilineata. 24 25 628 The LDA analysis indicated that C. unilineata has generally a more angulated shape on 26 27 629 the outer aperture lip of the last whorl, indicating a more concave subsutural ramp in 28 29 30 630 comparison to other Cryptogemma. However, we note that the convex hulls (Fig. 1) of 31 32 631 C. unilineata and C. powelli sp. nov. are greatly overlapping, so the strong carinated 33 34 632 brown-orange subsutural band remains the best suited character for distinguishing C. 35 36 37 633 unlineata from C. powelli. 38 39 634 40 41 635 List of COI diagnostic sites (position: character state): [172: G, 307: T, 361: G, 370: C] 42 43 636 44 45 46 637 Distribution: From East Africa to Central Pacific (Hawaiian and Society islands) (Fig. 47 48 638 7D), from ~300 to ~800 meters of depth (Fig. 5). 49 50 639 51 52 53 640 54 55 641 Cryptogemma periscelida (Dall, 1889) 56 57 642 (Fig. 10I-K) 58 59 60 27 Zoological Journal of the Linnean Society Page 28 of 65

1 2 3 4 643 5 6 7 644 Pleurotoma periscelida Dall, 1889. 283m (USS Albatross expedition, st. 2143, collected 8 9 645 on March 23 1884) and (196m, off Hatteras, North Carolina) 10 11 646 12 13 14 647 Remarks: 15 16 648 The protoconch is unknown, as it is consistently eroded in all examined material, even 17 18 649 in younger specimens. 19 20 650 Radula medium long (part of youngest section lost), more than 1.9 mm in length (>0.22 21 For Review Only 22 23 651 of AL), composed of more than 50 transverse rows of teeth. Marginal teeth 107-116 µm 24 25 652 long (mean 110 µm, n = 5, or 1.25% of AL), duplex. Anterior (inner) 0.4 of tooth length 26 27 653 solid, narrow lanceolate in dorsal view, in posterior part major and accessory limbs 28 29 30 654 medium broadly bifurcating, accessory limb with clear constriction and bent at less than 31 32 655 half tooth length, thin, nearly same length and width as major limb. Central formation 33 34 656 with long, narrow, sharp carinated cusp and lateral flaps with distinct posterior and 35 36 37 657 postero-lateral margins. Flaps completely fused with cusp (Fig. 7D). 38 39 658 This species has been described from the Gulf of Mexico, although specimens from 40 41 659 Suriname and from the MNHN recent expedition GUYANE 2014 (French Guyana) 42 43 660 were also collected. This is the only ‘Gemmula’-like species to be found exclusively in 44 45 46 661 the Atlantic Ocean, and it occurs in a 200-800 m depth range. 47 48 662 49 50 663 List of COI diagnostic sites (position: character state): [352: C, 412: G, 655: C] 51 52 53 664 54 55 665 Distribution: Found in the Gulf of Mexico and further south to French Guyana (Fig. 56 57 666 4B), from ~200 to ~500m of depth (Fig. 5). 58 59 60 28 Page 29 of 65 Zoological Journal of the Linnean Society

1 2 3 4 667 5 6 7 668 Cryptogemma powelli sp. nov. 8 9 669 (Fig. 11) 10 11 670 12 13 14 671 Type material: Holotype MNHN-IM-2013-68787; Paratype 1, MNHN IM-2007-40795, 15 16 672 paratype 2, MNHN-IM- 2007-40765, all live collected and processed for DNA 17 18 673 extraction. 19 20 674 21 For Review Only 22 23 675 Type locality: New-Caledonia, South-West of Ile des Pins, 22°48'S, 167°15'E, 449-465 24 25 676 meters depth, sand and debris (Expedition KANACONO, st. DW4697). 26 27 677 28 29 30 678 Etymology: Named after the New Zealand malacologist A.W.B. Powell (1901-1987), 31 32 679 who contributed enormously to the systematics of Conoidea, and in particular, who 33 34 680 revised the family Turridae in 1964 and described several species of Cryptogemma. 35 36 37 681 38 39 682 Description (holotype): Shell narrowly fusiform, with high spire, and medium-long 40 41 683 siphonal canal. Protoconch conical, eroded, with a diameter of 1.125 mm, of about 4 42 43 684 convex whorls. Teleoconch of 9 whorls; suture shallow, impressed. Shell height 38 mm, 44 45 46 685 aperture height 10.9 mm and shell diameter 12.5 mm. Whorls strongly shouldered, with 47 48 686 slightly concave shoulder slope, very weakly convex, nearly cylindrical periphery. 49 50 687 Spiral sculpture of fine, broadly spaced, wavy subequal spiral cords on subsutural ramp, 51 52 53 688 10 on last whorl, uppermost being much more pronounced than others, colored light 54 55 689 orange-brown. Sinus cord strongly gemmate, with gemmules clearly bisected (39 on the 56 57 690 body whorl). Spiral cords becoming thicker on whorl periphery, sometimes nodulose, 58 59 60 29 Zoological Journal of the Linnean Society Page 30 of 65

1 2 3 4 691 some notably wider than others. Axial sculpture of fine growth lines. Last whorl shortly 5 6 7 692 constricted to long siphonal canal, with 32 cords below sinus, of which about 20 on 8 9 693 canal. Aperture irregular oval, outer lip thin, simple. Anal sinus moderately deep, U- 10 11 694 shaped. Shell color straw-yellow, subsutural cord orange-brown, gemmae slightly 12 13 14 695 lighter than background. 15 16 696 Radula medium long, around 3.1 mm in length (0.28 of AL), composed of 75 transverse 17 18 697 rows of teeth. Marginal teeth 129-135 µm long (mean 133 µm, n = 5, or 1.22% of AL), 19 20 698 duplex. Anterior (inner) 1/3 of tooth length solid, lanceolate in dorsal view, in posterior 21 For Review Only 22 23 699 part major and accessory limbs broadly bifurcating, accessory limb with clear 24 25 700 constriction at half tooth length, thin, about half length and width of major limb. Central 26 27 701 formation with medium long and medium broad sharp carinated cusp and lateral flaps 28 29 30 702 with distinct posterior and postero-lateral margins. Flaps not completely fused with cusp 31 32 703 (Fig. 7H). 33 34 704 35 36 37 705 Remarks: This species corresponds to PSH 15 in Puillandre et al. (2012). Cryptogemma 38 39 706 powelli sp. nov. exhibits similar intraspecific variation in shell shape to that of its 40 41 707 congeners, based on its convex hull area (Fig. 1). Studied specimens vary in shades of 42 43 708 shell color, from light yellowish to light orange, occasionally with a patchy pattern. In 44 45 46 709 terms of both size and shape, C. powelli can in some cases be undistinguishable from C. 47 48 710 unilineata (Fig. 1). The morphometric analysis shows considerable overlap of the 49 50 711 measured shell features between the two species, with the main distinction between 51 52 53 712 them expressed by the second axis. Interpretation of this result suggests that the C. 54 55 713 powelli on average possess a more concave curvature of the outer aperture lip, 56 57 714 compared to C. unilineata. Besides, distinction of C. powelli and C. unilineata mostly 58 59 60 30 Page 31 of 65 Zoological Journal of the Linnean Society

1 2 3 4 715 relies on the degree of pronunciation of the subsutural cord – it is typically not 5 6 7 716 significantly stronger than the succeeding cords in the former species, but is always 8 9 717 thick and gemmate in the latter. The subsutural cord is, however, thicker in younger 10 11 718 specimens, rendering the distinction of juvenile C. powelli and C. unilineata specimens 12 13 14 719 problematic. 15 16 720 17 18 721 List of COI diagnostic sites (position: character state): [112: G, 322: T, 433: A] 19 20 722 21 For Review Only 22 23 723 Distribution: The confirmed distribution of this species based on sequenced specimens 24 25 724 ranges from the Indian Ocean to the Central Indo-Pacific (Fig. 4F), from ~400 to ~600m 26 27 725 of depth. 28 29 30 726 31 32 727 Discussion 33 34 728 Using an integrative taxonomy approach, by combining DNA-based species 35 36 37 729 delimitation methods with classical morphological taxonomic methods, we reevaluated 38 39 730 the Cryptogemma group studied in Puillandre et al. (2012) with new sequences – in 40 41 731 some cases from type material – resulting in eight delimited species. Among the eight 42 43 732 delimited species, C. phymatias, C. praesignis, C. timorensis and C. aethiopica 44 45 46 733 comprise 20 of the available species names, with C. powelli the only one described as 47 48 734 new. Our results contrast with the usual tendency in this family toward the discovery of 49 50 735 new species (e.g. Puillandre et al., 2017; Abdelkrim et al., 2018b). All the synonymized 51 52 53 736 species names in this group were established before the ‘molecular era of taxonomy’, 54 55 737 reinforcing the recommendation – yet not always followed (Bouchet & Strong, 2010) – 56 57 738 of systematically relying on molecular analysis, combined with other data, to delimit 58 59 60 31 Zoological Journal of the Linnean Society Page 32 of 65

