Notes on Afonsoconus Tucker & Tenorio, 2013 (Gastropoda, Conidae), with description of a new species from the Southwestern Indian Ocean

. Although cone snails are among the most studied group of gastropods, new species are still regularly described. Here, we focus on Afonsoconus Tucker & Tenorio, 2013, a lineage that includes only two species from the Indo-Pacific Ocean. The analysis of molecular (partial mitochondrial cox1 gene sequences) and morphological (shell and radular tooth) characters revealed that the samples collected by dredging in deep water during a recent expedition carried out in the Mozambique Channel are different from the samples collected in the Pacific Ocean. We thus introduce here a new species, Afonsoconus crosnieri sp. nov., from the SW Indian Ocean including records from the Mozambique Channel, the Comoros and Glorieuses Islands, Madagascar, South Africa and reunion Island.


Introduction
During the last decade, no less than 199 new species of cone snails (Gastropoda, Conoidea) have been described, making the 2010's, although still not finished, the most fruitful decade ever in terms of species description for cone snails (WoRMS editorial board 2018). Thus, even though cone snails are among the most studied group of marine molluscs, mainly because their fascinating shell diversity and the deadly venoms they produce have long attracted the attention of shell collectors and toxinologists, it seems As a result of the dredging carried out by the royal Danish research Ship Galathea in shallow water (58-85 m) off raoul Island, Kermadec Islands in 1952 (Powell 1958), a shell of moderate size, rather slender with a narrowly conical spire and carinated at the shoulder was sampled. This was considered a new species, and described with the name Conus (Dauciconus) bruuni Powell, 1958 (Fig. 1F-H).
The original generic or subgeneric placements of the species kinoshitai and bruuni were solely based on apparent morphological shell similarities with other taxa. This did not necessarily imply a phylogenetic affinity with the respective type species of each genus/subgenus. Röckel et al. (1995) treated all species  Kuroda, 1956, holotype, Kii, Japan, ca 100 fathoms, 71.0 mm (NSMN NC-H329). B. Afonsoconus kinoshitai (Kuroda, 1956), specimen from Formosa Strait, Taiwan,78.4 mm (INHS 44626). C. Conus (Virgiconus) tamikoae Shikama, 1973, holotype, north of Senkaku Island, Japan, 84.0 mm (KPMY 5602). D. Afonsoconus kinoshitai, specimen from Taiwan, 51.1 mm (MJT). E. Afonsoconus kinoshitai f. calliginosus (Shikama, 1979), specimen from Philippines, 55.3 mm (EM). F. Conus bruuni Powell, 1958, holotype, off raoul Island, Kermadec Islands, 29°13´ S, 177°57´ W, 75-85 m, 43.5 mm (NHMD-91131, previously ZMUC-GAS-808). G-H. Afonsoconus bruuni (Powell, 1958). G. Specimen from south New Caledonia, 223 m, 47.4 mm (MNHN). H. Specimen from Banc Cryptélia, Norfolk ridge, New Caledonia, 180-220 m, 48.3 mm (MNHN). I. radular tooth of A. kinoshitai from specimen 1D. J. radular tooth of A. bruuni, Atheris voucher specimen from New Caledonia, S L 68 mm (CP080507AB). Scale bars = 10 mm, unless otherwise stated. of cone snails as members of one single genus, Conus. Alternatively, Tucker & Tenorio (2009) proposed a new classification for the recent and fossil cone snails based upon shell and radula morphologies and available molecular data. Preliminary examination of the radular teeth of kinoshitai and bruuni (rolán & raybaudi-Massilia 1994a(rolán & raybaudi-Massilia , 1994b (Fig. 1I-J) suggested a close relationship between these two taxa, which were provisionally placed by Tucker & Tenorio (2009) in the genus Asprella Schaufuss, 1869 along with a rather large number of other species. The genus Asprella thus defined turned out to be polyphyletic (Puillandre et al. 2014). It was split into several genera that were more consistent with the phylogenetic relationships among the different