HERRMANN, KANTOR & BOUCHET, 2018

HERRMANN, KANTOR & BOUCHET SP. NOV.

Type data: Holotype: MNHN IM-2013-40655, New Caledonia, Kouakoué Canyon , EXBODI Stn 3821, 21°53′S, 166°50′E, 211–440 m, lv, 24.25 mm ( Fig. 21A, B) GoogleMaps . Paratypes: New Caledonia, EXBODI, Stn 3800, East coast off Thio, 21°32′S, 166°22′E, 279–282 m, 1 lv (Paratype 3, 21.5 mm); Stn 3807, East coast off Toupeti, 21°43′S, 166°36′E, 352–372 m, 1 lv (Paratype 2, 26.9 mm); Stn 3810, East coast off Toupeti, 21°44′S, 166°38′E, 384–385 m, 2 lv (Paratype 1, 27.35 mm; Paratype 4, 24.0 mm); Stn 3814, Passe du Solitaire, 21°49′S, 166°44′E, 331–344 m, 1 lv (Paratype 5, 20.4 mm) GoogleMaps .

Other material examined: New Caledonia, EXBODI Stn 3795, 21°32′S, 166°21′E, 240-245m, 1 lv (16.4 mm); Stn 3800, 21°32′S, 166°22′E, 279-282m, 12 lv (16.7, 17.1, 17.3, 15.0, 16.2, 16.2, 16.55, 16.8, 15.05, 23.1, 17.65, 17.8 mm); Stn 3805, 21°42′S, 166°34′E, 302 m, 1 lv (20.9 mm); Stn 3806, 21°42′S, 166°34′E, 307–309 m, 7 lv (22.0, 18.4, 15.9, 16.3, 14.0, 9.15, 9.8 mm); Stn 3807, 21°43′S, 166°36′E, 352–372 m, 11 lv (22.6, 20.1, 21.05, 20.7, 20.0, 19.2, 17.4, 20.9, 18.1, 18.2, 19.65 mm); Stn 3810, 21°44′S, 166°38′E, 384–385 m, 8 lv (25.5, 25.85, 29.3, 25.5, 27.4, 21.65, 25.5, 23.3 mm); Stn 3814, 21°49′S, 166°44′E, 331–344 m, 6 lv (22.8, 21.5, 20.6, 21.6, 18.9, 14.1 mm); Stn 3821, 21°53′S, 166°50′E, 211–440 m, 2 lv (25.85 and 25.15 mm); Stn 3823, 21°55′S, 166°55′E, 246–255 m, 2 lv (15.5 and 14.5 mm); Stn 3825, 21°58′S, 166°59′E, 349–405 m, 1 lv (16.2 mm); Stn 3828, 22°00′S, 167°01′E, 300–302 m, 1 lv (16.4 mm); Stn 3829, 22°02′S, 167°05′E, 350–360 m, 1 lv (23.0 mm); TERRASSES Stn 3095, 22°02′S, 167°06′E, 320–380 m, 1 lv (21.5 mm) GoogleMaps .

Papua New Guinea, BIOPAPUA Stn CP 3634, 07°29′S, 147°31′E, 279–290 m, 1 dd (21.75 mm) GoogleMaps .

Description: Shell small to medium sized (holotype 24.25 mm), elongate-fusiform to almost biconical, heavily sculptured, with narrow aperture and stepped appearance of late teleoconch whorls. Protoconch pointed, cyrtoconoid, translucent, of three or more slightly convex, glossy whorls. Protoconch/teleoconch transition distinct. Teleoconch of about nine whorls; suture canaliculated. Subsutural ramp forming distinct shelf, flat or even slightly inclined inside, giving teleoconch whorls a subcylindrical profile. Whorl periphery slightly convex, early spire whorls sculptured with three strong, closely set, spiral cords overriding low indistinct axial ribs to form series of prominent, axially aligned, beads. On third teleoconch whorl, adapical spiral cord splits into two, and succeeding spire whorls sculptured with four gemmate cords, adapical cord bordering subsutural ramp wavy, slightly narrower than others, delineated from second cord by narrower interspace. On later teleoconch whorls, interspaces between gemmate spiral cords gradually broadening, axial ribs becoming more distinct and sculpture rather reticulate. Last adult and penultimate whorls with 26 axial ribs each. Interspaces between axial ribs and spiral cords forming deep quadrangular depressions with microsculpture of fine co-axial growth lines, sometimes retaining fragments of periostracum forming squamiform projections. Shell base extended to moderately long, tapering, slightly notched siphonal canal. Four gemmate spiral cords on adapical portion of last adult whorl, and 12 cords on shell base and siphonal canal.

