Molecular and morphological evidence of the occurrence of the Norwegian skate Dipturus nidarosiensis (Storm, 1881) in the Mediterranean Sea

Abstract Fourteen specimens of the Norwegian skate, Dipturus nidarosiensis (Rajiformes, Rajidae), were caught off the Sardinian coasts (Central Western Mediterranean Sea) in 2005–2008 between 600 and 1420 m of depth. Their identification has been confirmed by the sequencing of three regions of the mtDNA (16SrDNA, control region and cytochrome c oxidase subunit 1) and comparison of the obtained sequences with that of three species of Dipturus (D. batis, D. oxyrinchus and D. nidarosiensis) from the Mediterranean Sea and the adjacent North-eastern Atlantic Ocean. A simple PCR-RFLP assay has been developed for an easy, reliable and robust identification of these skate species. A morphological comparison of the Norwegian skate with congeneric species is given in order to help future identifications. This is the first record of D. nidarosiensis in the Mediterranean Sea; the possibility of recent or ancient but unnoticed occurrence of the Norwegian skate in the region is discussed.


Introduction
Skates (order Rajiformes, family Rajidae) are an extremely diverse group of fishes, characterized by a high morphological conservatism (McEachran & Dunn 1998). In recent years, the number of nominal species has increased exponentially, with more species described in the last 60 years than in the previous 200 years and another 50Á100 species still to be described (Ebert & Compagno 2007).
Dipturus (Rafinesque, 1810) is the second most speciose genus within Rajidae, with most of its species described in the last 40 years. Five species of Dipturus and an undescribed species are known from the eastern Atlantic, and two of these also occur in the Mediterranean Sea. In the eastern Atlantic region, Dipturus batis (Linnaeus, 1758), Dipturus oxyrinchus (Linnaeus, 1758), Dipturus linteus Fries, 1839, Dipturus doutrei (Cadenat, 1960) and Dipturus nidarosiensis (Storm, 1881) are present, together with a further species (Dipturus sp.) which has yet to be formally described ). Two of these species, D. batis and D. oxyrinchus, occur in the Mediterranean Sea (Serena 2005).
According to the List Categories and Criteria of the IUCN, a recognized objective system for assessing the global risk of extinction for species (Vié et al. 2008), three out of five described Dipturus are considered threatened. The Norwegian skate Dipturus nidarosiensis is listed as near threatened (Stehmann 2008). It is known as the only endemic species of Dipturus to the NE Atlantic area. Despite its common name, D. nidarosiensis is a very rare species even in the Norwegian area where today it is very rarely caught (Williams et al. 2008). Like other large skates, D. nidarosiensis has a low reproductive rate and is vulnerable to trawl fisheries; therefore, this species needs careful monitoring.
Unfortunately, catch and landing data for skates are often of poor quality because of a general lack of species-specific recording (Johnston et al. 2005). Skates are difficult to identify, misidentifications can be quite common (Daan 2001), and this is especially true for the NE Atlantic Dipturus species. Actually, except for the more visually distinct species such as D. nidarosiensis, there is still concern regarding the validity of the skate identifications in the entire NE Atlantic area (Williams et al. 2008). This taxonomic impediment can hamper the assessment, conservation and management of global fish biodiversity (Ward et al. 2009).
On deep trawl surveys and commercial fishing operations off Sardinia (Central Western Mediterranean Sea) in 2005Á2008, 14 Dipturus specimens were collected between 600 and 1420 m. As their external features did not match with the description of the two species known for the Mediterranean Sea, the molecular genetic technique of DNA barcoding was utilized for their identification.
To facilitate the identification of the Dipturus skates found in the Mediterranean Sea, a rapid PCR-RFLP assay was developed on an amplified segment of the mitochondrial DNA Control Region.
Morphological comparisons with other species of Dipturus were performed to provide useful data in future identifications.