1 2 3 4 739 species. However, our study also shows that using only molecular delimitation of 5 6 7 740 species while ignoring some key specimens such as types would limit knowledge of 8 9 741 intra-specific morphological variability, thus obscuring also the ability to relate such 10 11 742 variability to extrinsic parameters, such as geographic and bathymetric ranges. In this 12 13 14 743 study, morphological study and taxonomic expertise on types resulted in the extension 15 16 744 of intra-specific variability (Fig. 1; e.g. C. aethiopica types). This might be explained 17 18 745 because, when describing a new species, there can be a strong subjective effect on the 19 20 746 choice of the type specimens (usually the largest of a lot). The recent and promising 21 For Review Only 22 23 747 success in sequencing DNA obtained from empty shells (Villanea et al., 2016; 24 25 748 Sarkissian et al., 2017) should assist in attributing names to molecularly-defined species 26 27 749 by sequencing type specimens. However, sequencing shells implies destructive 28 29 30 750 sampling, and will therefore possibly not be accepted by many museum curators. 31 32 751 Furthermore, identifying and attributing material to species within a species complex 33 34 752 still requires considerable taxonomic investigation. 35 36 37 753 By exploring the intraspecific variability of the shell, in isolation or in relation to 38 39 754 interspecific variability, geographical and bathymetric ranges, or ecological data, we 40 41 755 here emphasize a few remarks that may provide directions for the future study of this 42 43 756 group. 44 45 46 757 Geographical intra-specific variability. In C. aethiopica, for which many species were 47 48 758 synonymized, there are significant geographical variations in the spire whorls. 49 50 759 Specimens from the West Indian Ocean were found to have very smooth spire whorls, 51 52 53 760 while specimens from the central Indo-Pacific were found with either two or three well- 54 55 761 marked gemmate cords. Considering the few available discrete shell features, these 56 57 762 distinguishable characteristics were used by authors to justify the creation of several 58 59 60 32 Page 33 of 65 Zoological Journal of the Linnean Society

1 2 3 4 763 species names. No such intraspecific variation in shell sculpture was found in the other 5 6 7 764 species of Cryptogemma, although other geographically-structured morphological 8 9 765 variability has been observed in the Turridae genus Lophiotoma (Puillandre et al. 2017), 10 11 766 with shell color varying from one side of the Indo-Pacific to the other. 12 13 14 767 Ecological variability. It is shown (e.g. Bierne et al., 2003; Hauquier et al., 2017) that 15 16 768 some benthic species have clear preferences with regards to the substrate on which they 17 18 769 thrive (e.g. sandy/muddy vs. rocky bottom). However, other species are generalists with 19 20 770 respect to the substratum, as seems to be the case for the Cryptogemma species. All 21 For Review Only 22 23 771 species were found both on hard and soft substrate, with or without organic matter (e.g. 24 25 772 leaves or sunken wood). A remarkable case is C. tessellata, for which the three 26 27 773 available trawl images showed three very different environments (Fig. 12), and, even 28 29 30 774 more remarkably, correlated with shell morphology (Fig. 10C-D); specimens found in 31 32 775 sandy or spongy environments exhibited a light pale color ground while the individual 33 34 776 found in a volcanic rock bottom has a reddish color ground. However, the holotype of 35 36 37 777 this species shows an intermediate pattern and was collected “in mud and sand” 38 39 778 (Powell, 1964). More material is needed to properly study the potential effects of 40 41 779 substrate on shell color. 42 43 780 Sexual dimorphism. C. praesignis is the first Turridae turrid species for which apparent 44 45 46 781 sexual dimorphism of the shell has been documented (Kantor & Sysoev, 1991). The 47 48 782 (mature) females possess one or several tertiary notches that is probably linked to 49 50 783 fertilization according to the authors, since mature males possess an extremely large and 51 52 53 784 muscular penis. The authors proposed the hypothesis that these notches appear 54 55 785 periodically during the breeding period, mainly based on the observation of several 56 57 786 successive notches on the shells. 58 59 60 33 Zoological Journal of the Linnean Society Page 34 of 65

1 2 3 4 787 Bathymetric zonation. Bathymetric segregation between sister species has been shown 5 6 7 788 in molluscan species (e.g. Clague et al., 2012). In the Turridae, it has been found that a 8 9 789 pair of sister species, Lophiotoma abbreviata and L. brevicaudata, show distinct 10 11 790 bathymetric preferences, with the former observed at an average depth of 3 m and the 12 13 14 791 latter at an average depth of 15 m (Puillandre et al., 2017). The present study 15 16 792 demonstrates that, at a much wider scale, a similar case was found in the C. 17 18 793 praesignis/C. phymatias species pair, with the former species occurring between ~300 19 20 794 to ~1400 m and the latter between ~1400 to ~3000m depth (Fig. 5). This example could 21 For Review Only 22 23 795 be added to the list of models used for studying speciation processes in benthic species 24 25 796 with high dispersal capabilities and large geographic distributions, for which a 26 27 797 topographic barrier would not satisfactorily explain genetic isolation (see Etter et al., 28 29 30 798 2005; Zardus et al., 2006). 31 32 799 Geographical ranges. While an Indo-Pacific distribution is common in benthic species, 33 34 800 but so far no – molecularly supported - cases of Pacific-Atlantic distribution have been 35 36 37 801 found in benthic gastropods. There are cases of cosmopolitan pelagic gastropods: 38 39 802 Glaucus atlanticus (Churchill et al., 2014), Clione limacina (Jennings et al., 2010) and 40 41 803 Fiona pinnata (Trickey et al., 2016). To this extent, C. phymatias could be the first 42 43 804 documented, molecularly supported, case of a benthic gastropod with a Pacific-Atlantic 44 45 46 805 distribution (Fig. 4A). Cases of such molecularly confirmed cosmopolitan benthic 47 48 806 species have been found in other phyla such as Annelida (Meyer et al., 2008; Sun et al., 49 50 807 2016; Tomioka et al., 2016), Arthropoda (Casanova et al., 1998; Havermans et al., 51 52 53 808 2013) and Algae (Macaya & Zuccarello, 2010). However, human-mediated introduction 54 55 809 was strongly suggested for some of these cases, e.g. annelids. Nothing is known of the 56 57 810 biological features of C. phymatias that would enable such a wide dispersal of larvae. 58 59 60 34 Page 35 of 65 Zoological Journal of the Linnean Society

1 2 3 4 811 Protoconch variability. As reported in Bouchet & Warren (1979; 1994), many deep- 5 6 7 812 water gastropods’ larval shells indicate a planktotrophic larval development, which is 8 9 813 generally associated with good dispersal abilities and gene flow. The Cryptogemma 10 11 814 species are no exception to this, with protoconch varying from 4 to 5.25 whorls. 12 13 14 815 However, no correlation was found between the number of whorls of protoconch in a 15 16 816 specimen and its associated depth, implying that, from a certain threshold (e.g. 4), the 17 18 817 number of whorls is not the only factor determining the ability to disperse. Furthermore, 19 20 818 in other neogastropod species complex for which there is a comparable number of 21 For Review Only 22 23 819 protoconch whorls (e.g. Sanders et al., 2017) the study showed a strict ocean-delimited 24 25 820 (e.g. Indian only, Pacific only) distribution. As for C. phymatias, the (eroded) 26 27 821 protoconch of the holotype of B. carynae is to our knowledge the only one preserved 28 29 30 822 (but slightly eroded) in this species, and it shows no major difference 31 32 823 (diameter/height/number of whorls) from other Cryptogemma species with regards to 33 34 824 this character that could explain its very large Pacific-Atlantic distribution. 35 36 37 825 Morphological stasis in teleoconch. C. praesignis and C. timorensis, previously 38 39 826 assigned to Ptychosyrinx in Puillandre et al. (2012), show considerable similarity in 40 41 827 shell sculpture despite not being sister species, and the two taxa would be difficult to 42 43 828 distinguish without their significant difference in size. These two species represent a 44 45 46 829 clear case of morphological homogeneity in Conoidea, for which other examples have 47 48 830 been reported (Kantor et al., 2018). Large-scale morphometrics correlated with 49 50 831 phylogeny may serve to better characterize and understand the conserved ‘Gemmula- 51 52 53 832 like’ form across geological time. Some specimens from the Mio-Pliocene have been 54 55 833 attributed to species of Cryptogemma (see ‘Taxonomy’), however, an incommensurable 56 57 834 number of ‘Ptychosyrinx-like’ or ‘Gemmula-like’ Turridae from the Paleogene will 58 59 60 35 Zoological Journal of the Linnean Society Page 36 of 65