Aperture narrow, elongated; outer lip wavy, convex in adapical portion and straight below mid-height. Inside of outer lip smooth. Inner lip strongly calloused, sometimes reflected, retaining weak reticulate sculpture adapically, with four closely set fine columellar folds at mid-height. Coloration uniformly cream, protoconch white.

Radula (of Paratype 5) 0.62 mm long, 0.13 mm wide, of about 75 rows. Rachidian narrow, about 20 µm wide, bearing eight strong, moderately long, pointed cusps. Laterals attaining 47 µm in width, with straight anterior margin bearing 16+ strong, pointed, rather widely set cusps.

Distribution: New Caledonia, Papua New Guinea, 255– 384 m.

Etymology: The species epithet refers to the type locality.

Remarks: While the specimens from New Caledonia show a fine reticulate sculpture pattern on the late teleoconch whorls, the specimen from Papua New Guinea is characterized by denser axial ribs and in overall shell morphology is somewhat intermediate between such typical G. neocaledonica sp. nov. and Fusidomiporta ponderi sp. nov. The latter species is the one that resembles most G. neocaledonica sp. nov., but it can be differentiated by its notably stronger spiral cords and rather fusiform shell. The combination of elongate-biconical shell with subcylindrical whorl profile and characteristic reticulate pattern on late whorls allows for easy recognition of G. neocaledonica sp. nov. Other species of Gemmulimitra gen. nov. differ either in shell shape or in sculpture pattern, and G. avenacea , despite being the closest to G. neocaledonica sp. nov. in our multi-gene analyses, shows no morphological resemblance to it whatsoever.

IMPLICATIONS OF MORPHOLOGICAL STUDIES FOR THE TAXONOMY OF THE MITRIDAE

According to our earlier results ( Fedosov et al., 2015), the families Mitridae , Charitodoronidae and Pyramimitridae form a well-supported phylogenetic group, which we designate as a superfamily Mitroidea , although they are rather heterogeneous morphologically, as summarized in Table 4. The relationships between the three mitroidean families are clearly established, but the affinities of Mitroidea remain uncertain. The phylogenetic analysis of Fedosov et al. (2015) suggested that Mitroidea may be a sister group to Conoidea; this affinity, however, is not supported by any shared morphological features and needs to be further investigated.

The morphological distinctiveness of the Mitridae has been appreciated by many authors and, when anatomical descriptions scattered in many papers are brought together, the family ranks among the better studied among neogastropods. Beside multiple illustrations of radulae ( Risbec, 1928; Cernohorsky, 1970, 1976, 1991), data on the general body morphology and/ or foregut anatomy are available, among Mitrinae , for Mitra mitra ( Ponder, 1972) , Ziba carinata ( Simone & Turner, 2010) , Pseudonebularia cucumerina ( Risbec, 1928) and Episcomitra zonata ( Vayssière, 1901) ; among Strigatellinae , for Strigatella paupercula ( Ponder, 1972) , S. retusa and S. scutulata ( Risbec, 1928) ; among Imbricariinae , for Imbricariopsis conovula ( Ponder, 1972) ; among Cylindromitrinae , for Pterygia crenulata ( Risbec, 1928) ; among Isarinae , for Isara cornea ( Harasewych, 2009) ; and for Atrimitra idae ( West, 1990) , A. catalinae ( West, 1991) and Condylomitra tuberosa (herein), that we treat as incertae sedis. All the species that were studied with sufficient attention revealed an epiproboscis (referred to as ‘ tube à venin ’ by Risbec, 1928), either in combination with a radula of underived morphology (in most of the species mentioned) or with a uniserial radula ( Pterygia crenulata , Condylomitra tuberosa ). Meanwhile, an epiproboscis is not present in the Charitodoronidae , the Neogastropoda clade closest to the Mitridae . There is no doubt that the epiproboscis represents an autapomorphy of the Mitridae and can be used for the anatomical circumscription of the family.

