Caloria elegans (Alder and Hancock, 1845) is a nudibranch belonging to the family Facelinidae Bergh, 1889. The genus Caloria was erected by Trinchese (1888) with the description of Caloria maculata that is currently accepted (Gofas 2015) as a synonym of Caloria elegans Alder and Hancock, 1845 (Sartori and Gofas 2014). The original description of C. maculata by Trinchese (1888) is very detailed but it lacks the figure of the entire animal. The rhinophores are described in the text in two sections: "I rinofori a sezione trasversa circolare, sparsi di lievi ripiegature, ma non perfoliati.” (The rhinophores have a circular transverse section and scattered slight folds, but they are not perfoliate) (p. 291); and "I rinofori sottili, bianco-trasparenti, presentano delle rughe irregolari. Una striscia bianco-opaca si estende sulla faccia anteriore dei loro due terzi superiori.” (The thin rhinophores, white-transparent, show some irregular wrinkles. A white-matt strip on their anterior side extends to the upper two-third) (p. 292). These descriptions fit rhinophores possessing wrinkles or slight folds, but not lamellae, and they can be considered equivalent to the ones described for Eolis elegans by Alder and Hancock (1845, 1851). The latter authors, while redescribing the species (1851) clearly depicted the rhinophores (i.e. dorsal tentacles) as "of moderate length, stoutish, erect, tapering at the top and wrinkled transversely, of a pale fawn colour or buff, with a streak of white in front near the apex.” In this case, "wrinkled transversely” clearly indicates that they are not lamellate rhinophores, and in the depicted animal (1851: pl. 17, figs 2, 3) these wrinkles are very evident (Fig. 1A). Picton (1979) traced back the systematic history of C. elegans, redescribing this taxon with specimens from the Bristol Channel (UK) and considering C. maculata as its junior synonym. In the redescription, the rhinophores are not reported as lamellate ("Posterior surfaces of rhinophores covered with small irregularly distributed papillae”) and this character is clearly depicted in his figures (Picton 1979: figs 1A, B) (Fig. 1B). Later, Schmekel and Portmann (1982) described C. elegans, referring to its rinophores as "cylindrical and smooth”. This diversity in C. elegans rhinophores surface (smooth, wrinkled or papillate) was related by Thompson (1988) to the particular ontogenetic stage of the specimens and not to a typical diagnostic feature of the species. In fact, he wrote "The fawn coloured rhinophores... the rear surfaces... are in the adult specimens covered with small, irregularly distributed papillae (hard to detect or absent in the juveniles)”. This issue has been also briefly noticed by Cattaneo-Vietti et al. (1990): "rinofori che possono essere lisci o finemente papillati” (rhinofores can be smooth or finely papillated).

Another taxon synonymized with C. elegans is Acanthopsole quatrefagesi Vayssière, 1888, described based on specimens from the Mediterranean Sea, Villefranche-Sur-Mer (France), which was included in the genus Facelina Alder and Hancock, 1855 by Pruvot-Fol (1951). In 1888 Vayssière described A. quatrefagesi with recently preserved samples but no figures of the whole animal or the rhinophores. The original description (Vayssière 1888: plate VII, figs 137-143) includes cerata, penis, jaws and radular teeth. Facelina quatrefagesi was described as showing lamellate rhinophores: "Les rhinophores, moins longs que les tentacules labiaux, possédaient del lamelles olfactives transversales perpendiculaires, sortes d’anneaux sur les deux tiers supérieurs de leur longueur.” (The rhinophores, shorter than the labial tentacles, possessed perpendicular transverse olfactory lamellae, kinds of rings on the upper two-thirds of their length). The shape of the rhinophores is routinely used as a diagnostic character in almost all other nudibranchs. Based on this feature, F. quatrefagesi should not be considered as a synonym of C. elegans, as proposed by Picton (1979) and accepted in successive papers (Cattaneo and Barletta 1984, Sabelli et al. 1990). Recently, Trainito and Doneddu (2014) proposed F. quatrefagesi and C. elegans as separate species, mostly considering the morphology of their rhinophores. According to these authors, these two species often show very similar chromatic patterns (Fig. 2), and although some pigmentation differences can be observed (Trainito and Doneddu 2014: 106), their separation is mainly based on the rhinophore morphology, lamellate in F. quatrefagesi and smooth or papillated in C. elegans (Fig. 2). These authors also specified that F. quatrefagesi is very constant in colour, while C. elegans displays a certain degree of variability (Fig. 2A, B).

