All tissue samples were extracted from specimens that were collected as part of other biological studies carried out with appropriate permissions from local authorities: Comisión de Investigaciones Científicas de la Provincia de Buenos Aires, Argentina. (CIC, Ref. Leg. 56/S06); Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET, Fdo iBOL Argentina/2008) and Universidad Nacional de Mar del Plata, Argentina (UNMdP, EXA 407/08).

A total of 577 specimens of fishes collected in Argentine waters, representing 125 species, 98 genera, and 63 families were analyzed. 77 of these species were collected during research cruises conducted by the National Institute for Fisheries Research and Development (INIDEP) in the Argentine Sea, whereas the other 49 species came from researches conducted by the Laboratorio de Ictiología of Universidad Nacional de Mar del Plata in Mar Chiquita coastal lagoon and off Mar del Plata coast (Figure 1, Table S1). Methods followed recommendations of FISH-BOL [24]. Vouchers were morphologically identified following the identification reliability levels 1 and 2 according to the Fish-BOL collaborator's protocol [25]; Level 1: “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” and Level 2: “identification made with high degree of confidence at all levels—specimen identified by a trained identifier who had prior knowledge of the group in the region or used available literature to identify the specimen”. Some of the specimens were deposited in the fish collection of INIDEP; the other specimens studied were kept only as e-voucher (See Table S1). All sequence assemblies, electropherogram (trace) files, primer sequences and specimen provenance data were deposited in the publicly accessible “Fishes of Argentina” Project (code FARG) on the Barcode of Life Database (BOLD, [44]). This included digital images of morphological voucher specimens, sex and ontogenetic stage (post larval, juvenile or adult), total and standard length as well as GPS coordinates for all specimen collection localities.

Muscle tissue samples were taken from whole specimens and genomic DNA extracted according to the protocol of Ivanova et al. [45]. Amplification of the 5′ barcode region of COI was first attempted using FF2d_t1/FR1d_t1 primer combination and C_FishF1t1/C_FishR1t1 primer cocktails [46]. The primer combination C_FishF1t1 contained two primers (FishF2_t1/VF2_t1), and C_FishR1t1 also contained two primers (FishR2_t1/FR1d_t1) (Table 1). All primers were appended with M13 tails to facilitate sequencing. PCR reactions were performed in 96-well plates. The reaction master mix consisted of 825 µl water, 125 µl 10× 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 (5 U/µl) was prepared for each plate, and each well contained 10.5 µl mixture and 2 µl genomic DNA. The PCR reaction profile was comprised of an initial step of 2 min at 95°C and 35 cycles of 30 sec at 94°C, 40 sec at 52°C, and 1 min at 72°C, with a final extension at 72°C for 10 min. Amplicons were visualized on 2% agarose E-Gel® 96-well system (Invitrogen). Sequencing reactions used M13 forward and reverse primers using the BigDye® Terminator v.3.1 Cycle Sequencing Kit (Applied Biosystems, Inc.), and the reaction profile was comprised of an initial step of 2 min at 96°C and 35 cycles of 30 sec at 96°C, 15 sec at 55°C, and 4 min at 60°C. Products were directly sequenced using an ABI 3730 capillary sequencer according to manufacturer's instructions. For specimens that failed to amplify using the primer combinations above, C_VF1LFt1/C_ VR1LRt1 primer combinations [46] were tried. C_VF1LFt1 consisted of four primers (VF1_t1/VF1d_t1/LepF1_t1/VFli_t1), and C_VR1LRt1 also comprised four primers (VR1_t1/VR1d_t1/LepR1_t1/VRli_t1) (Table 2). The PCR reaction profile consisted of 1 min at 95°C and 35 cycles of 30 sec at 94°C, 40 sec at 50°C, and 1 min at 72°C, with a final extension at 72°C for 10 min. All other procedures followed those above. DNA sequences were aligned with SeqScape v.2.1.1 software (Applied Biosystems, Inc.). Sequence divergences were calculated using the Kimura two parameter (K2P) distance model [47], and unrooted NJ phenograms based on K2P distances were created using BOLD. K2P model was used because data set covers a large range of taxa spanning many orders and mtDNA is subject to mutational saturation at this level. Eventhough there are several distance models that take into account this issue, K2P is one of the simplest and commonest model used for describing differentiation among species using COI. On the other hand, being K2P the standard model used in barcode studies allows a better comparison with other barcode studies. Other NJ trees analysing individual genera were constructed using MEGA 4.1 [48] and were bootstrapped 500 times to provide percentage bootstrap values for branch points. Sequence data are available on both BOLD and GenBank (Table S2).