1 2 3 4 835 probably remain unattributed to a clade until a more comprehensive understanding, 5 6 7 836 characterization and formalization of the family is achieved. 8 9 837 Radular morphology. The radula is an important morphological character in conoidean 10 11 838 taxonomy. In Turridae, the radula is seemingly quite similar between congeneric species 12 13 14 839 with few exceptions, such as for Iotyrris (Abdelkrim et al, 2018b). Cryptogemma 15 16 840 presently is one of few genera of Turridae for which the radula has been examined for 17 18 841 every representative species. Morphology of the marginal duplex teeth is rather similar 19 20 842 among all species, except for slight variation in the development of the accessory limb; 21 For Review Only 22 23 843 either shorter or nearly the same in length as the major limb (C. timorensis), or much 24 25 844 narrower or nearly of the same width as the major limb (C. tesselata). However, we 26 27 845 note this with the caveat that the shape of the duplex teeth as inferred from SEM images 28 29 30 846 may differ considerably due to the position of the tooth, thus rendering comparison of 31 32 847 different specimens and species imperfect. Additionally, there are large variations 33 34 848 between species in the morphology of the central formation in terms of the shape of the 35 36 37 849 central cusp and the degree of development and shape of the lateral flaps, in some cases 38 39 850 with very distinct elevated edges (C. tesselata), in others with the flaps being barely 40 41 851 distinguishable (Cryptogemma phymatias). Another potentially important character is 42 43 852 the relative length of the radula, which may differ greatly between otherwise 44 45 46 853 morphologically close species (e.g. C. praesignis and C. timorensis), number of 47 48 854 transverse rows of teeth (in Cryptogemma varying from 45 to 99) and the relative length 49 50 855 of the marginal teeth (from 1.22% of AL in C. powelli to 1.9% in C. praesignis). 51 52 53 856 Unfortunately, little has been published on relative tooth size in different Turridae 54 55 857 lineages and other groups of Conoidea to date. 56 57 858 58 59 60 36 Page 37 of 65 Zoological Journal of the Linnean Society

1 2 3 4 859 The above-mentioned characteristics of the Cryptogemma species have been revealed 5 6 7 860 by the morphological and molecular delimitation of species, a practice strongly advised 8 9 861 for future studies on the Turridae. In particular, it is shown here that morphological 10 11 862 differences associated with geographical or ecological diversity within the distribution 12 13 14 863 range of a species may lead to erroneous taxonomic conclusions. Unexpectedly wide 15 16 864 geographic distributions in particular, such as those documented herein, further add to 17 18 865 the challenges of conoidean taxonomy, notably where morphologically heterogenous 19 20 866 taxa are concerned. This is because material potentially seen as too widely separated 21 For Review Only 22 23 867 geographically to be considered relevant (or where geographically well-separated taxa 24 25 868 have not even been considered), may in fact represent conspecific material, a notion 26 27 869 supported by the numerous synonymizations herein. Combined with the fact that 28 29 30 870 sampling of the deep sea is infrequent and geographically severely incomplete 31 32 871 (potentially masking the presence of morphological clines which results in the over- 33 34 872 emphasis on few, morphologically discrete forms), future studies on deep-sea Conoidea 35 36 37 873 must take such wide distributions into account when embarking upon taxonomic 38 39 874 investigation. 40 41 875 Finally, the study of shell variability may serve to produce hypotheses regarding the 42 43 876 evolutionary processes that lead to such observed diversity. Indeed, the fact that the 44 45 46 877 actual, molecular-based Cryptogemma species were previously split by ante-molecular 47 48 878 taxonomists in several species each, has led us to a re-evaluation and re-interpretation of 49 50 879 the intraspecific morphological diversity that we could have neglected otherwise. In this 51 52 53 880 regard, the words of Darwin (1857) "It is good to have hair-splitters & lumpers." make 54 55 881 sense: the splitter reminds lumpers that intraspecific variability should not be ignored, 56 57 58 59 60 37 Zoological Journal of the Linnean Society Page 38 of 65

1 2 3 4 882 and the lumper reminds splitters that intraspecific variability should not be over- 5 6 7 883 interpreted. 8 9 884 10 11 885 References 12 13 14 886 15 16 887 Abdelkrim J, Aznar‐Cormano L, Buge B, Fedosov A, Kantor Y, Zaharias P, Puillandre 17 18 888 N. 2018a. Delimiting species of marine gastropods (Turridae, Conoidea) using RAD 19 20 21 889 sequencing in an integrativeFor taxonomy Review framework. Only Molecular Ecology 27: 4591–4611. 22 23 24 890 Abdelkrim J, Aznar-Cormano L, Fedosov AE, Kantor YI, Lozouet P, Phuong MA, 25 26 891 Zaharias P, Puillandre N. 2018b. Exon-Capture-Based Phylogeny and Diversification of 27 28 892 the Venomous Gastropods (Neogastropoda, Conoidea). Molecular Biology and 29 30 31 893 Evolution 35: 2355–2374. 32 33 34 894 Barnard KH. 1958. Contribution to the knowledge of South African marine Mollusca. 35 36 895 Part 1. Gastropoda; Prosobranchiata: Toxoglossa. Annals of the South African Museum. 37 38 896 Annale van die Suid-Afrikaanse Museum. 44: 73–163. 39 40 41 897 Bickford D, Lohman DJ, Sodhi NS, Ng PKL, Meier R, Winker K, Ingram KK, Das I. 42 43 44 898 2007. Cryptic species as a window on diversity and conservation. Trends in Ecology & 45 46 899 Evolution 22: 148–155. 47 48 49 900 Bierne N, Bonhomme F, David P. 2003. Habitat preference and the marine-speciation 50 51 901 paradox. Proceedings of the Royal Society of London. Series B: Biological Sciences 52 53 54 902 270: 1399–1406. 55 56 57 58 59 60 38 Page 39 of 65 Zoological Journal of the Linnean Society

1 2 3 4 903 Bonhomme V, Picq S, Gaucherel C, Claude J. 2014. Momocs: outline analysis using R. 5 6 7 904 Journal of Statistical Software 56: 1–24. 8 9 10 905 Bouchet P, Kantor YI. 2004. New Caledonia: The major centre of biodiversity for 11 12 906 volutomitrid molluscs (Mollusca: Neogastropoda: Volutomitridae). Systematics and 13 14 907 Biodiversity 1: 467–502. 15 16 17 18 908 Bouchet P, Kantor YI, Sysoev A, Puillandre N. 2011. A new operational classification 19 20 909 of the Conoidea (Gastropoda). Journal of Molluscan Studies 77: 273–308. 21 For Review Only 22 23 910 Bouchet P, Strong E. 2010. Historical Name-Bearing Types in Marine Molluscs: An 24 25 911 Impediment to Biodiversity Studies? In: Polaszek A, ed. Systema Naturae 250 - The 26 27 28 912 Linnaean Ark. CRC Press, 63–74. 29 30 31 913 Bouchet P, Waren A. 1979. Planktotrophig larval development in deep-water 32 33 914 gastropods. Sarsia 64: 37–40. 34 35 36 915 Bouchet P, Warén A. 1994. Ontogenetic migration and dispersal of deep-sea gastropod 37 38 916 larvae. Reproduction, larval biology, and recruitment of the deep-sea benthos. 39 40 41 917 Columbia University Press, New York: 98–118. 42 43 44 918 Bouckaert R, Vaughan TG, Barido-Sottani J, Duchêne S, Fourment M, Gavryushkina 45 46 919 A, Heled J, Jones G, Kühnert D, Maio ND, Matschiner M, Mendes FK, Müller NF, 47 48 920 Ogilvie HA, Plessis L du, Popinga A, Rambaut A, Rasmussen D, Siveroni I, Suchard 49 50 51 921 MA, Wu CH, Xie D, Zhang C, Stadler T, Drummond AJ. 2019. BEAST 2.5: An 52 53 922 advanced software platform for Bayesian evolutionary analysis. PLOS Computational 54 55 923 Biology 15: e1006650. 56 57 58 59 60 39 Zoological Journal of the Linnean Society Page 40 of 65

