Molecular phylogeny of the marine snail genus Haminoea (Gastropoda, Cephalaspidea): A framework to study marine diversity and speciation

Haminoea are herbivorous, coastal snails occurring in temperate and tropical waters of the Atlantic and Eastern Pacific oceans, with one species present in temperate South Africa (Indian Ocean). The genus is taxonomically difficult as several available nominal species were introduced based on shell descriptions alone, or described based on subtle differences in morpho‐anatomical features, without a phylogenetic molecular framework. Fifteen species are currently accepted as valid in recent scientific literature and field guides (eight Eastern Atlantic, one temperate Indian Ocean, four Western Atlantic and three Eastern Pacific). Here we generate the first complete phylogeny (Bayesian and Maximum Likelihood) of this genus based on multilocus molecular data (COI, 12S rRNA, 16S rRNA, 28S rRNA) using a taxon set accumulated over a period of 15 years, coupled with species delimitation analyses methods (ABGD, ASAP, bPTP) and morpho‐anatomical studies. The goal of this study is to provide insights into the taxonomy, phylogenetic relationships and geographical distributions of species while generating a framework for future systematic reviews of the genus, as well as to study speciation and historical biogeography. Our results rendered four possible hypotheses of species diversity: with 14, 15, 19 and 20 candidate species and point to the fact that several taxa presently regarded as valid might be conspecific (e.g. H. orteai–H. templadoi–H. exigua; and H. alfredensis–H. antillarum–H. orbignyana), while highlighting the existence of a complex of four or five species often identified as H. elegans. Pervasive nomenclatural problems in the genus, including with the type species H. hydatis, are highlighted and discussed.


| INTRODUCTION
The marine gastropods of the genus Haminoea Turton & Kingston in Carrington, 1830 have been long considered to have a worldwide distribution, inhabiting temperate and tropical shorelines (Burn & Thompson, 1998;Malaquias & Cervera, 2006;Rudman, 1971).Yet, Oskars et al. (2019) and Oskars and Malaquias (2019) found Haminoea to be a paraphyletic assemblage of five distinct evolutionary lineages; three mostly tropical and restricted to the Indo-West Pacific (Haloa Pilsbry, 1921, Lamprohaminoea Habe, 1952and Bakawan Oskars & Malaquias, 2020a), one confined to Australasian (Papawera Oskars & Malaquias, 2020b) and Haminoea proper, geographically restricted to the Atlantic Ocean (including the Mediterranean Sea), the Eastern Pacific Ocean and with a single lineage represented in temperate stretches of the Indian Ocean coastline of South Africa.
The small, globose, semi-translucent and thin shells of these five genera are similar in shape, colour and size, which coupled with a lack of broad comparative morphological studies and phylogenetic frameworks, led the majority of authors to accept Haminoea as the only valid genus across the world (e.g.Burn & Thompson, 1998;Cervera et al., 2004;Gosliner et al., 2008;Malaquias & Cervera, 2006;Thompson, 1981;Valdés et al., 2006).The modern concept of Haminoea was proposed by Oskars et al. (2019) and Oskars and Malaquias (2019) who identified several diagnostic features to separate it from their Indo-West Pacific closely related genera, namely the higher number of lateral radular teeth, the presence of a muscular penis and a Hancock organ with a perfoliate structure.
Species of Haminoea are herbivorous and live predominantly in estuaries and coastal lagoons, where they are often found on seagrass, algae or sandy-muddy bottoms, but can also occur on rocky shores in tidepools or shallow depths always among algal mats (Boulch-Bleas, 1983;Malaquias et al., 2002Malaquias et al., , 2004Malaquias et al., , 2009;;Rudman, 1971).
In total, 15 species of Haminoea are currently accepted as valid in current scientific literature and field guides.However, in a literature search we were able to identify 48 nominal species, most of them of uncertain taxonomic status because of short and ambiguous species descriptions based only on shells, which are similar in shape, colour and dimensions (e.g.Leach, 1852 for H. dilatata; A. Adams, 1850 for H. glabra; Baker & Hanna, 1927 for H. angelensis;Petuch, 1987 for H. taylorae).
Furthermore, even among the 'well-established' species, there are questions about the taxonomic status of several of them.For example, the definition of the type species of the genus-H.hydatis-is problematic.This species was described by Linnaeus (1758) based on shells (unclear if only one or several) from the Mediterranean Sea but later assumed by various authors to be conspecific with specimens occurring between the British Isles and the Adriatic Sea, and characterized by having a smooth shell, a bilobed prostate separated by a constricted region and a radula with the first lateral tooth denticulated (Pruvot-Fol, 1954;Tchang, 1931;Thompson, 1981;Thompson & Brown, 1976;Vayssière, 1885).Another case is the species name H. elegans introduced by Gray (1825) based on shells from the British Isles and the Mediterranean Sea, yet, the name is commonly attributed to one of the tropical western Atlantic species (e.g.Caballer et al., 2015;Malaquias, 2014;Marcus, 1976;Marcus & Marcus, 1967;Redfern, 2001;Valdés et al., 2006) and also to spiralled shells occurring in tropical West Africa (Gabon, Republic of the Congo, São Tomé and Príncipe, Angola; Bernard, 1984;Martínez & Ortea, 1997;Rolán & Ryall, 1999).Likewise, the name H. ovalis is commonly employed to designate animals with tiny orange or yellow dots on the body occurring in the Eastern Pacific, between Mexico and Peru (Behrens & Hermosillo, 2005;Hermosillo et al., 2006;Oskars & Malaquias, 2019;Valdés & Camacho-Garcia, 2004); nonetheless, H. ovalis was described by Pease (1868) from Tahiti and was recently reassigned to the genus Lamprohaminoea by Oskars and Malaquias (2020c), who confirmed the species to be widespread in the Indo-West Pacific and absent from the Eastern Pacific.In fact, in a previous study, Oskars & Malaquias (2019, as Haminoea sp.1 475) showed that dotted orange haminoeids from Peru were phylogenetically related to all other Atlantic and Eastern Pacific Haminoea species.
In the present study, we generate the first complete phylogeny of the genus Haminoea based on multilocus molecular characters using a taxon set accumulated over a period of 15 years, which we believe to likely cover the entire diversity of the genus and include a comprehensive geographical coverage of the distribution of species.The main goals of this paper are to define the number of species in Haminoea and provide insights on their taxonomy, phylogenetic relationships and geographical distributions while establishing a framework for future detailed systematic reviews and studies on speciation and historical biogeography of this genus.1).