Samples
Fourteen specimens of Dipturus sp. were collected in the Sardinian Channel ( Figure 1) from 2005 to 2008: 9 adult females, 1 adult male, 2 sub-adult males, and 2 juvenile males. Twelve of these individuals were caught during experimental deep bottom trawl surveys carried out by the DBAE (Dipartimento di Biologia Animale ed Ecologia, University of Cagliari) from 800 to 1420 m depth, and two further specimens were caught in the same area by commercial trawl fishermen at 600 m depth and delivered to the DBAE laboratory. All captures were made in the area delineated by the coordinates N 38858/38848 Á E 09846/09820. After capture, the specimens were frozen on board, transferred to the laboratory, measured, and photographed dorsally and ventrally (Figure 1). Due to the very large size of the adult specimens, the whole body of only two adults (one adult male and one adult female) and of the two young specimens were preserved and kept for further inspections at the DBAE Zoological collection, Museum of Zoology, University of Cagliari (Table I).
Mediterranean specimens (n 012) of long-nosed skate D. oxyrinchus, caught around Sardinia during 2005 and 2006 in Medits (Bertrand et al. 2002) and Grund (Relini 2000) experimental trawl surveys at depths from 130 to 660 m, were analysed. After the measurements, photography and tissue sampling, one female and one male of D. oxyrinchus were preserved and kept at the DBAE Zoological collection (Table I).
Other samples of D. oxyrinchus, D. batis and D. nidarosiensis specimens were collected in the NE Atlantic by S. Iglesias and colleagues; all but two of the Atlantic specimens are preserved in the Muséum National d'Histoire Naturelle of Paris (MNHN) ( Table I).

Morphology
Measurements were recorded for all the Mediterranean specimens. Colour was recorded for fresh individuals. Some external measurements were made, as described by Hubbs & Ishiyama (1968) and Stehmann (1995) (Table II and Supplementary  Tables S1 and S2. Supplementary tables are available on the supplementary content tab of this articles page on www.informaworld.com/mbr).

Genetic analyses
Muscle tissues from the underside of the pectoral fins were sampled from frozen specimens then stored in absolute ethanol at 48C and kept in the genetic tissue repository at the University of Cagliari (DBAE) while the Atlantic tissue samples were stored at the 'Station de Biologie Marine de Concarneau' (BPS) ( Table I).
Genomic DNA was extracted according to a salting-out method (Miller et al. 1988). All samples were analysed using polymerase chain reaction (PCR) and direct DNA sequencing. PCR was used to selectively amplify the 16S ribosomal mitochondrial gene (16SrDNA) using the primers described in Palumbi et al. (1991), the control region (CR) using the primers ElDloopF and RajinaeP7r (Valsecchi et al. 2005), the 3'-end of the cytochrome oxidase subunit I (COI) with the primers RajaCOIf and RajaCOIr (Alvarado Bremer et al. 2005). Individual haplotypes were aligned with CLUSTAL-W implemented in MEGA version 4 (Tamura et al. 2007).
To estimate the genetic distances between pairs of DNA sequences we used the Kimura 2-parameter distance (DK2P; Kimura 1980) with the deletion of ambiguous or missing bases in pairwise sequence comparisons. Within-and among-clade distances were calculated and neighbor-joining (NJ) phylogenetic analyses were carried out in MEGA 4.0. Support for nodes for NJ analyses was assessed by non-parametric bootstrapping (1000 replicates).
To facilitate the identification of the Dipturus skates found in the Mediterranean Sea, a rapid PCR-RFLP assay was developed based on the combined restriction activity of two endonucleases DdeI (Invitrogen) and TaqI (Invitrogen) on the amplified segment of the mitochondrial DNA Control Region.