1 2 3 4 924 Casanova JP, De Jong L, Faure E. 1998. Interrelationships of the two families 5 6 7 925 constituting the Lophogastrida (Crustacea: Mysidacea) inferred from morphological and 8 9 926 molecular data. Marine Biology 132: 59–65. 10 11 12 927 Chan KO, Grismer LL, Brown RM. 2018. Comprehensive multi-locus phylogeny of 13 14 928 Old World tree frogs (Anura: Rhacophoridae) reveals taxonomic uncertainties and 15 16 17 929 potential cases of over- and underestimation of species diversity. Molecular 18 19 930 Phylogenetics and Evolution 127: 1010–1019. 20 21 For Review Only 22 931 Churchill CKC, Valdés Á, Foighil DÓ. 2014. Molecular and morphological systematics 23 24 932 of neustonic nudibranchs (Mollusca : Gastropoda : Glaucidae : Glaucus), with 25 26 27 933 descriptions of three new cryptic species. Invertebrate Systematics 28: 174–195. 28 29 30 934 Clague GE, Jones WJ, Paduan JB, Clague DA, Vrijenhoek RC. 2012. Phylogeography 31 32 935 of Acesta clams from submarine seamounts and escarpments along the western margin 33 34 936 of North America. Marine Ecology 33: 75–87. 35 36 37 38 937 Dall WH. 1918. Notes on the nomenclature of the mollusks of the family Turritidae. 39 40 938 Proceedings of the United States National Museum, 54, 313–333. 41 42 43 939 Darwin Correspondence Project, “Letter no. 2130,” accessed on 31 July 2019, 44 45 940 http://www.darwinproject.ac.uk/DCP-LETT-2130 46 47 48 941 Dayrat B. 2005. Towards integrative taxonomy. Biological Journal of the Linnean 49 50 51 942 Society 85: 407–417. 52 53 54 943 Dayrat B, Tillier A, Lecointre G, Tillier S. 2001. New Clades of Euthyneuran 55 56 944 Gastropods (Mollusca) from 28S rRNA Sequences. Molecular Phylogenetics and 57 58 59 945 Evolution 19: 225–235. 60 40 Page 41 of 65 Zoological Journal of the Linnean Society

1 2 3 4 946 Dejaco T, Gassner M, Arthofer W, Schlick-Steiner BC, Steiner FM. 2016. 5 6 7 947 Taxonomist’s Nightmare … Evolutionist’s Delight : An Integrative Approach Resolves 8 9 948 Species Limits in Jumping Bristletails Despite Widespread Hybridization and 10 11 949 Parthenogenesis. Systematic Biology 65: 947–974. 12 13 14 950 Edgar RC. 2004. MUSCLE: multiple sequence alignment with high accuracy and high 15 16 17 951 throughput. Nucleic Acids Research 32: 1792–1797. 18 19 20 952 Etter R, Rex MA, Chase MR, Quattro JM. 2005. Population Differentiation Decreases 21 For Review Only 22 953 with Depth in Deep-Sea Bivalves. Evolution 59: 1479–1491. 23 24 25 954 Fedosov AE, Puillandre N, Achaz, G. 2019. Revisiting use of DNA characters in 26 27 28 955 taxonomy with MolD-a tree independent algorithm to retrieve diagnostic nucleotide 29 30 956 characters from monolocus datasets. bioRxiv, 838151. 31 32 33 957 Fišer C, Robinson CT, Malard F. 2018. Cryptic species as a window into the paradigm 34 35 958 shift of the species concept. Molecular Ecology 27: 613–635. 36 37 38 959 Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R. 1994. DNA primers for 39 40 41 960 amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan 42 43 961 invertebrates. Molecular Marine Biology and Biotechnology: 6. 44 45 46 962 Galindo LA, Puillandre N, Strong EE, Bouchet P. 2014. Using microwaves to prepare 47 48 963 gastropods for DNA barcoding. Molecular Ecology Resources 14: 700–705. 49 50 51 52 964 Giardina CR, Kuhl FP. 1977. Accuracy of curve approximation by harmonically related 53 54 965 vectors with elliptical loci. Computer Graphics and Image Processing 6: 277–285. 55 56 57 58 59 60 41 Zoological Journal of the Linnean Society Page 42 of 65

1 2 3 4 966 Hasegawa K. 2009. Upper bathyal gastropods of the Pacific coast of northern Honshu, 5 6 7 967 Japan, chiefly collected by R/V Wakataka-maru. Deep-sea fauna and pollutants off 8 9 968 Pacific coast of northern Japan. Natl. Mus. Nat. Sci. Monograph, 39, 225-383. 10 11 12 969 Hauquier F, Leliaert F, Rigaux A, Derycke S, Vanreusel A. 2017. Distinct genetic 13 14 970 differentiation and species diversification within two marine nematodes with different 15 16 17 971 habitat preference in Antarctic sediments. BMC Evolutionary Biology 17: 120. 18 19 20 972 Havermans C, Sonet G, d’Acoz C d’Udekem, Nagy ZT, Martin P, Brix S, Riehl T, 21 For Review Only 22 973 Agrawal S, Held C. 2013. Genetic and Morphological Divergences in the Cosmopolitan 23 24 974 Deep-Sea Amphipod Eurythenes gryllus Reveal a Diverse Abyss and a Bipolar Species. 25 26 27 975 PLOS ONE 8: e74218. 28 29 30 976 Iglésias SP, Toulhoat L, Sellos DY. 2010. Taxonomic confusion and market 31 32 977 mislabelling of threatened skates: important consequences for their conservation status. 33 34 978 Aquatic Conservation: Marine and Freshwater Ecosystems 20: 319–333. 35 36 37 38 979 Jennings RM, Bucklin A, Ossenbrügger H, Hopcroft RR. 2010. Species diversity of 39 40 980 planktonic gastropods (Pteropoda and Heteropoda) from six ocean regions based on 41 42 981 DNA barcode analysis. Deep Sea Research Part II: Topical Studies in Oceanography 43 44 45 982 57: 2199–2210. 46 47 48 983 Jovelin R, Justine JL. 2001. Phylogenetic relationships within the polyopisthocotylean 49 50 984 monogeneans (Platyhelminthes) inferred from partial 28S rDNA sequences. 51 52 985 International Journal for Parasitology 31: 393–401. 53 54 55 986 Kabat AR. 1996. Molluscan types of the Albatross expeditions to the Eastern Pacific 56 57 58 987 described by WH Dall (1908): Bulletin of the Museum of Comparative Zoology. 59 60 42 Page 43 of 65 Zoological Journal of the Linnean Society

1 2 3 4 988 Kalyaanamoorthy S, Minh BQ, Wong TKF, von Haeseler A, Jermiin LS. 2017. 5 6 7 989 ModelFinder: fast model selection for accurate phylogenetic estimates. Nature Methods 8 9 990 14: 587–589. 10 11 12 991 Kantor YI, Fedosov AE, Puillandre N. 2018. New and unusual deep-water Conoidea 13 14 992 revised with shell, radula and DNA characters. Ruthenica 28. 15 16 17 18 993 Kantor YI, Puillandre N. 2012. Evolution of the Radular Apparatus in Conoidea 19 20 994 (Gastropoda: Neogastropoda) as Inferred from a Molecular Phylogeny. Malacologia 55: 21 For Review Only 22 995 55–90. 23 24 25 996 Kantor YI, Sysoev AV. 1991. Sexual Dimorphism in the Apertural Notch of a New 26 27 28 997 Species of Gemmula (Gastropoda: Turridae). Journal of Molluscan Studies 57: 205– 29 30 998 209. 31 32 33 999 Kapli P, Lutteropp S, Zhang J, Kobert K, Pavlidis P, Stamatakis A, Flouri T. 2017. 34 35 1000 Multi-rate Poisson tree processes for single-locus species delimitation under maximum 36 37 38 1001 likelihood and Markov chain Monte Carlo. Bioinformatics 33: 1630–1638. 39 40 41 1002 Kumar S, Stecher G, Li M, Knyaz C, Tamura K. 2018. MEGA X: Molecular 42 43 1003 Evolutionary Genetics Analysis across Computing Platforms. Molecular Biology and 44 45 1004 Evolution 35: 1547–1549. 46 47 48 1005 Lefébure T, Douady CJ, Gouy M, Gibert J. 2006. Relationship between morphological 49 50 51 1006 taxonomy and molecular divergence within Crustacea: Proposal of a molecular 52 53 1007 threshold to help species delimitation. Molecular Phylogenetics and Evolution 40: 435– 54 55 1008 447. 56 57 58 59 60 43 Zoological Journal of the Linnean Society Page 44 of 65