| Sampling of taxa
Outgroup taxa consisted of species from two additional genera, namely Haloa (represented by four species) and Lamprohaminoea (represented by one species).The trees were rooted with Smaragdinella, a genus closely related to Haminoea (Oskars et al., 2019).In total, this study includes 206 specimens (94 EA Haminoea, 69 WA Haminoea, 18 EP Haminoea, 5 tWIO Haminoea and 16 outgroup taxa) and a total of 608 sequences, of which 426 were newly generated for this study (Table 1).

| DNA extraction, amplification and sequencing
DNA was extracted from tissue obtained from the foot or parapodial lobes using the Qiagen DNeasy Blood and Tissue Kit (catalogue no.69504) following the protocol recommended by the manufacturer.For small specimens with shell height between 2 and 3 mm, the whole specimen was digested and hard parts such as the shell, radula and gizzard plates were collected for morphological examination.
For samples that did not amplify with Qiagen Taq, additional 25 μL reactions were set with TaKaRa Ex Taq Polymerase HS (250 U) (Cat.number: RR006A), following the protocol described by Oskars et al. (2015).For some    samples, the amount of MgCl 2 and DNA was increased, and the volume of water adjusted accordingly in the PCR cocktail.In addition, 10x dilutions of DNA extractions were attempted for samples that did not yield results with all previous approaches.The quality and quantity of PCR products were assessed by gel electrophoresis following standard methods (see Eilertsen & Malaquias, 2013).Successful PCR products were purified according to the EXO-SAP method described by Eilertsen and Malaquias (2013).Sequence reactions were run on an ABI 3730XL DNA Analyser (Applied Biosystems).