species (Tucker & Tenorio 2013). The species kinoshitai and bruuni were placed in the new genus Afonsoconus Tucker & Tenorio, 2013. In the subsequent classification of Conidae proposed by Puillandre et al. (2015) based upon the molecular phylogeny, Afonsoconus was given subgeneric rank within Conus. According to the reconstructed phylogeny of Puillandre et al. (2014), Afonsoconus is the sister group to fish-eating species placed in the (sub)genus Textilia Swainson, 1840 ( Fig. 2B-D), and this relationship was highly supported. Irrespective of its subjective ranking as a genus or a subgenus, the supraspecific taxon Afonsoconus includes species that form a monophyletic group. The species in this group exhibit characteristic radular features which can be considered true synapomorphies that allow immediate separation from the species in their sister group Textilia. According to Tucker & Tenorio (2013) and WoRMS editorial board (2018), the (sub)genus Afonsoconus thus includes two extant species, and no species of Afonsoconus have been reported as fossils (Tucker & Tenorio 2013). The species included in (sub)genus Afonsoconus occur in the Indo-Pacific region, ranging from Japan to New Zealand (Kermadec Islands) through Taiwan, the Philippines, the Solomon Islands and New Caledonia (röckel et al. 1995;Monnier et al. 2018). The occurrence of Afonsoconus kinoshitai in the Indian ocean (Mozambique, Madagascar and reunion Islands) has been cited in the literature (rolán & raybaudi-Massilia 1994a;röckel et al. 1995).
In 2017, the Muséum national d'Histoire naturelle (MNHN) carried out the oceanographic expedition BIoMAGLo (https://expeditions.mnhn.fr/campaign/biomaglo) aboard the rV Antéa, within the framework of the Tropical Deep Sea Benthos program (Corbari et al. 2017). The main objective of the BIoMAGLo campaign was to explore biodiversity and study the deep marine ecosystems of the islands of Mayotte, Glorieuses and Comoros in the Indian ocean, and to highlight their distinctness or affinities with other regions of the Mozambique Channel. It also pursued the analysis of assemblages of species in relation to the diversity of deep habitats, and the connectivity between the African and north/ south coasts of Madagascar to determine the isolation level of the region for selected model species. During this research cruise, the deep benthic fauna in the Mayotte-Glorieuses zone was surveyed, dredging from 80 m to a depth of 1070 m. Several live specimens of a cone snail initially identified as Conus (Afonsoconus) kinoshitai were collected. Several empty shells of apparently the same species had previously been collected in the course of other MNHN expeditions to the Mozambique Channel such as MIrIKY (https://expeditions.mnhn.fr/campaign/miriky), carried out in 2009 to the northwest of Madagascar. Additional specimens had been taken in Banc du Leven in the course of other French dredging campaigns carried out in 1969 and 1973. A number of specimens come from fishermen's nets from different areas in southern Mozambique, South Africa and reunion Island. Many of these are nowadays in the private collections of amateur shell collectors in Europe. The sequencing of a fragment of the cox1 gene for the live-collected kinoshitai -like individuals from the BIoMAGLo Expedition showed a significant genetic divergence from typical Afonsoconus kinoshitai specimens from the Pacific ocean, and also from Afonsoconus bruuni from New Caledonia. The molecular data, in conjunction with comparative analyses of shell and radula characters are consistent with the hypothesis that the members of the A. kinoshitai group originating from the SW Indian Ocean actually correspond to a distinct species. Here, we summarize the main morphological features that characterize the members of genus Afonsoconus, and introduce the new species from the SW Indian Ocean with the name Afonsoconus crosnieri sp. nov.