1 2 3 4 1009 Macaya EC, Zuccarello GC. 2010. Dna Barcoding and Genetic Divergence in the Giant 5 6 7 1010 Kelp Macrocystis (laminariales)1. Journal of Phycology 46: 736–742. 8 9 10 1011 Martin K. 1935. Oligocaene Gastropoden von Buton. Rijksmuseum van Geologie en 11 12 1012 Mineralogie. 13 14 15 1013 Meyer A, Bleidorn C, Rouse GW, Hausen H. 2008. Morphological and molecular data 16 17 18 1014 suggest a cosmopolitan distribution of the polychaete Proscoloplos cygnochaetus Day, 19 20 1015 1954 (Annelida, Orbiniidae). Marine Biology 153: 879–889. 21 For Review Only 22 23 1016 Nguyen LT, Schmidt HA, von Haeseler A, Minh BQ. 2015. IQ-TREE: A Fast and 24 25 1017 Effective Stochastic Algorithm for Estimating Maximum-Likelihood Phylogenies. 26 27 28 1018 Molecular Biology and Evolution 32: 268–274. 29 30 31 1019 Pante E, Schoelinck C, Puillandre N. 2015. From Integrative Taxonomy to Species 32 33 1020 Description: One Step Beyond. Systematic Biology 64: 152–160. 34 35 36 1021 Powell AWB. 1964. The family Turridae in the Indo-Pacific. Part 1. The subfamily 37 38 1022 Turrinae. Indo-Pacific mollusca. 1: 227–346. 39 40 41 42 1023 Powell AWB. 1967. The family Turridae in the Indo-Pacific, Part 1a. The Turrinae 43 44 1024 concluded. Indo-Pacific mollusca. 1: 409–444 45 46 47 1025 Puillandre N, Cruaud C, Kantor YI. 2010. Cryptic species in Gemmuloborsonia 48 49 1026 (Gastropoda: Conoidea). Journal of Molluscan Studies 76: 11–23. 50 51 52 1027 Puillandre N, Fedosov AE, Zaharias P, Aznar-Cormano L, Kantor YI. 2017. A quest for 53 54 55 1028 the lost types of Lophiotoma (Gastropoda: Conoidea: Turridae): integrative taxonomy in 56 57 1029 a nomenclatural mess. Zoological Journal of the Linnean Society 181: 243–271. 58 59 60 44 Page 45 of 65 Zoological Journal of the Linnean Society

1 2 3 4 1030 Puillandre N, Kantor YI, Sysoev A, Couloux A, Meyer C, Rawlings T, Todd JA, 5 6 7 1031 Bouchet P. 2011. The dragon tamed? A molecular phylogeny of the Conoidea 8 9 1032 (Gastropoda). Journal of Molluscan Studies 77: 259–272. 10 11 12 1033 Puillandre N, Modica MV, Zhang Y, Sirovich L, Boisselier MC, Cruaud C, Holford M, 13 14 1034 Samadi S. 2012. Large-scale species delimitation method for hyperdiverse groups. 15 16 17 1035 Molecular Ecology 21: 2671–2691. 18 19 20 1036 Robba E, Sartono S, Violanti D, Erba E. 1989. Early Pleistocene gastropods from 21 For Review Only 22 1037 Timor. Memorie di Scienze Geologiche 41: 61–113. 23 24 25 1038 Ronquist F, Huelsenbeck JP. 2003. MrBayes 3: Bayesian phylogenetic inference under 26 27 28 1039 mixed models. Bioinformatics 19: 1572–1574. 29 30 31 1040 Sanders MT, Merle D, Bouchet P, Castelin M, Beu AG, Samadi S, Puillandre N. 2017. 32 33 1041 One for each ocean: revision of the Bursa granularis species complex (Gastropoda: 34 35 1042 Tonnoidea: Bursidae). Journal of Molluscan Studies 83: 384–398. 36 37 38 1043 Sarkissian CD, Pichereau V, Dupont C, Ilsøe PC, Perrigault M, Butler P, Chauvaud L, 39 40 41 1044 Eiríksson J, Scourse J, Paillard C, Orlando L. 2017. Ancient DNA analysis identifies 42 43 1045 marine mollusc shells as new metagenomic archives of the past. Molecular Ecology 44 45 1046 Resources 17: 835–853. 46 47 48 1047 Simon C, Franke A, Martin A. 1991. The Polymerase Chain Reaction: DNA Extraction 49 50 51 1048 and Amplification. In: Hewitt GM,, In: Johnston AWB,, In: Young JPW, eds. NATO 52 53 1049 ASI Series. Molecular Techniques in Taxonomy. Berlin, Heidelberg: Springer Berlin 54 55 1050 Heidelberg, 329–355. 56 57 58 59 60 45 Zoological Journal of the Linnean Society Page 46 of 65

1 2 3 4 1051 Sun Y, Wong E, Tovar-Hernández MA, Williamson JE, Kupriyanova EK. 2016. Is 5 6 7 1052 Hydroides brachyacantha (Serpulidae : Annelida) a widespread species? Invertebrate 8 9 1053 Systematics 30: 41–59. 10 11 12 1054 Sysoev A. 1996. Deep-sea conoidean gastropods collected by the John Murray 13 14 1055 Expedition, 1933-34. Bulletin of the Natural History Museum. 62: 1–30. 15 16 17 18 1056 Thiele J. 1929. Handbuch der systematischen weichtierkunde. Jena: G. Fischer. 19 20 21 1057 Todd JA, Rawlings TA.For 2014. AReview review of the Polystira Only clade—the Neotropic’s largest 22 23 1058 marine gastropod radiation (Neogastropoda: Conoidea: Turridae sensu stricto ). Zootaxa 24 25 1059 3884: 445–491. 26 27 28 1060 Tomioka S, Kondoh T, Sato-Okoshi W, Ito K, Kakui K, Kajihara H. 2016. 29 30 31 1061 Cosmopolitan or Cryptic Species? A Case Study of Capitella teleta (Annelida: 32 33 1062 Capitellidae). Zoological Science 33: 545–554. 34 35 36 1063 Trickey JS, Thiel M, Waters JM. 2016. Transoceanic dispersal and cryptic diversity in a 37 38 1064 cosmopolitan rafting nudibranch. Invertebrate Systematics 30: 290–301. 39 40 41 42 1065 Tucker JK. 2004. Catalog of Recent and fossil turrids (Mollusca: Gastropoda). Zootaxa 43 44 1066 682: 1–1295. 45 46 47 1067 Villanea FA, Parent CE, Kemp BM. 2016. Reviving Galápagos snails: ancient DNA 48 49 1068 extraction and amplification from shells of probably extinct endemic land snails. 50 51 52 1069 Journal of Molluscan Studies 82: 449–456. 53 54 55 1070 Will KW, Mishler BD, Wheeler QD. 2005. The Perils of DNA Barcoding and the Need 56 57 1071 for Integrative Taxonomy. Systematic Biology 54: 844–851. 58 59 60 46 Page 47 of 65 Zoological Journal of the Linnean Society