In this connection, the morphology of the Caribbean Pleioptygma helenae , as addressed by Quinn (1989), deserves special attention. Quinn’s description of the foregut anatomy raises many questions; in particular, the position of the buccal mass is not described explicitly, as well as the nature of the ‘ proboscis bulb ’, and we believe that the homologies of these organs may have been misinterpreted by him. This assumption is further reinforced by the fact that the two specimens dissected by Quinn displayed some variation in foregut morphology, which may be an artefact of poor fixation (with the use of rhum suggested by the author!). Some parts of the description are of special interest. In particular, the ‘proboscis bulb attached to the cephalic cavity floor by a broad, rather thick band of muscles originating in the foot’ ( Quinn, 1989: 14) more likely refers to the large buccal mass and a radula/odontophore retractor muscle. Then the peculiar introvert, which ‘invaginates and runs back through the outer tube/sheath and enters the proboscis bulb [?=buccal mass]’ may not be anything but an epiproboscis.

Undoubtedly, the anatomy of Pleioptygma helenae is peculiar and its homologies need to be clarified based on investigations of additional material. However, given its rarity and the small chance of obtaining live-taken specimens in the near future, we dare reassess the systematics of Pleioptygma based on currently available data. Several lines of evidence support a placement of Pleioptygma as a separate subfamily within the Mitridae , including (1) the mitriform shell, (2) the radula with rachidian and laterals of about equal morphology – similar to the one in Domiporta and (3) the presence of a structure that, based on topology and morphology, is closely comparable to the mitrid epiproboscis, and most likely is an epiproboscis.

EVOLUTION OF RADULAR MORPHOLOGY IN THE MITRIDAE

Our studies revealed only two major radula types in the family Costellariidae , with a very simple scenario of two independent transitions from plesiomorphic to derived ( Fedosov et al., 2017). The situation is incomparably more complex in the Mitridae . Here, we recognize six major types of radula; each is referred to a genus that typically represents it ( Fig. 41).

1 – Mitra type: radula with laterals notably wider than rachidian tooth and bearing multiple equal or subequal cusps. This underived radula morphology is widespread across the Mitridae tree. It characterizes the family Charitodoronidae , most Mitrinae (except the Domiporta group and some Pseudonebularia ), most Isarinae , as well as the incertae sedis genera Probata , Carinomitra , Atrimitra ( Cernohorsky, 1970; West 1990, 1991) and Dibaphimitra ( Bayer, 1942) .

1a – Strigatella type: radula with slightly curved laterals bearing short robust cusps on their medial convex portion and lacking cusps laterally; rachidian always with central unpaired cusp. This type of radula is found in all Strigatellinae (except in the S. lugubris – S. coronata clade). The Strigatella type of radula represents a modified Mitra type, and intermediate morphologies with Strigatella -like laterals but different rachidian are found outside Strigatellinae in Neotiara nodulosa ( Mitrinae ) and Cancilla schepmani ( Imbricariinae ).

1b – Nebularia type: a slightly modified version of the Mitra type, with a characteristic rachidian bearing only five cusps, the central unpaired cusp being notably enlarged. This type of morphology characterizes the genus Nebularia as circumscribed herein.

2 – Profundimitra type: very wide rachidian, roughly attaining the width of laterals and bearing equal number of cusps. This type of radula characterizes the mitrine genera Profundimitra and Fusidomiporta and is also found in some Domiporta (but not in its type species, D. filaris ), Pseudonebularia maesta and, outside Mitrinae , in the genus Pleioptygma ( Quinn, 1989) .

3 – Imbricaria type: radula characterized by moderately wide multicuspidate laterals, with one of the medial cusps notably enlarged compared to the others. Cusps of laterals are also differentiated in size, with a central unpaired cusp retained, although often reduced in size. This type of radula is found in all Imbricaria species except I. fulgetrum .

3a – Cancilla type: it resembles the Imbricaria type, but the laterals are more than twice as wide as the rachidian, and the central unpaired cusp of the rachidian is absent. This type of radula is found in Cancilla isabella ( Salisbury & Huang, 2015) and other species of this genus studied herein, as well as in Imbricariopsis , Subcancilla erythrogramma and Imbricaria fulgetrum .

4 – Scabricola type: radula characterized by strongly modified laterals in which the lateralmost cusp is greatly enlarged to form a robust spine. Typically, the rachidian is comb-like ( Scabricola variegata , Swainsonia spp. ) but it may also bear a single strong cusp. This type of radula is found in species of Scabricola and Swainsonia .

