DNA Barcoding Southwestern Atlantic Skates: A 20-Year Effort in Building a Species Identification Library

Abstract

The skate fauna in the Southwest Atlantic Ocean (SWA; 34–55° S) is represented by ~32 species, many of which share external features that have led to misidentifications and deficient fishery statistics. The use of DNA barcoding to discriminate SWA skate species was explored after 20 years of surveys. COI sequences were subjected to distance-based neighbor-joining (NJ), maximum likelihood (ML), barcode index number (BIN), automatic barcode gap discovery (ABGD), and nucleotide diagnostic character (NDC) analyses. For widely distributed species, a haplotype network was built. Overall, 187 specimens and 31 egg cases from 26 skate species were barcoded. NJ and ML analyses showed that nearly all species exhibited unique barcodes or clusters of closely related haplotypes, except for Psammobatis normani/P. rudis and Dipturus trachyderma/D. argentinensis. The first pair was discriminated by NCD. BIN analysis recovered 17 groups, whereas ABGD recovered 23, better reflecting taxonomic diversity. In summary, 24 species were resolved by COI. Phylogeographic signals were observed for Amblyraja doellojuradoi and Zearaja brevicaudata. Compiling our results with data from BOLD, almost all the species occurring in the area possess barcodes, contributing to completing and curating the BOLD reference library, which constitutes an important tool for resolving taxonomic issues, tracing fishery products, and performing eDNA biomonitoring.

1. Introduction

Skates (Chondrichthyes, Rajiformes) constitute a monophyletic and cosmopolitan group of cartilaginous fishes, with representatives in all oceans. They inhabit cold and temperate waters from the coast to depths of more than 4000 m. Some species also occur in Arctic and Antarctic waters and at least one species lives primarily in brackish/freshwater habitats [1,2]. The order accounts for nearly 300 species and is composed of two highly diverse families (Rajidae and Arhynchobatidae) and two smaller families (Anacanthobatidae and Gurgesiellidae) [3,4,5,6,7,8]. This group has life history traits such as relatively slow growth, late age at sexual maturity, long incubation periods, and relatively high longevity that gives them low resilience to overfishing as targeted or by-catch species [9,10,11]. Despite their high diversity, several species share external characters that hinder the correct identification by non-specialists and therefore may lead to error-prone or deficient classification in fishery statistics [12,13,14].

The skate fauna in the Argentine continental shelf (ACS) is represented by two families, Arhynchobatidae and Rajidae, comprising 24 nominal species [15,16,17,18,19]. Furthermore, six species of skates have been recorded from the deep waters off the ACS (beyond 800 m): the thickbody skate Amblyraja frerichsi (Krefft, 1968), the Antarctic starry skate Amblyraja georgiana (Norman, 1938), the whiteleg skate Amblyraja taaf (Meissner, 1987), the dark-belly skate Bathyraja meridionalis Stehmann, 1987, the butterfly skate Bathyraja papilionifera Stehmann, 1985, and the whitemouth skate Bathyraja schroederi (Krefft, 1968) [15,16,20]. In addition, some reported skates such as Eaton’s skate Bathyraja eatonii and commander skate Bathyraja maccaini have also been recorded north of 55°S [21]. This brings the total number of species present in the Southwestern Atlantic (SWA) to 32 species, which have been reviewed by the IUCN and categorized as critically endangered (2), endangered (4), vulnerable (4), near-threatened (4), least concern (15), and data-deficient (3) (Table S1).

Skates are increasingly exploited as target species because traditional bony fish stocks from commercial fisheries are being drastically depleted [22,23]. Skates are a common component of the demersal fish community along the South American continental shelf and have become a concern in Argentina because of the considerable and increasing catches in recent decades due to international demand. The annual capture increased from 761 tons to a maximum of 26,957 tons from 1992 to 2008, and then it continually decreased down to 6551 tons in 2022 [12,16,24,25,26]. Unfortunately, no specific identification is given to most skate captures, and are labeled as “Rayas nep”, meaning “no specified skates”, in the statistics of Argentine’s skates’ fishery [12].

Over the last years, several molecular studies have been conducted on skates in order to support morphological research [27,28] and to resolve taxonomic uncertainty [29]. Some of these works included the use of DNA barcoding as a standardized molecular marker to aid in taxonomic approaches [30,31,32,33,34,35,36,37,38,39,40]. The Fish Barcode of Life Initiative (FISH-BOL; [41]) seeks to establish a mitochondrial 5’ cytochrome c oxidase subunit I (COI) reference sequence library for the molecular identification of fishes worldwide, following a common protocol that includes links to voucher specimens [42]. In this sense, some regional initiatives were conducted specifically regarding chondrichthyan species [43,44]. The primary goal of barcoding focuses on the assembly of reference sequence libraries derived from expert-identified voucher specimens in order to develop reliable molecular tools for species identification in nature [45].

The use of DNA barcoding in fishes can facilitate subsequent species identification by non-specialists, help highlight specimens that represent a range expansion of known species, flag previously unrecognized (e.g., cryptic) species, and enable identifications where traditional methods are not applicable (e.g., fillets, egg cases, and neonates). As of February 2025, at least 190 skate species from 33 genera have been barcoded (https://v4.boldsystems.org/ (accessed on 7 February 2025)), representing >60% of all nominal species [46,47].

DNA barcoding surveys on SWA fishes began in 2005 [35]. The study included the assessment/ability of DNA barcoding to identify 21 skate species from the ACS. Posteriorly, Ribeiro et al. [37] published a DNA barcode reference sequence library for marine fishes from the coastal region of southern Brazil. The latter added several new barcodes for bony fish but no new barcodes for skates. Given the low availability of barcode data during these years, these studies were performed from a local perspective. In addition, several species lacked barcode data and a unique and unambiguous molecular tag. On the other hand, these studies did not consider the use of new taxonomic tools for the identification of molecular operational taxonomic units (MOTUs) such as the barcode index number (BIN) or the automatic barcode gap discover (ABGD) analyses [48,49]. Recently, 38 skate species collected from four major ocean areas of the Atlantic were barcoded and taxonomically analyzed [44]. Nevertheless, the study did not cover the southern tip of South America, and included only five species from the SWA, adding just one skate species to previous studies. After 20 years of DNA barcoding surveys on the SWA skate fauna, a well-curated COI barcode library for the area is here presented. The aims of this paper were as follows: (1) to probe the ability of DNA barcoding to discriminate skate species and the role that BINs play in this respect; (2) to analyze the taxonomic species diversity within each BIN; (3) to evaluate the utility of DNA barcoding for identifying skate egg cases; and (4) to discuss and highlight possible misidentifications recorded in the literature regarding COI sequences of skates.

