The Atlantic bluefin tuna, Thunnus thynnus, is a commercially important species that has been severely over-exploited in the recent past. Although the eastern Atlantic and Mediterranean stock is now showing signs of recovery, its current status remains very uncertain and as a consequence their recovery is dependent upon severe management informed by rigorous scientific research. Monitoring of early life history stages can inform decision makers about the health of the species based upon recruitment and survival rates. Misidentification of fish larvae and eggs can lead to inaccurate estimates of stock biomass and productivity which can trigger demands for increased quotas and unsound management conclusions. Herein we used a molecular approach employing mitochondrial and nuclear genes (CO1 and ITS1, respectively) to identify larvae (n = 188) collected from three spawning areas in the Mediterranean Sea by different institutions working with a regional fisheries management organization. Several techniques were used to analyze the genetic sequences (sequence alignments using search algorithms, neighbour joining trees, and a genetic character-based identification key) and an extensive comparison of the results is presented. During this process various inaccuracies in related publications and online databases were uncovered. Our results reveal important differences in the accuracy of the taxonomic identifications carried out by different ichthyoplanktologists following morphology-based methods. While less than half of larvae provided were bluefin tuna, other dominant taxa were bullet tuna (Auxis rochei), albacore (Thunnus alalunga) and little tunny (Euthynnus alletteratus). We advocate an expansion of expertise for a new generation of morphology-based taxonomists, increased dialogue between morphology-based and molecular taxonomists and increased scrutiny of public sequence databases.
Citation: Puncher GN, Arrizabalaga H, Alemany F, Cariani A, Oray IK, Karakulak FS, et al. (2015) Molecular Identification of Atlantic Bluefin Tuna (Thunnus thynnus, Scombridae) Larvae and Development of a DNA Character-Based Identification Key for Mediterranean Scombrids. PLoS ONE 10(7): e0130407. https://doi.org/10.1371/journal.pone.0130407
Academic Editor: Martin Castonguay, Institut Maurice-Lamontagne, CANADA
Received: December 4, 2014; Accepted: May 6, 2015; Published: July 6, 2015
Copyright: © 2015 Puncher et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: Data are within the manuscript, in the NCBI GenBank repository, accession numbers KT003822 to KT003924, and the Barcode of Life database, accession numbers MLRV001-15 to MLRV084-15.
Funding: This work was carried out under the provision of the ICCAT Atlantic Wide Research Program for Bluefin Tuna (GBYP), funded by the European Community (Grant SI2/542789), Canada, Croatia, Japan, Norway, Turkey, United States (NMFS NA11NMF4720107), Chinese Taipei, and the ICCAT Secretariat (see: http://www.iccat.int/GBYP/en/Budget.htm). The contents of this paper do not necessarily reflect the point of view of ICCAT or of the other funders. This work was co-funded by the Marine Ecosystem Health and Conservation (MARES) Joint Doctorate programme selected under Erasmus Mundus and coordinated by Ghent University (FPA 2011-0016). The funders had no role in study design, data analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Atlantic bluefin tuna (BFT, Thunnus thynnus) are the largest of tunas roaming the world's oceans and magnificent creatures that have fed and fascinated humankind for millennia. Vast herds of BFT ply the cold waters of the North Sea and North Atlantic Ocean for prey in the winter, returning to spawning grounds in the Mediterranean Sea and Gulf of Mexico in the spring and early summer. Fisheries have long taken benefit of these migratory patterns, slaughtering multitudes of tuna by hook, gaff, harpoon and net as they migrated through coastal waters to feed Roman Legions, burgeoning principalities, fishery empires and modern multinational corporations. This unrelenting appetite has recently brought stocks to the brink of collapse. Gone are the large BFT that spawned in the Black Sea and swelled the market places of Istanbul up until the mid 1980s [1,2]. BFT stocks that inspired a revolution in fishing gear in the North Sea, sustaining a French, Norwegian, German, Dutch and Danish fleet for several decades crashed in 1963 without warning . In the 1960s, a massive shoal of BFT that appeared off the coast of Brazil was quickly targeted by Japanese long-liners and faded into the annals of history in as little as 7 years [3,4]. After many years of decline, the fisheries organization responsible for managing BFT stocks, the International Commission for the Conservation of Atlantic Tunas (ICCAT), finally enforced a rigorous recovery plan and stocks have started showing signs of recovery; although, the rate of recovery remains highly uncertain [5,6].
Currently, ICCAT manages BFT stocks as two populations: one which spawns in the Gulf of Mexico and forages in the North Western Atlantic (NWA) and a second that spawns in the Mediterranean Sea and forages in the Mediterranean and North Eastern Atlantic (NEA) . Despite evidence suggesting that these two populations mix, ICCAT manages them separately, dividing their ranges along the 45° longitude. Novel insights developed from an array of technologies including satellites tags, genetics and microchemistry suggests that the population structure of BFT is much more complex [8–14]. If regional population structuring exists, it is paramount for the welfare of the species that it be maintained, in order to conserve genetic biodiversity and evolutionary potential. An accurate and confident model of the population structure of BFT and the factors that affect their distribution is key to their continued viability.
In the past 15 years, several molecular techniques have been used in an effort to develop a more accurate vision of BFT population structure and dynamics in line with the results developed by electronic tagging campaigns and traditional ecological knowledge (summary and references in . Unfortunately, the results of these studies have been inconclusive and often contradictory. Due to the highly migratory nature of BFT, some research groups investigating the species' genetic population structure are now using only young tuna for their research as it is widely assumed that eggs, larvae and tuna of less than a few months age do not disperse far from their point of origin.
The location and abundance of early life stage fish is also used to improve stock assessments and our understanding of BFT spatial dynamics. Indices developed from the abundance of eggs and larvae collected from spawning sites have been used over 120 times to adjust and substantiate stock assessments of 18 different species within five teleost families throughout the world . For decades, scientists from ICCAT member nations have been using larval indices generated from surveys conducted in the Gulf of Mexico to calibrate Virtual Population Analyses (VPAs) of western Atlantic BFT [17–19]. In 2013, the first standardized BFT larval indices for a Mediterranean spawning site were published based on larval surveys conducted by the Oceanographic Institute of Spain (IEO) around the Balearic Islands in the western Mediterranean . Temporal shifts in BFT larvae abundance and condition can also provide important information about recruitment success, relative to short and long term environmental changes [21–23]. Surveys that have monitored the distribution of tuna larvae have shown that changes in relative abundances of different species are directly influenced by hydrodynamics [21,24–29]. In the context of a rapidly changing environment, our ability to properly identify and monitor fish species throughout their life history is critical for effective wildlife management and conservation efforts [30,31].