| Phylogenetic analyses
Geneious (v.R11, Kearse et al., 2012) was used to inspect, edit, and assemble the chromatograms of the forward and reverse DNA strands.All sequences were blasted in GenBank to check for contamination.Single gene sequences were aligned with Muscle (Edgar, 2004) implemented in Geneious.Alignments were trimmed to a position where at least 50% of the sequences had nucleotides and missing positions at the ends were coded as missing data (?).All sequences were deposited in GenBank (Table 1).
Bayesian inference analyses (BI) using MrBayes (Huelsenbeck & Ronquist, 2001) were run through the portal CIPRES Science gateway V.3.3 (https://www.phylo.org)on the initial single gene datasets (Appendix S2-S8) and all-genes concatenated dataset (Figure 1, Appendices S9 and S10; 2492 bp).For the ribosomal genes, the datasets selected for concatenation were those that yielded the best-resolved trees with higher node support.All samples with sequences available for two or more genes were used in the concatenation analysis.In addition, samples with a single gene from unique geographical localities or with a unique phylogenetic position in the single gene trees were also included in the concatenated dataset.The analyses used three parallel runs of 5 million generations for the single gene analyses and 15 million generations for the concatenated dataset, with sampling every 100 generations.The concatenated dataset was partitioned by gene and each partition was run under the best-fit model of evolution.Convergence of runs was inspected in Tracer v1.7 (Rambaut et al., 2018) with a burn-in set to 25% by comparing the likelihood of trees drawn by the independent runs.Posterior probabilities (PP) higher than 0.95 were considered statistically significant (Alfaro et al., 2003;Huelsenbeck et al., 2001).A Maximum Likelihood analysis (ML) of the concatenated dataset was run with the RAxML (v.8.2; Stamatakis, 2014) plug-in implemented in Geneious.The analysis was partitioned by gene and run under the 'rapid bootstrapping and search for best scoring ML tree' algorithm, using a random starting tree and the model GTR + G + I with 1000 bootstrap (BS) replicates.Bootstrap values higher than 75% were considered significantly supported (Felsenstein, 1985).Consensus phylograms were converted to graphics in FigTree v1.3.1 (Rambaut & Drummond, 2009).

| Molecular species delimitation analyses
We used the DNA sequences of the COI gene to evaluate candidate species by using the Automatic Barcode Gap Discovery delimitation method (ABGD) (Puillandre et al., 2012) and the Assembling Species by Automatic Partitioning (ASAP) (Puillandre et al., 2021) under default settings and three different models of molecular evolution (Jukes-Cantor (JC69), Kimura TS/TV = 2.0 (K80), Simple Distance).In addition, we used the bPTP method (Poison tree processes) on the same COI dataset.This method is intended to delimiting species that are consistent with the phylogenetic species concept and model speciation in terms of the number of substitutions (Zhang et al., 2013).

| Haplotype network analyses
Haplotype networks were generated based on the COI DNA sequences for the groups recognized by the phylogenetic analyses as putative candidate species, but which were T A B L E 2 COI uncorrected genetic p-distances between and within groups of Haminoea.Figures are depicted as percentages (%).Distances within groups showed in bold font.

| Morpho-anatomical analysis
The shell and anatomical features of selected specimens were studied to aid in interpreting the taxonomic status of some problematic lineages or to address complex taxonomic cases resulting from the molecular phylogenetic analyses (see Section 4 for details).
The shells were gently separated from the animals with the aid of forceps.The male reproductive system, gizzard and buccal bulb were dissected by a dorsal incision through the cephalic shield.Shells were photographed with a digital DSLR camera equipped with a macro lens and strobe lights.Shell height (H) was measured with a digital Vernier calliper.The reproductive system was drawn using a stereo microscope fitted with a drawing tube and the penial sheath was removed to expose the penial papilla.The gizzard and buccal bulb were placed in a solution containing 180 μL buffer ATL with 20 μL of proteinase K solution (both from the Qiagen DNeasy® Blood and Tissue Kit) and incubated at 56°C at night in order to clean the gizzard plates, jaws and radulae.The penial papillae, gizzard plates, jaws and radulae were mounted on metallic stubs using carbon sticky tabs and then sputter-coated with gold-palladium for scanning electron microscopy (SEM).Prior to sputter coating and SEM, the gizzard plates and penial papillae were dehydrated with Hexamethyldisilazane (HMDS) by covering each sample inside small square watch glasses and left to dry between 30 min and 1 h inside a fume hood.All samples were scanned and imaged with a Fei Quanta 450 scanning electron microscope.
The COI gene analyses rendered 19 groups putatively compatible with candidate species of Haminoea (Appendix S2).All groups but one (H.orbignyana; PP = 0.88), received maximum or nearly maximum support.The 12S rendered 18 groups, but clade support was comparatively lower and often below statistical thresholds; the group missing is a singleton only represented in the COI dataset (Haminoea sp.256 Croatia) (Appendix S2).The 16S tree rendered 15 groups and clustered together with no support (PP = 0.72) four groups recognized in the COI analyses (groups 13 + 14 + 15 + 16).None of these four groups formed supported sub-clades.Only group 13 (H.alfredensis) was nearly supported (PP = 0.91) but one representative branched apart (Appendix S5).The 28S gene tree was the less resolved with several of the groups recognized by the mitochondrial gene analyses, rendered non-monophyletic (Appendix S8).On the contrary, the concatenated analyses rendered the same 19 groups as the COI analysis, with two groups represented by singletons (Haminoea sp.256 [Croatia; group 2] and Haminoea sp.543 [Spanish Mediterranean; group 3]), one group (group 14) with moderate support (PP = 0.95) and all remaining 16 groups with maximum support (PP = 1) (Figure 1, Appendices S9 and S10).
The bPTP analysis suggested 20 candidate species, corresponding to the same 19 groups rendered by the COI and concatenated analyses, yet group 17 (H.'elegans 3') represented by two samples was inferred to correspond to two putative species (sample 266 from Abaco, Bahamas and sample 568 from Eleuthera, Bahamas; Table 1,  Appendices S2 and S13).