2. Materials and Methods

2.1. Sample Collection

Specimens and egg cases were collected from the Argentine continental shelf (ACS) between 2005 and 2018 (Figure 1, Table S2). Different sources were employed for this study. Firstly, we used sequences from our BOLD project FARG (fishes of Argentina) (www.boldsystems.org), which includes specimens collected prior to 2009 from the Argentine continental shelf (ACS) and published by Mabragaña et al. [35]. Secondly, we obtained sequences from specimens collected on board the oceanographic vessel “Puerto Deseado” of CONICET, during research cruises conducted from 2009 to 2014 on the ACS and slope, designed for studying the biodiversity of marine fauna. Samples were collected using two bottom trawls, a bottom trawl net (135 mm mesh in the wings, and 60 mm in the cod end; vertical height 3.7 m, horizontal opening 10 m) and a shrimp net (50 mm mesh in the wings, and 20 mm in the cod end; vertical height 1 m, horizontal opening 4 m). The tow duration was between 20 and 30 min and the trawling speed was from 1.5 to 3 knots. Furthermore, samples obtained from artisanal ships off the Mar del Plata coast (Argentina) during 2010–2012 and from fishery landings at a fish processing plant in Mar del Plata were used. Additionally, selected public DNA sequences of some SWA skates (Amblyraja taaf, Bathyraja meridionalis, B. eatonii, and B. maccaini) were gathered from BOLD.

Vouchers were morphologically identified following the identification reliability level 1 according to the Fish-BOL collaborator’s protocol [42] as follows: “highly reliable identification—specimen identified by (1) an internationally recognized authority of the group, or (2) a specialist that is presently studying or has reviewed the group in the region in question”. We followed Fricke et al. [47,50] for species names and higher classification.

A piece of tissue was taken from representatives of each species and preserved in 96% ethanol for subsequent molecular analysis. Voucher specimens were labeled, photographed, formalin-fixed (with further alcohol long-term preservation), and housed as vouchers in the fish collection of “Instituto de Investigaciones Marinas y Costeras (IIMyC)-CONICET-Universidad Nacional de Mar del Plata”, Argentina.

2.2. DNA Extraction, COI Amplification, and Sequencing

DNA extraction, polymerase chain reaction (PCR), and sequencing of the 5’ region of a fragment of the mitochondrial cytochrome c oxidase subunit I gene (COI) were performed following standard DNA barcoding protocols [51] coupled with primers and primer cocktails developed for fishes [52,53]. For these new samples, DNA extraction and amplification of COI were performed at the Argentine International Barcode of Life Laboratory (IIMyC, CONICET, Mar del Plata, Argentina).

Sequencing was performed at the Advanced Analysis Center’s Genomics Facility (College of Biological Sciences, University of Guelph, Ontario Canada), in the Canadian Centre for DNA Barcoding (CCDB) at the Biodiversity Institute of Ontario, (University of Guelph, ON, Canada), and in Macrogen (Seoul, Republic of Korea).

Amplification of the 5′ region of COI, corresponding to base positions 6474 to 7126 of the Danio rerio mitochondrial genome [54], was first attempted using FF2d_t1/FR1d_t1 primer combination and C_FishF1t1/C_FishR1t1 primer cocktails [51]. The primer combinations C_FishF1t1andC_FishR1t1 both contained two primers (FishF2_t1/VF2_t1 and FishR2_ t1/FR1d_t1, respectively). PCR reactions were performed in 96-well plates. The reaction master mix consisted of 825 μL water, 125 μL 106 buffer, 62.5 μL MgCl2 (25 mM), 6.25 μL dNTP (10 mM), 6.25 μL each primer (0.01 mM), and 6.25 μL Taq DNA polymerase (5U/μL). This mixture was prepared for each plate, and each well contained 10.5 μL of solution and 2 μL of genomic DNA. The PCR reaction profile comprised an initial step of 2 min at 95 °C, and 35 cycles of 30 s at 94 °C, 40 s at 52 °C, and 1 min at 72 °C, with a final extension at 72 °C for 10 min. For specimens that failed to amplify using the primer combinations mentioned above, the primer combinations C_VF1LFt1/C_ VR1LRt1 [53] consisting of VF1_t1/VF1d_t1/LepF1_t1/VFli_t1 and VR1_t1/VR1d_t1/LepR1_t1/VRli_t1 primer sets, respectively, were tried. All primers were appended with M13 tails to facilitate sequencing. Amplicons were visualized on a 2% agarose E-Gel H 96-well system (Invitrogen Waltham, MA, USA). Sequencing reactions applied M13 forward and reverse primers using the BigDyeH Terminator v.3.1 Cycle Sequencing Kit (Applied Biosystems Inc. Waltham, MA, USA), and the reaction profile comprised an initial step of 2 min at 96 °C and 35 cycles of 30 s at 96 °C, 15 s at 55 °C, and 4 min at 60 °C. Products were directly sequenced using an ABI 3730 capillary sequencer, according to the manufacturer’s instructions.

2.3. Data Analysis

DNA sequences were aligned with SeqScape v.2.1.1 software (Applied Biosystems, Inc.) and further visually double-checked. Barcode sequences were subjected to distance-based, diagnostic character (maximum likelihood), and spectral clustering (BIN) analyses.

2.3.1. Distance-Based Analyses

Distance measures were employed to assess intraspecific, intrageneric, and intrafamily variation. Sequence divergences were calculated using both the p distance and the Kimura two-parameter (K2P) distance model [55] in order to facilitate comparison with previous studies. The nearest-neighbor distance (NND) distribution analysis, i.e., the minimum genetic distance between a species and its closest neighbor species, was also performed using the “barcoding gap analysis” tool from BOLD. Maximum intraspecific distance in axis X was plotted against the minimum distance to the NN for each species in axis Y to infer the presence of a “barcode gap”. Negative values for axis Y showed no resolution in the barcode gap.

2.3.2. Cluster and Phylogenetic Analyses

For cluster analysis, the Tamura 3-parameter with Gamma distribution (T92 + G) model was chosen as it was determined as the best-fit model under the Akaike information criterion for neighbor-joining (NJ) and maximum likelihood (ML) analyses. Phylogenetic relationships among haplotypes were estimated by maximum likelihood (ML) analyses using MEGA 12 [56], with 500 bootstrap replicates to estimate node support values for the resulting phylogeny.

2.3.3. Lineage Delimitation Analysis

The barcode index number (BIN) was used to estimate the number of species directly from the barcode records, and the congruence of these estimates with the distance-based and character-based approaches were evaluated. BIN analysis algorithmically clustered barcode sequences to create operational taxonomic units (OTUs) that closely reflected species groupings. BINs were automatically assigned by BOLD and assessed using the “BIN discordance report” analysis tool [48]. This tool labeled a BIN as “concordant” when it comprised sequences attributed to the same species, and “discordant” when it comprised sequences of different species. BOLD was also used to explore the genetic divergence between barcode records of given species of the Argentine continental shelf and slope with other available barcode sequences for the corresponding species from other oceans.

Lineage limits were also explored using the automatic barcode gap discovery method (ABGD) [49]. The ABGD automatically finds the distance at which a barcode gap occurs and sorts the sequences into putative species based on this distance. Therefore, as in BIN analysis, it is applicable as an independent tool without a priori species hypothesis, and it provides insight into whether the taxonomic identification based on morphological features has any genetic support. The ABGD was run with the default settings (P min = 0.001, P max = 0.1, steps = 10, X relative gap width = 1.5, Nb bins = 20) and K2P distance on the ABGD web server (https://bioinfo.mnhn.fr/abi/public/abgd/). An additional ABGD analysis with a minimum relative gap width of 0.5, and P max of 0.01 was set and run, following Rosso et al. [57]. For those species with low genetic divergence, species boundaries were also explored by using nucleotide diagnostic characters (NDCs).