Unfortunately, early life stage fishes are often inaccurately identified by inexperienced technicians [32, 33]. These errors can lead to a misunderstanding of the spatial distribution of species, confusion over life history traits and population dynamics, inaccurate estimations of recruitment rates, survivorship and stock biomass, and potentially disguise the collapse or recovery of localized spawning sites [34,35]. The potential causes for misidentifications of tuna larvae are several: 1) Identification of tuna eggs and the larvae of some tuna species, using morphological characteristics alone, is very challenging, requiring an in depth knowledge of taxonomy, patience and experience [36–38], 2) samples are often badly damaged during collection or as a result of preservation [39–41], 3) some tuna identification keys are inaccurate and require updating, and 4) expert taxonomists in general are few and in demographic decline [42–44].
Due to the difficulties of identifying tuna larvae researchers occasionally outsource the task to distant laboratories such as the Sea Fisheries Institute, Plankton Sorting and Identification Center in Poland [17,19,22,45]; however, misidentification of larvae has occurred at that facility in the past as well (F. Alemany personal communication). Others have simply resorted to assigning scombrids to lower taxonomic levels [22,46]. Advances in molecular techniques and genetic barcoding now offer another solution to this problem.
There are a variety of ways in which researchers using genetic barcodes analyze their data in order to identify their specimens. The most commonly used tools for sequence association in barcoding studies are: 1) alignment with voucher sequences from online databases using Basic Local Alignment Search Tool (BLAST) tools provided by National Center for Biotechnology Information or the ID system provided by the Barcode of Life Database (BOLD), 2) Neighbour-Joining trees, and 3) classification using a molecular key of characteristic attributes. The BLAST program uses a heuristic algorithm to identify the sequences contained in GenBank that are most similar to the query sequence provided by the user . The ID System by BOLD employs a Hidden Markov Model (HMM) for alignment construction and returns only sequence matches that are less than 1% divergent from the query sequence . Neighbour-Joining (NJ) trees are ubiquitous among barcoding publications. They use a hierarchical clustering method to construct a phenogram based on a distance matrix of similarity between reference and query sequences. These distance matrices can be constructed by various methods which can impact species identification accuracy and measures of confidence. The Kimura 2-parameter model (K2P) for NJ trees has become the default model for most fish species identification studies; however, researchers have recently been challenging this assumption [49–53].
This study describes how larvae collected from the Strait of Sicily, Western Ionian Sea and Levantine Sea were acquired from three different institutions for genomic analysis within ICCAT's Atlantic wide research programme for bluefin tuna (GBYP). All larvae had been provisionally identified as Thunnus thynnus by technicians using morphology-based methods. All larvae were barcoded using a 650bp fragment of the CO1 gene and identified to species in an effort to assess the accuracy of identification. We have also compared the effectiveness of various methods used for associating sample sequences to reference or voucher sequences. We review the overall effectiveness of the Neighbour-Joining tree approach and compare two methods used for distance matrix construction (p-distance vs. K2P). Finally, we develop and assess a character-based key which uses unique genetic characteristics in much the same way as taxonomic keys that identify organisms based on diagnostic morphological features.
Materials and Methods
Strait of Sicily.
Larval tows were performed by Istituto per l’Ambiente Marino Costero of the National Research Council of Italy (IAMC-CNR) in the waters off Sicily’s southern coast (35°30’N-38°90’N, 12°38’E-15°10’E; Fig 1[A]), on board the R/V “Urania”, during 17–21 July 2011 and 5–19 July 2012 using two Bongo nets with 40 and 90cm diameters equipped with 1 mm black mesh. Bongo nets were towed obliquely from the surface to 100 m and back to the surface at two knots. All larvae were preserved in 96% ethanol and transported to the laboratory where they were identified to family, genus or species level when possible; using various taxonomic keys [54–58]. A total of 88 larvae with a mean length of 4.9 ± 1.5 mm and provisionally identified as tunas were sent to the GenoDREAM laboratory at the University of Bologna for genetic barcoding. An additional 4 non-scombrid larvae were also provided as outliers.
Capo Passero, Ionian Sea.
OCEANA’s 2008 larval survey aboard the R/V Marviva Med was the first BFT larval survey undertaken by an NGO team in the Mediterranean . Larval tows were conducted 15 July—11 August 2008 using a bongo 90 net with a quadrangular mouth opening equipped with 500 μm mesh for horizontal surface plankton tows east of Sicily (36°30’N-37°33’N, 15°35’E-15°59’E; Fig 1[B]). The net was towed at 2–2.5 knots at the surface for a constant duration of 10 minutes. Upon retrieval of nets from the sea, all larvae were immediately preserved in ethanol. All larvae were then dispatched to experts at IEO facilities in Palma de Mallorca where they were identified to species by experts. A total of 58 larvae identified as T. thynnus were dispatched to the GenoDREAM laboratory in November 2013 for genetic analysis.
A larval cruise was conducted by Istanbul University during 20–24 June 2012 along the southern coast of Turkey (36°07’N-36°10’N, 33°33’E-33°47’E; Fig 1[C]). Larvae were collected from surface waters using a Bongo net of 90 cm diameter and 1 mm mesh size, towed at 2–2.5 knots for 10 minutes. All captured larvae were immediately preserved in 96% ethanol and identified to species level on-board using microscopes and the taxonomic key by Richards (2005). A total of 38 larvae with a mean length of 6.9 ± 1.5 mm were conditionally identified as T. thynnus. These larvae were then bisected and the caudal sections were dispatched to GenoDREAM lab for genetic verification.
All larvae used in this study were collected from the Mediterranean Sea using plankton nets and sacrificed via immersion in 96% ethanol, according to standard larval survey practices. Permission to conduct the larval surveys and all legal permits were granted by the International Commission for the Conservation of Atlantic Tuna (ICCAT) and the General Fisheries Commission for the Mediterranean. ICCAT is the intergovernmental organization responsible for the management of bluefin tuna in the Atlantic Ocean and Mediterranean Sea and all research was coordinated through their “Atlantic-wide research programme for bluefin tuna” (GBYP). ICCAT issued a recommendation (Rec. 11–06) allowing the parties involved in this research to collect and sacrifice larvae for the purposes of genetic research as well as ship samples from one country to another (Certificate No. ICCAT RMA12-049).
All 188 scombrid and non-scombrid larvae were digested overnight in a Proteinase K solution and genomic DNA was extracted from each and purified using Promega’s Wizard SV96 Genomic DNA Purification kit and vacuum manifold. Fragments of the CO1 gene (~ 650 bp) were amplified (PCR) using FishF2 (5’-TCGACTAATCATAAAGATATCGGC- AC-3’) and FishR2 (5’-ACTTCAGGGTGACCGAAGAATCAGAA-3’) primers first published by Ward et al. . PCR reactions were performed in 50 μL volume consisting of 1x PCR Buffer, 1.0 μM of each primer, 160 μg/mL of BSA, 0.4 mM of dNTPs, 1.5 mM of MgCl2, 2.5 U/mL of Invitrogen Taq polymerase and ~100 ng of template DNA. PCR conditions consisted of 94°C for 3 min, 35 cycles of 30 sec at 94°C, 30 sec at 52°C, and 30 sec at 72°C, with a final extension at 72°C for 3 min. Due to the introgression of the mitochondrial genome of T. alalunga into the T. thynnus gene pool [41,60–63], all larvae that were identified as T. alalunga using the CO1 gene were then barcoded using sequences from the ITS1 region. DNA extractions from archived tissue samples of adult T. thynnus (2) and T. alalunga (2) were used as reference standards for all further analyses. Fragments of the ITS1 gene (~680bp) were amplified using the ITS1F (5’-TCCGTAGGTGAACCTGCGG-3’) and ITS1R (5’-CGCTGCGTTCTTCATCG-3’) primers designed by Chow et al. . All other PCR reagents were the same as above and conditions were as follows: 94°C for 3 min, 35 cycles of 30 sec at 94°C, 30 sec at 50°C, and 30 sec at 72°C, with a final extension at 72°C for 3 min.