5 – Neocancilla type: radula characterized by a rachidian with a pair of very robust and blunt central cusps, and laterals bearing few short and robust cusps in their medial portions. This type of radula is known in all species of Neocancilla . The radulae of Scabricola olivaeformis and S. coriacea have morphologies somewhat intermediate between the Scabricola type and the Neocancilla type.

6 – Pterygia type: very narrow uniserial radulae remarkable by the complete loss of laterals. Rachidians of varying morphology, but usually flattened, with serrated margins. Radulae of this type are found in all species of the genera Pterygia and Condylomitra , and in the Strigatella lugubris – Strigatella coronata clade of Strigatella .

Although the diversity of radular morphologies in the family Mitridae is impressive, the understanding of its evolution is greatly hampered by homoplasies. Radula morphology does not clearly align with the inferred phylogenetic groupings of the Mitridae ( Fig. 41) because of an ubiquitous retention, in advanced lineages, of the plesiomorphic state and because of multiple convergences. In 15 genera of Mitridae , all studied species presented an underived Mitra - type radula. Seven genera appeared heterogeneous in radula morphology, combining several general types. Thus, specific apomorphies can be identified in only few cases, and in even fewer cases the apomorphic state is shared by all members of the lineage. Only eight genera ( Profundimitra , Fusidomiporta , Neocancilla , Pterygia , Nebularia , Condylomitra and, with some reservations, Imbricariopsis and Swainsonia ) are supported by distinct apomorphies in radular morphology. Among them, only Nebularia and Neocancilla show autapomorphic radula types. Undoubtedly, the subfamily Imbricariinae shows the greatest diversity of radular morphologies, and the pattern of radular evolution is most obvious and consistent in that subfamily. The progressive differentiation of cusps on both the laterals and the rachidian is noteworthy, along with a general tendency to the reduction of the number of cusps. The radulae of Cancilla spp. are closest to the underived Mitra type, and alternate courses of radula transformation are observed in the Imbricaria and Scabricola – Neocancilla clades.

In addition to a blurred phylogenetic signal of radula characters, our understanding of radula evolution in the Mitridae is impeded by a lack of evidence on the functionality of different types of radula, as there are no data on the possible adaptive value of different morphologies and selection pressures that have led to the emergence of the observed diversity. Further studies on the functional morphology and biochemistry of mitrid secretions would in this respect open new perspectives.

TAXONOMY OF MITRIDAE : HISTORICAL CONSIDERATIONS AND NEW ARRANGEMENT

The current Mitridae species list (as indexed in World Register of Marine Species. consulted on 19 August 2017) comprises 402 accepted Recent species, plus the three new species described in the present work. The baseline of Mitridae systematics is the fundamental revisions by Walter Cernohorsky, who first dealt with the subfamily Mitrinae ( Cernohorsky, 1976) , and later with the Imbricariinae and Cylindromitrinae ( Cernohorsky, 1991) . Cernohorsky’s monographs had a profound impact on mitrid taxonomy and systematics, despite the general lumping attitude prevailing in his time, which resulted in an extensive taxonomic graveyard for many of the species-level taxa he accepted as valid. Several cases of abusive synonymization are demonstrated in the present study. For instance, Mitra morchii A. Adams, 1855 , considered by Cernohorsky a synonym of Cancilla isabella , is, based on Huang & Salisbury (2017) and on our data, not related to Cancilla or even to Imbricariinae . Another example is Mitra millepunctata G. B. Sowerby III, 1889 synonymized ( Cernohorsky, 1976) with Domiporta carnicolor despite notable disparity in shell proportions, and later described as Mitra terryni Poppe, 2008 ( Herrmann, 2017). Altogether the number of valid species of Mitridae was greatly diminished by Cernohorsky, but the rate of species description increased notably in the following decades and over 100 species (i.e. almost 30% of the currently accepted number of species) were described since 1991. This burst of activity in mitrid taxonomy is also notable for being almost entirely accounted for by the amateur community: only ten out of 112 species described in the last 25 years were described by academics. Amateurs perhaps more reluctantly establish new supraspecific taxa, and only two genera Calcimitra Huang, 2011 and Magnamitra Huang & Salisbury, 2017 were established in the last 25 years, whereas 65 species described over this period were originally placed in Mitra .