1 2 3 4 1072 Zardus JD, Etter RJ, Chase MR, Rex MA, Boyle EE. 2006. Bathymetric and geographic 5 6 7 1073 population structure in the pan-Atlantic deep-sea bivalve Deminucula atacellana 8 9 1074 (Schenck, 1939). Molecular Ecology 15: 639–651. 10 11 1075 12 13 14 1076 15 16 1077 Legends 17 18 1078 19 20 1079 Figure 1: Projection of specimens along the two first discriminant axes. Convex hulls 21 For Review Only 22 23 1080 are drawn to easily observe the range of each species in the morphospace. Triangles are 24 25 1081 type specimens, with names of species indicated; the background color corresponds to 26 27 1082 the species attribution from the literature and the border color to the LDA prediction. 28 29 30 1083 Type specimens were not included in the LDA training set. Orange segments 31 32 1084 correspond to the projection of coefficients multiplied by the standard deviation of each 33 34 1085 raw variable (PC axes and shell length). Outlines of most extreme shapes are drawn 35 36 37 1086 with their respective MNHN numbers. 38 39 1087 40 41 1088 Figure 2: Bayesian Tree (MrBayes) for the COI. Posterior probabilities (>0.95) are 42 43 1089 only shown for specific and inter-specific nodes. Abbreviations next to each specimen 44 45 46 1090 number refer to main marine provinces: WIP: Western Indo-Pacific; CIP: Central Indo- 47 48 1091 Pacific; EIP; Eastern Indo-Pacific; EP: Eastern Pacific; WA: Western Atlantic. 49 50 1092 Alignment provided in Appendix 1. 51 52 53 1093 54 55 56 57 58 59 60 47 Zoological Journal of the Linnean Society Page 48 of 65

1 2 3 4 1094 Figure 3: Bayesian Tree (MrBayes) of the COI, 12S and 28S concatenated. Posterior 5 6 7 1095 probabilities and bootstrap values are shown for each node. Alignment provided in 8 9 1096 Appendix 2. 10 11 1097 12 13 14 1098 Figure 4: Map showing the species distributions. Filled circles represent localities of 15 16 1099 sequenced specimens; full triangles represent localities of type specimens. 17 18 1100 19 20 1101 Figure 5: Candlestick chart of depth distributions of the eight species. The high and low 21 For Review Only 22 23 1102 value correspond respectively to the minimum and maximum values of depth recorded 24 25 1103 for each species, while the open and close values correspond to the confirmed minimum 26 27 1104 and maximum values of depth. Confirmed minimum and maximum values of depth 28 29 30 1105 correspond respectively to the highest depth value of ending of trawl and the highest 31 32 1106 depth value of beginning of trawl. 33 34 1107 35 36 37 1108 Figure 6: Cryptogemma phymatias (Watson, 1886). (A) Lectotype of Gemmula 38 39 1109 benthima Dall, 1908, USNM 123089, Gulf of Panama. (B) Paralectotype of Gemmula 40 41 1110 benthima Dall, 1908, USNM 123087, Gulf of Panama. (C) Holotype of Bathybermudia 42 43 1111 carynae Haas, 1949, FMNH 31656, Bermuda. (D) Lateral view of the protoconch of B. 44 45 46 1112 carynae (E) Holotype of Pleurotoma phymatias Watson, 1886, NHMUK 1887.2.9.957, 47 48 1113 Philippines. (F-G) AMS T0077, IN2017_V03 - Sampling the Abyss, Jervis 49 50 1114 Commonwealth Marine Reserve. (H) MNHN-IM-2013-49928, R/V Tansei-maru, KT- 51 52 53 1115 12-32, Okinawa. (I) MNHN-IM-2009-13476, AURORA 2007, Philippines. 54 55 1116 56 57 58 59 60 48 Page 49 of 65 Zoological Journal of the Linnean Society

1 2 3 4 1117 Figure 7: Radulae of studied Cryptogemma. (A) Cryptogemma praesignis 5 6 7 1118 (Smith,1895). (B-C) Cryptogemma phymatias (Watson, 1886). (B) MNHN 8 9 1119 uncatalogued, Biocal, New-Caledonia, st. CP23 (C) USNM 857019, Venezuela. (D) 10 11 1120 Cryptogemma periscelida (Dall, 1889), R/V Pelican, 28º23.935’N, 89º22.508’W, 675- 12 13 14 1121 765 m. (E) Cryptogemma aethiopica (Thiele, 1925), MNHN-IM-2013-50175, 15 16 1122 DONGSHA 2014, Taïwan. (F) Cryptogemma tessellata (Powell, 1964), MNHN-IM- 17 18 1123 2007-40775, EBISCO, Chesterfield Islands. (G) Cryptogemma unilineata (Powell, 19 20 1124 1964), MNHN-IM-2013-61844, ZhongSha 2015, Taïwan. (H) Cryptogemma powelli 21 For Review Only 22 23 1125 sp. nov., MNHN-IM-2013-68787, KANACONO, New Caledonia. (I) Cryptogemma 24 25 1126 timorensis (Tesch, 1915), MNHN-IM-2013-09864, PAPUA NIUGINI, Papua. Scale 26 27 1127 bars 50 μm. 28 29 30 1128 31 32 1129 Figure 8: (A-K) Cryptogemma praesignis (Smith,1895) and (L-P) Cryptogemma 33 34 1130 timorensis (Tesch, 1915). (A) Holotype of Pleurotoma praesignis Smith, 1895, ZSIC, 35 36 37 1131 Sri Lanka. (B) MNHN-IM-2009-13477, AURORA-2007, Philippines. (C) MNHN-IM- 38 39 1132 2013-62903, BIOMAGLO, Iles Glorieuses. (D) Holotype of Pleurotoma lobata 40 41 1133 Sowerby III, 1903, NHMUK 1903.7.27.49, South Africa. (E) Holotype of Ptychosyrinx 42 43 1134 lordhoweensis, ZMMU Lc-14532, Lord Howe Rise. (F) Syntype of Pleurotoma 44 45 46 1135 bisinuata von Martens, 1901, ZMB 683, off East Africa. (G) Holotype of Pleurotoma 47 48 1136 microscelida Dall, 1895, USNM 127122, Hawaiian Islands. (H). Holotype of 49 50 1137 Pleurotoma rotatilis von Martens, 1902, ZMB 60070, off East Africa. (I) MNHN-IM- 51 52 53 1138 2013-62893, BIOMAGLO, Iles Glorieuses. (J) MNHN-IM-2013-62900, BIOMAGLO, 54 55 1139 Iles Glorieuses. (K) Lateral view of the protoconch of MNHN-IM-2013-62893. (L) 56 57 1140 Holotype of Ptychosyrinx timorensis teschi Powell, 1964, USNM 239068, Borneo. (M) 58 59 60 49 Zoological Journal of the Linnean Society Page 50 of 65

1 2 3 4 1141 MNHN-IM-2009-14860, ATIMO VATAE, South of Madagascar. (N) MNHN-IM- 5 6 7 1142 2007-40928, SALOMON 2, Salomon Islands. (O) MNHN-IM- 2013-58793, 8 9 1143 KAVIENG, Papua. (P) Lateral view of the protoconch of MNHN-IM-2009-14935. 10 11 1144 12 13 14 1145 Figure 9: Cryptogemma aethiopica (Thiele, 1925). (A) Holotype of Pleurotoma 15 16 1146 aethiopica Thiele 1925, ZMB 109376, off East Africa. (B) Holotype of Pleurotoma 17 18 1147 fusiformis Thiele, 1925, ZMB 109388, off Sumatra. (C) MNH- IM-2009-14979, 19 20 21 1148 ATIMO VATAE, SouthFor of Madagascar. Review (D) Lateral Only view of the protoconch of MNHN- 22 23 1149 IM-2013-48115, KANADEEP, Chesterfield Islands. (E) MNHN-IM- 2009-29301, 24 25 1150 EXBODI, New Caledonia. (F) MNHN-IM- 2009-14778, ATIMO VATAE, South of 26 27 1151 Madagascar. (G) Holotype of Pinguigemmula luzonica Powell, 1964, USNM 237784, 28 29 30 1152 Philippines. (H) Holotype of Pinguigemmula okinavensis McNeil, 1960, UNSM MO 31 32 1153 562851, Miocene or Pliocene of Okinawa. (I) MNHN-IM-2013-19950, PAPUA 33 34 1154 NIUGINI, Papua. (J) Holotype of Pinguigemmula philippinensis Powell, 1964, USNM 35 36 37 1155 237563, Philippines. 38 39 1156 40 41 1157 42 43 44 1158 Figure 10: (A-D) Cryptogemma tessellata (Powell, 1964), (E-H) Cryptogemma 45 46 1159 unilineata (Powell, 1964) & (I-K) Cryptogemma periscelida (Dall, 1889). (A) Holotype 47 48 1160 of Gemmula tessellata Powell, 1964, AWMM MA71042, Hawaiian Islands. (B) Lateral 49 50 1161 view of the protoconch of MNHN-IM-2013-48252, KANADEEP, Chesterfield Islands. 51 52 53 1162 (C) MNHN-IM-2009-24975, EXBODI, New Caledonia. (D) MNHN-IM- 2007-40775, 54 55 1163 EBSICO, Chesterfield Islands. (E) Holotype of Gemmula congener unilineata Powell, 56 57 1164 1964, BPBM 8929, Hawaiian Islands. (F) Lateral view of the protoconch of MNHN- 58 59 60 50 Page 51 of 65 Zoological Journal of the Linnean Society