| Haplotype network analyses
The haplotype network of groups 4 + 5 + 6 formed by samples from the eastern and central Mediterranean Sea was well structured with 14 haplotypes and three recognizable haplogroups separated by 28 substitutions (between groups 4 and 6) and 21 substitutions (between groups 5 and 6).Only one case of shared haplotypes was detected in group 6 between samples from Spain and France (Figure 2).The haplotype network of groups 13 + 14 + 15 + 16 with samples from Europe, West Africa, temperate South Africa and Caribbean Sea, includes 19 haplotypes and four recognizable haplogroups connected through hypothetical haplotypes (black circles; Figure 3).The highest number of substitutions among haplogroups was 23 between group 13 (H.alfredensis) and 15 (H.'antillarum 2').The haplotype network of groups 18 + 19 formed by samples from Brazil, Caribbean Sea and Cape Verde Islands includes 21 distinct haplotypes and two recognizable haplogroups separated by 36 substitutions (Figure 4).
F I G U R E 2 COI haplotype network produced with the TCS method in PopART for groups 4, 5 and 6.Colours of circles refer to the geographic origin of each haplotype.The relative size of circles is proportional to the number of sequences of that same haplotype.Black circles refer to hypothetical haplotypes and black bars to mutations.

| The Haminoea orbignyana- alfredensis-antillarum complex
Probably, the most surprising result of this study is the hypothetical conspecificity of the Eastern Atlantic and Mediterranean Sea species H. orbignyana, with H. alfredensis from the temperate Indian Ocean shores of South Africa, and the Western Atlantic lineages of H. antillarum (groups 15 and 16).Despite the fact that our phylogenetic analyses (Figure 1) rendered these four lineages F I G U R E 3 COI haplotype network produced with the TCS method in PopART for groups 13, 14, 15 and 16.Colours of circles refer to the geographic origin of each haplotype.The relative size of circles is proportional to the number of sequences of that same haplotype.Black circles refer to hypothetical haplotypes and black bars to mutations.monophyletic, the genetic distances among them are comparatively lower varying between a minimum of 2.4% (H.alfredensis and H.orbignyana) and a maximum of 4.5% (H.antillarum group 16 and H. orbignyana), with the genetic distance between the two lineages of H. antillarum estimated at 3.9%.These four lineages are only supported by the COI gene, whereas the 12S and 28S genes rendered support for the clade containing the specimens of H. antillarum from Yucatan (group 15), and the latter gene also provided support for the clade with specimens of H. alfredensis (Figure 1, Appendices S2, S3, S8, S10, S11 and S11).None of the four lineages was supported by the 16S gene analysis, which nevertheless clustered all representatives together but with low node support (PP = 0.72).
The species H. orbignyana, H. antillarum and H. alfredensis are well established in the literature, yet they were never studied in a comparative framework.A closer look at the literature together with our own preliminary data on the morphology of specimens reveals that all these nominal species share a pear-shaped smooth shell and a body colouration characterized by dense black pigmentation along the edges of the cephalic shield and parapodial lobes.In contrast, H. antillarum has mildly denticulated inner lateral radular teeth, whereas in H. alfredensis and H. orbignyana these teeth are smooth.Likewise, whereas in the latter two species, the proximal lobe of the prostate is wider, conferring the prostate an acorn-like shape, in H. antillarum seems to be the opposite with the distal lobe wider compared to the proximal one (Gosliner, 1987;Macnae, 1962;Malaquias & Cervera, 2006;Marcus & Marcus, 1967;Thompson, 1981;Valdés et al., 2006; personal observations), but this requires further anatomical investigations in order to be confirmed.