A concise review of changes in mitrid genus-level taxonomy was presented by Cernohorsky (1970), in a study that itself contained the description of four new genera and subgenera ( Domiporta , Dibaphimitra , Neocancilla and Sohlia ). In the present study, we have re-assessed the genus-level systematics of Recent Mitridae based on a combination of molecular and morphological data. The genus Charitodoron is segregated in the newly established family Charitodoronidae . The 26 genera of the revised Mitridae comprise six subfamilies: Mitrinae (with 14 genera), Strigatellinae (with the single genus Strigatella ), Imbricariinae (with six genera), Cylindromitrinae (with the genera Pterygia and Nebularia ), Isarinae new subfamily (with the genera Isara and Subcancilla ) and Pleioptygmatinae (with the single genus Pleioptygma ); seven genera, Atrimitra Dall, 1918 , Carinomitra gen. nov., Condylomitra gen. nov., Dibaphimitra Cernohorsky, 1970 , Magnamitra Huang & Salisbury, 2017 , Vicimitra Iredale, 1929 and Probata Sarasúa, 1989 are treated as incertae sedis. Undoubtedly the most revolutionary change in the taxonomy of the Mitridae is the falling apart of Mitra with the genus in its former taxonomic extension now reassigned to 14 genera, of which six are new. Other noteworthy changes are the transfer of the formerly mitrine Nebularia to Cylindromitrinae ; of Strigatella to its own, newly recognized, subfamily Strigatellinae ; of the formerly imbricariine genera Ziba and Domiporta to Mitrinae ; and of Subcancilla to the newly established subfamily Isarinae . The genera Mitra and Ziba now include a much reduced species diversity, as all the Indo-Pacific species earlier placed in Ziba are now transferred to Imbricaria . The contents of Imbricaria is also expanded as a result of the transfer of the Indo-Pacific species of Subcancilla , which appear unrelated to the New World species of that genus.

The 32 genera now recognized in the family Mitridae increase considerably the previously accepted genus-level diversity (19 genera, including Charitodoron , indexed in WoRMS as of August 2017). Still, the placement of several sequenced species remains uncertain. Lineage 3 represented by MNHN IM-2007-35623 and lineage 16 represented by MNHN IM-2007-30270 did not cluster with any other lineage and were not successfully sequenced for genetic markers other than COI. Thus, they at present remain ‘hanging’ in the list of unallocated Mitridae and may potentially represent two more new genera.

PHYLOGEOGRAPHIC PATTERNS IN THE MITRIDAE

It has been widely known that the Indo-West Pacific harbours the greatest diversity of mitrid species and lineages, and therefore a good representation of IP localities at various depths was a primary requirement to our sampling. Eighty eight of the 103 inferred species of Mitridae and Charitodoronidae in our data set originate from Indo-Pacific localities. Besides, three species were sampled from the Mediterranean and NE Atlantic, two from West Africa, four from the Caribbean and six from the Panamic province ( Fig. 42). Twenty-three mitrid genera were sampled in the Indo-Pacific, of which 19 do not occur outside the Indo-Pacific according to our data. Moreover, the subfamilies Imbricariinae and Cylindromitrinae , as circumscribed herein, are represented solely by Indo-Pacific forms. At least three genera ( Episcomitra , Isara and Ziba ) are found in West Africa, with the first two also inhabiting the Mediterranean. The New World fauna of Mitridae includes at least eight genera in four subfamilies: Mitrinae ( Neotiara ), Strigatellinae ( Strigatella ), Isarinae ( Isara , Subcancilla ) and Pleioptygmatinae ( Pleioptygma ), plus the genera Atrimitra , Dibaphimitra and Probata that we treat as incertae sedis. The six genera Atrimitra , Dibaphimitra , Neotiara , Pleioptygma , Probata and Subcancilla are currently thought to be endemic to Panamic and/or the Caribbean, although a close affinity of Panamic and Caribbean species was herein confirmed only in the genera Neotiara and Subcancilla .

Of all studied genera, Charitodoron probably demonstrates the narrowest range, being confined to deep waters in the Mozambique Chanel and off South Africa. According to some authors (e.g. Obura, 2012), starting in the Eocene, this area served as a main refuge for relict lineages of Tethyan origin which, for some reason, did not give rise to new radiations. The long branch that separates Charitodoron on the molecular tree, some characters such as the underived morphology (with only the radula indicating its relatedness to Mitridae ), and its low diversity in the Recent fauna, all point to the relict nature of Charitodoron . The paucispiral, bulbous protoconch of Charitodoron indicates non-planktotrophic development, which correlates well with its restricted distribution.