After receiving all CO1 sequences from Macrogen Europe (Amsterdam, Netherlands), they were aligned in MEGA6 using the ClustalW algorithm and trimmed to 612bp.
BLAST and BOLD search engines
All larval sequences were converted to FASTA format and submitted to a nucleotide BLAST through the NCBI website (http://www.ncbi.nlm.nih.gov/). The first five sequence matches with highest similarity to database references (max ident) were analyzed for species consistency. Similarly, all sequences were submitted to BOLD's Identification System (http://www.boldsystems.org/) for comparison with all species level barcode records of animals (148,815 sequences as of 09 Nov 2014). All matches were analyzed using the percent similarity score.
Neighbour joining tree analysis
Ten reference sequences from each of thirteen scombrid species were downloaded from BOLD (n = 122) and GenBank (n = 8) (S1 Appendix). Sequences were mined from GenBank only when the number of sequences on BOLD were insufficient. Aside from the BFT reference sequences, eight of the species occur in the Mediterranean Sea and may have larvae associated with those of BFT (Auxis rochei, Auxis thazard, Scomber colias, Scomber scombrus, Euthynnus alletteratus, Katsuwonus pelamis, Sarda sarda, T. alalunga,). The remaining four species are closely related to the other species and were included in order to rule out the occurrence of genetic introgression (Scomber japonicus, Thunnus albacares, Thunnus maccoyii, Thunnus atlanticus). Reference sequences were sourced from throughout each species’ geographic distribution in order to have ample representation of genetic variation. Various phenograms (all sequences, only reference sequences, all larvae and reference sequences belonging to Mediterranean species) were built using the Neighbour Joining method  with both the Kimura-2-parameter (K2P) distance model  and p-distances  for distance matrix construction. The statistical support of each node was tested using bootstrap analysis  with 1000 replications in MEGA6 . Selected trees were modified using FigTree (http://tree.bio.ed.ac.uk/software/figtree/).
Assignment by characteristic attribute key
All variable sites of an alignment containing all reference and larvae sequences were highlighted using MEGA6. This simplified alignment was inspected for loci containing unique nucleotides that were diagnostic of particular species. The character-based key constructed here differs from those constructed by  in that it requires first assignment to genus and then species. In this way we have used nucleotides that may not be unique to single species or genera. Rather we are providing a multi-step assignment tool which uses multiple characteristics, much like a taxonomic key. According to the terminology used by , we have selected “pure characteristic attributes” (PCAs), or specific nucleotides located at variable sites that are unique to single clades and thus diagnostic for single clades. Some of these characteristic attributes stand alone and are called “simple CAs”, whereas other loci should be used in combination and are thus called “compound CAs”. Once the key was constructed all larval sequences were assigned to species accordingly.
DNA extractions from all larvae were successfully amplified, sequenced and analyzed. Sequences from 84 larvae with photographic records were uploaded to the Barcode of Life Database (Project: MLRV; Accession numbers: MLRV001-15 to MLRV084-15). All other sequences were uploaded directly to the GenBank database (KT003822 to KT003924). A few unexpected challenges were encountered concerning the reference sequences downloaded from both BOLD and GenBank. One T. thynnus reference sequence (BOLD: GBGCA443-10; GenBank: GQ414572) affiliated with the work of  corresponds with CO1 sequences of T. alalunga. Within that work, the authors clearly state that this sequence belongs to a BFT with introgressed mtDNA from albacore; however, no mention of this is associated with the sequences itself in either database. They also describe how two sequences belonging to Pacific-like T. thynnus (Atlantic bluefin with introgressed T. orientalis mitochondria) were used and published in GenBank (GQ414570 and GQ414573). Both of these sequences were later incorporated into the BOLD database (GBGCA445-10 and GBGCA442-10, respectively); however, in both databases GQ414570 is identified as T. thynnus, while QQ414573 is identified as T. orientalis. Another sequence featured in , which they referred to as T. orientalis, has since been uploaded to GenBank (GQ414566) and BOLD (GBGCA449-10) under the title of T. thynnus. BOLD and BLAST queries, as well as our CA key, clearly identify this as a T. orientalis sequence. It is likely that the authors have confused sequences and their IDs at some point (confirmed by J. Viñas). Finally, a single T. orientalis sequence (GenBank: JN097817; BOLD: GBGCA1390-13) uploaded by several researchers from the South Korean National Fisheries Research and Development Institute matches those of T. thynnus.
For all but two sequences, BLAST provided a species match with an identity similarity higher than 99% (Table 1). The median value for maximum similarity scores across the type five selected matches for each larva was 100%. The remaining two larvae from the Strait of Sicily received identity similarity scores of 93% and 95% for sequences belonging to bullet tuna (Auxis rochei, Scombridae) and T. thynnus, respectively. Overall, only 42% of larvae were identified as T. thynnus. In fact, nearly as many (39%) were identified as bullet tuna. The larvae collected in the Strait of Sicily were composed of five different scombrid species with only 21 identified as T. thynnus. Although the samples from the Levantine Sea contained fewer taxa, none of the larvae provided were identified as T. thynnus. All larvae collected from the Ionian Sea, offshore from Capo Passero were identified as T. thynnus and received identity similarity scores of 100%. The four non-scombrid larvae from Sicily were identified as Chromis sp., picarel (Spicara smaris, Centracanthidae), greater weever (Trachinus draco, Trachinidae), and pygmy lanternfish (Lampanyctus pusillus, Myctophidae). Five larvae from the Levantine Sea were assigned to two additional outlier taxa: brown comber (Serranus hepatus, Serranidae) (2) and common pandora (Pagellus erythrinus, Sparidae) (3). All larvae provisionally identified as albacore using CO1 sequences and later barcoded with the ITS1 gene clustered with T. alalunga standards (data not shown), thereby ruling out the possibility of false species identification due to hybridization/mtDNA introgression.
All larvae were identified by BOLD with a confidence score of 99–100%, aside from the two individuals collected the Strait of Sicily discussed above for which a “no match” was returned. No match was given by BOLD because of the program's 1% divergence threshold. All other specimen identifications were equal to those provided by BLAST.
