Reassessing North Eastern Atlantic‐Mediterranean species of Trapania (Mollusca, Nudibranchia)

Trapania is the second largest genus belonging to the family Goniodorididae, of which most of the species are reported from Indo‐Pacific waters. To date, there are nine species of Trapania distributed along the temperate coasts of the East Atlantic Ocean and Mediterranean Sea: Trapania fusca, Trapania graeffei, Trapania hispalensis, Trapania lineata, Trapania maculata, Trapania orteai, Trapania pallida, Trapania sanctipetrensis and Trapania tartanella. However, the validity of some of these species has been problematic due to uncertain taxonomic characteristics used for the differentiation of the species. The genus Trapania has a very uniformly external morphology and very similar internal anatomy. As a consequence, the features most commonly used to differentiate species have been the colour pattern of the body and the morphology of the radula. In the present study, we perform a morphological and molecular revision of the East Atlantic‐Mediterranean species of the genus Trapania. Morphological analyses include dissections and scanning electron microscope photographs of radulae, labial cuticles and penises. Molecular work includes phylogenetic, species delimitation and haplotype network analyses. Our results bring doubt on the taxonomic characteristics used so far, suggesting that the richness of the North Eastern Atlantic‐Mediterranean species has been overestimated. Trapania hispalensis, T. lineata and T. pallida are shown to belong to the same taxa, with Trapania lineata as senior synonym.

by phylogenetic analyses based on morphological data (Gosliner & Fahey, 2008). Trapania is characterized by having a reduced mantle margin, with a single pair of curved extra-rhinophoral and extra-branchial appendages. The rhinophores are lamellate, and there are three tripinnate gill branches forming a semicircle around the anus. The radular formula is N × 1.0.1, and the jaws have elements. The penis is armed (Gosliner & Fahey, 2008;Kress, 1968;Rudman, 1987). The species of Trapania show a uniform body plan, with external morphology and internal features, such as the reproductive system, being very similar between species. Therefore, the main characteristics used to differentiate species have been the radula and the colour of the body (Cervera et al., 2000;Edmunds, 2009;Gosliner & Fahey, 2008;Moro & Ortea, 2015;Rudman, 1987).
However, in some cases, the intraspecific variability and interspecific similarities of these characteristics have also been questioned. Some species have very similar radulae, such as Trapania graeffei (Bergh, 1880) and Trapania tartanella (Kress, 1968) or Trapania squama Gosliner & Fahey, 2008, Trapania toddi Rudman, 1987, Trapania melaina Gosliner & Fahey, 2008 and Trapania euryeia Gosliner & Fahey, 2008(Gosliner & Fahey, 2008. In contrast, intraspecific variability has been reported between different specimens of the same species, or even between different teeth within the same radula, as example in the species Trapania maringa Er. Marcus, 1957 or Trapania maculata Haefelfinger, 1960 (Cervera & García-Gómez, 1989a;Haefelfinger, 1960;Marcus, 1957;Templado et al., 1988). These variations could question the validity of the shape of the radula as a taxonomic characteristic to difference species, leaving the colour pattern of the species as the main useful feature. However, the variability in the colour of European species has also been questioned by some authors, waiting for future studies to clarify the validity or synonymy of some species (Edmunds, 2009;Rudman, 1987;Templado et al., 1988).
In order to clarify the biodiversity of the North Eastern Atlantic-Mediterranean species of Trapania, and reviewing the characteristics used to differentiate them, specimens of eight of the nine species that inhabit the area (according to their previous identifications) were studied. Morphological analyses were performed by dissections. We included scanning electron microscope photographs of the radulae, labial cuticles and penial spines. Moreover, partial sequences of mitochondrial and nuclear genes were obtained, and a phylogenetic species delimitation and haplotype network analyses were carried out.
Morocco and Italy (Table S1). Also, the type material of some species was loaned by different Museums (Table S1). The newly collected specimens matched with eight morphospecies according to their initial identifications, including Trapania cf. fusca, Trapania hispalensis, Trapania lineata, Trapania maculata, Trapania orteai, Trapania pallida, Trapania sanctipetrensis and Trapania tartanella. Most of the specimens were preserved in 96% ethanol, and some of them were preserved in formalin. Specimens examined were deposited at the Museo Nacional de Ciencias Naturales, Madrid (Spain). Holotypes of T. lineata, T. maculata and T. pallida, preserved in Bouin's fixative, were loaned by the Natural History Museum Basel (Switzerland). The neotype of T. tartanella, preserved in 4% formalin, was loaned by the Museo Nacional de Ciencias Naturales, Madrid (Spain) ( Table S1). Initial sampling for molecular analyses included a total of 21 specimens. In addition, 15 taxa were added from GenBank, seven belonging to the family Goniodorididae, seven species belonging to the superfamily Onchidoridoidea, and the pleurobranch Berthella martensi (Pilsbry, 1896) was used as outgroup (Table 1).