1 2 3 4 1165 IM-2007-40786, NORFOLK 2, New Caledonia. (G) MNHN-IM-2009-14953, ATIMO 5 6 7 1166 VATAE, South of Madagascar. (H) MNHN-IM-2007-38340, MAINBAZA, Mozambic 8 9 1167 channel. (I) Lectotype of Pleurotoma periscelida Dall, 1889, USNM 87391, Columbia. 10 11 1168 (J) MNHN-IM- 2013-56285, GUYANE 2014, French Guyana. (K) MNHN-IM- 2013- 12 13 14 1169 56283, GUYANE 2014, French Guyana. 15 16 1170 17 18 1171 19 20 1172 Figure 11: Cryptogemma powelli sp. nov. (A-C) Holotype, MNHN-IM- 2013-68787, 21 For Review Only 22 23 1173 KANACONO, New Caledonia. (D) Lateral view of the protoconch of the paratype 1 24 25 1174 MNHN-IM- 2007-40765, EBISCO, Chesterfield Islands. (E) Paratype 2, MNHN-IM- 26 27 1175 2007-40795, Norfolk 2, New Caledonia. (F) MNHN-IM-2013-55832, TAIWAN 2013, 28 29 30 1176 Taïwan. (G-I) Paratype, MNHN-IM- 2007-40765, EBISCO, Chesterfield Islands. 31 32 1177 33 34 1178 Figure 12: Pictures of trawls from stations were Cryptogemma tessellata (Powel, 1964) 35 36 37 1179 specimens were found. 38 39 1180 40 41 1181 42 43 1182 Supplementary Material 44 45 46 1183 47 48 1184 Figure S1: Result of the leave-one-out cross-validation procedure that was used to find 49 50 1185 the number of PC axes leading to the best percentage of species predictions. 51 52 53 1186 54 55 1187 Figure S2: Results of the ABGD, mPTP and GMYC analysis 56 57 1188 58 59 60 51 Zoological Journal of the Linnean Society Page 52 of 65

1 2 3 4 1189 Figure S3: Results of the morphometric analysis without taking size in account. 5 6 7 1190 8 9 1191 Table S1: Material examined in this study 10 11 12 13 14 15 16 17 18 19 20 21 For Review Only 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 52 Page 53 of 65 Zoological Journal of the Linnean Society

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 For Review Only 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 53 Zoological Journal of the Linnean Society Page 54 of 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Projection of specimens along the two first discriminant axes. Convex hulls are drawn to easily observe the 19 range of each species in theFor morphospace. Review Triangles are type Only specimens, with names of species indicated; 20 the background color corresponds to the species attribution from the literature and the border color to the 21 LDA prediction. Type specimens were not included in the LDA training set. Orange segments correspond to 22 the projection of coefficients multiplied by the standard deviation of each raw variable (PC axes and shell 23 length). Outlines of most extreme shapes are drawn with their respective MNHN numbers. 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 55 of 65 Zoological Journal of the Linnean Society

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 Bayesian Tree (MrBayes) for the COI. Posterior probabilities (>0.95) are only shown for specific and inter- 46 specific nodes. Abbreviations next to each specimen number refer to main marine provinces: WIP: Western 47 Indo-Pacific; CIP: Central Indo-Pacific; EIP; Eastern Indo-Pacific; EP: Eastern Pacific; WA: Western Atlantic. Alignment provided in Appendix 1. 48 49 205x289mm (300 x 300 DPI) 50 51 52 53 54 55 56 57 58 59 60 Zoological Journal of the Linnean Society Page 56 of 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 Bayesian Tree (MrBayes) of the COI, 12S and 28S concatenated. Posterior probabilities and bootstrap values 24 are shown for each node. Alignment provided in Appendix 2. 25 26 155x79mm (300 x 300 DPI) 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 57 of 65 Zoological Journal of the Linnean Society

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 Map showing the species distributions. Filled circles represent localities of sequenced specimens; full 46 triangles represent localities of type specimens. 47 188x297mm (300 x 300 DPI) 48 49 50 51 52 53 54 55 56 57 58 59 60 Zoological Journal of the Linnean Society Page 58 of 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Candlestick chart of depth distributions of the eight species. The high and low value correspond respectively 34 to the minimum and maximum values of depth recorded for each species, while the open and close values 35 correspond to the confirmed minimum and maximum values of depth. Confirmed minimum and maximum 36 values of depth correspond respectively to the highest depth value of ending of trawl and the highest depth 37 value of beginning of trawl. 38 39 276x230mm (300 x 300 DPI) 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 59 of 65 Zoological Journal of the Linnean Society

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 Cryptogemma phymatias (Watson, 1886). (A) Lectotype of Gemmula benthima Dall, 1908, USNM 123089, 46 Gulf of Panama. (B) Paralectotype of Gemmula benthima Dall, 1908, USNM 123087, Gulf of Panama. (C) 47 Holotype of Bathybermudia carynae Haas, 1949, FMNH 31656, Bermuda. (D) Lateral view of the protoconch of B. carynae (E) Holotype of Pleurotoma phymatias Watson, 1886, NHMUK 1887.2.9.957, Philippines. (F-G) 48 AMS T0077, IN2017_V03 - Sampling the Abyss, Jervis Commonwealth Marine Reserve. (H) MNHN-IM-2013- 49 49928, R/V Tansei-maru, KT-12-32, Okinawa. (I) MNHN-IM-2009-13476, AURORA 2007, Philippines. 50 51 189x231mm (300 x 300 DPI) 52 53 54 55 56 57 58 59 60 Zoological Journal of the Linnean Society Page 60 of 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 Radulae of studied Cryptogemma. (A) Cryptogemma praesignis (Smith,1895). (B-C) Cryptogemma phymatias (Watson, 1886). (B) MNHN uncatalogued, Biocal, New-Caledonia, st. CP23 (C) USNM 857019, 40 Venezuela. (D) Cryptogemma periscelida (Dall, 1889), R/V Pelican, 28º23.935’N, 89º22.508’W, 675-765 m. 41 (E) Cryptogemma aethiopica (Thiele, 1925), MNHN-IM-2013-50175, DONGSHA 2014, Taïwan. (F) 42 Cryptogemma tessellata (Powell, 1964), MNHN-IM-2007-40775, EBISCO, Chesterfield Islands. (G) 43 Cryptogemma unilineata (Powell, 1964), MNHN-IM-2013-61844, ZhongSha 2015, Taïwan. (H) Cryptogemma 44 powelli sp. nov., MNHN-IM-2013-68787, KANACONO, New Caledonia. (I) Cryptogemma timorensis (Tesch, 45 1915), MNHN-IM-2013-09864, PAPUA NIUGINI, Papua. Scale bars 50 μm. 46 210x210mm (300 x 300 DPI) 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 61 of 65 Zoological Journal of the Linnean Society

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 (A-K) Cryptogemma praesignis (Smith,1895) and (L-P) Cryptogemma timorensis (Tesch, 1915). (A) 46 Holotype of Pleurotoma praesignis Smith, 1895, ZSIC, Sri Lanka. (B) MNHN-IM-2009-13477, AURORA-2007, 47 Philippines. (C) MNHN-IM-2013-62903, BIOMAGLO, Iles Glorieuses. (D) Holotype of Pleurotoma lobata Sowerby III, 1903, NHMUK 1903.7.27.49, South Africa. (E) Holotype of Ptychosyrinx lordhoweensis, ZMMU 48 Lc-14532, Lord Howe Rise. (F) Syntype of Pleurotoma bisinuata von Martens, 1901, ZMB 683, off East 49 Africa. (G) Holotype of Pleurotoma microscelida Dall, 1895, USNM 127122, Hawaiian Islands. (H). Holotype 50 of Pleurotoma rotatilis von Martens, 1902, ZMB 60070, off East Africa. (I) MNHN-IM-2013-62893, 51 BIOMAGLO, Iles Glorieuses. (J) MNHN-IM-2013-62900, BIOMAGLO, Iles Glorieuses. (K) Lateral view of the 52 protoconch of MNHN-IM-2013-62893. (L) Holotype of Ptychosyrinx timorensis teschi Powell, 1964, USNM 53 239068, Borneo. (M) MNHN-IM-2009-14860, ATIMO VATAE, South of Madagascar. (N) MNHN-IM- 2007- 40928, SALOMON 2, Salomon Islands. (O) MNHN-IM- 2013-58793, KAVIENG, Papua. (P) Lateral view of the 54 protoconch of MNHN-IM-2009-14935. 55 56 210x293mm (300 x 300 DPI) 57 58 59 60 Zoological Journal of the Linnean Society Page 62 of 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 63 of 65 Zoological Journal of the Linnean Society