Neighbour-joining tree analysis
Neighbour joining tree analysis using CO1 sequences produced well-defined clusters of candidate larvae with reference sequences (Figs 2 and 3). All larvae clustered with reference taxa in the same manner as the BLAST and BOLD results. Phenograms containing only the larvae and reference sequences for species found in Mediterranean Sea showed lowest bootstrap probability (BP) for branch nodes separating the true tunas (BFT and albacore) from the other scombrids (BP = 33–35) and a clade containing the true tunas and E. alletteratus from the other species (BP = 30–34). The node containing the A. rochei reference sequences and 74 larvae was the least stable with BPs of 53–59. All other nodes had BPs > 80. Phenograms based on distance matrices using p-distances (Fig 4) had consistently higher BP values than those built with K2P distances (Fig 3). P-distance based phenograms also tended to exclude larval sequences from clusters, as is the case with one larva in the Auxis spp. cluster in Fig 2. When the larval sequences were removed from the alignments, all BP values increased (Fig 4). Predictably, BP values for the A. rochei (ΔBP = 39) and T. thynnus (ΔBP = 17) nodes increased most dramatically, as they were the two clusters to increase in sequence size most dramatically. When T. atlanticus, T. maccoyii, T. obesus and T. albacares reference sequences were included in the alignments, the neighbour-joining tree based on p-distances failed to differentiate clusters for each and combined all with the same cluster alongside the T. thynnus and related larvae sequences (BP = 61; Fig 5). The topology of this same clade in the corresponding K2P-based phenogram changed somewhat but the composition remained the same, albeit the BP value was much lower (BP = 35).
The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches . The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the tree. The evolutionary distances were computed using the p-distance method  and are in the units of the number of base differences per site. The analysis involved 280 nucleotide sequences. All ambiguous positions were removed for each sequence pair. There were a total of 612 nucleotide positions in the final dataset.
The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches . The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the tree. The evolutionary distances were computed using the Kimura 2-parameter model  and are in the units of the number of base differences per site. The analysis involved 280 nucleotide sequences. All ambiguous positions were removed for each sequence pair. There were a total of 612 nucleotide positions in the final dataset.
The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches . The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the tree. The evolutionary distances were computed using the p-distance method  and are in the units of the number of base differences per site. The analysis involved 91 nucleotide sequences. All ambiguous positions were removed for each sequence pair. There were a total of 612 nucleotide positions in the final dataset.
The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches . The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the tree. The evolutionary distances were computed using the p-distance method  and are in the units of the number of base differences per site. The analysis involved 330 nucleotide sequences. All ambiguous positions were removed for each sequence pair. There were a total of 612 nucleotide positions in the final dataset.
The ITS1 NJ tree constructed from sequences of voucher specimens and larvae identified as T. alalunga using CO1 did not reveal any introgressed BFT.
Construction of a character-based key uncovered 68 nucleotides capable of distinguishing all reference taxa. As the primary purpose of our research was the isolation of T. thynnus larvae from the rest of the catch, Thunnus spp. sequences were the first to be isolated using four diagnostic nucleotides (Table 2). Using these nucleotides a total of 90 larvae were identified as members of the Thunnus genus, representing less than half of all larvae. Six bases separating T. alalunga reference sequences from the other five Thunnus species identified eleven albacore tuna among the larvae. A single thymine located at position 231 of our alignment was capable of isolating all T. thynnus references from the other species of the same genus. When these criteria were applied to the larvae, a total of seventy-eight larvae were identified as bluefin tuna. One outstanding individual was identified as T. maccoyii, T. atlanticus, T. albacares or T. obesus using these criteria. Since none of these species can be found in the Mediterranean Sea, this is likely an example of genetic introgression. Three bases, found to be discriminatory for the Auxis species of the Mediterranean, identified seventy-four larvae among the remaining samples. All of these Auxis candidate larvae were identified as A. rochei using eight bases that were found to discriminate between A. rochei and A. thazard reference sequences. Using six diagnostic loci, fourteen Euthynnus alletteratus larvae were identified. Eleven characteristic loci set the Scomber spp. apart from the other species and ten loci were used to identify a single larva. Six diagnostic bases were found for Katsuwonus pelamis and Sarda sarda but none of the larvae analyzed were identified as members of either species. Since the character-based identification key was developed for the identification of scombrids only, the remaining nine larvae could not be identified.
(Mis)Identification of larvae
The 11 species identified among the samples are typical of plankton surveys conducted in the region and during this sampling season. BFT larvae are commonly associated with larvae of other scombrids, having been previously captured with dense concentrations of bullet tuna off the coasts of Tunisia [70,71], Sicily  and the Balearic Islands . They have also been found alongside albacore tuna [21,26] and Atlantic black skipjack (Euthynnus alletteratus, Scombridae) . Among these species, BFT are found in proportionally higher concentrations in deep offshore waters beyond shelf breaks [25,70]. Researchers in Italy  captured L. pusillus larvae in the Strait of Sicily in June-July 2000. Larvae of S. hepatus, Chromis chromis (Pomacentridae), P. erythrinus and T. draco have been captured in the Aegean during June of 2003–2006 . Alemany et al.  have encountered all the species that we have identified in the Balearic Islands, aside from P. erythrinus and Atlantic chub mackerel (Scomber colias, Scombridae). Accurate identification of these samples to species level is a testament to the versatility of the genetic marker used and the potential of the ever-expanding resources of the Barcode of Life Database.
The high number of correctly identified BFT larvae within the Sicilian samples was expected, since large quantities of T. thynnus are typical of the Sicilian Channel, western Ionian and southern Tyrrhenian Seas [74,75]. To date, BFT larvae have been found in highest concentration in Mediterranean waters around Sicily, the Balearic Islands and the southern coast of Turkey [23,24,76,77]. It comes as no surprise that all 58 larvae provided by the IEO and OCEANA-Marviva project were correctly identified, since their staff are leading experts in the field of ichthyoplanktology. However, the fact that no BFT larvae were provided amongst the 38 larvae received from the Levantine Sea calls into question all previous publications on the subject of BFT reproduction in that area. For example, Oray and Karakulak  captured 121 bluefin tuna (T. thynnus), 94 bullet tuna (A. rochei) and 22 Atlantic black skipjack (E. alletteratus) larvae in the northern Levantine basin during larval surveys conducted during 5–18 June 2004. Their publication established a benchmark in the literature by which many assumptions of BFT movements, reproductive behaviour and population structuring has been made. However, during that survey, the researchers, capture protocols and larvae identification methods were the very same that procured the misidentified larvae discussed herein. The possibility that larvae from the Levantine Sea have been misidentified in the past, demands a review of the timing, location and extent to which BFT are spawning in the eastern Mediterranean. If BFT spawning areas are indeed limited in number, then their accurate identification and subsequent conservation from over-exploitation, habitat alteration and pollution is critically important .