Conversely, the distribution of Isara is the widest among mitrid genera: this is the only genus recorded in more than two major zoogeographical regions of the shelf (as defined by Briggs & Bowen, 2012) – Indo-West Pacific, East Atlantic and West Atlantic ( Fig. 42) – and it may be also present in the East Pacific (see remarks under Isara ). Isara species contribute significantly to mitrid diversity in the peripheral Indo-Pacific (South Australia, presumably South Africa) (see Fig. 8), but they are rare and not really diverse in the Central Indo-West Pacific. In our understanding, Isara is undoubtedly an old and underived lineage of Mitridae , and this pattern may be interpreted in two different ways. According to one possible scenario, it was once widely distributed in tropical seas and subsequently was replaced by younger and derived lineages in the Central Indo-West Pacific while maintaining its diversity in peripheral areas. The other possible scenario suggests that the primarily Indo-Pacific Isara , once ‘forced’ into subtropical waters by growing competition with younger evolutionary lineages of Mitridae , has adapted to new temperatures and, through this, was capable of spreading beyond the biogeographical limits of the Indo-Pacific. Although most known species of Strigatella occur in the Indo-Pacific, an amphi- Pacific distribution characterizes this genus, with at least one species inhabiting shallow water in the Panamic province ending up close to the Indo-Pacific species in both molecular and morphological characters. Finally, if ‘ Mitra ’ hebes is related to the species of Domiporta , as suggested by shell and radular morphology, the range of that genus would also include West Africa.

The generally low overlap in regional lists of mitrid genera may be interpreted as a consequence of the relatively late major diversification of mitrid lineages, dating back to the time when modern biogeographical barriers were already established, and largely separate evolutionary radiations having taken place in (1) Late Tethys – Paratethys – Indo-Pacific, (2) West Africa and (3) the New World.

We have attempted to circumscribe the bathymetric distribution of the newly delineated genera, based primarily on the locality data of sequenced specimens. Given the patchiness of our data, we arbitrarily divided the sampled depth range from 0 to about 1800 m into four depth intervals: (1) from 0 to 40 m, (2) from 41 to 80 m, (3) from 81 to 300 m and (4)> 300 m.

Of the 103 species of Mitridae studied herein based on molecular characters, 72 were sampled from the 0- to 40-m interval, seven from the 41- to 80-m interval, nine from the 81- to 300-m interval, and 20 from depths in excess of 300 m. This accounting does not, however, strictly reflects changes in species richness with depths, as it is strongly biased by the distribution of collecting efforts ( Fig. 43). Intertidal and upper subtidal zones (at diveable depths) as well as outer slopes from about 200 down to about 1800 m were sampled incomparably better than the mesophotic and abyssal zones. Therefore, our data do not necessarily demonstrate a drop in species richness in the 41- to 300-m interval. Nevertheless, despite the sampling bias, there is an obvious pattern of greater mitrid diversity in shallow water, decreasing with depth. This pattern can be explained by a greater diversity of habitats, often fragmented and intermixed at a small scale, in shallow-water ecosystems. Another related factor is the multitude of biological interactions in these shallow habitats, driven by the complex ecological structure of reefs and associated habitats, providing a wealth of niches to be explored by mitrids and their preys.

The bathymetric distribution of the newly delineated mitrid genera is shown in Figure 43 based on our results (black vertical bars) and literature (grey vertical bars). Because of the very fragmentary data, some of the displayed results are inconsistent, like the disjunct bathymetric range of Gemmulimitra and Subcancilla . We in fact fully expect that many genera, which in our results are restricted to the 0- to 40-m-depth interval, in fact reach deeper, probably down to 80– 120 m. Nevertheless, some general conclusions can be drawn from the observed bathymetric distribution of the genera. The representation of mitrid genera in shallow water is notably higher than at depths in excess of 100–150 m, which mirrors the distribution of species diversity discussed above. Whereas Gemmulimitra , Imbricaria , Isara , Roseomitra and Subcancilla have extensive bathymetric ranges, with congeneric species sampled from the intertidal down to 300–400 m, the mitrid fauna from the greatest depths is essentially represented by specialized lineages (treated here as genera), that do not occur in shallower water. The genera Calcimitra , Cancilla , Cancillopsis , Fusidomiporta and Profundimitra represented in our data set by 15 species (plus two species of Charitodoron ) and the genus Eumitra (not included in our phylogenetic studies) constitute several separate radiations that thrive in the deep sea.