2.3.4. Comparative Analysis of COI Sequences of All Skates Registered in the SWA (BIN Discordance Report)

The public library of BINs in BOLD was also used to scrutinize whether the literature on the DNA barcoding of skates had incorporated different BINs under unique nominal taxa (i.e., to ascertain taxonomic conflicts among barcode studies conducted by different teams of researchers). For those species sharing the same BIN, we additionally explored their COI sequences for diagnostic characters with the tool available in BOLD.

2.3.5. Population Structure

For those species with large distribution and more than 20 sequences (Amblyraja doellojuradoi and Zearaja brevicaudata), nucleotide (π) and haplotype (h) diversities were calculated for mitochondrial sequence data using the program DnaSP v. 4.10.9 [58]. A median-joining network was generated from mitochondrial haplotypes using the program Network v. 4.0.1.7. Additionally, pairwise fixation indices (FSTs), analysis of molecular variance (AMOVA), and neutrality tests were computed in ARLEQUIN 3.5 [59].

2.3.6. Repository

All sequence assemblies, electropherogram (trace) files, primer sequences, and specimen provenance data were deposited in the “South Western Atlantic Skate” (Project code: SWAS) project in the Barcode of Life Database (BOLD, http://www.boldsystems.org (accessed on 7 February 2025) [60]). This included digital images of morphological voucher specimens, sex and ontogenetic stages (juvenile or adult), total and standard lengths, and GPS coordinates for all specimen collection localities.

3. Results

Overall, 218 samples belonging to 26 different species from families Arhynchobatidae and Rajidae were successfully barcoded. No stop codons, insertions, or deletions were found in any of the amplified sequences, showing that all of them constituted functional mitochondrial COI sequences. Five species were represented by only one sequence. The average nucleotide frequencies were G (16.47%), C (24.24%), A (25.48%), and T (33.819%).

The p distances averaged just 0.19% within species, but averaged 3.52% within genera and 14.45% within families. Similarly, the K2P genetic distances averaged 0.19% within species, 3.65% within genera, and 16.44% within families (

Table 1. a. Summary of distribution of sequence divergence at each taxonomic level (p distance). Dist—distance. b. Summary of distribution of sequence divergence at each taxonomic level (K2P distance).

a
	n	Taxa	Comparisons	Min Dist (%)	Mean Dist (%)	Max Dist (%)	SE Dist (%)
Within Species	211	20	1404	0.00	0.19	1.46	0.00
Within Genus	178	5	2518	0.15	3.52	8.36	0.00
Within Family	217	1	19,514	0	14.45	20.49	0.00
b
	n	Taxa	Comparisons	Min Dist (%)	Mean Dist (%)	Max Dist (%)	SE Dist (%)
Within Species	211	20	1404	0.00	0.19	1.48	0.00
Within Genus	178	5	2518	0.15	3.65	8.98	0.00
Within Family	217	1	19,514	0	16.44	24.57	0.00

). A barcode gap was observed in 86.4% of all analyzed species (Figure 2;

Table 2. Barcode gap analysis showing the distribution of distances within each species and the distance to the nearest neighbor of each species. Where the species is a singleton, N/A is displayed for intraspecific values. Distances are highlighted with an asterisk (*) if the nearest neighbor is less than 2% divergent, or when the distance to the nearest neighbor is less than the max intraspecific distance.

Species	Mean Intra-Sp	Max Intra-Sp	Nearest Neighbor	Nearest Species	Distance to NN
Amblyraja doellojuradoi	0.24	1.12	FARG568-09	Zearaja brevicaudata	11.19
Atlantoraja castelnaui	N/A	0	FARG427-08	Atlantoraja platana	7.89
Atlantoraja cyclophora	0.19	0.31	FARG426-08	Atlantoraja platana	4.22
Atlantoraja platana	0	0	FARG475-08	Atlantoraja cyclophora	4.22
Bathyraja albomaculata	0.09	0.16	FARG117-06	Bathyraja macloviana	0.31 *
Bathyraja brachyurops	0.2	0.47	FARGB450-11	Bathyraja macloviana	2.67
Bathyraja cousseauae	0.15	0.15	FARG501-08	Bathyraja papilionifera	1.24 *
Bathyraja griseocauda	0	0	FARG501-08	Bathyraja papilionifera	0.93 *
Bathyraja macloviana	0.42	0.95	FARG145-06	Bathyraja albomaculata	0.31 *
Bathyraja magellanica	0.18	0.46	CEGAR056-14	Bathyraja sp.	2.71
Bathyraja multispinis	0.04	0.15	FARG501-08	Bathyraja papilionifera	1.55 *
Bathyraja papilionifera	N/A	0	FARG258-06	Bathyraja griseocauda	0.93 *
Bathyraja scaphiops	0.11	0.34	FARG172-06	Bathyraja cousseauae	1.89 *
Bathyraja sp.	N/A	0	FARG172-06	Bathyraja cousseauae	1.61 *
Dipturus argentinensis	0,09	0,31	SWAS007-24	Dipturus trachyderma	0 *
Dipturus trachyderma	N/A	0	FARG019-06	Dipturus argentinensis	0 *
Psammobatis bergi	0.04	0.18	FARG463-08	Psammobatis lentiginosa	2.73
Psammobatis extenta	N/A	0	SWAS001-24	Psammobatis rutrum	4.71
Psammobatis lentiginosa	0.41	1.48	FARG737-09	Psammobatis bergi	2.73
Psammobatis normani	0.15	0.6	FARG479-08	Psammobatis rudis	0.15 *
Psammobatis rudis	0.09	0.33	FARG390-08	Psammobatis normani	0.15 *
Rioraja agassizii	0.3	0.77	FARG475-08	Atlantoraja cyclophora	8.77
Sympterygia acuta	0.25	0.71	FARG215-06	Sympterygia bonapartii	7.36
Sympterygia bonapartii	0.02	0.18	FARG716-09	Sympterygia acuta	7.36
Zearaja brevicaudata	0.17	0.62	FARG127-06	Dipturus argentinensis	2.79

). The absence of a barcode gap (i.e., the distance to the nearest neighbor was smaller than the max intraspecific distance) was observed in Bathyraja macloviana, Dipturus argentinensis, Psammobatis rudis, and P. normani.

3.1. Cluster, Phylogenetic, and Lineage Delimitation Analyses

The NJ and ML tree showed a strong concordance between morphological identification and COI sequence clustering. Indeed, nearly all species formed a cohesive cluster of shared haplotypes, with high bootstrap support (Figure 3 and Figure 4). The only exceptions were the following pairs of species: Psammobatis rudis/P. normani and Dipturus argentinensis/D. trachyderma, whose sequences could not be separated from each other. The ML tree also showed that species were clustered in two main clades corresponding to the families Arhynchobatidae and Rajidae (Figure 4). Moreover, all genera were grouped following tribe assignment (Arhynchobatini, Bathyrajini, Riorajini, Amblyrajini, and Rajini).