| Phylogenetic analyses
Successful sequences were assembled and edited using SeqMan II software (DNAStar, Madison, WI, USA). The alignment was done using MEGA7 (Kumar et al., 2016). Protein-coding sequences were translated into amino acids for confirmation of alignment using the genetic code invertebrate mitochondrial DNA for COI and universal code for H3. All sequences were blasted in GenBank to check for contamination. Sequences were trimmed to 658, 449 and 328 pair bases for COI, 16S and H3, respectively. The evolutionary models were selected using jModelTest2 on XEDE (2.1.6) available at CIPRES Science Gateway (Miller et al., 2010), for 16S gene and for each codon position of COI and H3, under Bayesian information criteria (BIC) (Schwarz, 1978). Evolutionary models for COI were TrN+G, TPM1uf+I and HKY+G for the first, second and third codon position, respectively. The TPM3uf+G evolutionary model was selected for 16S. For H3 gene, TIM2+I, JC and TPM2uf+G were selected for the first, second and third codon position, respectively. Bayesian inference (BI) analysis was performed using the software package MrBayes on XSEDE (3.2.7a), available at CIPRES Science Gateway (Miller et al., 2010), for ten million generations, four independent runs and sampling frequency of 1000. Nodes were considered supported by posterior probabilities ≥0.96 (Alfaro et al., 2003). Maximum likelihood (ML) analysis was performed using the software package RAxML-NG (Kozlov et al., 2019), with a bootstrapping cut-off of 0.03 implemented. Nodes were considered statistically significant by bootstraps values ≥75 (Hillis & Bull, 1993). Phylogenetic trees were conducted for individual genes and for concatenate of a minimum of two (COI+16S) (COI+H3) to three genes (COI+H3+16S). The tree obtained was shown using FigTree (Rambaut, 2009). To obtain the final tree, the results were edited in Adobe Photoshop CC 2014. Nodes with posterior probabilities and bootstraps values not supported were collapsed.

| Haplotype network analysis
Haplotype network analysis was conducted for Trapania lineata, using the program PopArt v. 1.7 (Leigh & Bryant, 2015). Empty positions at both ends of the alignment were removed, leaving a final alignment with 565 bp of 12 taxa for COI. Trait file with geographic area codes was created with excel, assigning a value of 0 when specimens were absent and 1 when specimens were present. Alignment and trait files were imported to PopArt, and the TCS algorithm analysis (Clement et al., 2000) was conducted. Mutations are shown as one-step edges. The haplotype network was edited using PopArt v.1.7 and Adobe Photoshop CC 2014.

| Morphological examination
The external morphology was examined from photographs of living specimens and laboratory observations. Specimens were fully dissected for detailed study, and the digestive and reproductive systems were drawn under a Nikon SMZ-1500 dissecting microscope with a camera lucida attached. The buccal bulb was submerged in 10% NaOH to dissolve the tissue and musculature surrounding the radula and labial cuticle. Then, these structures were risen in distilled water. Penises and labial cuticles were critically dried using hexamethyldisilazane, and all structures were examined under a Hitachi S3000N scanning electron microscope (SEM) at 'Servicio Interdepartamental de Investigación', Autonomous University of Madrid, Spain.