DNA extractions and use of molecular markers
It is noteworthy that we were able to extract high quality DNA, and identify to species level, larvae that had been archived in ethanol at room temperature for over 5 years. This possibility should be taken into consideration for all collections containing similarly preserved wildlife specimens. Additionally, the molecular markers used were effective at identifying all larvae to species level. The use of CO1 for sample identification has been criticized, as it lacks the capacity of discrimination among the Neothunnus tribe and the Pacific and Atlantic BFT [59,62]. Since none of the members of the Neothunnus tribe or T. orientalis are present in the Mediterranean Sea, this was not a concern for our study. Our reconstruction of NJ trees including the additional non-Mediterranean Thunnus spp. sequences were also unable to reliably distinguish the members of that genera, aside from T. alalunga which consistently clustered independently from the others. Some of the earliest molecular work using a suite of allozymes also found albacore to be most divergent and uncovered very little divergence between T. albacares, T. maccoyii and T. orientalis . DNA sequences from the mitochondrial control region suggest that BFT is a sister taxa of the southern bluefin tuna and that the albacore and Pacific bluefin tuna (PBFT) form a divergent monophyletic clade . The high number of diagnostic loci contained in our CO1 sequences which discriminate between the albacore and the other Thunnus species support this claim. Clearly, the level of divergence between species is dependent upon the character used for comparison. For example, the high level of divergence between the Pacific and Atlantic bluefin tuna shown when analyzing the mitochondrial control region  vanishes when comparing ITS1 gene sequences . This could be a result of historical hybridization of the PBFT with albacore which resulted in the transfer of the albacore mtDNA genome into the PBFT line .
The presence of erroneous or misleading reference sequences in both GenBank and BOLD is both troublesome and concerning. Surely, it was not the intention of the founders of these databases that users should have to check the origin and associated publications of every sequence. The presence of sequences belonging to hybrid organisms in genetic reference databases is confusing and requires additional traceability and documentation.
BLAST and BOLD assignment
The results generated by BOLD's global alignments and BLAST's local alignments were consistent with one another. The identity similarity threshold used by BOLD prevented the identification of two larvae from the Strait of Sicily, whose identity were later confirmed by BLAST, NJ trees and our character-based identification key. Lowenstein et al.  complained that BOLD performed poorly during their attempts to identify species used in sushi but the database was still in its infancy then and it has come a long way since and now contains over 4 million collected from 146 countries. During this period of rapid growth, BOLD began featuring sequences mined from GenBank in their Public Data Portal. We have found several sequences that are in clear violation of the data standards that were established when BOLD was first introduced. The founders of BOLD have clearly stated that sequences do not undergo any kind of centralized review and that the quality of data featured in the database is ultimately dependent upon the individuals that have uploaded the data . At the moment, more than half of the Thunnus thynnus sequences featured in BOLD have been mined from GenBank. This hybridization of the two databases, without BOLD's once lauded traceability standards, threatens to undermine the reputation and usefulness of BOLD.
The NJ trees accurately identified larvae to species level for taxa with reference sequences included in the alignments. Clearly, for this approach to work, query larvae must cluster with voucher sequences; an obvious disadvantage when compared to the BLAST and BOLD approaches. The use of NJ trees for species identification has recently come under fire from various sources [49,53]. An entire section focused on NJ trees has been featured in a recent publication entitled “The seven deadly sins of DNA barcoding” . A major disadvantage associated with NJ trees surfaces when individual sequences are assigned to the space between two reference clusters. On these researchers are forced to retreat to lower taxonomic levels . By increasing the number of sequences used for each reference taxa, one can decrease the frequency of these ambiguous outcomes . Our results suggest that the models used for the assembly of distance matrices are also important for the reduction of ambiguous results. Although the K2P model is the most widely used model for NJ tree construction in fish barcoding studies, we have found that distance matrices based on p-distance provide higher BP values. Recent critical reviews of NJ tree construction agree that identification of species using NJ trees based on K2P distances can be inappropriate to the task and more suitable, less complex models can prove more effective [51,52]. In fact, Collins et al. , wrote that K2P “was without exception a poorly approximating model at the species level”. Why the K2P model is so widely used in fish barcoding studies is a mystery. Srivathsan and Meier  suggest that the widespread use of K2P is a result of its use by early barcoding proponents who wanted to highlight the extreme differences between species. Perhaps researchers are simply following the examples of their peers or that of BOLD which uses the K2P model for their taxa identification trees . Regardless of their weaknesses, phenograms do provide attractive graphic representations of the results.
Realizing the faults inherent in the NJ tree approach, some researchers have found that a character-based method of specimen identification has proven more appropriate to the task [68,84]. Paine et al.  constructed a character-based key for identification of degraded tissue samples using reference sequences from 17 species of the Scombridae common to the Western Atlantic Ocean. Our molecular key differs significantly, since the molecular key of Paine et al.  begins with position 575 of our alignment, thereby providing only 37 bases for comparison. The molecular key of Lowenstein et al.  was generated from alignments shifted only 40 base pair positions towards the 3' end of the CO1 sequence. The molecular key of Lowenstein et al.  includes Thunnus species only, as they were making efforts to develop a tool for seafood traceability. Interestingly, they did not include in their study the skipjack tuna, Katsuwonus pelamis, a globally cosmopolitan fish in all of the world's oceans and by far the world’s most important tuna fishery. They, too, discovered the same individual nucleotide that distinguishes T. thynnus from all other tunas. Many of the diagnostic loci discovered by Lowenstein et al.  no longer function after the inclusion of the Scombrus spp. and Auxis spp. sequences. Character-based keys for scombrid species have also been developed for the mitochondrial control region and ITS1 [60,61].
Misidentification of early life stage fishes has already occurred in various commercial species, leading to inaccurate estimations of spawning stock biomass [35,85]. We have shown here, for the first time that tuna larvae collected from the Mediterranean Sea have been misidentified by larval survey crews. If the larvae and eggs of BFT are to be used to improve stock assessments in the future, it is imperative that problems associated with species identification are resolved. Fishery independent data, such as larval abundance, are certainly welcome for the betterment of stock assessments where traditional fishery data has lost its credibility; however, scientific rigour and quality control must also accompany this data or we run the risk of repeating the mistakes of the past. Genetic barcoding is a legitimate technique that can support species identification and play a crucial role in fisheries management efforts. We suggest that all routine fisheries work involving larvae should make use of both taxonomists and geneticists in order to ensure both accuracy of results and efficient use of financial resources. We call upon the few taxonomic world experts to update the identification keys associated with fish species of economic and conservation concern, embrace the digital community and pass their knowledge onto new generations through training courses, either in situ or with the various Information and Communications Technology platforms now available. Most barcoding efforts are dependent on online databases for voucher sequences; therefore, it is crucial that quality standards are upheld if the barcoding effort is to retain its legitimacy. Enough doubt has been cast on the inappropriate use of NJ trees for specimen identification purposes that they are now regarded by many simply as an attractive communication tool; however, they are still being widely used. Conversely and despite nearly two decades of use in scombrid identification, character-based keys are still not used as a reliable and recognized tool. Perhaps it is the user-friendly automated style of NJ trees that have kept phenograms popular, despite their fallibility. Several efforts are being made to automate the generation of character-based keys and their use as specimen identification tools. The BOLD System now features a Diagnostic Characters analysis suite and R has a package (Spider) dedicated to their use . Character-based keys also have potential for translation into microarrays and high throughput NGS genotyping platforms. Certainly, standardized high throughput genotyping tests will have profound impacts on the future of larvae identification and fisheries surveys.