Interpretation of the fossil mitrid taxa is difficult, primarily due to the fact that similar shell features, such as shell shape or sculpture pattern, were demonstrated to have evolved convergently in unrelated Recent lineages of the family. The inferred topology of the Mitridae tree, with unresolved early polytomy, further complicates tracing the evolution of shell features and the placement of fossil forms. Cernohorsky (1970, 1976) suggested that the most Recent genera can be traced back to Miocene; however, given the drastic rearrangement of the genus-level taxonomy of the family herein, the validity of Cernohorsky’s statement needs to be revisited. Whereas the Oligocene Clifdenia is here considered a possible relative of the Recent Calcimitra , relationships of the genera Dentimitra Koenen, 1890 , Fusimitra Conrad, 1855 and Pseudocancilla Staadt, 1913 are highly speculative. The early Miocene Austroimbricaria gracilior (Ihering, 1897) , with a low spire and strong columellar folds, is clearly a mitrid; however, it is unclear to which of the conchologically similar Pterygia or Imbricariopsis it may be close.

WORLD DIVERSITY OF MITRIDAE

Of 103 species included in our molecular data set, 89 were identified with confidence – these constitute 22% of the described diversity of the family Mitridae . Of the remaining 15 species, four were either not identified (‘sp.’) or were attributed to a described species with an indication of some disparity in morphological or molecular characters (referred to as ‘cf.’); ten proved to be unnamed – although three of them showed some similarity to described species and were allocated to tentative species complexes (referred to as ‘aff .’). Remarkably, of the 60 mitrid species from the Indo-Pacific sampled in the 0- to 40-m-depth interval, no new species were identified with confidence. The proportion of unidentified/undescribed species grows with increasing depth and, of 20 species from depths greater than 300 m in the analysis, only 11 were identified with confidence, and seven (i.e. 35%) were undoubtedly new, all representing previously undescribed lineages of Mitrinae . Three of these species were here described as Profundimitra taylori gen. et sp. nov., Fusidomiporta ponderi gen. et sp. nov. and Gemmulimitra neocaledonica gen. et sp. nov. Thus, an estimate of 30–100 species of Mitridae remaining to be described from the deep waters of the Indo-Pacific seems reasonable to us. Recognition of new deep-water mitrid species may be hampered by the high intraspecific variability of shell sculpture and, simultaneously, rampant convergences leading to hardly distinguishable shells in not closely related species. Likewise, the radula appears generally useless for species delimitation, as in most cases it retains plesiomorphic morphology, and the significance of slight variations in the number of cusps is unknown. Thus, molecular data will remain essential for a reliable recognition of new deep water taxa of Mitridae .

Whereas a significant proportion of the deep-water mitrids remains undescribed, the overall diversity of the family in deep-water pales in comparison with the well-documented diversity of the family in shallow water. These contrasting proportions of described Mitridae in shallow vs. deep water can be viewed as a result of recent efforts by collectors and amateur taxonomists. Since the first half of the 19 th century and the iconic monograph by Reeve (1844–1845), the comparatively easily accessible species from shallow water are collected steadily throughout many Indo-Pacific locales and then carefully scrutinized by a dedicated community of amateurs, leading to the present day state of mitrid taxonomy, with a pretty well-inventoried shallow-water fauna. Undoubtedly, there are areas with high local endemism that still hold undescribed species, especially in peripheral locales (e.g. southern Australia, South and East Africa and the Arabian Sea), but this is not the general situation.

The diversity of mitrids in the Mediterranean is low, and all three species known from that sea were included in the present study. Conversely, we have included only a limited number of species from West Africa and the New World, which makes any judgment on diversity in these regions untimely. Two Panamic species in our data set were not confidently identified but were represented by subadults, the identification of which is always troublesome, and the uncertainty of our identifications cannot be viewed as an indication of undescribed species. A more thorough sampling in both the Panamic and Caribbean region is definitely required to assess the proportion of undescribed species in these areas and uncover the relationships of their endemic phylogenetic lineages.