Morphology
The specimens of skate caught in Sardinia clearly belonged to the genus Dipturus Rafinesque 1810 in having the combination of the following characters: a long, hard rostral cartilage (length more than 60% of the dorsal head length), disc rhomboid and nearly free of denticles with few thorns (Compagno 1999),   anterolateral margin of disc concave, line connecting tip of snout to anterior aspect of lateral corner of disc not intersecting margin of disc after rhomboid disc (Stehmann & Bü rkel, 1984) and a total length greater than 55 cm when adult (Ishihara 1987). The Sardinian Dipturus specimens were distinct from both species of the genus thought to occur in the Mediterranean in colour pattern, pattern of tail thorns, snout length, number of tooth rows, total length, and depth occupied (Table II and Figure 2). However, the Sardinian specimens appeared similar or identical to D. nidarosiensis which is endemic to the NE Atlantic, apart from the reduced number of median thorns on the tail (presumably lost during fishing operations as suggested by the conspicuous scars) (Table II). In adults, both dorsal and ventral surfaces were uniformly dark (Figure 1a, b, e, f). The dorsal surface of juveniles was medium brown to greyÁbrownish, in some specimens darker; the ventral surface was dark brown, darker than the dorsal surface (Figure 1c, d). This is the main character that easily distinguishes D. nidarosiensis from the other species of Dipturus present in the Mediterranean/ Eastern Atlantic; in the latter the underside is not uniformly coloured and often whitish or paler than dorsal surface (Figure 2). The violet skate D. doutrei has a similar colour pattern to D. nidarosiensis; dark brown both dorsally and ventrally. However, D. doutrei is smaller (up to 100 cm TL), with fewer thorns on tail (from 11 to 20, small, sometimes indistinct) and fewer tooth rows (upper jaw 29Á34) (Stehmann 1995) (Table II). Therefore, like D. nidarosiensis, the Sardinian specimens morphologically did not correspond to other eastern Atlantic Dipturus.

DNA sequence analyses
Haplotype codes and GenBank Accession numbers for the three genes analysed (16SrDNA, CR and COI) are listed in Table I.
At least 600 bp of 16SrDNA nucleotide sequences were determined from Mediterranean D. oxyrinchus and Dipturus sp. individuals and compared with 16SrDNA sequences from Atlantic specimens of D. oxyrinchus, D. nidarosiensis and D. batis. In Atlantic D. nidarosiensis the variation in the 16SrDNA fragment defined two haplotypes (16Shap1 and 16Shap2, which differed from each other for one indel); all the 14 Mediterranean specimens of Dipturus sp. showed a single 16SrDNA haplotype (16Shap2). There was only one 16SrDNA haplotype in Atlantic D. oxyrinchus (16Shap3); all but one of the D. oxyrinchus from Sardinia shared the same 16Shap3 haplotype which was differentiated by two transitions from the 16Shap4 haplotype of the About 640 bp of nucleotide sequence for the CR fragment from D. oxyrinchus, Dipturus sp., D. nidarosiensis and D. batis individuals, both of Mediterranean and Atlantic origin, were determined.
Two different CR haplotypes were found for D.
nidarosiensis (CRhap1 Á 2), four CR haplotypes were found for Dipturus sp. (CRhap1 Á 4), two CR haplotypes for D. batis (CRhap 11 Á 12) and five CR haplotypes for D. oxyrinchus specimens (CRhap5, 7 Á 10). In particular, CRhap8 and CRhap7/9 were identical to the homologous portion of the D. oxyrinchus haplotypes by Valsecchi et al. (2005). The sequencing of three mitochondrial fragments (16SrDNA, CR and COI) showed that the Dipturus sp. specimens caught in the sea around Sardinia shared the same sequences as those from Dipturus nidarosiensis specimens from the NE Atlantic and therefore they were classified as this species.
The three species of Dipturus (the common skate, the longnose skate and the Norwegian skate) can be easily discriminated by molecular means, since 16SrDNA, CR and COI species-specific haplotypes were found for all of them. Nevertheless, the analysis of sequences indicated that the three markers were not informative in the same way. Both the CR and COI fragments were considerably more variable than the 16SrDNA. Numbers of differences and Kimura (1980) genetic distances within and between species are reported in Table III. Overall, the Kimura mean genetic distances among the three species of Dipturus were very low for the 16SrDNA (DK2P% 00.98%), low for the COI sequences (DK2P% 03.45%) and CR (DK2P% 04.57%). In all cases the mean intraspecies divergence resulted in lower than mean inter-species divergence. In particular, for the COI sequences the mean interspecies divergence was largely higher than 10 times the mean intra-specific variation (DK2P% 00.16%); the standard divergence threshold to delimit species proposed by Hebert et al. (2004), known as the '10-fold rule', was largely overcome.
Besides the distance between sequences, phylogenetic trees were also reconstructed to assess the relationship between the Dipturus sp. sequences and their neighboring sequences. All the phylogenetic analyses realized for the three markers separately (data not shown) produced trees with similar topologies where all specimens were assigned to the same clades. Dipturus sp. individuals clustered together with D. nidarosiensis specimens. Dipturus nidarosiensis, D. oxyrinchus and the Atlantic D. batis appeared as three well-separate clades, supported by high bootstrap values (Figure 3).