The mean interspecific distance (K2P) was 3.65. However, some species exhibited low interspecific divergence (<2%, Table 1). This was reflected in the spectral clustering analysis. Indeed, not all specific clusters were assigned to a different BIN. Moreover, DNA barcoding revealed 17 different OTUs among the 26 species morphologically identified in our analyses (Figure 3). Other than P. rudis/P. normani (BIN BOLD:AAB7451) and Dipturus argentinensis/D. trachyderma (BOLD:AAB5857), most species of Bathyraja (B. albomaculata, B. cousseauae, B. griseocauda, B. macloviana, B. multispinis, B. papilionifera, B. scaphiops, and Bathyraja sp.) were assigned to the same BIN (BOLD:AAA8067).

Conversely, the analysis of the ABGD data set using the default parameters resulted in six recursive partitions that ranged from 76 (P max = 0.001) to 16 groups in two partitions (P max = 0.00129, 0.0215). Two partitions presented 26 groups (P max = 0.0167, 0.00278), and another one 24 (P max = 0.00464). Some of the partitions were consistent with BIN analysis (15 BINs vs. 16 ABGD groups), while the partition with 24 groups reflected much better the taxonomic discrimination. The only pairs of species that could not be resolved in this partition were the following: (i) Psammobatis rudis/P.normani, (ii) Bathyraja macloviana/B. albomaculata, and (iii) Dipturus trachyderma/D. argentinensis. Unlike BIN analysis, ABGD could resolve the discrimination of almost all Bathyrajid species (Figure 3). When the minimum relative gap width was set to 0.5 and the P max to 0.01, analysis resulted in seven recursive partitions that ranged from 81 (P max= 0.001, 0.00129) to 24 groups (P max = 0.00464, 0.00599). The other three partitions presented 27 groups (P max = 0.00167, 0.00215, 0.00278). The partitions with 24 groups showed the same taxonomic discrimination as that found using the default parameters (Figure 3). In addition, each Bathyraja species, other than clustering separately, did not share haplotypes and possessed specific NDCs that allowed the discrimination of all of them (

Table 3. Nucleotide diagnostic characters of Bathyraja species from the Southwest Atlantic.

Species/Site	34	58	85	86	97	163	184	197	208	232	268	298	340	352
Base	G	G	A	A	T	G	C	G	G	T	T	C	C	A
B. albomaculata	.	.	.	.	.	.	.	.	.	.	.	.	.	.
B. cousseauae	.	.	.	.	.	.	.	.	.	.	.	.	.	.
B. griseocauda	.	.	.	.	.	.	.	.	.	.	.	.	.	.
B. macloviana	.	.	.	.	.	.	.	.	.	.	.	.	.	.
B. meridionalis	.	.	.	.	.	.	.	.	.	.	.	.	.	.
B. multispinis	.	C	.	.	.	.	.	A	.	C	.	.	.	.
B. papilionifera	.	.	.	.	.	A	.	.	.	.	.	.	.	.
B. scaphiops	A	.	.	.	.	.	T	.	.	.	.	T	G	T
B. eatoni	.	.	.	.	.	.	.	.	.	.	C	.	.	.
B. maccaini	.	.	G	.	.	.	.	.	.	.	.	.	.	.
Bathyraja sp.	.	A	T	T	G	.	.	.	A	.	.	.	.	G
Species/Site	397	409	415	439	463	469	490	520	529	568	598	603	610	628
Base	G/T	C	C	T	A	A	A	C	A	C	C	C	C	G
B. albomaculata	.	.	.	C	.	.	.	.	.	.	.	.	.	.
B. cousseauae	.	.	.	.	.	.	.	.	.	.	.	.	T	.
B. griseocauda	.	T	.	.	.	.	.	.	.	.	.	.	.	.
B. macloviana	.	.	.	.	G	.	.	.	.	.	.	.	.	.
B. meridionalis	.	.	.	.	.	.	G	.	.	.	T	.	.	.
B. multispinis	A	.	T	.	.	.	.	.	.	T	.	.	.	.
B. papilionifera	C	.	.	.	.	.	.	.	.	.	.	A	.	.
B. scaphiops	.	.	.	.	.	.	.	T	.	.	.	.	.	A
B. eatoni	.	.	.	.	.	C	.	.	.	T	.	.	.	T
B. maccaini	.	.	.	.	.	.	.	.	T	.	T	.	.	.
Bathyraja sp.	.	.	.	.	.	.	.	.	.	.	-	-	-	-

). One of the groups retrieved in all the analyses corresponded to Dasyatis hypostigma (outgroup); therefore, ABGD recovered twenty-three skate groups.

Even though Psammobatis rudis and P. normani could not be separated in the ML or NJ tree, neither by BIN nor ABGD analysis, a deep assessment of their sequences showed no shared haplotypes between them and presented one NDC that allowed to differentiate them. Site #456 was always distinctive for each species, being represented by a C in P. rudis and a T in P. normani.

3.2. Comparative Analysis of COI Sequences of All Skates Registered in SWA (BIN Discordance Report)

Considering the available public data in BOLD, 30 species from SWA had barcodes. Overall, 17 BINs were recovered for these species (

Table 4. Summary of Southwest Atlantic skate species with available sequences in BOLD, and main information containing each BIN (barcode index number). * This is a misidentification detected in BOLD. Atlantoraja cyclophora has its own BIN. See details when describing the BIN AAB1883.

Species	BIN	Average/Max Distance P-Dist (%)	Distance to the Nearest P-Dist (%)	Nearest Member	BIN
Amblyraja doellojuradoi Amblyraja georgiana	ABZ5141 210 [186 public]	0.43%/2.4%	1.94%	Amblyraja radiata	AAA4500
Amblyraja taaf
Atlantoraja castelnaui	AAB4961 35 (29)	0.11%/0.49%	7.13%	Atlantoraja platana	AAB1884
Atlantoraja cyclophora	AAB1882 36 (29)	0.26%/0.96%	3.56%	Atlantoraja platana	AAB1884
Atlantoraja platana	AAB1884 29 (24)	0.06%/0.336%	3.94%	Atlantoraja cyclophora	AAB1882
Bathyraja albomaculata	AAA8067 307 (256)	1.86%/4.65%	1.92%	Bathyraja interrupta	ABY6293
Bathyraja cousseauae
Bathyraja eatonii
Bathyraja griseocauda Bathyraja maccaini
Bathyraja macloviana
Bathyraja meridionalis
Bathyraja multispinis
Bathyraja papilionifera
Bathyraja scaphiops
Bathyraja sp.
Bathyraja brachyurops	ABX5438 9 (9)	0.19%/0.47%	2.23%	Bathyraja sp.	AAA8067
Bathyraja magellanica	AAF0384 5 (5)	0.18%/0.46%	2.23%	Bathyraja sp.	AAA8067
Dipturus argentinensis	AAB5857 216(210)	0.61%/2.34%	2.69%	Dipturus sp.	AAB4430
Dipturus trachyderma
Psammobatis bergi	AAF6480 10 (9)	0.03%/0.18%	2.66%	Psammobatis lentiginosa	AAD8610
Psammobatis extenta	AGQ0364 (1)	N/A
Psammobatis lentiginosa	AAE0763 10 (1)	0.18%/0.35%	4.59%	Psammobatis normani	AAB7451
Psammobatis normani	AAB7451 83 (69)	0.26%/1.28%	3.41%	Psammobatis lentiginosa	AAD8610
Psammobatis rudis
Rioraja agassizii	AAB1883 29 (23)	0.34%/1.12% Rioraja agassizii (26), Atlantoraja cyclophora (3) *	8.06%	Atlantoraja cyclophora	AAB1882
Sympterygia acuta	AAD5559 23 (23)	0.24%/0.77%	7.54%	Sympterygia bonapartii	AAE2829
Sympterygia bonapartii	AAE2829 13 (13)	0.02%/0.18%	3.93%	Sympterygia brevicaudata	ADN1186
Zearaja brevicaudata	AAB5856 94 (43)	0.24%/1.5%	2.7%	Dipturus argentinensis	AAB5856