Even if our molecular results based on the species delimitation analyses and genetic distances suggest the occurrence of a single ubiquitous species with amphi-Atlantic distribution encompassing the Iberian Peninsula, the Mediterranean Sea, West Africa including the Canary Islands, the temperate shores of South Africa in the Indian Ocean and the western Atlantic along the Yucatan Peninsula, Florida and Bermuda, this warrants caution and further corroboration by conchological and morphoanatomical data.As highlighted above, H. antillarum seems to be characterized by relevant anatomical differences from the digestive and reproductive systems, and even if genetic distances are comparatively low, this could be due to different evolutionary rates between species of Haminoea.
On the contrary, and even in the absence of sound data on the duration of the pelagic larval stage of Haminoea (Schaefer, 1996), the confirmed occurrence of specimens attributed to H. elegans (group 18) on both sides of the Atlantic (Figure 1; only 0.3% different in the COI gene; Table 2) supports a high dispersal capability, at least in some species of the genus (Martínez & Ortea, 1997; current study as H. 'elegans 1' [group 18]; Figure 1).Thus, we cannot discard that representatives of the orbignyanaalfredensis-antillarum complex may have larvae with high dispersal capability favouring gene flow between distant populations.However, we must admit that the genetic distance between the two putative lineages of H. antillarum from nearby locations, namely the Yucatan side of the Gulf of Mexico (group 15) and the Florida Keys/Florida F I G U R E 4 COI haplotype network produced with the TCS method in PopART for groups 18 and 19.Colours of circles refer to the geographic origin of each haplotype.The relative size of circles is proportional to the number of sequences of that same haplotype.Black circles refer to hypothetical haplotypes and black bars to mutations.
Peninsula-Bermuda (group 16), estimated at 3.9% and 14 substitutions between these two haplogroups (Figure 3) challenges this view.Even if the prevalent ocean current system in the area suggests putative connectivity between Yucatan and the Florida Peninsula through the Loop Current (Gyory et al., 2011), faunal breaks between tropical Florida and the more temperate/sub-tropical Gulf of Mexico have been documented for several groups of molluscs and fish (Briggs, 1974;Lee & Ó Foighil, 2004;Mikkelsen & Bieler, 2000;Reeb & Avise, 1990), likely reflecting seasonal changes in the current systems and water temperatures oscillations.These factors may hinder gene flow between the Yucatan and the Florida-Bermuda populations, creating periods of temporary isolation that could explain the observed genetic discontinuity.
Another interesting aspect is the sister relationship between the lineages H. orbignyana (Eastern Atlantic) and H. alfredensis (temperate South Africa).The cold water of the Benguela current established at the end of the Miocene (Siesser, 1980) is regarded as a strong barrier for temperate and tropical marine coastal species isolating the faunas of the Atlantic and Indian Oceans, while at the same time, paleontological and morphological evidence suggests that this barrier was sporadically bridged by several coastal invertebrate organisms (Briggs, 1995;Vermeij & Rosenberg, 1993).Few molecular evidence for dispersal from the Atlantic into the Indian Ocean is still available.This is the case for reef fish (Floeter et al., 2008;Rocha et al., 2005) and sea slugs (Churchill et al., 2014;Golestani et al., 2019), which seem to have taken advantage of the disruption of the Benguela and Agulhas currents during warmer interglacial periods of the Pleistocene.Because Haminoea is a genus of Atlantic and Eastern Pacific affinity, the sister relationship between H. orbignyana and H. alfredensis is more parsimoniously explained as being the result of dispersal of larvae or H. orbignyana into the Indian Ocean during these warmer periods, with the establishment of viable populations followed by isolation after the reestablishment of the current system.