PCR-RFPL
The sequential digestion of CR PCR-products with the two endonucleases DdeI and TaqI resulted in species-specific pattern of fragments that can be used to distinguish the three species of Dipturus.
After digestion of the CR PCR-amplified fragments, D. nidarosiensis, D. batis and D. oxyrinchus DNAs could be readily distinguished through agarose gel electrophoresis (Figure 4).

Discussion
Morphologically, the Sardinian specimens appeared similar or identical to D. nidarosiensis and distinct from the other Dipturus species from the Mediterranean and eastern Atlantic.
In the past, the number of Dipturus species in the Mediterranean Sea was questioned and morphological discrimination between them has been regarded as a very difficult task for non-specialists because of the lack of readily scorable diagnostic characteristics. Doubts about the validity of historical identifications of D. batis in the Mediterranean region are reported (Dulvy et al. 2006), because this species could potentially be confused with D. oxyrinchus, despite morphological and colour differences (Tortonese 1956;Dulvy et al. 2006). Unfortunately, to the best knowledge of the authors, pictures are not available in referenced publications for any D. batis caught in the Mediterranean Sea. The only visual representation of Mediterranean D. batis specimens is the drawing by Tortonese (1956) (Figure 5a). Furthermore, from a preliminary survey in the European Museum and Ichthyologic Collections the voucher specimens collected in the Mediterranean and classified as D. batis Figure 3. NJ tree based on the K2P genetic distances of COI sequences. Near the nodes are reported only the 50% bootstrap values. GenBank accession number is indicated near the species names. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. Branches leading to Leucoraja erinacea and L. ocellata haplotypes were collapsed to a single terminal tip (full triangles in figure). Sequences extracted from the complete mitochondrial genome sequences of Okamejei kenojei and Amblyraja radiata were used as outgroups. , whose accuracy in species-specific identification is somehow questionable. Sequence divergence of the three regions of mtDNA sequenced (16SrDNA, control region and cytochrome c oxidase subunit I) indicated that Sardinian specimens were different from both the common skate, D. batis and the longnose skate, D. oxyrinchus but identical to D. nidarosiensis specimens.
Among the three mitochondrial markers, the 600 base pair 16SrDNA sequences of D. nidarosiensis exhibit very low sequence divergence from all other Dipturus species (0.84% from D. oxyrinchus and 1.52% from D. batis). The 560 base pair COI DNA 'barcodes' derived from specimens of D. nidarosiensis exhibit greater sequence divergence; the level of pairwise sequence divergence for D. nidarosiensis/D. oxyrinchus (4.57%) and D. nidarosiensis/D. batis (6.07%) was well above the threshold (10) inter-/intra-specific divergence) proposed by Hebert et al. (2004) and Lefébure et al. (2006) for true species. The average COI sequence divergence between Dipturus spp. (DK2P03.45%) was about 20 times that within species; it was low relative to values reported among other chordates (Hebert et al. 2003) and fishes (Ward et al. 2009), but appears typical of other Chondrichthyes. For instance, the average COI sequence divergence between species was 5.5% and within species was 0.2% for the Dipturus/Zearaja complex (Ward et al. 2008). Among species of Bathyraja average interspecies distance was even lower, 3.6% (Spies et al. 2006) and 2.9% for Leucoraja erinacea and L. ocellata (Alvarado Bremer et al. 2005). This very limited degree of separation could be explained by the well known slow nucleotide substitution rates in the COI gene in sharks, seven-to eightfold lower than in primates or ungulates (Martin et al. 1992;Martin & Palumbi 1993). Finally, the 640 base pair CR DNA 'barcodes' derived from specimens of D. nidarosiensis exhibit the greatest sequence divergence from all other Dipturus species (DK2P07.74% from D. oxyrinchus and DK2P 07.77% from D. batis). Because of its high nucleotide polymorphism, the CR could represent a solid marker for species identification. Furthermore, based on the CR sequence variation among the three species of Dipturus, a very simple PCR-RFLP assay has been developed to routinely identify in the laboratory samples of adults, juveniles or eggs.
Taken together, these independent morphological and molecular observations serve to corroborate one another and thus provide strong evidence for the recognition of D. nidarosiensis as a new species in the Chondrichthyan fauna list for the Mediterranean area.
In the present study, the first record of the Norwegian skate in the Mediterranean Sea is reported. In view of the small sampling effort invested so far in deep areas of the Mediterranean Sea and the relatively few data known for the deepsea biodiversity, the record of D. nidarosiensis in the  Sardinian waters represents an important contribution and allows for an update of the Mediterranean deep-sea elasmobranch biodiversity.
In reality, the presence of skates resembling to the Norwegian skate can be found in old books describing the Italian marine fauna. Doderlein (1885) listed among the species of Rajidae a third species other than the common skate 'batis' and the longnose skate 'oxyrinchus'. This is described as a dark skate, brown both dorsally and ventrally, named Raja macrorhynchus Rafinesque 1810, the same species is also included in Bonaparte (1832) as Laeviraja macrorhynchus (Figure 5b). Therefore, it is possible that representatives of the Norwegian skate could have been caught in the past but they were unnoticed.
The alternative hypothesis is that D. nidarosiensis could represent a new species, entered from the Atlantic in recent years. In fact, over 500 alien species (macrophytes, invertebrates and fishes) have been recorded thus far in the Mediterranean Sea (Galil 2007); however, the great majority of these records refer to coastal benthic species (Bianchi 2007). In effect, during interglacial periods, characterized by a temperate to warm climate as today, the superficial current entering the Mediterranean allows the colonisation of subtropical species from the Senegalese province while the outgoing bottomcurrent made the arrival of deep-sea denizens difficult (Emig & Geistdoerfer 2004). Therefore, the most likely hypothesis is that bathyal D. nidarosiensis has inhabited the Mediterranean Sea at least since the last glacial period, when many bathyal species from the Atlantic (of temperate/boreal origin) entered the basin, favoured by the inverted direction of the bottom-currents through the strait of Gibraltar (Emig & Geistdoerfer 2004).
Apart from speculations on the origin of the Norwegian skates in the Mediterranean, this paper confirms the effectiveness of molecular techniques for identification of species in a group of fish where morphological characters alone makes species identification difficult. Despite the known limited sequence divergence among species that in the past did not always allow for the discrimination among closely related skates (Valsecchi et al. 2005;Spies et al. 2006), mitochondrial sequences obtained here resulted in valid DNA markers for a clear separation of the species studied. The availability of DNA sequences and of an easy PCR-RFLP assay for the three species of Dipturus will be of great help in the future, assisting both in the accurate identification of adults of the genus and in the identification of eggs and young skates, whose morphological characters may not be fully reliable (Daan 2001).