No stop codon, insertions or deletions were found in any of the amplified sequences, showing that all of them constitute functional mitochondrial COI sequences. Of the 651 bases, 354 nucleotid sites were variable and most substitutions occurred in the third nucleotide position within codons (61.3%). All the amplified sequences were larger than 600-bp, the limit typically observed for NUMTs (nuclear DNA sequences originating from mtDNA) [49]. As reported by Ward et al [24], in FISHBOL review, NUMTs do not appear to be a concern for fish barcoding.

A total of 577 specimens, representing 125 species, 98 genera and 63 families were barcoded (Table S1). Average nucleotide frequency were C (27.25%), T (30.74%), A (24.04%) and G (17.97%). Nearly all species exhibited unique barcode haplotypes or cohesive clusters of very closely related haplotypes, which permitted the discrimination of at least 95.2% of species (Appendix S1). The K2P genetic distances averaged just 0.23% within species, but averaged 4.04% within genera, 16.99% within families, 24.07% within orders and 25.21% within classes, increasing value with taxonomic level (Figure 2). Hence, overall, there was a 17-fold more pronounced difference among congeneric species than among conspecific individuals.

Cases of shared haplotypes were detected in six of the species analyzed, including the following pairs: the skates Psammobatis rudis and P. normani, the anchovies Engraulis anchoita and Anchoa marinii, and between the oreos Allocyttus verrucosus and Pseudocyttus maculatus (Figures 3, 4). Results observed within the genus Psammobatis will be discuss later. The anchovies Engraulis anchoita (n = 3) and Anchoa marinii (n = 3), could not be separated by COI even though they are two valid species of Engraulidae. These species have external morphological features that allow differentiating each other; however when specimens are not well preserved it is more difficult to identify them from images. Unfortunately these specimens were not retained and only the e-vouchers (digital images) have been recorded to reliable re-checked the identification. From images it can be seen that specimens identified as Engraulis anchoita clearly represent this species because of the broad silver stripe along flank, the large size of the specimens (>14 cm TL), and the large eye size in relation to snout. However those specimens identified as Anchoa marini are not well preserved (most lateral scales are missing, and both dorsal and anal fins can not be recognized), so it is difficult from images to identify them properly. Therefore, the lack of discrimination at molecular level between these pairs of species is likely a problem of misidentification. Regarding the zero divergence between Oreosomatid species from Argentina, this case is likely to be a misidentification. Allocyttus verrucosus and Pseudocyttus maculatus are the Oreosomatid species recorded in the southwest Atlantic [33], [50], [51]. Re-examination of e-vouchers allow identifying both specimens as Allocyttus verrucosus, because in both the first dorsal spine is much shorter than the second spine (the opposite occurred in P. maculatus). On the other hand, Ward, [52] in a general analysis of DNA barcoding divergence among fishes worldwide, showed that specimens of Allocyttus verrucosus, Allocyttus niger and Pseudocyttus maculatus from the Indian Ocean and Tasmanian waters were clearly identified using COI sequences and were separated from each other in the NJ tree. However, specimens of A. verrucosus and P. maculatus from Argentina clustered together and near those of A. niger from Indian Ocean. Therefore it is possible that both individuals from Argentina are actually A. niger, a species that had never been reported for the southwest Atlantic Ocean [33], [50], [51] although Yearsley et al [53] in a book of Australian seafood shows a point in the southwest Atlantic as part of the distribution of this species. Further morphological and molecular studies are needed to clarify the actual taxonomic identification of these nominal species.

The maximum intraspecific genetic variation was below 1% in 93.6% of the species analysed except for the following species: Cynoscion guatucupa (1.72%, n = 5), Squalus acanthias (1.71%, n = 11), Helicolenus lahillei (1.24%, n = 13), Parona signata (1.09%, n = 6), Iluocoetes fimbriatus (1.09%, n = 11) and Prionotus nudigula (1.09%, n = 4). In addition, deep intraspecific divergences were found between individuals of Prepilus paru (12.3%, n = 2), this result will be discuss later. As observed by Hebert et al. [3], species congeneric K2P distances were much higher than 2%, except between some species of the genus Bathyraja, but each of these species formed cohesive units and were separated from each other in the NJ tree (Figure 3). From the nine species of Bathyraja barcoded, only three of them presented divergent values >2%, whereas in the remain six, values were relatively lower, being the lowest within B. albomaculata and B. macloviana (D = 0.61%) (Figure 3). The relatively low divergence found among species of Bathyraja was also found by Spies et al. [54] studying the genetic variability in 15 species of skates from Alaska. From the 12 species of the genus Bathyraja, four presented distances values relatively low (D<1%), even though all are morphologically different species.