). Information regarding distance values and the nearest neighbor of each BIN is also displayed in Table 4. The BIN discordance analysis, taking into account not only our observations but also other barcode data in BOLD, showed a priori 6 discordant BINs, 10 concordant BINs, and 1 singleton BIN (Table 4). The discordant BINs were BOLD:ABZ5141, BOLD:AAA8067, BOLD:AAB7451, BOLD:AAB5857, BOLD:AAB5856, and BOLD:AAB1883. A closer analysis within each BIN showed different reasons for the observed discordance.

BIN BOLD:ABZ5141 included specimens from six different Amblyraja species (Amblyraja hyperborea, A. jenseni, A. doellojuradoi, A. georgiana, A. badia, A. taaf), alongside specimens identified as Bathyraja spinicauda and Cruriraja durbanensis. Probably, the latter corresponded to contaminations or mislabeling, considering that Amblyraja, Bathyraja, and Cruriraja belonged to families Rajidae, Arhynchobatidae, and Gurgesiellidae, respectively. In addition, specimens of B. spinicauda were included in another BIN (BOLD:AAA8067) and, although there were no other barcodes for C. durbanensis, Cruriraja species were included in two different BINs (BOLD:AAE9408 for C. parcomaculata and BOLD:AAB7055 for C. hulleyi). On the other hand, the presence of several species of Amblyraja indicated that the BIN algorithm was not sufficient to discriminate them. A deep analysis of sequences from the species of SWA (A. doellojuradoi, A. georgiana, and A. taaf) showed sites that were species-specific (NDC) (

Table 5. Nucleotide diagnostic characters for Amblyraja species from the Southwest Atlantic.

Species/Site	178	316	337	346	532	577	595
Amblyraja. doellojuradoi Amblyraja georgiana	G A	CG T	C T	A G	A G	C A	T C
Amblyraja taaf	A	T	C	G	G	A	C

).

Similarly, BIN BOLD:AAA8067 included specimens from 22 different Bathyraja species (Bathyraja abyssicola, B. albomaculata, B. aleutica, B. cousseauae, B. eatonii, B. griseocauda, B. irrasa, B. leucomelanos, B. maccaini, B. macloviana, B. meridionalis, B. multispinis, B. murrayi, B. papilionifera, B. scaphiops, B. shuntovi, B. spinicauda, B. spinosissima, B. richardsoni, B. taranetzi, B. violacea), specimens identified as Bathyraja n. sp. dwarf, Bathyraja n. sp. blond, Bathyraja sp., Bathyraja sp. P, Bathyraja n. sp. eatonii, and Rhinoraja longicauda. The latter probably represented a contamination or mislabeling because specimens of R. longicauda were included in a different BIN. Sequences of Bathyraja n. sp. dwarf came from Smith et al. [31] and corresponded to a recently resurrected species, Bathyraja arctowskii (Dollo 1904) [61]. Regarding the sequences of Bathyraja sp., one of them corresponded to B. arctowskii as it clustered with sequences of Bathyraja n. sp. dwarf. The other three sequences, called “Bathyraja sp. P”, clustered together with specimens of B. murrayi. Finally, sequences named Bathyraja n. sp. eatonii and Bathyraja n. sp. blond came from Smith et al. [31] and, as pointed out by the aforementioned authors, they probably corresponded to two undescribed species of Bathyraja from the Southern Ocean, since they each formed one cohesive cluster distinct from all other Bathyraja species. The pairwise distribution in BIN BOLD:AAA8067 was very large in comparison with other BINs, and it overlapped with the pairwise distribution of the nearest neighbor that corresponded to B. interrupta (BIN BOLD:ABY6293). In this case, BIN analysis was inconsistent with the species.. A deep analysis of sequences from the species of SWA (Bathyraja albomaculata, B. cousseauae, B. eatonii, B. griseocauda, B. maccaini, B. macloviana, B. meridionalis, B. multispinis, B. papilionifera, B. scaphiops, B. eatonii, and B. maccaini) showed several sites that were species-specific (NDC) (Table 3). In this case, the BIN analysis was not sufficient to effectively discriminate these Bathyraja species.

BIN AAB1883 mainly included specimens of Rioraja agassizii from Argentina (Mabragaña et al. [35] and present study), from Brazil (public BIN in BOLD), and some Atlantoraja cyclophora from Brazil (Ribeiro et al. [37]). However, A. cyclophora had its own BIN (AAB1882), containing specimens from both Argentina (Mabragaña et al. [35] and present study) and Brazil (Public BIN in BOLD, Tinti unpublished data). Unfortunately, no photographs of specimens of A. cyclophora from Ribeiro et al. [37] were available. In addition, no sequences of R. agassizi have been mentioned [37]. Therefore, in this case, the species did not have two different BINs, rather there was a misidentification. Summarizing this BIN could be reclassified as concordant.

BIN BOLD:AAB5857 included sequences of Dipturus argentinensis, D. trachyderma (from Chile and Argentina), Zearaja chilensis from Chile, Z. nasuta from New Zealand, and Z. maugeana from Australia. This BIN contained a total of 270 sequences, 234 of which were mined from GenBank and corresponded to all species except Z. nasuta. Sequences of D. argentinensis clustered together with a sequence of D. trachyderma from Argentina and two from Chile (mined from Genbank). This cluster also included sequences of Dipturus sp. from Argentina and an unknown locality. These sequences were mined from GenBank, where they were identified as D. argentinensis. On the other hand, the remaining sequences of D. trachyderma from Chile constituted three different clusters. Sequences of Z. chilensis formed two clusters, one of which included sequences of D. maugeanus. These sequences were also mined from GenBank. Finally, sequences of Z. nasuta formed a cohesive cluster. Unfortunately, for those sets of sequences mined from Genbank, no available photos were available to allow specimen identification. These results would indicate that, for these species, the BIN algorithm was not sufficient to discriminate them. In addition, BIN BOLD:AAB5856 included specimens identified as Zearaja brevicaudata, Zearaja chilensis, Dipturus brevicaudatus, and D. chilensis. This BIN contained a total of 94 sequences, 56 of which were mined from GenBank. All sequences came from specimens from SWA. None of these specimens were labeled as different species form cohesive clusters (they were mixed).