| Phylogenetic and species delimitation analyses
We successfully obtained sequences for the 21 taxa initially sampled (Table 1). The highest intraspecific distance between two specimens of same species were 2.1% for COI, 1.3% for 16S and 0.5% for H3 (  Figure  1), henceforth referred as Trapania sanctipetrensis; (ii) specimens of Trapania lineata Trapania pallida and Trapania hispalensis were considered the same species, with a p-distance of 2.1%, 1.3% and 0% in COI, 16S and H3 genes, respectively (Table 2; Figure 1), supporting the synonymy of T. pallida and T. hispalensis with T. lineata. Trapania maculata as well as Trapania orteai were confirmed as valid species (Figure 1). On the contrary, bPTP species delimitation analysis performed in COI divided the specimens of T. orteai and T. lineata in different taxa. Specimens of T. orteai from Morocco were considered different taxa from those collected in Spain (maximum COI p-distances 2.1%). Regarding T. lineata, bPTP analysis divided the species in f taxa. Two of these taxa included morphotypes of T. lineata and the synonymized T. hispalensis. The two specimens of the morphotype corresponding to the synonymized T. pallida were considered one different taxon. The phylogenetic tree obtained based on concatenate gene sequences supported the monophyly of the genus by grouping Trapania taxa into a well-supported clade (BI = 1; ML = 100) (Figure 1). Although only five species of Trapania were included, some relationships between species were well-supported by Bayesian Inference analysis. The first clade gathered the East Atlantic-Mediterranean Trapania species, separated from T. reticulata Rudman, 1987 (BI = 0.98). Amongst them, T. sanctipetrensis and T. orteai joined as sister species (BI = 0.99), and these two species clustered with T. lineata (BI = 1).

| Haplotype network analysis
The haplotype network included the 12 specimens of Trapania lineata, grouped as the same taxa in ABGD and ASAP species delimitation analyses for COI and 16S, and 16S bPTP. The analysis included specimens with the white morphotype of Trapania pallida from Ireland, white specimens with yellow appendages, rhinophores, gills, oral tentacles and tail, corresponding with the morphotype of Trapania hispalensis, from the Atlantic coast of France, Portugal, Spain and Morocco, as well as the Mediterranean coast of Morocco, and specimens of T. lineata from the Mediterranean coast of Spain and Italy (Figure 2; Table  S1). The network showed 10 haplotypes, of which only two were shared between specimens, each of them by two individuals (Figure 2). One of the haplotypes were shared by a specimen of T. lineata with thin lines along the body, from Menorca (Spain), and a specimen with T. hispalensis morphotype from Capghir (Morocco). The second haplotype was shared by two specimens of T. lineata collected in Girona (Spain). The overall network showed a lack of geographic structure, with no apparent relationship between the T. hispalensis morphotype and T. lineata, as well as the different locations. However, the haplotypes corresponding to the specimens from Ireland were grouped in a same clade, being the morphotype of T. pallida the most distant with T. hispalenis morphotype and T. lineata, with six mutation sites between both groups.

| Morphological analyses
Results of morphological analyses are shown in Appendix S1, including redescription of species, drawings of internal organs ( Figure S1) and scanning electron photographs of radulae, labial cuticles and penises ( Figures  S2-S4). Morphological analyses supported the results obtained based on molecular data. The study of the external T A B L E 2 COI, 16S and H3 gene pairwise uncorrected p-distances (%) within and amongst species of Trapania morphology and internal anatomy of Trapania lineata, and the synonymized Trapania hispalensis and Trapania pallida showed that the main difference between them was the colour pattern ( Figure 3), which turned to be a colour variability within the senior synonym species T. lineata. The radula morphology showed some intraspecific variations not only amongst specimens, but also between different teeth within the same radula (see Systematic Results: Appendix S1). Another species with intraspecific variation in the radula was Trapania maculata. However, this species, like Trapania orteai, did not show remarkable variations in the colour pattern (Figure 4a-d).
Trapania orteai showed slight differences in the size of the spots that cover the body, being larger in some specimens than in others. The second species with differences in colour pattern was Trapania sanctipetrensis (Figure 4ef). We studied one specimen from Huelva that matched with the poor original description of Trapania fusca. However, morphological and molecular results showed that the specimen from Huelva was T. sanctipetrensis, whose studied specimen was collected in the type locality, Sancti Petri, Cádiz (Spain). This work has been registered under Zoobank Accession: urn:lsid:zoobank. org:pub:A0417F75-5074-4CE2-9A47-2BEE96BA5FFF.