| The Haminoea elegans complex
Haminoea elegans is characterized by having whitish translucent spiralled shells and it has been regarded as widely distributed in the Western Atlantic throughout the Gulf of Mexico and the Caribbean Sea southwards to Brazil (Valdés et al., 2006) with records in West Africa between the Gulf of Guinea and Angola (Bernard, 1984;Martínez & Ortea, 1997;Rolán & Ryall, 1999).However, the attribution of the name 'elegans' to this tropical amphi-Atlantic species stems certainly from a misidentification perpetuated in the literature over time.The name H. elegans was introduced by Gray (1825) based on spiralled shells from the British Isles and the Mediterranean Sea, and the name is most certainly a junior synonym of Haminoea navicula, the only European species with a deeply spiralled shell (Malaquias & Cervera, 2006).
Our results showed the existence of cryptic diversity in this 'species' with specimens provisionally ascribed by us to H. elegans splitting in four (or five) clades of possible species status (groups 7, 17-19; Figure 1, Appendices S2, S9, S10 and S13).Representatives of groups 17, 18 and 19 clustered together with maximum support, whereas group 7 branched off elsewhere in the tree (Figure 1, Appendix S2, S9 and S10).
If, in contrast, our results unequivocally support group 7 as a good species, they are not conclusive about the eventual status of group 17, with one of the species delimitation analysis (bPTP), suggesting the possible occurrence of two lineages in this group.However, none of the single gene and combined analyses retrieved reciprocal monophyly between sub-clades within group 17.When present, the sub-clades are not statistically supported (Appendices S3-S10).
The results are also not entirely conclusive about the conspecificity of groups 18 and 19 (see Section 3 -theme 3.2).Groups 18 and 19 are the only two in the complex with a genetic distance between themselves below 10%, but still moderately high (= 7.5%).Moreover, in the haplotype network analysis, they were separated by 36 substitutions, the largest number of substitutions between putative conspecific groups among all our haplotype network analyses (Figure 4).
There are several names available in the literature that could be regarded as previous attempts to describe the shells variability in the H. elegans complex (e.g.H. guildingii (Swainson, 1840) [shells globose with visible spiral striae], H. petitii (d 'Orbigny, 1841) [shells lacking or with inconspicuous spiral striae], H. succinea (Conrad, 1846) [shells cylindrical with tightly arranged spiral striae], H. taylorae (Petuch, 1987) [shells globose with numerous fine spiral striae]).These names have been in part considered synonyms of H. elegans (MolluscaBase, 2022;Valdés et al., 2006) or hardly used in scientific literature, but our results show the need to carefully re-evaluate the status of these names since some of them may apply to lineages revealed by our analyses.
The only study that provided a comparative analysis of the various types of shells of 'H.elegans' in the Western Atlantic was by Redfern (2013: 266-268).This author recognized five different types of whitish shells that could be associated with H. elegans; four with spiral striae and one apparently smooth.One of these forms was named by Redfern (2013)  Here we provide for the first time a phylogenetic framework to properly explore the diversity of the Haminoea 'elegans' species-complex.Yet confirming whether our four (or five) candidate species correspond to the shell types identified by Redfern (2013) and are compatible with the names available in the literature, requires additional taxonomic work based on detailed analyses of conchological and morpho-anatomical characters.

| The Haminoea hydatis- fusari complex
Another difficult case consists of groups 4, 5 and 6 in our phylogeny (Figure 1), which were rendered a single candidate species by the ABGD and ASAP molecular species delimitation methods.These three clades received maximum or nearly maximum support in both BI and ML analyses, but interestingly if considered together as a single clade the support lowers to 0.84 (PP) and 59% (BS), although this seems to be mostly influenced by the 16S gene data (Appendix S5).On the contrary, the haplotype network analysis (Figure 2) recovered the three groups as distinct, separated by 28 substitutions (between groups 4 and 6) and 21 substitutions (between groups 5 and 6) and showed a lack of shared haplotypes.Genetic distances were moderately high, ranging between 5.1% (between groups 5 and 6), 6.2% (between groups 4 and 6) and 7.4% (between groups 4 and 5).
This larger clade, including the three groups (4 + 5 + 6), contains only specimens from the Mediterranean Sea and one from the Eastern Atlantic island of Selvagem Grande (Madeira Archipelago).They are all characterized by a distinct anatomical feature among Haminoea, namely a prostate with a constricted zone between the proximal and distal lobes (see Thompson, 1988).This feature has been described for the type species of the genus H. hydatis (Talavera et al., 1987;Tchang, 1931;Thompson, 1976Thompson, , 1981Thompson, , 1988) and H. fusari (Álvarez et al., 1993).According to the literature these two species are basically distinguished by the presence of denticulated inner lateral teeth in H. hydatis (Talavera et al., 1987;Tchang, 1931;Vayssière, 1885) while they are smooth in H. fusari (Álvarez et al., 1993).
Haminoea hydatis is the type species of the genus described by Linnaeus (1758) based on shells from the Mediterranean Sea.The type specimen illustrated in the webpage of the Linnean Collections, London (https:// linne an-online.org/16897/#?s=0&cv=0&z=0.0365%2C-0.0109%2C1.232%2C1.503), is a shell about 9 mm in height with a smooth surface.Vayssière (1885) studied specimens from the Gulf of Marseille on the Mediterranean French coast with smooth shells and a radula with inner lateral teeth denticulated, which he identified as H. hydatis.Later, Tchang (1931) described the male reproductive system of specimens from the same region as having a prostate with the proximal and distal lobes separated by a narrow tubular region, and Talavera et al. (1987) mentioned a smooth, cylindrical pointed penis.Thus, progressively it became established in the scientific literature the idea that H. hydays (originally only known from shells) was characterized by having smooth shells, radulae with denticulated inner lateral teeth and a prostate with a narrow region separating the two lobes.This view was reinforced by the fact that up until the end of the first half of the 20th century, the European fauna of Haminoea was basically restricted to two accepted species; either H. hydatis with its small smooth shells or H. navicula with larger and deeply spiralled shells.
However, several species with smooth shells accepted as valid (see Introduction) were described during the second half of the last century, one of them (H.fusari) also with a prostate with two lobes separated by a narrow region, but with smooth radular inner lateral teeth (Álvarez et al., 1993).But the study of the holotype of H. fusari (MNCN 15.05/5356) revealed in fact the presence of mostly smooth inner lateral teeth, but interestingly some of them had the lower half denticulated.Intraspecific radular variability was described by Malaquias and Cervera (2006) for H. navicula and might occur also in specimens identified as H. fusari.This would basically make the two species anatomically undistinguishable and thus likely conspecific, rendering the name H. fusari a junior synonym of H. hydatis.In addition, the colour patterns of specimens in groups 4 and 5 are alike (data not available for group 3), with large unpigmented peri-ocular areas, dark upper sides of the parapodial lobes and fine bright-white dots along the edge of the cephalic shield, which further supports their conspecificity (see Figure 1).