BIN BOLD:AAB7451 included specimens of Psammobatis normani and P. rudis, and no other BIN existed for any of these species, indicating that, in this case, the BIN algorithm was not sufficient to discriminate them. As was previously stated, NDC was the only molecular way to discriminate them, using COI sequences. This BIN also included public sequences of P. scobina from Chile (mined from Genbank), and three sequences identified as Psammobatis sp. from the Malvinas Islands. None of these specimens had photographs or additional data. None of these species formed cohesive clusters (they were mixed). A deep analysis of the public sequences showed that, from the 39 specimens of P. scobina, 21 shared the specific NDC for P. normani in site #456; from which, 14 of them shared the same sequence haplotype of P. normani from SWA. On the other hand, six specimens possessed the specific NDC of specimens of P. rudis from SWA. Finally, 12 P. scobina sequences had an exclusive NDC, a C in site #603, and all of them shared the same haplotype. As no photographs were available for P. scobina specimens, these results could indicate that, within the sequences identified as “P. scobina”, sequences of P. normani and P. rudis from Chilean waters were also included. Regarding the specimens from the Malvinas Islands, one of the sequences possessed the specific NDC of P. rudis and the remaining two had the specific NDC of P. normani.

3.3. Population Structure of Some Widespread Species

Multiple records in two species (Amblyraja doelojuradoi and Zearaja brevicaudata) were obtained from different sites in SWA. Therefore, they were assessed for evidence of phylogeographic signals in the COI sequence variation. For each species, three groups were generated (

Table 6. Summary of genetic parameters for the mtDNA COI by species, divided by zones. References: n—number of samples; Nh—number of haplotypes; S—number of segregating sites; Hd—haplotype diversity and Tajima’s D with p-values. * Significant differentiation at 5% level.

Species	Group	n	Nh	S	Hh	Tajima’s D	p-Value
Amblyraja doellojuradoi	Buenos Aires shelf	10	5	6	0.6667	−1.79631 *	0.015
	Buenos Aires deep	3	2	1	0.6667	0.00001	0.984
	Patagonian shelf	9	4	2	0.4167	−0.58325	0.303
Zearaja brevicaudata	Buenos Aires shelf	12	4	4	0.6818	−0.74109	0.259
	San Jorge Gulf shelf	7	4	3	0.8	0.60031	0.705
	Patagonian shelf	6	2	1	0.4762	0.55902	0.844

). Amblyraja doellojuradoi showed 10 COI haplotypes and three groups were generated as follows: (i) Buenos Aires shelf, (ii) Buenos Aires deep, and (iii) Patagonian shelf. The haplotype genealogy showed one ancestral haplotype shared by all three populations, with a substantial number of private haplotypes found in every population. Only one haplotype was shared between the three populations (Figure 5A). AMOVA showed no significant global differentiation (Fst = 0.0162, p = 0.441) for the full data set. Neutrality tests based on Tajima’s D were negative for the Buenos Aires shelf population, indicating balancing selection or population growths. In the other populations, the Tajima’s D values were not significant (

Table 7. Pairwise population differentiation. Genetic differentiation (FST) of Amblyraja doellojuradoi populations (A) and Zearaja brevicaudata populations based on COI. BAS—Buenos Aires shelf; BAD—Buenos Aires deep; PS—Patagonian shelf; SJSG—San Jorge Gulf shelf. * Significant differentiation at 5% level.

A		BAS	BAD	PS
	BAS	-
	BAD	0.05263	-
	PS	−0.06557	0.02567	-
B		BAS	SJSG	PS
	BAS	-
	SJSG	0.13216 *	-
	PS	−0.03526	0.23179 *	-

).

For Zearaja brevicaudata, six COI haplotypes were identified. AMOVA showed significant global differentiation (Fst = 0.189, p = 0.015) for the full data set. Three groups were generated as follows: (i) Buenos Aires shelf (ii) San Jorge Gulf shelf, and (iii) Patagonian shelf. The haplogroups showed a slightly deep star-like topology, indicating recent expansion. The haplotype network showed two dominant haplotypes, separated by only two mutational steps. The most common haplotype represented 12 individuals from the three groups. The haplotype derived from this one was separated by only two mutational steps (from the San Jorge Gulf shelf). The second most common haplotype (two mutational steps away from the first one) was represented by seven individuals. This haplotype had three derived private haplotypes (two from the Buenos Aires shelf and one from the San Jorge Gulf shelf) (Figure 5B). COI data showed significant pairwise population differentiation, with the highest fixation index found between the Buenos Aires shelf and the San Jorge Gulf shelf (Fst = 0.133), and between the Patagonian shelf and the San Jorge Gulf shelf (Fst = 0.231), both values were significant (p < 0.05). A lower index was found for the Buenos Aires shelf vs. the Patagonian shelf, with Fst = −0.03, not significant. The Tajima’s D neutrality test ranged between 0.55 and 0.74 for all three groups regardless of assignment, and none were significant (Table 7).

3.4. Skate Egg Case Identification

Thirty-one egg cases corresponding to a priori eight species and two unidentified species collected from benthos containing embryos in different stages of development were identified by means of DNA barcoding (Figure 6). Molecular results agreed with previous morphological identification (

Table 8. Skate egg cases collected from the Southwest Atlantic identified by means of DNA barcoding.

Prior Morphological Identification	Sample Size	Species Matched
Amblyraja doellojuradoi	3	Yes
Bathyraja brachyurops	2	Yes
Psammobatis lentiginosa	4	Yes
Psammobatis normani	5	Yes
Psammobatis rudis	3	Yes
Rioraja agassizii	1	Yes
Sympterygia acuta	2	Yes
Zearaja brevicaudata	9	Yes
Unidentified	1	Rioraja agassizii
Unidentified	1	Bathyraja sp.

). However, one egg case collected on the beach could not be identified because it lacked any diagnostic character, but DNA barcoding allowed its assignment to Rioraja agassizii. In addition, one undescribed egg case, collected from the continental slope in the Mar del Plata Canyon, was assigned to the genus Bathyraja within BIN BOLD:AAA8067, but could not be assigned to any barcoded species at the regional or global level; the closest match was B. spinosissima (98.92%), a species from the North Atlantic Ocean [4]. The egg case was relatively large; its length (without horns) was 98.4 mm and its width was 59.2 mm. The capsule’s surface had a rough texture to the touch and lacked attachment fibers. An embryo at a middle developmental stage (st. 28/29 following Vazquez et al. [62]) was found inside the egg case (Figure 6C).