Bernardo Patti and Angelo Bonanno contributed to the collection of larvae in the Sicilian Channel.
Conceived and designed the experiments: GNP. Performed the experiments: GNP. Analyzed the data: GNP. Contributed reagents/materials/analysis tools: GNP HA FA ACa IKO FSK GB ACu SM FT. Wrote the paper: GNP HA ACa FA FT.
- 1. Mather FJ, Mason JM, Jones AC. Historical Document: Life History and Fisheries of Atlantic Bluefin Tuna NOAA Technical Memorandum National Marine Fisheries Service; 1995.
- 2. Karakulak S, Oray I, Corriero A, Deflorio M, Santamaria N, Desantis S, et al. Evidence of a spawning area for the bluefin tuna (Thunnus thynnus L.) in the eastern Mediterranean. J Appl Ichthyol. 2004; 20(4): 318–320.
- 3. Fromentin JM, Powers JE. Atlantic bluefin tuna: population dynamics, ecology, fisheries and management. Fish Fish. 2005; 6: 281–306.
- 4. Fromentin JM, Reygondeau G, Bonhommeau S, Beaugrand G. Oceanographic changes and exploitation drive the spatio‐temporal dynamics of Atlantic bluefin tuna (Thunnus thynnus). Fish Oceanogr. 2014; 23(2): 147–156.
- 5. Fromentin JM, Bonhommeau S, Arrizabalaga H, Kell LT. The spectre of uncertainty in management of exploited fish stocks: The illustrative case of Atlantic bluefin tuna. Mar Policy. 2014b; 47: 8–14.
- 6. ICCAT Report of the standing committee on research and statistics (SCRS) (Madrid, Spain, 29 September to 3 October 2014); 2014.
- 7. Block BA, Teo SLH, Walli A, Boustany A, Stokesbury MJ, Farwell CJ et al. Electronic tagging and population structure of Atlantic bluefin tuna. Nature. 2005; 434: 1121–1127. pmid:15858572
- 8. Rooker JR, Secor DH, De Metrio G, Schloesser R, Block BA, Neilson JD. Natal homing and connectivity in Atlantic bluefin tuna populations. Science. 2008; 322: 742–744. pmid:18832611
- 9. Rooker JR, Arrizabalaga H, Fraile I, Secor DH, Dettman DL, Abid N, et al. Crossing the line: Migratory and homing behaviors of Atlantic bluefin tuna. Mar Ecol Prog Ser. 2014; 504: 265–276.
- 10. Galuardi B, Royer F, Golet W, Logan J, Neilson J, Lutcavage M. Complex migration routes of Atlantic bluefin tuna (Thunnus thynnus) question current population structure paradigm. Can J Fish Aquat Sci. 2010; 67: 966–976.
- 11. Cermeño P, Tudela S, Quílez-Badia G, Trápaga SS, Graupera E. New data on bluefin tuna migratory behaviour in the western and central Mediterranean Sea. ICCAT Col Vol Sci Pap. 2012; 68: 151–162.
- 12. Aranda G, Abascal FJ, Varela JL, Medina A. Spawning behaviour and post-spawning migration patterns of Atlantic bluefin tuna (Thunnus thynnus) ascertained from satellite archival tags. PloS ONE. 2013; 8(10): e76445. pmid:24098502
- 13. Riccioni G, Landi M, Ferrara G, Milano I, Cariani A, Zane L, et al. Spatio-temporal population structuring and genetic diversity retention in depleted Atlantic bluefin tuna of the Mediterranean Sea. Proc Natl Acad Sci. 2010; 107:2102–2107. pmid:20080643
- 14. Riccioni G, Stagioni M, Landi M, Ferrara G, Barbujani G, Tinti F. Genetic structure of bluefin tuna in the Mediterranean Sea correlates with environmental variables. PLoS ONE. 2013; 8(11): e80105. pmid:24260341
- 15. ICCAT Report of the 2013 bluefin meeting on biological parameters review (Tenerife, Spain, 07 May – 13 May, 2013); 2013. pp. 1–75.
- 16. Stratoudakis Y, Bernal M, Ganias K, Uriarte A. The daily egg production method: recent advances, current applications and future challenges. Fish Fish. 2006; 7(1): 35–57.
- 17. Scott GP, Turner SC, Churchill GB, Richards WJ, Brothers EB. Indices of larval bluefin tuna, Thunnus thynnus, abundance in the Gulf of Mexico; modelling variability in growth, mortality, and gear selectivity. B Mar Sci. 1993; 53: 912–929.
- 18. Scott GP, Turner SC. Updated index of bluefin tuna (Thunnus thynnus) spawning biomass from Gulf of Mexico ichthyoplankton surveys. ICCAT Col Vol Sci Pap. 2003; 55: 1123–1126.
- 19. Ingram GW Jr, Richards WJ, Lamkin JT, Muhling B. Annual indices of Atlantic bluefin tuna (Thunnus thynnus) larvae in the Gulf of Mexico developed using delta-lognormal and multivariate models. Aquat Living Resour. 2010; 23(01): 35–47.
- 20. Ingram GW Jr, Alemany F, Álvarez D, García A. Development of indices of larval bluefin tuna (Thunnus thynnus) in the western Mediterranean Sea. ICCAT Col Vol Sci Pap. 2013; 69(2): 1057–1076.
- 21. Alemany F, Quintanilla L, Velez-Belchi P, Garcia A, Cortés D, Rodriguez JM, et al. Characterization of the spawning habitat of Atlantic bluefin tuna and related species in the Balearic Sea (western Mediterranean). Prog Oceanogr. 2010; 86: 21–38.
- 22. Lindo-Atichati D, Bringas F, Goni G, Muhling B, Muller-Karger FE, Habtes S. Varying mesoscale structures influence larval fish distribution in the northern Gulf of Mexico. Mar Ecol Prog Ser. 2012; 463: 245–257.
- 23. García A, Cortés D, Quintanilla J, Ramirez T, Quintanilla L, Rodríguez JM, et al. Climate‐induced environmental conditions influencing interannual variability of Mediterranean bluefin (Thunnus thynnus) larval growth. Fish Oceanogr. 2013; 22(4): 273–287.
- 24. García A, Alemany F, Velez-Belchí P, López-Jurado JL, Cortés D, de la Serna JM, et al. Characterization of the bluefin tuna spawning habitat off the Balearic Archipelago in relation to key hydrographic features and associated environmental conditions. ICCAT Col Vol Sci Pap. 2005; 58: 535–549.
- 25. Alemany F, Deudero S, Morales-Nin B, López-Jurado JL, Jansa J, Palmer M, et al. Influence of physical environmental factors on the composition and horizontal distribution of summer larval fish assemblages off Mallorca island (Balearic archipelago, western Mediterranean). J Plankton Res. 2006; 28: 473–487.