Material and methods
Most of the material studied here was previously deposited in institutional repositories. Descriptions and measurements are based on shells oriented in the traditional way: spire up with the aperture facing the viewer. The taxonomy used in the present work follows Tucker & Tenorio (2013), with the updates and modifications included in Puillandre et al. (2015). Specimens were collected by dredging in deep water during campaigns carried out by the MNHN (expeditions.mnhn.fr) in the Mozambique Channel and northwest Madagascar aboard the rV Antéa and Miriky, namely BIoMAGLo 2017 and MIrIKY respectively, at depth ranges of 80 to 1100 m. Some specimens were taken in Banc du Leven in the course of other French dredging campaigns carried out in 1969 and 1973. Specimens in private collections come from local fishermen in most cases. Preserved specimens of other cone snail species used in the phylogenetic analyses were collected by the MNHN expeditions EXBoDI and TErrASSES in New Caledonia, and SANTo 2006 in Vanuatu. Distribution maps were generated with GeoMapApp (http://www.geomapapp.org), using the general bathymetric map of the oceans as a default basemap.
We describe shell morphology using the terminology established in Röckel et al. (1995). We also used the procedure described in röckel et al. (1995) for counting the number of protoconch whorls. Taxonomy 472: 1-20 (2018)   6 For morphometric comparisons, adult shells selected among available specimens in the collections of the MNHN and other sources (private collections) were measured with a digital caliper, and the measurements rounded to 0.1 millimeter. All the measurements are in a spreadsheet, deposited as electronic supporting information (Appendix). For comparison of shell morphometry, we performed analysis of the covariance (ANCoVA) for different shell parameters, namely maximum diameter (MD), height of the maximum diameter (HMD) and spire height (SH), using species hypotheses as factor and shell length (S L ) as covariate. Additionally, compared the mean values of S L statistically using t-and u-tests. Statistical tests were carried out using STATGrAPHICS XVII-X64, after all the measurement sets passed the normality tests.

European Journal of
We used the terminology for radular morphology of Tucker & Tenorio (2009), and the abbreviations in Kohn et al. (1999) andraybaudi-Massilia (2002). The radular sac was dissected from the cone snail and soft parts were digested in concentrated aqueous potassium hydroxide for 24 hours. The resulting mixture was then placed in a Petri dish and examined with a binocular microscope. The radular teeth were removed with fine tweezers, rinsed with distilled water, then mounted on a slide using Aquatex (Merck) Mounting Medium, and examined under a compound microscope. Figure photos were obtained with a CCD camera attached to the microscope.
DNA was extracted using the Epmotion 5075 robot (Eppendorf), following the manufacturers' recommendations. A fragment of the cytochrome oxidase subunit I (cox1) was amplified using universal primers LCo1490/HCo2198 (Folmer et al. 1994). PCr reactions were performed in 25 µl, containing 3 ng of DNA, 1 × reaction buffer, 2.5 mM MgCl 2 , 0.26 mM dNTP, 0.3 mM of each primer, 5% DMSo, and 1.5 units of Qbiogene Q-Bio Taq. Amplification consisted of an initial denaturation step at 94°C for 4 min, followed by 35 cycles of denaturation at 94°C for 30 s, annealing at 50°C for cox1, followed by extension at 72°C for 1 min. The final extension was at 72°C for 5 min. PCR products were purified and sequenced by the Eurofins sequencing facility. Specimens are registered in the MNHN collections and sequences were deposited in BoLD and GenBank (Table 1).
Additional sequences from the Afonsoconus clade and from closely related species (Textilia), as well as one sequence of Fraterconus distans (Hwass in Brugière, 1792), used as outgroup, were downloaded from GenBank (Table 1). As no indels were detected, cox1 sequences were aligned manually. A phylogenetic tree was reconstructed using MrBayes 3.2 (Huelsenbeck & ronquist 2001), with two runs, each consisting of three Markov chains of 10 000 000 generations each, with a sampling frequency of one tree each 1000 generations. Each codon position of the cox1 gene was treated as an unlinked partition, each following a GTr model, with a gamma-distributed rate variation across sites approximated in four discrete categories and a proportion of invariable sites. Convergence of each analysis was evaluated using Tracer 1.7 (rambaut et al. 2018) to check that ESS values were all greater than 200. A consensus tree was then calculated after omitting the first 25% of trees as burn-in. Kimura 2-parameter (K2P) genetic distances were calculated using MEGA 6 (Tamura et al. 2013