4. Discussion

Fish barcoding studies in Argentine waters started shortly after the use of a portion of the COI gene as a global system to identify organisms was proposed [63] and were led by Dr. Díaz de Astarloa at the FCEyN, University of Mar del Plata, Mar del Plata, Argentina. The first surveys were carried out almost exclusively in marine and estuarine habitats, and allowed—after three years of field surveys—the molecular identification of 125 taxa, including 21 skate species [35]. During the following years, new collections of fishes were studied, including in freshwater [57,64] and Antarctic habitats [65]. After 20 years of surveys, an ichthyological collection with more than 5000 specimens belonging to more than 700 species was built at the Instituto de Investigaciones Marinas y Costeras (FCEyN, Universidad Nacional de Mar del Plata, CONICET). This effort resulted in 100 additional sequences from 17 skate species, 5 of which had not been previously registered by Mabragaña et al. [35], including Psammobatis extenta, P. bergi, and Bathyraja sp., barcoded ex novo in the present study, totalizing 26 SWA skate species with their barcode sequence. Other initiatives in Atlantic and Antarctic waters (e.g., [19,31,37,44]) provided DNA barcodes for skate species occurring in the area. In this sense, when compiling our results with these data available from BOLD, currently about 30 skate species inhabiting SWA have barcodes, which represent almost all of the species registered in the area. Some of these species (Amblyraja taaf, A. georgiana, Bathyraja eatonii, and B. maccaini) lack barcodes derived from samples from SWA but have been barcoded from the Southern Ocean [31,60].

As pointed out in the results section, the phylogenetic analysis (ML) showed that all analyzed species were clustered in two main clades corresponding to the families Arhynchobatidae and Rajidae, and genera were grouped following tribe assignment (Arhynchobatini, Bathyrajini, Riorajini, Amblyrajini, and Rajini), in agreement with the results obtained by Crobe et al. [44], combining COI with Nd2 sequences. Even though mitochondrial marker (COI) is not routinely used as a reliable marker for phylogenetic relationships, the results obtained in this work are in accordance with current systematic classification in skates [46].

Distance-based and cluster analysis coupled with lineage delimitation approaches using COI (BIN, ABGD) allowed the identification of almost all of the skate diversity recorded in SWA. Furthermore, pairs of species with very low genetic divergence and no discrimination using the aforementioned analysis (Bathyraja macloviana/B. albomaculata and Psammobatis rudis/P. normani) could be identified by NDC. The only pair of species that could not be discriminated from each other using COI were D. trachyderma from D. argentinensis. This issue will be discussed later. Although there was a 19-fold more pronounced difference among congeneric species than among conspecific individuals, low sample sizes in some species may have biased genetic distance calculations and reduced the reliability of intraspecific variation estimates. In this sense, it is important to note that some species represented by a single specimen in this study (e.g., A castelnaui, P. rutrum) had several barcodes available in BOLD and their intraspecific distance was within the average range observed in our data.

Groups generated by ABGD better reflected the nominal species than BIN analysis. Indeed, ABGD could resolve the distinction of 20 of the 26 species analyzed, whereas BIN could resolve only 14. This was particularly evident within the genus Bathyraja, in which several species of the genus were included in the same BIN, whereas ABGD could resolve the discrimination of almost all Bathyrajid species. Similar results were obtained by Crobe et al. [44] for Atlantic skates. However, other studies found that the BIN system outperformed ABGD in discriminating closely related species [48,66]. Among all the analyses performed in the present work, BIN turned out to be the most conservative. Particularly, our data showed three discordant BINs, including species of genera Bathyraja, Psammobatis, and Dipturus. Furthermore, if all data available in BOLD were included, the BIN discordance report showed a priori six discordant BINs named BOLD:ABZ5141, BOLD:AAA8067, BOLD:AAB7451, BOLD:AAB5856, BOLD:AAB5857, and BOLD:AAB1883. From the assessment of the BIN discordance report, relevant taxonomic and nomenclatural issues can be highlighted.

Relatively low genetic divergences among Amblyraja spp. (BOLD:ABZ5141) and some Bathyraja species (BOLD:AAA8067) were previously reported for North Pacific species of these genera [30] and also for Antarctic Bathyraja species [31], suggesting recent speciation given the low mutation rate of COI. Currently, 10 species of Amblyraja are recognized. Weigmann [5] highlighted the possible conspecifics of several Amblyraja species (A. badia, A. frerichsi, A. georgiana, A. jenseni, A. reversa, A. robertsi, and A. taaf) with A. hyperborea based on the almost non-existence of interspecific morphological differences among them. This issue leads to the need to delve deeper into integrative taxonomic studies on these species distributed worldwide in cold deep waters. Low genetic divergence between phylogenetic closely related species has been reported for many neotropical fishes [57,67,68].

Longnose skate genera Zearaja and Dipturus were grouped into two BINs. One of them (BOLD:AAB5856) contained sequences from SWA identified as Zearaja brevicaudata and Z. chilensis, but a careful revision of specimens behind these sequences revealed that all of them corresponded to the former species. For many years, Z. chilensis was believed to inhabit both the Southwest Atlantic and Southeast Pacific Oceans. But, recently, through a molecular and morphological study, Gabbanelli et al. [39] found that the specimens from both oceans corresponded to different species, leading to the resurrection of Z. brevicaudata from the synonym of Z. chilensis. Therefore, in this BIN there were sequences that were identified as Z. chilensis, before Gabbanelli et al. [39], but they actually corresponded to Z. brevicaudata. In addition, there were sequences labeled as Dipturus brevicaudatus and D. chilensis, which obeyed to the existing controversy about the validity of these genera, which is still unresolved [3,40,69,70]. In the present study, Zearaja is considered as the valid genus, according to the Catalogue of Fish [47]. Therefore, this BIN could be reclassified as concordant. On the other hand, Zearaja brevicaudata, which possesses a wide distributional range in SWA, showed populational structuring at the COI level. Even though COI is not a populational marker, our results are quite in concordance with those provided by Irigoitia et al. [71], who suggested the presence of three different stocks for this species based on parasites assemblages: one in the northern Argentine Sea, and two from north and south Patagonian waters.

Moreover, several species of Dipturus and Zearaja were included in BIN BOLD:AAB5857. Sequences of D. argentinensis and D. trachyderma clustered together consistently with NJ and ML analyses. At the same time, these species could not be separated by ABGD analysis. Recently, these species were synonymized by Figueroa et al. [19] and they assigned D. argentinensis as a junior synonym of D. trachyderma. This was made on the basis of two mitochondrial markers (COI and nd2) and limited morphological studies. In this respect, it is important to note that strong morphological differences do exist between these species [17,72]. Dipturus argentinensis presents dermal denticles restricted to the snout and one row of caudal thorns, while D. trachyderma presents a large amount of dermal denticles covering both surfaces of the body and five rows of caudal thorns. It is remarkable that, although morphological characteristics of adult specimens of D. argentinensis are not known, these features allowed to differentiate both species in juvenile stages and similar sizes. In this regard, the authors compared only adult and preadult specimens of D. trachyderma with juvenile specimens of D. argentinensis, and synonymized the two species without an exhaustive analysis of the morphology and without addressing the sudden change in dermal denticles and thorn patterns that would be expected to happen if they were conspecific [19]. Furthermore, it was previously reported that mitochondrial markers were not enough to discriminate between closely related fish species, i.e., the silverside fishes of Odontesthes (Atherinopsidae) [68], contrasting with genomic analysis, which, in fact, allowed its discrimination [73,74]. Therefore, the hypothesis of a recent speciation should be taken into account and explored through comprehensive integrative taxonomic studies, comparing specimens covering the entire ontogenetic range (from juvenile to adults), through morphogeometric analyses and nuclear markers in order to evaluate the co-specificity of these species.