| The eastern Pacific species
In the Eastern Pacific coastlines of North and Central America, there are three species of Haminoea commonly recognized as valid between Alaska and Panama, namely H. ovalis, H. virescens and H. vesicula (Behrens & Hermosillo, 2005;Hermosillo et al., 2006;Valdés & Camacho-Garcia, 2004).For this region, our analyses recognized lineages compatible with these three species.The species H. virescens (group 11) with pear-shaped shells and monolobated prostates (Gibson & Chia, 1989;Valdés, 2019;personal observations) split off as sister to the eastern Atlantic/Indian Ocean complex H. alfredensis-H.orbignyana-H.antillarum, although with no support (PP = 0.83, BS = 59%; Figure 1).A similar pattern was found for the species H. vesicula (group 8) characterized by a globose-quadrate shell and bilobed prostate (Gibson & Chia, 1989;Valdés, 2019;personal observations), which was rendered sister to the eastern European H. orteai, but again with no support (PP = 0.51, BS = 45%; Figure 1).This phylogenetic pattern with putative speciation across these two oceanic realms could be explained by processes related to the uplift of the Isthmus of Panama, which separated the Atlantic from the eastern Pacific around 3 Mya (Coates & Obando, 1996), however, the low support values hamper any sound explanation.
Finally, we retrieved a lineage morphologically compatible with what has been named in recent literature as H. ovalis (Behrens & Hermosillo, 2005;Hermosillo et al., 2006;Valdés & Camacho-Garcia, 2004).However, as explained in the Introduction this name applies to an Indo-West Pacific species in the genus Lamprohaminoea (Oskars & Malaquias, 2020c;Pease, 1868).This Eastern Pacific species has globose smooth shells and a body doted by abundant tiny orange dots.We could not find any available name fitting the features known for this lineage, which thus may represent an undescribed species.& Cervera, 2006).The species H. orteai was the first described European species characterized by a penis with an apical crest with 10 lamellae and the peculiarity of lacking an unpigmented periocular area in the cephalic shield (Talavera et al., 1987).However, our observations revealed that the description of this latter feature is not entirely accurate; in fact, like all other species of the genus, H. orteai has an unpigmented periocular area yet rounded and of a much smaller diameter.This can even be seen in the original description of the species (Talavera et al., 1987: 66, figure 15).Later, two additional species also with penises with apical crests were described for European waters, namely H. templadoi (García et al., 1991) and H. exigua (Schaefer, 1992).These two species are morphoanatomically very similar to H. orteai when it comes to the radula, the male reproductive system, the shell and the reduced diameter of the periocular area.Haminoea templadoi was described as having a shell with transverse folds interconnecting longitudinal growth lines and a radula with the first two inner lateral teeth denticulated (García et al., 1991: 396, figures 2, 3), but the study of the holotype (MNCN 15.05/854) showed a radula with only the inner lateral teeth denticulated.Moreover, according to our interpretation, the distinct shell structure likely relates to the comparatively larger size of the shell (H = 21.6 mm).A thorough systematic review of species combining detailed morphological work and the current phylogenetic framework is necessary to address the putative conspecificity of the species H. orteai, H. templadoi and H. exigua.

| Notes on other Atlantic species
Three additional and unidentified species were rendered by our analyses; Haminoea sp.[group 12] from Brazil, and two species represented by single individuals, namely Haminoea sp.[group 2] from Croatia and Haminoea sp.
[group 3] from Roses, Girona, Spain (Mediterranean Sea).The specimens from Brazil externally resemble H. 'ovalis' from the eastern Pacific and are interestingly part of a fully supported clade which includes the later species (Figure 1).The remaining two Mediterranean species might represent undescribed taxa, but this requires additional work to be confirmed.The colourations of these species are quite unique among European species with more or less uniform orangish and brownish background colour patterns (Figure 1).