Diagnosis
Shell (Fig. 1A-H). Elongated conical to cylindrical shell; spire low and conical in shape; posterior notch deep and cords present on whorl tops; columella twisted, but without anterior notch; shell and spire coloration variable; operculum small, ovate-shaped; periostracum thin and translucent, with multiple fine spiral rows of small tufts.
RadulaR tooth (Fig. 1I-J). Narrow and elongated, with a large to medium relative size; waist indistinct; anterior section equal or slightly longer than the posterior section; tooth serrated with a fairly long row of small serrations; terminating cusp small; barb and blade very short; blade barely twice as long as barb; base large; basal spur present; basal ligament present (not shown in Fig. 1I-J).

Geographic distribution
The species included in the genus occur in the Indo-Pacific region.

Geologic range
recent.

Remarks
Afonsoconus is here treated as a genus, following Tucker & Tenorio (2013) and Monnier et al. (2018), but Puillandre et al. (2014) ranked it as a subgenus within Conus.
There are currently two species included in genus Afonsoconus (WoRMS editorial board 2018). A number of taxon names associated with A. kinoshitai are considered synonyms (forms). These are tamikoanus Shikama, 1973, calliginosus Shikama, 1979and brontodes Shikama, 1979, and were already presented in the Introduction (vide supra). The name Conus (Chelyconus) wistaria Shikama, 1970 has occasionally been associated to A. kinoshitai especially among amateur shell collectors, but the name actually applies to a color form of Pionoconus fulmen (röckel et al. 1995;Filmer 2012;Tucker & Tenorio 2013).
The food habits of the species in Afonsoconus are not known, but the radular morphology ( Fig. 1I-J) suggests that they prey on worms. Based upon conotoxin analysis, it has been inferred that A. kinoshitai is a piscivorous species ( . However, this assumption is not supported either by direct observation of prey capture nor by the morphology of the respective radular teeth of these species, which are more consistent with a vermivorous feeding mode (Tucker & Tenorio 2013). Several conotoxins have been identified for A. kinoshitai, most notably the μ-conotoxin μ-KIIIA (Bulaj et al. 2005;Zhang et al. 2007;Khoo et al. 2009). This conotoxin blocks mammalian neuronal tetrodotoxin (TTX) resistant voltage-gated sodium channels (VGSCs) and is a potent analgesic (Bulaj et al. 2005;Zhang et al. 2007;McArthur et al. 2011).