In order to definitively assess the taxonomy of these two nominal species, individuals from different Tyrrhenian localities (western Mediterranean Sea) were sampled and investigated with an integrative taxonomy approach—using both anatomical and molecular methods—aimed at increasing the robustness of species delimitation (Modica et al. 2014).

The mitochondrial cytochrome oxidase subunit I (COI) marker is routinely used for species delimitation analyses and is now considered a very useful barcoding marker for molluscs in general (Furfaro et al. 2016b), due to its fast-evolving rate of mutation. The partial 16S marker is a non-coding mitochondrial gene that shows the capability to fold in a 2D secondary structure that could be diagnostic for species delimitation (Lydeard et al. 2000, Salvi et al. 2010, Salvi et al. 2014) and as an additional tool for phylogenetic analyses. The H3 is a nuclear gene characterized by a slow evolving rate (Padula et al. 2016) and it is commonly used to explore the relationships occurring within higher taxonomic levels.

We used partial sequences of two mitochondrial (COI and 16S rDNA) and one nuclear marker (histone H3), newly produced and retrieved from GenBank to clarify species boundaries between the two taxa under study.

Samples were collected by SCUBA diving at Mediterranean and Atlantic localities. Specimen data, including collection localities, accession numbers and references, are reported in Table 1. Vouchers are stored at the Department of Biology and Biotechnologies, ‘La Sapienza’ University (Rome, Italy) (Table 1). Hundreds of individuals were observed in situ or indirectly by photographs. Sequences from 13 individuals (ascribed to the family Facelinidae and the outgroup) were used for the molecular analyses.

Microscopical observations of the radulae and jaws (Fig. 3A, D) from specimens of C. elegans and F. quatrefagesi revealed no noteworthy differences. All the radulae observed had formula (0.1.0.)×20 and were very similar to each other in the rachidian tooth (Fig. 3B-D) and also in the denticulated jaws (Fig. 3A, C).

The reproductive systems showed differences in the length of the ampulla duct and in the shape of the terminal portions of the prostate. In particular, the ampulla of C. elegans specimens was connected through a duct shorter than that of F. quatrefagesi. Furthermore, the prostatic portion of the C. elegans reproductive system had a constriction before the deferent duct, while it expanded gradually into the vas deferens in F. quatrefagesi (Fig. 3E, F). No other major differences were observed.

Cerata of both species possess an apical portion, right behind the typical black spot, which is contractile and may lead to specimens showing cerata with different lengths. This characteristic is visible in Picton’s drawing (1979) and was also recognized in specimens observed in vivo during this study.

Stereomicroscope observations on the rhinophores of C. elegans specimens revealed that, according to their stretching, they can appear i) smooth when elongated to their maximum length (Fig. 4A); ii) with clearly visible papillae when slightly contracted (compare Fig. 4B with Fig. 1B); or iii) markedly wrinkled when considerably contracted (compare Fig. 4C with Fig. 1B).

Molecular analyses yielded 32 new sequences, which were added to the 34 sequences retrieved from GenBank (Table 1). The complete dataset consisted of 66 sequences (15 for the nuclear H3, 18 for the mitochondrial 16S and 21 for the COI barcode marker) obtained from 22 specimens (including the outgroup). The H3 (=294 bp), 16S (=420 bp) and COI (=588 bp) concatenated alignment consisted of 1302 positions and a total number of 244 variable sites.

The best-fitting substitution models for H3, 16S and COI were K80+I, HKY+G and TPM1uf+I+G, respectively.