In this study, the DNA barcodes of Psammobatis extenta and P. bergi were sequenced and published for the first time, thus adding to previous efforts [35,44], and allowing us to recover the COI sequences for all the species of this genus present in SWA. These species clustered independently from each other in the ML, NJ, BIN, and ABGD analyses, except for P. rudis, and P. normani, which clustered together in all cases, as previously observed by Mabragaña et al. [35]. Although the sequences of these two species were almost indistinguishable from each other, they did not share common haplotypes and the NDC found in #456 enabled correct identification. Additionally, morphological [75] and egg case characteristics validated P. rudis and P. normani as different nominal species [76,77]. Naylor et al. [28] provided sequences of the Nd2 gene for several specimens of Psammobatis collected in the Malvinas Islands, identified only to the genus level. Considering the location site and geographic distribution of species from this genus [78,79,80], these specimens probably corresponded to P. rudis or P. normani. No additional genetic markers were available for skates of the genus Psammobatis. On the other hand, P. rudis and P. normani were also distributed in the Pacific Ocean alongside P. scobina [81], and BIN analysis showed that the three species shared the same BIN. However, P. scobina sequences were mined from GenBank, and no images were available, thus leading to possible misidentifications. In this sense, NDC analysis seemed to indicate that only the 12 sequences identified as P. scobina that had NDC in site #603 actually corresponded to that species, while the other 27 were misidentifications with P. rudis and P. normani specimens from Chilean waters. The results obtained in the present work highlight the need for further molecular studies, principally to discriminate P. rudis, P. normani, and P. scobina, and lead to a hypothesis of recent speciation for these species.

Finally, all analysis performed with our data (even BIN analysis) allowed the identification of Rioraja agassizii and Atlantoraja cyclophora. However, some inconsistency was observed when including additional data from BOLD. Indeed, the BIN assigned to R. agassizii was discordant due to the presence of sequences of A. cyclophora, which appeared to be a misidentification. On the one hand, A. cyclophora had its own private BIN (BOLD:AAB1882), containing specimens from both Argentina (Mabragaña et al. [35] and present study) and Brazil (Public BIN in BOLD, Tinti unpublished data). On the other hand, juveniles of both species shared some color features, which could easily lead to misidentification of the specimens [82,83]. Unfortunately, no photographs of specimens of A. cyclophora nor sequences of R. agassizii from Ribeiro et al. [37] were available. Therefore, in this case, it was more likely a misidentification rather than the species having two different BINs. Considering this information, this BIN can be reclassified as concordant.

The probable misidentifications detected in some BOLD records during the course of this survey highlight the necessity for combined efforts to continuously maintain and improve the BOLD reference library. Several records were taken from GenBank and lacked images and GPS positions or even did not have voucher specimens deposited in ichthyological collections. This issue was fundamental, because database inaccuracies might affect not only the possibility for resolving taxonomic issues but also broader applications (e.g., traceability of fishery products [84,85,86], or eDNA studies [87]).

The barcode analysis that was performed was a useful tool that contributed to the taxonomic identification of skate egg cases from the Southwestern Atlantic and the identification of an egg case without clear taxonomic characters. Moreover, it allowed assigning to the genus Bathyraja an undescribed deep sea egg case. In this regard, three species of Bathyraja were known to occur in the deep waters of SWA, B. meridionalis, B. papilionifera, and B. schroederi. The DNA barcode of the latter species was unknown; however, the sequence for the Bathyraja sp. did not match B. meridionalis or B. papilionifera. In addition, the known distribution of B. schroederi in SWA included the continental slope off Uruguay and the Malvinas Islands [88]. Finally, B. schroederi was a medium-sized skate (maximum size: 130 cm total length) [4]. All this information suggests that the unknown egg case may correspond to B. schroederi. This taxonomic issue was discussed by Vazquez et al. [88], but, at that time, no molecular data were available to reach a conclusion. Finally, DNA barcoding confirmed that the A. doellojuradoi species lays egg cases at the limit of its known depth distribution, supporting the existence of a nursery site for this species, as indicated by Vazquez et al. [89].

Deficiencies in catch statistics, such as those reported for SWA elasmobranchs [12,90], constitute a serious problem for both scientists and fishery managers to obtain a true picture of the status of skate species. One of the limiting factors is the lack of accurate identification of morphologically similar species. In this respect, DNA barcoding emerges as a useful molecular tool for species identification. Taking into account that almost all SWA species have barcodes, incorporating this technique as a regular step in fishery management will result in an improvement in catch data records and, hence, in fishery statistics. Moreover, DNA barcoding has resulted in a powerful technique to reveal the bycatch of endangered batoid species in Brazilian waters [90], and for the identification of illegal shark and ray fin trading in Australian waters [33]. On the other hand, DNA barcoding has proven valuable for the identification of seafood mislabeling in different countries [84,90,91,92,93,94,95], including the use of threatened skates and sharks (e.g., Atlantoraja castelnaui, Galeorhinus galeus, Mustelus schmitii) as replacement bony fishes in Argentine markets [84]. The mislabeling of fish species has not only financial implications but also could be a significant threat to human health and the conservation of threatened species [93,96,97].

Several studies emphasize the importance of well-curated and comprehensive DNA barcode reference libraries. In this respect, many initiatives throughout the world have contributed to the global barcode reference sequence library for fishes (FiSHBOL) within BOLD, including the Northeast Atlantic and Bering Sea [36,38], Northwest Atlantic [34], Bering Sea [30], Mediterranean Sea [43], Southwest Pacific (Australia) [32], Southwest Atlantic [35,37], and Southern Ocean [31]. Recently, Crobe et al. [44] established the ELASMO ATL project within the Barcode of Life Data System (BOLD), providing a valuable resource for accurate species identification and evolutionary studies of Atlantic skates. Similarly, our efforts in compiling a substantial ichthyological collection and generating DNA barcodes for SWA skates contribute to this global endeavour, facilitating more precise taxonomic assessments and conservation strategies.

5. Conclusions

Compiling our results with the available data from BOLD, about 30 skate species inhabiting SWA have barcodes sequences, representing almost all the species occurring in the area. Moreover, almost all species may be identified using this short fragment of mitochondrial DNA (COI). These results highlight the utility of DNA barcoding as a reliable complement for skate biodiversity studies, based on the identification of specimens or egg cases, hence supporting classical taxonomy. On the other hand, a complete and curated reference library constitutes a basic and necessary tool not only for resolving taxonomic issues but also for the traceability of fishery products, and for biomonitoring biodiversity through eDNA studies.