Phylogenetic analyses
Afonsoconus is recovered as a monophyletic group with high support (Posterior Probability PP = 1) (Fig. 3). The Afonsoconus clade is sister to the Textilia clade (Puillandre et al. 2014), which contains fish-eating species characterised by their polished and shining subcylindrical to cylindrical shells (Fig. 2), and by their harpoon-shaped radular teeth (Fig. 2). Afonsoconus is clearly split in three subclades, each of them fully supported (PP = 1), and with high genetic distances between them (> 8%). Conversely, genetic distances within each subclade are all < 1%, except between the two samples of kinoshitai, with a genetic distance of 5.4%. The three subclades correspond to different geographic regions, one with specimens from the Philippines, another with specimens from New Caledonia, and a third one containing the specimens from the Mozambique Channel (BIoMAGLo expedition). The specimens from the Philippines and New Caledonia correspond respectively to the species A. kinoshitai and A. bruuni. According to the phylogenetic relationships and the genetic distances, the specimens from the Mozambique Channel deserve specific status, and this new species is hereby introduced. It is interesting to note that the observed p-distance between the two specimens of A. kinoshitai from GenBank (sequences FJ937341.1 and KJ550543.1) is consistent with a separation at the species level, as found for other species of cone snails (e.g., Duda et al. 2008;Puillandre et al. 2011). Both specimens come from the Philippines, and one of them (sequence KJ550543.1) appears labelled in GenBank as Conus kinoshitai tamikoae (= tamikoanus). Given the fact that the tamikoanus from Japan/China is a synonym (form) of A. kinoshitai, as recognised by its author in Shikama (1979), the results of the phylogeny actually suggest that there may be at least two different species of Afonsoconus in the Philippines. If we accept that the specimen associated with the sequence FJ937341.1 is A. kinoshitai, the other one would be a putative new species, morphologically similar to the form tamikoanus according to its label. The specimens from the Philippines labelled as tamikoanus are treated as a subspecies of A. bruuni in raybaudi-Massilia (2008), or as a full species in Monnier et al. (2018). It is likely that the tamikoanus-like specimen in GenBank is a representative of the taxon featured in raybaudi-Massilia (2008) and in Monnier et al. (2018). unfortunately, no voucher specimen or photo thereof is associated with the GenBank sequence KJ550543.1, so any further taxonomical claim on this matter would be merely speculative at this stage.    (Fig. 4A-B).
Shell. Moderately large. Maximum length: 83.8 mm. Shell profile narrowly conical, with convex sides adapically, and straight below. Spire of moderate height, of straight or very slightly convex outline. Multispiral protoconch with about three whorls, yellowish, glossy and translucent (Fig. 4D). First four teleoconch whorls weakly tuberculated (Fig. 4E), with tubercles becoming obsolete on fifth whorl, being absent in later whorls. occasionally, the tubercles may fuse together forming a ridge over the  (2018) 14 suture, producing a spire with a slightly stepped aspect. Sutural ramp flat or slightly concave, with five increasing to eight spiral cords. Shoulder subangulate to rounded.
teleoconch. Early teleoconch whorls white or yellowish. Late teleoconch whorls white with light brown irregular blotches and flecks. Ground colour white, pale yellow or pale violet. Last whorl overlaid with brown flammules or blotches, often fused forming spiral bands. There are two broad white, sparsely patterned spiral bands immediately above and below mid-body. Basal quarter and shoulder area also predominantly white and sparsely patterned. In addition, reddish-brown fine interrupted spiral lines and dots present in variable amounts, more evident in sparsely patterned areas. Columella white, callous and twisted. Aperture white or pale purplish, rather narrow adapically, widening abapically. Posterior notch rather deep. Periostracum yellow-brown, thin and translucent, with fine spiral rows of small tufts. Small and ovate operculum present.
RadulaR teeth (examined in holotype (Fig. 4H) and in paratype MNHN IM-2013-62925 (Fig. 4G)). 35 to 45 teeth in radular sac. radular tooth medium-sized: its total length relative to shell length S L /T L = 42-48. Waist indistinct. Anterior portion equal to or slightly longer than posterior section of tooth (T L /AP L = 1.8-2.0). With one barb opposed to a pointed, very short blade, which covers 12-14% of anterior portion of tooth. Blade only slightly larger than barb, which covers 8-9% of anterior portion. Tooth serrated, with one long row composed of 40-60 small denticles, ending in a small, pointed terminating cusp. Base large, with small but prominent sharp spur pointing upwards. Basal ligament present (not shown in Fig. 4G-H).

Distribution and habitat
Known from the Mozambique Channel including the Comoros, Iles Glorieuses, southern Mozambique, South Africa (Kwazulu-Natal coast) and NW Madagascar, between 180 and 314 m depth. Also present in réunion Island (Fig. 6).