- 26. Aguilar R, Lastra P. Bluefin tuna larval survey: 2008 Oceana-MarViva Mediterranean Project. 2009; 76 pp.
- 27. Mariani P, MacKenzie BR, Iudicone D, Bozec A. Modelling retention and dispersion mechanisms of bluefin tuna eggs and larvae in the northwest Mediterranean Sea. Prog Oceanogr. 2010; 86: 45–58.
- 28. Reglero P, Ciannelli L, Alvarez-Berastegui D, Balbín R, López-Jurado JL, Alemany F. Geographically and environmentally driven spawning distributions of tuna species in the western Mediterranean Sea. Mar Ecol Prog Ser. 2012; 463: 273–284.
- 29. Muhling BA, Reglero P, Ciannelli L, Alvarez-Berastegui D, Alemany F, Lamkin JT, et al. Comparison between environmental characteristics of larval bluefin tuna Thunnus thynnus habitat in the Gulf of Mexico and western Mediterranean Sea. Mar Ecol Prog Ser. 2013; 486: 257–276.
- 30. Costa FO, Carvalho GR. The Barcode of Life Initiative: synopsis and prospective societal impacts of DNA barcoding of fish. Life Sci Soc Pol. 2007; 3(2): 29–40.
- 31. D’Alessandro EK, Sponaugle S, Serafy JE. Larval ecology of a suite of snappers (family: Lutjanidae) in the Straits of Florida, western Atlantic Ocean. Mar Ecol Prog Ser. 2010; 410: 159–175.
- 32. Vecchione M, Mickevich MF, Fauchald K, Collette BB, Williams AB, Munroe TA, et al. Importance of assessing taxonomic adequacy in determining fishing effects on marine biodiversity. ICES J Mar Sci. 2000; 57: 677–681.
- 33. Ko HL, Wang YT, Chiu TS, Lee MA, Leu MY, Chang KZ, et al. Evaluating the accuracy of morphological identification of larval fishes by applying DNA barcoding. PLoS ONE. 2013; 8(1): e53451. pmid:23382845
- 34. Armstrong MJ, Connolly P, Nash RDM, Pawson MG, Alesworth E, Coulahan PJ, et al. An application of the annual egg production method to estimate the spawning biomass of cod (Gadus morhua L.), plaice (Pleuronectes platessa L.) and sole (Solea solea L.) in the Irish Sea. ICES J Mar Sci. 2001; 58(1): 183–203.
- 35. Fox CJ, Taylor MI, Pereyra R, Villasana MI, Rico C TaqMan DNA technology confirms likely overestimation of cod (Gadus morhua L.) egg abundance in the Irish Sea: implications for the assessment of the cod stock and mapping of spawning areas using egg‐based methods. Mol Ecol. 2005; 14(3): 879–884. pmid:15723679
- 36. Richards WJ, Pothoff T Analysis of the taxonomic characters of young scombrid fishes, genus Thunnus. In: The early life history of fish. (Ed. Blaxter J.H.S.), Springer, Berlin. 1974; pp. 623–648.
- 37. Kohno H, Hoshino T, Yasuda F, Taki Y. Larval melanophore patterns of Thunnus alalunga and T. thynnus from the Mediterranean. JPN J Ichthyol. 1982; 1–28.
- 38. Richards WJ Scombridae: mackerels and tunas. In: Early Stages of Atlantic Fishes: an Identification Guide for the Western Central North Atlantic (ed. Richards WJ). 2006; pp. 2187–2229. Taylor and Francis, Boca Raton, Florida.
- 39. Matsumoto WM, Ahlstrom EH, Jones S. On the clarification of larval tuna identification. Fish B-NOAA. 1972; 70: 1–12.
- 40. Richards WJ, Potthoff T, Kim JM. Problems identifying tuna larvae species (Pisces: Scombridae: Thunnus) from the Gulf of Mexico. Fish B-NOAA. 1990; 88: 607–609.
- 41. Paine MA, McDowell JR, Graves JE. Specific identification of western Atlantic Ocean scombrids using mitochondrial DNA cytochrome c oxidase subunit I (COI) gene region sequences. B Mar Sci. 2007; 80(2): 353–367.
- 42. Boero F. Light after dark: the partnership for enhancing expertise in taxonomy. Trends Ecol Evol. 2001; 16: 266. pmid:11301157
- 43. Hopkins GW, Freckleton RP. Declines in the numbers of amateur and professional taxonomists: implications for conservation. Anim Conserv. 2002; 5: 245–249.
- 44. Wilson EO The encyclopedia of life. Trends Ecol Evol. 2003; 18: 77–80.
- 45. Matarese AC, Spies IB, Busby MS, Orr JW. Early larvae of Zesticelus profundorum (family Cottidae) identified using DNA barcoding. Ichthyol Res. 2011; 58(2): 170–174.
- 46. Boehlert GW, Mundy BC. Vertical and onshore-offshore distributional patterns of tuna larvae in relation to physical habitat features. Mar Ecol Prog Ser. 1994; 107: 1–13.
- 47. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. "Basic local alignment search tool." J Mol Biol. 1990; 215: 403–410. pmid:2231712
- 48. Ratnasingham S, Hebert PDN. BOLD: The Barcode of Life Data System (http://www.barcodinglife.org). Mol Ecol Notes. 2007; 7: 355–364. pmid:18784790
- 49. Little DP, Stevenson DW. A comparison of algorithms for the identification of specimens using DNA barcodes: examples from gymnosperms. Cladistics. 2007; 23: 1–21.
- 50. Zhang AB, Muster C, LIANG HB, Zhu CD, Crozier R, Wan P, et al. A fuzzy‐set‐theory‐based approach to analyse species membership in DNA barcoding. Mol Ecol. 2011; 21(8): 1848–1863. pmid:21883585
- 51. Collins RA, Boykin LM, Cruickshank RH, Armstrong KF. Barcoding's next top model: an evaluation of nucleotide substitution models for specimen identification. Methods Ecol Evol. 2012; 3(3): 457–465.
- 52. Srivathsan A and Meier R (2012) On the inappropriate use of Kimura‐2‐parameter (K2P) divergences in the DNA‐barcoding literature. Cladistics 28(2): 190–194.
- 53. Collins RA, Cruickshank RH. The seven deadly sins of DNA barcoding. Mol Ecol Res. 2013; 13(6): 969–975.
- 54. Fahay MP. Guide to the early stages of marine fishes occurring in the western North Atlantic Ocean, Cape Hatteras to the southern Scotian Shelf. J Northwest Atl Fish Sci. 1983; 4: 423.
- 55. Fahay MP. Early Stages of Fishes in the Western North Atlantic Ocean, 1st edn (Vol. 1). Northwest Atlantic Fisheries Organization, Dartmouth, Nova Scotia; 2007.
- 56. Kawamura G, Masuma S, Tezuka N, Koiso M, Jinbo T, Namba K. Morphogenesis of sense organs in the bluefin tuna Thunnus orientalis. In The big fish bang. Proceedings of the 26th annual larval fish conference, Bergen. 2003; pp. 123–135.