All recursive steps (30 with p-distance, 22 with K2P) in the ABGD analyses of the COI alignment resulted in the same sequence repartitions with the two groups, C. elegans and F. quatrefagesi, representing two distinct Preliminary Species Hypothesis (Fig. 5) with 14.5% of COI genetic distance (Table 2).

The BI and ML analyses of single genes and partitioned concatenated datasets (H3+16S+COI; Fig. 5) (16S+COI; Supplementary Material Fig. S1) produced largely congruent tree topologies. The molecular analyses of the concatenated nuclear and mitochondrial dataset (Fig. 5) clearly demonstrated the presence of two monophyletic clades, one comprising C. elegans (Bpp=1, Bs=100) and the other F. quatrefagesi (Bpp=1, Bs=100). The sister group relationship of C. elegans and F. quatrefagesi was always highly supported (Bpp=1 ad Bs=100).

The mitochondrial 16S rRNA primary sequence examined is the 3’ half portion of the gene corresponding to domains IV and V (Gutell and Fox 1988, Gutell et al. 1993). Its derived secondary structure conforms to the canonical architecture also proved in molluscs (Lydeard et al. 2000, Salvi et al. 2010, 2014). Both domains IV and V show a high conservation in folding when compared among the species of Facelinidae analysed, with the exception of the variable L7 and L13 loops of domain V (Horovitz and Meyer 1995), which might prove to be diagnostic RNA barcoding regions (Fig. 6). The primary and secondary structures of 16S rRNA were compared between C. elegans and F. quatrefagesi. Three diagnostic nucleotide positions in the L7 stem are labelled by asterisks in Figure 6, and a more evident structural diversity in the apical part of the L13 stem (boxed in Fig. 6) was observed.

C. elegans and F. quatrefagesi are two ‘aeolid’ nudibranchs showing a very similar morphology. For this reason, in the past their taxonomic status was the subject of controversy and they were considered as synonyms. The population of F. quatrefagesi in the Marine Protected Area "Tavolara Punta Coda Cavallo” (Trainito and Doneddu 2015) prompted the reinstatement of this taxon based on rhinophore morphology (Trainito and Doneddu 2014, 2015). Slight differences in the morphology of rhinophores (shape of papilles) has proven to be useful to separate genera, as in the case of Baeolidia Bergh, 1888 and Limenandra Haefelfinger and Stamm, 1958 (Carmona et al. 2014). As outlined in the Introduction section, rhinophores of C. elegans have been described as smooth and papillate, but also as wrinkled. Actually, in C. elegans only, rhinophores can vary from smooth to papillate or wrinkled, according to their degree of contraction. Nevertheless, the shape of rhinophores (i.e. smooth/papillate/wrinkled in C. elegans and lamellated in F. quatrefagesi) holds as the only diagnostic phenotypical character to separate these two taxa, which indeed show very similar chromatic patterns (Trainito and Doneddu 2014).

In the present work, we applied a molecular approach, utilizing three genetic markers, to define species limits and infer a preliminary phylogenetic framework of these facelinids. All the molecular analyses returned the same congruent results with C. elegans and F. quatrefagesi as two different and well-separated species. Furthermore, the 16S RNA L7 and L13 domains proved to be highly diagnostic regions due to the presence of some nucleotide substitutions with a 2D structural diversity in this group of nudibranchs, which should be further investigated as an additional tool for species delimitation analyses.

C. elegans and F. quatrefagesi also share a largely overlapping distribution, both inhabiting the whole Mediterranean basin and the neighbouring eastern Atlantic Ocean. However, only C. elegans ranges up to the northeastern Atlantic Ocean (Fig. 7).

The case of these largely sympatric sibling species is a clear example of the power of the integrative taxonomy approach for studying cryptic diversity within marine invertebrates (Gosliner and Fahey 2011, Furfaro et al. 2016a), with molecular analyses that helped to frame morphological and anatomical observations.

The very preliminary molecular phylogenetic pattern retrieved in the present study revealed that the genera Facelina and Caloria are not monophyletic. This is an additional indication of the need for a thorough revision of the Facelinidae, with the redefinition of the genera of the family.