Remarks
The specimens of A. crosnieri sp. nov. form a monophyletic group, with large genetic distances with respect to the two other species, A. kinoshitai and A. bruuni (Fig. 3). Despite the overall similarities in shell characters, A. crosnieri sp. nov. can be separated from its sister species by shell morphometry. Thus, A. crosnieri sp. nov. and A. kinoshitai do not exhibit significant differences in shell length, but they do differ in rD, PMD and rSH. Analysis of the covariance (ANCoVA) for the shell parameters MD, HMD and SH, using species hypotheses as factor and shell length (S L ) as covariate, yielded statistically significant results (Table 2). In the case of A. bruuni, there are statistically significant differences in mean shell length with A. crosnieri sp. nov. There are no differences in PMD or rSH, but these two species do differ in rD: ANCoVA for MD, using species hypotheses as factor and S L as covariate indicates statistically significant differences (Table 3). Thus, A. crosnieri sp. nov. is narrower-bodied and has a higher spire than the conoid-cylindrical A. kinoshitai, whereas A. bruuni has a shell which is usually smaller in length and broader at the shoulder, with a more conical appearance. A discriminant function analysis (DFA), performed with shell length (S L ) and the shell morphometric parameters MD, HMD and SH as variables and species hypotheses as factor, correctly classified 100% of the specimens in the sample (Fig. 7). According to the standardized coefficients, discriminant function 1 (DF1) represents mainly decrease in S L and increase in MD, whereas discriminant function 2 (DF2) represents mainly decrease in S L and increase in HMD, with a smaller contribution of increasing SH. These results indicate that A. crosnieri sp. nov. can be separated with a high degree of certainty from A. kinoshitai and A. bruuni based on significant differences in size and shell shape. The radular teeth of the three species show similar morphological characters. In spite of these similarities in the general aspect of the radular teeth, there are differences in their relative sizes. Thus, the radular teeth of A. kinoshitai and A. bruuni exhibit rather large relative sizes, with S L /T L in the range of 25 to 30. The radular teeth examined for A. crosnieri sp. nov. are clearly smaller, with S L /T L in the range of 42 to 48. This difference might indicate radular tooth adaptation to a specific type of prey, most likely a particular group of worms, and constitutes another useful trait for the separation of A. crosnieri sp. nov. from its sister species.

Discussion
The introduction of the new taxon A. crosnieri sp. nov. elevates to three the number of species in Afonsoconus. However, it seems likely that the number of species in this group is actually understimated, and that there is a hidden biodiversity yet to be studied and properly identified. We have remarked upon the fact that the taxon A. kinoshitai from the Philippines most likely comprises two species at least, but additional preserved material with reliable locality data is necessary to accomplish the task of identification. There is firm evidence for the expansion of the range of Afonsoconus species to the Table 2. Comparison between morphometric parameters for Afonsoconus crosnieri sp. nov. (N = 14) and A. kinoshitai (Kuroda, 1956)  bruuni. This represents a very significant expansion of the range of Afonsoconus species, which covers large areas in the Pacific. There are also some records from Fiji and Tonga (MNHN expeditions BorDAu1 and 2).
The question remains open about whether the identity of these specimens of Afonsoconus species from the Southern Pacific can be verified. In the same fashion as the specimens from the SW Indian Ocean have turned out to be members of a new species, it is possible that the specimens collected in the remote Austral Islands might actually correspond to undescribed taxa within the genus. These specimens are currently under study, and will probably require the acquisition of further preserved material for a solid specific identification.  (Kuroda, 1956), A. bruuni (Powell, 1958) and A. crosnieri sp. nov. The analysis was performed using shell length (S L ) and the shell morphometric parameters MD, HMD and SH as variables, and species hypotheses as factor. Both DF1 and DF2 are statistically significant at the 95% probability level (DF1: 69.9 % relative percent, Wilks lambda = 0.0449, χ 2 = 138.085, df = 8, p = 0.0000; DF2: 30.1 % relative percent, Wilks lambda = 0.2946, χ 2 = 54.38, df = 3, p = 0.0000).  (Powell, 1958)