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 Cryptogemma aethiopica (Thiele, 1925). (A) Holotype of Pleurotoma aethiopica Thiele 1925, ZMB 109376, 40 off East Africa. (B) Holotype of Pleurotoma fusiformis Thiele, 1925, ZMB 109388, off Sumatra. (C) MNH- IM- 41 2009-14979, ATIMO VATAE, South of Madagascar. (D) Lateral view of the protoconch of MNHN-IM-2013- 42 48115, KANADEEP, Chesterfield Islands. (E) MNHN-IM- 2009-29301, EXBODI, New Caledonia. (F) MNHN-IM- 43 2009-14778, ATIMO VATAE, South of Madagascar. (G) Holotype of Pinguigemmula luzonica Powell, 1964, 44 USNM 237784, Philippines. (H) Holotype of Pinguigemmula okinavensis McNeil, 1960, UNSM MO 562851, Miocene or Pliocene of Okinawa. (I) MNHN-IM-2013-19950, PAPUA NIUGINI, Papua. (J) Holotype of 45 Pinguigemmula philippinensis Powell, 1964, USNM 237563, Philippines. 46 47 210x214mm (300 x 300 DPI) 48 49 50 51 52 53 54 55 56 57 58 59 60 Zoological Journal of the Linnean Society Page 64 of 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 (A-D) Cryptogemma tessellata (Powell, 1964), (E-H) Cryptogemma unilineata (Powell, 1964) & (I-K) 40 Cryptogemma periscelida (Dall, 1889). (A) Holotype of Gemmula tessellata Powell, 1964, AWMM MA71042, 41 Hawaiian Islands. (B) Lateral view of the protoconch of MNHN-IM-2013-48252, KANADEEP, Chesterfield 42 Islands. (C) MNHN-IM-2009-24975, EXBODI, New Caledonia. (D) MNHN-IM- 2007-40775, EBSICO, 43 Chesterfield Islands. (E) Holotype of Gemmula congener unilineata Powell, 1964, BPBM 8929, Hawaiian 44 Islands. (F) Lateral view of the protoconch of MNHN-IM-2007-40786, NORFOLK 2, New Caledonia. (G) 45 MNHN-IM-2009-14953, ATIMO VATAE, South of Madagascar. (H) MNHN-IM-2007-38340, MAINBAZA, Mozambic channel. (I) Lectotype of Pleurotoma periscelida Dall, 1889, USNM 87391, Columbia. (J) MNHN- 46 IM- 2013-56285, GUYANE 2014, French Guyana. (K) MNHN-IM- 2013-56283, GUYANE 2014, French 47 Guyana. 48 49 210x215mm (300 x 300 DPI) 50 51 52 53 54 55 56 57 58 59 60 Page 65 of 65 Zoological Journal of the Linnean Society

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Cryptogemma powelli sp. nov. (A-C) Holotype, MNHN-IM- 2013-68787, KANACONO, New Caledonia. (D) 35 Lateral view of the protoconch of the paratype 1 MNHN-IM- 2007-40765, EBISCO, Chesterfield Islands. (E) 36 Paratype 2, MNHN-IM-2007-40795, Norfolk 2, New Caledonia. (F) MNHN-IM-2013-55832, TAIWAN 2013, 37 Taïwan. (G-I) Paratype, MNHN-IM- 2007-40765, EBISCO, Chesterfield Islands. 38 39 192x167mm (300 x 300 DPI) 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Zoological Journal of the Linnean Society Page 66 of 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Review Only 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 Pictures of trawls from stations were Cryptogemma tessellata (Powel, 1964) specimens were found. 46 47 433x947mm (200 x 200 DPI) 48 49 50 51 52 53 54 55 56 57 58 59 60 Acknowledgments

This work was supported by the Service de Systématique Moléculaire (UMS 2700

CNRS MNHN), the CONOTAX project funded by the French National Research

Agency (grant number ANR-13-JSV7- 0013-01), the bilateral cooperation research funding from the Ministry of Science and Technology, Taiwan (grant number MOST

102-2923-B-002-001-MY3) and the French National Research Agency (grant number

ANR 12-ISV7-0005-01). The contribution of Y. Kantor and A. Fedosov was supported by the grant from Russian Science Foundation no. 16-14-10118-П (PI Y. Kantor), F.

Criscione and A. Hallan supported by a grant from the Australian Biological Resources

Study (ABRS grant RF217-57 to PI F. Criscione), and Y. Kano by the Japan Society for the Promotion of Science (KAKENHI grant nos 15H04412 and 18H02494).

The material in this paper originates from numerous shore-based expeditions and deep sea cruises, conducted respectively by MNHN and Pro-Natura International (PNI) as part of the Our Planet Reviewed programme (ATIMO VATAE, MAINBAZA,

GUYANE 2014, PAPUA NIUGINI, KAVIENG 2014; PI P. Bouchet)) and/or by

MNHN and Institut de Recherche pour le Développement (IRD) as part of the Tropical

Deep-Sea Benthos programme (BIOMAGLO, TARASOC, MIRIKY, AURORA 2007,

EBISCO, TAIWAN 2013, DongSha 2014, NANHAI 2014, BIOPAPUA, SALOMON

2, SALOMONBOA 3, CONCALIS, EXBODI, NORFOLK 2, TERRASSES,

ZHONGSHA 2015, KANACONO, KANADEEP, MADEEP; PIs Bertrand Richer de

Forges, Philippe Bouchet, Sarah Samadi, Nicolas Puillandre, Wei-Jen Chen, Tin-Yam

Chan). Scientific partners included the University of Papua New Guinea (UPNG);

National Fisheries College, Kavieng, Papua New Guinea; the National Museum of the

Philippines; Institut d'Halieutique et Sciences Marines (IH.SM), Université de Tuléar, Madagascar; University of Taipei and University of Keelung, Taiwan; Universidade

Eduardo Mondlane, Maputo; the Madagascar bureau of the Wildlife Conservation

Society (WCS); and Instituto Español de Oceanografia (IOE). Funders and sponsors included the Total Foundation, Prince Albert II of Monaco Foundation, Stavros

Niarchos Foundation, Richard Lounsbery Foundation, Vinci Entrepose Contracting,

Fondation EDF, European Regional Development Fund (ERDF), the Philippines Bureau of Fisheries and Aquatic Research (BFAR), the French Ministry of Foreign Affairs,

Fonds Pacifique and the Government of New Caledonia. Additional field work included

PANGLAO 2004, a joint project of MNHN and University of San Carlos, Cebu City.

All expeditions operated under the regulations then in force in the countries in question and satisfy the conditions set by the Nagoya Protocol for access to genetic resources.

The authors would like to thank Virginie Héros, Barbara Buge, and Julien Brisset

(MNHN) for their help in curating the vouchers; Alexander Sysoev (ZMMU) for photographs and tissue loan; Christine Zorn (ZMB) for photographs and type specimens loan; Ellen Strong, Mark. S. Florence and Kathy Hollis (USNM) for photographs and specimens loan; Jochen Gerber (FMNH) for photographs; Philippe Maestrati (MNHN) for photographs; Regina Kawamoto & Norine Yeung (BPBM) for photographs;

Severine Hannam (AWMM) for photographs; Andreia Salvador (NHMUK) for photographs;Jun Hashimoto (Nagasaki University) for organizing the Nagasaki-maru cruise N275; the Plateau Technique de Microscopie Électronique (PtME) of MNHN for

SEM pictures. Posthumous thank goes to Richard Kilburn for photograph of P. praesignis.