| Concluding remarks
The genus Haminoea is a difficult taxonomic group with many available names introduced based on shell descriptions alone or described based on their morpho-anatomical features outside a phylogenetic molecular framework, several of them grounded on subtle differences.
The current literature consensus accepts eight species as valid in the Eastern Atlantic (including the amphi-Atlantic H. 'elegans'), four in the Western Atlantic (including H. 'elegans'), three in the Eastern Pacific and one in temperate South Africa.As explained in the Discussion, we open the possibility that several of the Eastern Atlantic taxa, presently regarded as valid species, might be conspecific (e.g. H. orteai, H. templadoi, H. exigua).A re-evaluation of the literature and type material, preliminary anatomical work, combined with our molecular phylogenetic framework seem to indicate that several of the subtle differences used in the past to introduce new species might not be sound enough.On the contrary, there were several cases where our results were not sufficient to reach definitive conclusions about the taxonomic status of certain species (e.g.H. alfredensis, H. antillarum, H. elegans, H. orbignyana) and further morphological work is necessary to understand their diversity and draw a robust taxonomic hypothesis.A preliminary combination of conchological, morphological and phylogenetic data demonstrated that putative cases of cryptic diversity may, in fact, reflect previously detected differences in shell characters that led to the description of species currently considered invalid (e.g. the cryptic species complex H. 'elegans').
There are several paradigmatic examples of pervasive nomenclature confusion that need to be evaluated, such as the status of the type species H. hydatis, H. elegans-a name introduced for a European species but largely in use for western Atlantic animals, and H. ovalis-an IWP species in the genus Lamprohaminoea, but a name commonly used to also refer to an Eastern Pacific lineage.
With this work, progress was made to underhand the diversity of Haminoea snails and the relationships between species, and for the first time a phylogeny of the genus is presented.This new framework combined with a detailed study of shells, morpho-anatomy of wet specimens, revision of original descriptions and type material, can help solve the many issues remaining with the taxonomy of Haminoea snails in the Atlantic and Eastern Pacific Oceans.
3.3.Empty positions at both ends of the alignments were treated as missing data, yielding final alignments of 677 bp.The programme DnaSP v.5(Librado & Rozas, 2009) was used to identify the number and sequences of the different haplotypes.Notepad++ was additionally used to generate trait files with geographic area codes based on a binary coding where 0 stands for sample absent and 1 for sample present and to edit the file into nexus format (nex).Alignments and trait files were finally run in PopArt v. 1.7 (Population Analysis with Reticulate Trees;Leigh & Bryant, 2015) to create a standard tight compact spanning (TCS;Clement et al., 2002) network analysis to visualize the relationships and distances between the individual haplotypes from different groups and geographical areas.The TCS haplotype networks were edited in PopArt for more satisfying visualization.
Haminoea elegans proper and the other four Haminoea A, B, C and D. According to the author, Haminoea elegans and Haminoea sp.A are characterized by globose-quadrate opaque shells with wavy-spiral striae and a partially concealed involute spire; Haminoea sp.B by a more oval translucent shell, with numerous tightly arranged spiral striae and a spire concealed by a callus; Haminoea sp.C by a globose opaque smooth shell and a spire concealed by a callus; Haminoea sp.D by cylindrical translucent shells, with lightly impressed spiral striae and a spire concealed by a callus.

Taxon DNA extraction code Group No Locality Voucher No COI 12S rRNA 16S rRNA 28S rRNA
List of specimens of Haminoea and outgroups used for phylogenetic analyses, with sampling localities, voucher numbers and GenBank accession numbers (numbers with prefix 'OR' are novel sequences generated for this study).
T A B L E 1 Cartoon based on the Bayesian phylogeny of Haminoea species depicted in Appendix S9 and resulting from the combined analysis of the mitochondrial COI, 12S rRNA and 16S rRNA and nuclear 28S rRNA gene markers.Figures above branches are Bayesian posterior probabilities and those below branches are bootstrap values derived by maximum likelihood.The trees were rooted with Smaragdinella sp. and representatives of the genera Haloa and Lamprohaminoea were included as outgroups.Both rooting and outgroups were removed for clarity (see Appendices S9 and S10 for complete trees).Images groups 1, 7, 14, 19 by Manuel Malaquias; images groups 8, 10, 11 by Ángel Valdés; images groups 2, 4, 6, 9 courtesy of Jakov Prkic; image group 3 courtesy of Marina Poddubetskaia; image group 12 courtesy of Joana Bahia; image group 13 courtesy of George Branch; image group 15 courtesy of Jazmin Ortigosa; images groups 16, 17 courtesy of Colin Redfern; image group 18 courtesy of Marlo Krisberg.Image not available for group 15.