- 57. Potthoff T. Osteological development and variation in young tunas, genus Thunnus (Pisces, Scombridae), from the Atlantic Ocean. Fish B-NOAA. 1974; 72(2): 563–588.
- 58. Potthoff T, Richards WJ Juvenile bluefin tuna, Thunnus thynnus (Linnaeus), and other scombrids taken by terns in the Dry Tortugas, Florida. Fish B-NOAA. 1970; 20(2), 389–413.
- 59. Ward RD, Hanner R, Hebert PDN. The campaign to DNA barcode all fishes, FISH-BOL. J Fish Biol. 2009; 74: 329–356. pmid:20735564
- 60. Alvarado Bremer JR, Naseri I, Ely B. Orthodox and unorthodox phylogenetic relationships among tunas revealed by the nucleotide sequence analysis of the mitochondrial DNA control region. J Fish Biol. 1997; 50(3): 540–554.
- 61. Chow S, Nakagawa T, Suzuki N, Takeyama H, Matsunaga T. Phylogenetic relationships among Thunnus species inferred from rDNA ITS1 sequence. J Fish Biol. 2006; 68(A): 24–35.
- 62. Viñas J, Tudela S. A validated methodology for genetic identification of tuna species (genus Thunnus). PLoS ONE. 2009; 4(10): e7606. pmid:19898615
- 63. Viñas J, Gordoa A, Fernández-Cebrián R, Pla C, Vahdet Ü, Araguas RM. Facts and uncertainties about the genetic population structure of Atlantic bluefin tuna (Thunnus thynnus) in the Mediterranean. Implications for fishery management. Rev Fish Biol Fisher. 2011; 21: 527–541.
- 64. Saitou N, Nei M. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol. 1987; 4(4): 406–425. pmid:3447015
- 65. Kimura M. A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. J Mol Evol. 1980; 16: 111–120. pmid:7463489
- 66. Nei M, Kumar S. Molecular Evolution and Phylogenetics. Oxford University Press, New York, 2000.
- 67. Felsenstein J Confidence limits on phylogenies: an approach using the bootstrap. Evolution. 1985; 783–791.
- 68. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. MEGA6: Molecular Evolutionary Genetics Analysis version 6.0. Mol Biol Evol. 2013; 30: 2725–2729. pmid:24132122
- 69. Lowenstein JH, Amato G, Kolokotronis SO. The real maccoyii: identifying tuna sushi with DNA barcodes–contrasting characteristic attributes and genetic distances. PLoS ONE. 2009; 4(11): e7866. pmid:19924239
- 70. Giovanardi O, Romanelli M. Preliminary note on tuna larvae in samples from the coasts of the southern-central Mediterranean Sea collected by the MV Arctic Sunrise in June/July 2008. ICCAT Col Vol Sci Pap. 2010; 65: 740–743.
- 71. Koched W, Hattour A, Alemany F, García A, Said K. 2013 Spatial distribution of tuna larvae in the Gulf of Gabes (Eastern Mediterranean) in relation with environmental parameters. Mediterr Mar Sci. 2013; 14: 5–14.
- 72. Cuttitta A, Arigo A, Basilone G, Bonanno A, Buscaino G, Rollandi L, et al. Mesopelagic fish larvae species in the Strait of Sicily and their relationships to main oceanographic events. Hydrobiologia. 2004; 527(1): 177–182.
- 73. Isari S, Fragopoulu N, Somarakis S. Interannual variability in horizontal patterns of larval fish assemblages in the northeastern Aegean Sea (eastern Mediterranean) during early summer. Estuar Coast Shelf S. 2008; 79: 607–619.
- 74. Sanzo L. Uova e primi stadi larval de tonno (Orcynus thynnus Ltkn). Consiglio nazionale dell ricerche. Reale comitato talassografico italiano 189; 1932. [In Italian].
- 75. Piccinetti C, Piccinetti Manfrin G, Soro S. Résultats d’une campagne de recherche sur les larves de thonidés en Mediterranée. ICCAT Col Vol Sci Pap. 1997; 46, 207–214. [In French].
- 76. Tsuji S, Nishikawa Y, Segawa K, Hiroe Y. Distribution and abundance of Thunnus larvae and their relation to the oceanographic condition in the Gulf of Mexico and the Mediterranean Sea during May through August of 1994. ICCAT Col Vol Sci Pap. 1997; 46: 161–176.
- 77. García A, Alemany F, de la Serna JM, Oray I, Karakulak S, Rollandi L, et al. Preliminary results of the 2004 bluefin tuna larval surveys off different Mediterranean sites (Balearic Archipelago, Levantine Sea and the Sicilian Channel). ICCAT Col Vol Sci Pap. 2005; 58: 1261–1270.
- 78. Oray IK, Karakulak FS. Further evidence of spawning of bluefin tuna (Thunnus thynnus L., 1758) and the tuna species (Auxis rochei Ris., 1810, Euthynnus alletteratus Raf., 1810) in the eastern Mediterranean Sea: preliminary results of TUNALEV larval survey in 2004. J Appl Ichthyol. 2005; 21(3): 236–240.
- 79. Hyde JR, Lynn E, Humphreys R Jr, Musyl M, West AP, Vetter R. Shipboard identification of fish eggs and larvae by multiplex PCR, and description of fertilized eggs of blue marlin, shortbill spearfish, and wahoo. Mar Ecol Prog Ser. 2005; 286: 269–277.
- 80. Chow S, Kishino H. Phylogenetic relationships between tuna species of the genus Thunnus (Scombridae: Teleostei): inconsistent implications from morphology, nuclear and mitochondrial genomes. J Mol Evol. 1995; 41: 741–748. pmid:8587119
- 81. Elliott NG, Ward RD. Genetic relationships of eight species of Pacific tunas (Teleostei: Scombridae) inferred from allozyme analysis. Mar Freshwater Res. 1995; 46(7): 1021–1032.
- 82. Paine MA, McDowell JR, Graves JE. Specific identification using COI sequence analysis of scombrid larvae collected off the Kona coast of Hawaii Island. Ichthyol Res. 2008; 55(1): 7–16.
- 83. Ross HA, Murugan S, Li WLS. Testing the reliability of genetic methods of species identification via simulation. Syst Biol. 2008; 57(2): 216–230. pmid:18398767
- 84. Richardson DE, Vanwye JD, Exum AM, Cowen RK, Crawford DL. High‐throughput species identification: from DNA isolation to bioinformatics. Mol Ecol Notes. 2007; 7(2): 199–207.
- 85. Daniel LB III, Graves JE. Morphometric and genetic identification of eggs of spring-spawning sciaenids in lower Chesapeake Bay. Fish B-NOAA. 1994; 92(2): 254–261.
- 86. Brown SDJ, Collins RA, Boyer S. Spider: an R package for the analysis of species identity and evolution, with particular reference to DNA barcoding. Mol Ecol Resour. 2012; 12: 562–565. pmid:22243808