| DISCUSSION
The colour pattern is one of the most common features used to identify nudibranchs species (Korshunova et al., 2020;Layton et al., 2018;Padula et al., 2016). However, the incorporation of molecular analyses has shown that morphology can be misleading. Cryptic or pseudocryptic species have been discovered thanks to molecular analyses in several groups of nudibranchs, showing the existence of species that had gone unnoticed due to the great morphological similarity with another species (Korshunova et al., 2019;Layton et al., 2018;Martín-Hervás et al., 2021;Pola et al., 2014Pola et al., , 2019Sørensen et al., 2020;Toms et al., 2021). On the contrary, intraspecific variability within the same F I G U R E 1 Phylogenetic relationships (BI/ML) based on the concatenated mitochondrial (COI and 16S) and nuclear (H3) genes. Purple branches represent Trapania taxa. Trapania species newly sequenced are in bold. Different colours highlighted in bPTP, ABGD and ASAP species delimitation analyses for COI and 16S represent potential different taxa species has been observed and molecularly supported, showing that the same species can present varied morphotypes (Almada et al., 2016;Araujo et al., 2022;Layton et al., 2018;Padula et al., 2016;Paz-Sedano et al., 2017;Sørensen et al., 2020). As a consequence, the reliability of using colour pattern as a diagnostic character has already been questioned within nudibranchs (Layton et al., 2018).
In addition to this question, we here confirm the theories of some authors that the colour pattern of European species of Trapania should be taken with caution (Edmunds, 2009;Rudman, 1987), proving the intraspecific colour variability of the species Trapania lineata and Trapania sanctipentrensis.
Even though the bPTP species delimitation analysis based on COI sequences divided T. lineata into five different taxa and Trapania orteai into two taxa, ABGD and ASAP analyses on COI, and bPTP, ABGD and ASAP analyses on 16S supported the validity of both species. bPTP analysis has previously shown to over-splitting clades recognized in phylogenetic analyses (Layton et al., 2018;. In addition, results are also supported by COI p-distances. Some nudibranchs have shown interspecific genetic distances as low as 4.3% in Polycera (Sørensen et al., 2020) or even a value of 3.6% in Halgerda (Tibiriçá et al., 2018). The value here obtained is notably lower than these limits, with a higher COI p-distance within T. lineata and within T. orteai of 2.1% (Table 2). Notably, the higher value within T. lineata is the result of comparing specimens of the synonymized Trapania pallida , from Ireland with the remaining southernmost specimens. In addition, Irish specimens were the most distant in the haplotype network and clustered in the phylogenetic tree. These results reflect a reduced gene flow between populations, which may show an incipient speciation of northern specimens, where a lineage division is incomplete due to insufficient time of reproductive isolation. However, since Trapania pallida has recently reported on the Mediterranean coasts of France (Canes (Horst, 2010) and Cote d'Azur (Meudic et al., 2016)) and Italy (La Spezia (Trainito et al., 2018)), the isolation could break down and this morphotype could not evolve as a different species. It would be interesting to perform haplotypes network analyses including specimens of Trapania pallida morphotype collected in the Mediterranean Sea to clarify this possible incipient speciation of the morphotype. In T. orteai, the maximum p-distance corresponds to the comparation of specimens from Morocco vs. specimens from Spain. However, in the present study, we only included sequences of specimens from two different populations. Therefore, the taxon sampling is insufficient to consider an incipient speciation.  Regarding the morphology of the species, the North Eastern Atlantic-Mediterranean species of Trapania here included are well-described in their original description, as well as by other authors that posteriorly recorded these species (Ballesteros, 1985;Brown & Picton, 1976;Cervera & García-Gómez, 1989a, 1989bCervera et al., 2000;Gavaia et al., 2004;Haefelfinger, 1960;Kress, 1968Kress, , 1970Ortea et al., 1989;Tamsouri, 2014;Thompson & Brown, 1984). Our specimens match these descriptions (see Appendix S1). In addition, our study confirms the variability in the colour pattern of T. lineata and T. sanctipetrensis and adds new details of the penial spines in the species T. orteai, T. sanctipetrensis and T. tartanella. The presence of elongated, pointed penial spines has already been reported for these species (Cervera & García-Gómez, 1989b;Cervera et al., 2000;Ortea et al., 1989), including the difference in size between the spines located at the base and at the most distal part of the penis in T. orteai (Cervera & García-Gómez, 1989b). However, the change in elongated spines to rectangular spines with several cusps was not observed in any of them.
The evidence of the colour variability within Trapania species may rekindle some doubts about the synonymy of T. fusca, T. tartanella and T. graeffei, which most significant difference that supported the validity of these species was the colour pattern (Doneddu et al., 2020;Ortea et al., 1989). In addition, T. sanctipetrensis could be added to this possible synonymy due to the great similarity of its radula with T. tartanella and T. graeffei (Ortea et al., 1989). The four of them share a brown colour pattern with different intensity. However, in the present study, only a specimen T. sanctipetrensis was found suitable for molecular analyses collected from the type locality. Despite the concordance of the specimen collected from Huelva (Spain) with the species T. fusca, we preferred to retain the name of T. sanctipetrensis instead of considering it synonym of T. fusca until specimens of T. fusca are found in the type locality (Arcachon, France). Similarly, no anatomical differences were found between the species T. sanctipetrensis and the preserved neotype of T. tartanella. However, the neotype of T. tartanella was initially preserved in formalin; therefore, it does not allow molecular studies. Furthermore, T. tartanella was originally described from Naples (Italy), Mediterranean Sea, (Ihering, 1886) and the neotype was collected in Asturias (Spain) (Ortea et al., 1989). Although our study shows that the differences between the species cannot be based solely on the colour pattern, and the external morphology and internal anatomy do not show great differences, we prefer to keep the species T. fusca, T. tartanella, T. sanctipetrensis and T. graeffei as valid species until new specimens from the type locality of these species are collected and sequenced.
It is also worth mentioning that Pruvot-Fol (1954) illustrated a species of Trapania, Trapania lafonti, in her contribution to the Opisthobranch fauna from France, including a drawing of the dorsal view of the animal, teeth, penial spines and jaws. However, she did not include any description or other reference in the text, nor did she mention the authority. Pruvot-Fol (1954) cited specifically that 'Three species have been described; having found intermediaries between two of them, I am convinced that 'Drepania tartanella' is the youngster of 'Drepania fusca'. The third is also synonymous (variety of colouring). There is therefore only one species in Europe'. Immediately after, the author described Trapania fusca, indicating the species Drepania graeffei and Drepania tartanella as synomyms. The existence of three Trapania species until 1953 is also in agreement with MolluscaBase (2022), which only add Trapania japonica (Baba, 1935), from Japan, before that date. Therefore, the three species described up to that time by Pruvot-Fol (1954) were T. fusca, T. graeffei and T. tartanella, without signs of T. lafonti. We only found Trapania lafonti in this publication. It is possible that Pruvot-Fol (1954) used the name T. lafonti referring to T. fusca, alluding to the author of the species, Lafont. Therefore, Trapania lafonti must be considered as a not available name following the Zoological Nomenclature Code, which indicates that 'to be available, every new name published after 1930 must be accompanied by a description or definition […], or be accompanied by a bibliographic reference […] or be proposed expressly as a new replacement name (nomen novum) for an available name' (ICZN, 2000).
In this study, we here carried out a preliminary attempt to clarify the biodiversity of Trapania species that inhabit the templated coast of North Eastern Atlantic Ocean and the Mediterranean. We confirmed the validity of T. lineata, T. maculata and T. orteai, which inhabit both the Atlantic Ocean and the Mediterranean Sea, crossing the Strait of Gibraltar (Ballesteros et al., 2019;Cervera & García-Gómez, 1989b;Haefelfinger, 1960;Kress, 1968;Ortea & Urgorri, 1981). However, the species Trapania hispalensis, T. lineata and T. pallida belong to the same taxa, with Trapania lineata as senior synonym. In addition, the validity of the species T. fusca, T. graeffei, T. sanctipentrensis and T. tartanella remains pending to sequence of specimens from the type localities in future studies.
Girona, Jose Templado for specimens from Menorca and Joan Pereira for specimens from Ibiza. Also, we are grateful to the Natural History Museum Basel (Switzerland) to allow us to study the paratypes of the species Trapania lineata and Trapania maculata and to the Museo Nacional de Ciencias Naturales (Spain) to allow us to study the neotype of Trapania tartanella. Many thanks to Ángel Luque for his comments and help trying to clarify the location and identification of Trapania fusca and information about Trapania lafonti. Sampling and molecular studies were partially supported by the project 'Estudios de los Opistobranquios (Mollusca; Gastropoda) de la isla de Menorca (Islas Baleares, Mediterráneo Occidental)', Institut Menorqui d'Estudis, P.I: Juan Lucas Cervera with the collaboration of Marta Pola, the 'Investigación taxonómica de los moluscos de la provincia de Huelva', Consejería de Agricultura, Ganadería, Pesca y Desarrollo Sostenible, Junta de Andalucía, P.I: José Francisco Martín Álvarez and the 'Ayuda a los Grupos de Investigación del Plan Andaluz de Investigación, Desarrollo e Innovación (Grupos PAIDI), PAIDI RNM-213, P.I: Juan Lucas Cervera, with the collaboration of Marta Pola.